Small breakthrough (was: Re: Logic Analysers)

[Adrian Graham]

I've just done a little dance of joy.

Whilst looking at the code and trying to cross reference it against the
timing signals I was seeing for RD operations it became obvious that there
was an address mixup and a RET instruction was jumping back to the wrong
part of the code - a return to 0x005C was actually going back to 0x0054
which implied bit 4 could be having a problem.

I piggybacked a 4116-2 onto bit 4's chip and got a burst of activity then
what seemed like a crash, looking at the decoded address output I could
follow the code through several loops to a IN instruction where it stopped
and sure enough IO/M had gone high on the 8085 - first time I've seen that
happen :D

Replacing the chip and testing the old one in my 4116 tester showed it was
properly dead.

Onwards, ever onwards... Now I need to find out either what device should be
at 0xE3 or which particular chip on the IO/M path has a stuck input (there's
3) - given this machine's failure rate at blown gates I'll not be surprised
if this is the next fault.

Cheers folks!

On 04/02/2017 13:47, "Noel Chiappa"  wrote:

>> From: Jon Elson
> 
>> Any time you see really narrow glitches, especially when they are one
>> LA sample wide, you have no idea what they actually look like. The LA
>> detects that the pulse was there at the instant it sampled it, but you
>> don't know whether it was 5 ns wide, or 70 ns wide ... You also don't
>> know whether they were full-amplitude pulses or runts that just barely
>> crossed the logic threshold of the analyzer.
> 
> Which is why I always prefer to work with an LA _and_ a 'scope: the 'scope
> lets me see what the signals look like, how much noise/etc there is, etc,
> etc, while the LA can do other things - better triggering, capture longer
> time periods, etc.
> 
> (Now they have those fancy new digitial 'scope with capture capability, and
> you can get the best of both worlds with one box, but I guess they are still
> kind of pricy.)
> 
> But you can probably pick up an old 'scope for not much money on eBait. I
> can't imagine working on anything without one.
> 
> Noel

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Noel Chiappa]

> From: Jon Elson

> Any time you see really narrow glitches, especially when they are one
> LA sample wide, you have no idea what they actually look like. The LA
> detects that the pulse was there at the instant it sampled it, but you
> don't know whether it was 5 ns wide, or 70 ns wide ... You also don't
> know whether they were full-amplitude pulses or runts that just barely
> crossed the logic threshold of the analyzer. 

Which is why I always prefer to work with an LA _and_ a 'scope: the 'scope
lets me see what the signals look like, how much noise/etc there is, etc,
etc, while the LA can do other things - better triggering, capture longer
time periods, etc.

(Now they have those fancy new digitial 'scope with capture capability, and
you can get the best of both worlds with one box, but I guess they are still
kind of pricy.)

But you can probably pick up an old 'scope for not much money on eBait. I
can't imagine working on anything without one.

Noel

Re: Logic Analysers

[Adrian Graham]

On 04/02/2017 05:40, "dwight" <dkel...@hotmail.com> wrote:

> All the glitches are at the beginning of the ALE. There is nothing
> there that has any meaning. Things are changing at this time. Not
> every thing changes at the same rate. That is why they have an
> ALE to mark when the address is good. When high, the circuit address
> latch is open. When ALE goes low, it captures the address.
> You really should be looking at the processor timing diagram and
> understand what you are looking at.

I'm doing that right now as it happens, I've got a dead tree edition of the
MCS85 User Manual that has all the timings in and I'm trying to match them
with what the CPU is doing. What I've been thinking is happening is because
the traces I'm seeing don't match exactly on screen then there's a problem
but of course as long as the transitions are happening within the right time
frame there isn't a problem.

> 
> Technically a glitch at the beginning of the ALE can last until
> some nanoseconds before the falling edge and the circuit would
> work fine. These glitches are much shorter than the ALE and
> clearly not an issue.

So I'm looking too deeply, OK.

Back to the books for me then :)

A

> 
> From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Jon Elson
> <el...@pico-systems.com>
> Sent: Friday, February 3, 2017 6:36:27 PM
> To: gene...@classiccmp.org; Discussion@
> Subject: Re: Logic Analysers
> 
> On 02/03/2017 04:34 PM, Adrian Graham wrote:
>> On 03/02/2017 19:43, "Tony Duell" <ard.p850...@gmail.com> wrote:
>> 
>>> But that's why I said 'about'. I am doing order-of-magnitude calculations,
>>> not trying to design a delay line. I would estimate that between adjacent
>>> ICs on the same board you'd get a delay measured in 10's or 100's of
>>> picoseconds. That sort of order. So a 25MHz logic analyser, with an
>>> effective time resolution of 40ns (if that) is not going to show it.
>>> 
>>> There is no way you're going to get delays of 40ns between adjacent
>>> ICs on any reasonable PCB.
>> This is the sort of thing I mean:
>> 
>> http://www.binarydinosaurs.co.uk/STCExecutelA1checking.jpg
>> 
>> Watching the A1 address line (no triggers just sampling 6 points) and a
>> pulse appears at ROM4 on the falling edge of the ALE signal but not the
>> other 3 ROMs or the LS373 flip-flop that's demultiplexing the AD1 pin of the
>> 8085. While I was thinking about the possibility of propagation delay I
>> noticed this one:
>> 
>> http://www.binarydinosaurs.co.uk/STCExecutelA1checking2.jpg
>> 
>> Pulse missing from ROM3.
> First pic, pulses are missing from ROMs 1-3, seen on ROM4.
> But, those pulses on ROM4 are really narrow, and may be
> noise, or very narrow glitches.  Any time you see really
> narrow glitches, especially when they are one LA sample
> wide, you have no idea what they actually look like.  The LA
> detects that the pulse was there at the instant it sampled
> it, but you don't know whether it was 5 ns wide, or 70 ns
> wide (with a 40 ns sampling period).  You also don't know
> whether they were full-amplitude pulses or runts that just
> barely crossed the logic threshold of the analyzer.  So, I'm
> not sure what you've shown there actually represents a
> problem or not. Especially on the 2nd picture, the pulses
> you have highlighted really look like a single sample wide,
> and if the logic levels of the analyzer are not exactly the
> same, or other slight deviation, it could have missed a
> narrow glitch.  Anyway, on old 8-bit micro gear, there may
> be plenty of narrow glitches in the 40 ns range, but the
> operation of the chips is most likely NOT going to depend on
> the circuits responding to such glitches.  I think you are
> chasing your tail about these things, and missing a real
> malfunction that is not related to this.  Could be EPROM
> bits that have faded, one shot capacitors that have changed
> value or something.
> 
> Jon
> 

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Adrian Graham]

On 04/02/2017 02:36, "Jon Elson"  wrote:

>> http://www.binarydinosaurs.co.uk/STCExecutelA1checking2.jpg
>> 
>> Pulse missing from ROM3.
> First pic, pulses are missing from ROMs 1-3, seen on ROM4.
> But, those pulses on ROM4 are really narrow, and may be
> noise, or very narrow glitches.  Any time you see really
> narrow glitches, especially when they are one LA sample
> wide, you have no idea what they actually look like.  The LA

Gotcha, I forget that any circuit can be a haven of noise and like you say
the LA is just detecting that there's something there at that point in time.
The fact that the logic software even has a 'glitch filter' on each channel
should've given me a clue.

> the circuits responding to such glitches.  I think you are
> chasing your tail about these things, and missing a real
> malfunction that is not related to this.  Could be EPROM
> bits that have faded, one shot capacitors that have changed
> value or something.

I've tested all the caps apart from the .47uF decoupling ones and replaced
the dead or out-of-spec ones. This week I replaced all the ROMs with 'new'
2764s and changed the old sockets, I discovered that two of the original
ROMs had gone open circuit. Fortunately I'd dumped them beforehand.

I'll keep chipping away :)

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[dwight]

Remember what both Tony and I said earlier. You have to know

what is suppose to be happening. Just probing around, looking

for something funny looking is not usually very fruitful.

You really need to spend some time looking at data sheets.

Dwight



From: dwight <dkel...@hotmail.com>
Sent: Friday, February 3, 2017 9:40:48 PM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Logic Analysers


All the glitches are at the beginning of the ALE. There is nothing

there that has any meaning. Things are changing at this time. Not

every thing changes at the same rate. That is why they have an

ALE to mark when the address is good. When high, the circuit address

latch is open. When ALE goes low, it captures the address.

You really should be looking at the processor timing diagram and

understand what you are looking at.

Technically a glitch at the beginning of the ALE can last until

some nanoseconds before the falling edge and the circuit would

work fine. These glitches are much shorter than the ALE and

clearly not an issue.

Dwight



From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Jon Elson 
<el...@pico-systems.com>
Sent: Friday, February 3, 2017 6:36:27 PM
To: gene...@classiccmp.org; Discussion@
Subject: Re: Logic Analysers

On 02/03/2017 04:34 PM, Adrian Graham wrote:
> On 03/02/2017 19:43, "Tony Duell" <ard.p850...@gmail.com> wrote:
>
>> But that's why I said 'about'. I am doing order-of-magnitude calculations,
>> not trying to design a delay line. I would estimate that between adjacent
>> ICs on the same board you'd get a delay measured in 10's or 100's of
>> picoseconds. That sort of order. So a 25MHz logic analyser, with an
>> effective time resolution of 40ns (if that) is not going to show it.
>>
>> There is no way you're going to get delays of 40ns between adjacent
>> ICs on any reasonable PCB.
> This is the sort of thing I mean:
>
> http://www.binarydinosaurs.co.uk/STCExecutelA1checking.jpg
>
> Watching the A1 address line (no triggers just sampling 6 points) and a
> pulse appears at ROM4 on the falling edge of the ALE signal but not the
> other 3 ROMs or the LS373 flip-flop that's demultiplexing the AD1 pin of the
> 8085. While I was thinking about the possibility of propagation delay I
> noticed this one:
>
> http://www.binarydinosaurs.co.uk/STCExecutelA1checking2.jpg
>
> Pulse missing from ROM3.
First pic, pulses are missing from ROMs 1-3, seen on ROM4.
But, those pulses on ROM4 are really narrow, and may be
noise, or very narrow glitches.  Any time you see really
narrow glitches, especially when they are one LA sample
wide, you have no idea what they actually look like.  The LA
detects that the pulse was there at the instant it sampled
it, but you don't know whether it was 5 ns wide, or 70 ns
wide (with a 40 ns sampling period).  You also don't know
whether they were full-amplitude pulses or runts that just
barely crossed the logic threshold of the analyzer.  So, I'm
not sure what you've shown there actually represents a
problem or not. Especially on the 2nd picture, the pulses
you have highlighted really look like a single sample wide,
and if the logic levels of the analyzer are not exactly the
same, or other slight deviation, it could have missed a
narrow glitch.  Anyway, on old 8-bit micro gear, there may
be plenty of narrow glitches in the 40 ns range, but the
operation of the chips is most likely NOT going to depend on
the circuits responding to such glitches.  I think you are
chasing your tail about these things, and missing a real
malfunction that is not related to this.  Could be EPROM
bits that have faded, one shot capacitors that have changed
value or something.

Jon

Re: Logic Analysers

[dwight]

All the glitches are at the beginning of the ALE. There is nothing

there that has any meaning. Things are changing at this time. Not

every thing changes at the same rate. That is why they have an

ALE to mark when the address is good. When high, the circuit address

latch is open. When ALE goes low, it captures the address.

You really should be looking at the processor timing diagram and

understand what you are looking at.

Technically a glitch at the beginning of the ALE can last until

some nanoseconds before the falling edge and the circuit would

work fine. These glitches are much shorter than the ALE and

clearly not an issue.

Dwight



From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Jon Elson 
<el...@pico-systems.com>
Sent: Friday, February 3, 2017 6:36:27 PM
To: gene...@classiccmp.org; Discussion@
Subject: Re: Logic Analysers

On 02/03/2017 04:34 PM, Adrian Graham wrote:
> On 03/02/2017 19:43, "Tony Duell" <ard.p850...@gmail.com> wrote:
>
>> But that's why I said 'about'. I am doing order-of-magnitude calculations,
>> not trying to design a delay line. I would estimate that between adjacent
>> ICs on the same board you'd get a delay measured in 10's or 100's of
>> picoseconds. That sort of order. So a 25MHz logic analyser, with an
>> effective time resolution of 40ns (if that) is not going to show it.
>>
>> There is no way you're going to get delays of 40ns between adjacent
>> ICs on any reasonable PCB.
> This is the sort of thing I mean:
>
> http://www.binarydinosaurs.co.uk/STCExecutelA1checking.jpg
>
> Watching the A1 address line (no triggers just sampling 6 points) and a
> pulse appears at ROM4 on the falling edge of the ALE signal but not the
> other 3 ROMs or the LS373 flip-flop that's demultiplexing the AD1 pin of the
> 8085. While I was thinking about the possibility of propagation delay I
> noticed this one:
>
> http://www.binarydinosaurs.co.uk/STCExecutelA1checking2.jpg
>
> Pulse missing from ROM3.
First pic, pulses are missing from ROMs 1-3, seen on ROM4.
But, those pulses on ROM4 are really narrow, and may be
noise, or very narrow glitches.  Any time you see really
narrow glitches, especially when they are one LA sample
wide, you have no idea what they actually look like.  The LA
detects that the pulse was there at the instant it sampled
it, but you don't know whether it was 5 ns wide, or 70 ns
wide (with a 40 ns sampling period).  You also don't know
whether they were full-amplitude pulses or runts that just
barely crossed the logic threshold of the analyzer.  So, I'm
not sure what you've shown there actually represents a
problem or not. Especially on the 2nd picture, the pulses
you have highlighted really look like a single sample wide,
and if the logic levels of the analyzer are not exactly the
same, or other slight deviation, it could have missed a
narrow glitch.  Anyway, on old 8-bit micro gear, there may
be plenty of narrow glitches in the 40 ns range, but the
operation of the chips is most likely NOT going to depend on
the circuits responding to such glitches.  I think you are
chasing your tail about these things, and missing a real
malfunction that is not related to this.  Could be EPROM
bits that have faded, one shot capacitors that have changed
value or something.

Jon

Re: Logic Analysers

[Jon Elson]

On 02/03/2017 04:34 PM, Adrian Graham wrote:

On 03/02/2017 19:43, "Tony Duell"  wrote:


But that's why I said 'about'. I am doing order-of-magnitude calculations,
not trying to design a delay line. I would estimate that between adjacent
ICs on the same board you'd get a delay measured in 10's or 100's of
picoseconds. That sort of order. So a 25MHz logic analyser, with an
effective time resolution of 40ns (if that) is not going to show it.

There is no way you're going to get delays of 40ns between adjacent
ICs on any reasonable PCB.

This is the sort of thing I mean:

http://www.binarydinosaurs.co.uk/STCExecutelA1checking.jpg

Watching the A1 address line (no triggers just sampling 6 points) and a
pulse appears at ROM4 on the falling edge of the ALE signal but not the
other 3 ROMs or the LS373 flip-flop that's demultiplexing the AD1 pin of the
8085. While I was thinking about the possibility of propagation delay I
noticed this one:

http://www.binarydinosaurs.co.uk/STCExecutelA1checking2.jpg

Pulse missing from ROM3.
First pic, pulses are missing from ROMs 1-3, seen on ROM4.  
But, those pulses on ROM4 are really narrow, and may be 
noise, or very narrow glitches.  Any time you see really 
narrow glitches, especially when they are one LA sample 
wide, you have no idea what they actually look like.  The LA 
detects that the pulse was there at the instant it sampled 
it, but you don't know whether it was 5 ns wide, or 70 ns 
wide (with a 40 ns sampling period).  You also don't know 
whether they were full-amplitude pulses or runts that just 
barely crossed the logic threshold of the analyzer.  So, I'm 
not sure what you've shown there actually represents a 
problem or not. Especially on the 2nd picture, the pulses 
you have highlighted really look like a single sample wide, 
and if the logic levels of the analyzer are not exactly the 
same, or other slight deviation, it could have missed a 
narrow glitch.  Anyway, on old 8-bit micro gear, there may 
be plenty of narrow glitches in the 40 ns range, but the 
operation of the chips is most likely NOT going to depend on 
the circuits responding to such glitches.  I think you are 
chasing your tail about these things, and missing a real 
malfunction that is not related to this.  Could be EPROM 
bits that have faded, one shot capacitors that have changed 
value or something.


Jon

Re: Logic Analysers

[Jon Elson]

On 02/03/2017 01:35 PM, Mouse wrote:

the propagation delay as the signal gets to each pin (remember a
foot is about a nanosecond.  [...])

Not really.  A foot is about a light-nanosecond, yes, but
high-frequency signals in copper travel by skin effect, moving
significantly more slowly - somewhere around .6c, I think it is.

Well, actually, it depends on the impedance.  So, 50 Ohm 
coax cable is .6 C, twisted pair differential cable is 120 
Ohms, and the propagation velocity is a little above .7 I 
think.  Wires in open space are faster, around .8 or so, but 
the signal quality may be poor.


Jon

Re: Logic Analysers

[Chuck Guzis]

On 02/03/2017 04:10 PM, dwight wrote:
> I'm not sure you want to hide glitches. There are times
> 
> when you might want to see them.
> 
> It is more about knowing when a glitch has meaning and when it
> doesn't.

Indeed.  That's one of the the things that impressed me about the early
HP 1615 logic analyzer--it had a glitch detector.  Glitches can be
maddeningly difficult to find using traditional (i.e. 'scope) methods.

--Chuck

Re: Logic Analysers

[dwight]

I'm not sure you want to hide glitches. There are times

when you might want to see them.

It is more about knowing when a glitch has meaning and when it doesn't.

Dwight



From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Adrian Graham 
<wit...@binarydinosaurs.co.uk>
Sent: Friday, February 3, 2017 3:46:32 PM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Logic Analysers

On 03/02/2017 23:29, "dwight" <dkel...@hotmail.com> wrote:

> Adrian
>
>  What you see on the other select line is what is called a glitch.
>
> These are not that uncommon during the early part of the address.
>
> What is important is that there are no glitchs when ALE transitions.

Ah, ok, there's a glitch filter that I can apply to each channel, I'll
explore that.

Cheers!



> 
> From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Adrian Graham
> <wit...@binarydinosaurs.co.uk>
> Sent: Friday, February 3, 2017 2:34:18 PM
> To: General Discussion: On-Topic and Off-Topic Posts
> Subject: Re: Logic Analysers
>
> On 03/02/2017 19:43, "Tony Duell" <ard.p850...@gmail.com> wrote:
>
>> But that's why I said 'about'. I am doing order-of-magnitude calculations,
>> not trying to design a delay line. I would estimate that between adjacent
>> ICs on the same board you'd get a delay measured in 10's or 100's of
>> picoseconds. That sort of order. So a 25MHz logic analyser, with an
>> effective time resolution of 40ns (if that) is not going to show it.
>>
>> There is no way you're going to get delays of 40ns between adjacent
>> ICs on any reasonable PCB.
>
> This is the sort of thing I mean:
>
> http://www.binarydinosaurs.co.uk/STCExecutelA1checking.jpg
>
> Watching the A1 address line (no triggers just sampling 6 points) and a
> pulse appears at ROM4 on the falling edge of the ALE signal but not the
> other 3 ROMs or the LS373 flip-flop that's demultiplexing the AD1 pin of the
> 8085. While I was thinking about the possibility of propagation delay I
> noticed this one:
>
> http://www.binarydinosaurs.co.uk/STCExecutelA1checking2.jpg
>
> Pulse missing from ROM3.
>
> Given the paths on this board aren't massive and resistance is equal between
> all points when measured with a DMM (and all sockets have been replaced,
> traces checked etc) what else could I be looking at?
>
> --
> Adrian/Witchy
> Binary Dinosaurs creator/curator
> Www.binarydinosaurs.co.uk - the UK's biggest private home computer
> collection?
>
>

--
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Adrian Graham]

On 03/02/2017 23:29, "dwight" <dkel...@hotmail.com> wrote:

> Adrian
> 
>  What you see on the other select line is what is called a glitch.
> 
> These are not that uncommon during the early part of the address.
> 
> What is important is that there are no glitchs when ALE transitions.

Ah, ok, there's a glitch filter that I can apply to each channel, I'll
explore that.

Cheers!



> 
> From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Adrian Graham
> <wit...@binarydinosaurs.co.uk>
> Sent: Friday, February 3, 2017 2:34:18 PM
> To: General Discussion: On-Topic and Off-Topic Posts
> Subject: Re: Logic Analysers
> 
> On 03/02/2017 19:43, "Tony Duell" <ard.p850...@gmail.com> wrote:
> 
>> But that's why I said 'about'. I am doing order-of-magnitude calculations,
>> not trying to design a delay line. I would estimate that between adjacent
>> ICs on the same board you'd get a delay measured in 10's or 100's of
>> picoseconds. That sort of order. So a 25MHz logic analyser, with an
>> effective time resolution of 40ns (if that) is not going to show it.
>> 
>> There is no way you're going to get delays of 40ns between adjacent
>> ICs on any reasonable PCB.
> 
> This is the sort of thing I mean:
> 
> http://www.binarydinosaurs.co.uk/STCExecutelA1checking.jpg
> 
> Watching the A1 address line (no triggers just sampling 6 points) and a
> pulse appears at ROM4 on the falling edge of the ALE signal but not the
> other 3 ROMs or the LS373 flip-flop that's demultiplexing the AD1 pin of the
> 8085. While I was thinking about the possibility of propagation delay I
> noticed this one:
> 
> http://www.binarydinosaurs.co.uk/STCExecutelA1checking2.jpg
> 
> Pulse missing from ROM3.
> 
> Given the paths on this board aren't massive and resistance is equal between
> all points when measured with a DMM (and all sockets have been replaced,
> traces checked etc) what else could I be looking at?
> 
> --
> Adrian/Witchy
> Binary Dinosaurs creator/curator
> Www.binarydinosaurs.co.uk - the UK's biggest private home computer
> collection?
> 
> 

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[dwight]

Adrian

 What you see on the other select line is what is called a glitch.

These are not that uncommon during the early part of the address.

What is important is that there are no glitchs when ALE transitions.

Dwight



From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Adrian Graham 
<wit...@binarydinosaurs.co.uk>
Sent: Friday, February 3, 2017 2:34:18 PM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Logic Analysers

On 03/02/2017 19:43, "Tony Duell" <ard.p850...@gmail.com> wrote:

> But that's why I said 'about'. I am doing order-of-magnitude calculations,
> not trying to design a delay line. I would estimate that between adjacent
> ICs on the same board you'd get a delay measured in 10's or 100's of
> picoseconds. That sort of order. So a 25MHz logic analyser, with an
> effective time resolution of 40ns (if that) is not going to show it.
>
> There is no way you're going to get delays of 40ns between adjacent
> ICs on any reasonable PCB.

This is the sort of thing I mean:

http://www.binarydinosaurs.co.uk/STCExecutelA1checking.jpg

Watching the A1 address line (no triggers just sampling 6 points) and a
pulse appears at ROM4 on the falling edge of the ALE signal but not the
other 3 ROMs or the LS373 flip-flop that's demultiplexing the AD1 pin of the
8085. While I was thinking about the possibility of propagation delay I
noticed this one:

http://www.binarydinosaurs.co.uk/STCExecutelA1checking2.jpg

Pulse missing from ROM3.

Given the paths on this board aren't massive and resistance is equal between
all points when measured with a DMM (and all sockets have been replaced,
traces checked etc) what else could I be looking at?

--
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Adrian Graham]

On 03/02/2017 19:43, "Tony Duell"  wrote:

> But that's why I said 'about'. I am doing order-of-magnitude calculations,
> not trying to design a delay line. I would estimate that between adjacent
> ICs on the same board you'd get a delay measured in 10's or 100's of
> picoseconds. That sort of order. So a 25MHz logic analyser, with an
> effective time resolution of 40ns (if that) is not going to show it.
> 
> There is no way you're going to get delays of 40ns between adjacent
> ICs on any reasonable PCB.

This is the sort of thing I mean:

http://www.binarydinosaurs.co.uk/STCExecutelA1checking.jpg

Watching the A1 address line (no triggers just sampling 6 points) and a
pulse appears at ROM4 on the falling edge of the ALE signal but not the
other 3 ROMs or the LS373 flip-flop that's demultiplexing the AD1 pin of the
8085. While I was thinking about the possibility of propagation delay I
noticed this one:

http://www.binarydinosaurs.co.uk/STCExecutelA1checking2.jpg

Pulse missing from ROM3.

Given the paths on this board aren't massive and resistance is equal between
all points when measured with a DMM (and all sockets have been replaced,
traces checked etc) what else could I be looking at?

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

RE: Logic Analysers

[Rob Jarratt]

> 3) An HP1630. I forget which one, probably a 1630G. It does all I want. I was
> also AFAIK the last HP LA to have a proper component-level service manual.
> It's also a classic computer in its own right (6809 + 6829 MMU). Oddly the CRT
> is scanned vertically, I have no idea why.
> 


Hello Tony,

I bought a 1630G a while back. It came with pods, but didn't have the leads 
that plug into the pods. I am sure I could make something (in fact I intend to 
try it out this weekend if my cold doesn't get the better of me), but if you 
have any information on what to search for (model/part number) or know where I 
might buy some that would be really useful.

Regards

Rob

Re: Logic Analysers

[Paul Berger]

On 2017-02-03 5:05 PM, Adrian Graham wrote:

On 03/02/2017 20:38, "Paul Berger"  wrote:


As I have said before, the most important piece of test gear is a
brain.

-tony

...And if you don't have a schematic, you ring out the connections and
draw your own...

I'm not yet skilled enough to draw a schematic but I've drawn out a complete
layout of the board with all chips and traces in an open source design
package called Fritzing. Certainly a big help with wiring up the analyser.
These drawings are the only docs I have however... One thing I'd REALLY like
is a memory map but maybe the original designer of this machine will come up
trumps soon.

Well if that does not work out the approach I would take is work 
backwards from the chip select on the memory and I/O devices, things 
that may throw you off are if PLDs or ROMs are used as part of the 
logic.  Chip select logic is usually pretty straight forward.  As has 
been discussed before I/O on a 8085 may be in a seperate address space 
of might mapped into the I/O space.  IO/-M selects between the two 
address spaces.


Paul.

Re: Logic Analysers

[Paul Berger]

On 2017-02-03 4:41 PM, Tony Duell wrote:

...And if you don't have a schematic, you ring out the connections and draw
your own...

Given the number of times I've done that, I half-feel like mentioning
a grandmother
and sucking eggs ;-)

More seriously, to draw out a useful schematic -- not just one that
shows what is
connected to what, but also groups related sections together, also
requres a good
understanding of the device, of what is likely to be going on, etc. In
other words that
brain comes in handy...

-tony
Without a doubt the brain is the most important tool. in my day job 
of providing advise on fixing equipment, not component level mind you, 
you quickly learn who is equipped with this tool and who is not.


Paul.

Re: Logic Analysers

[Eric Smith]

On Fri, Feb 3, 2017 at 2:00 PM, dwight  wrote:

> I once worked with a device that came close to telling you
> that U15 was failing.
>
> It was called a signature analyzer.
> It was good as a first pass production tester.
>

It required that the device under test be put in a state where the signals
had predictable waveforms, since it basically was just computing a 16-bit
hash of samples of the waveforms, referenced to clock and start signals.
Typically the device under test would be put in a special signature test
mode to do that.


> It needed a pin bed to match the board under test.
>

For automated full-board testing, yes.  But signature testing was also used
with a probe on individual signals for field diagnostic purposes.

Re: Logic Analysers

[Adrian Graham]

On 03/02/2017 20:38, "Paul Berger"  wrote:

>> As I have said before, the most important piece of test gear is a
>> brain.
>> 
>> -tony
> ...And if you don't have a schematic, you ring out the connections and
> draw your own...

I'm not yet skilled enough to draw a schematic but I've drawn out a complete
layout of the board with all chips and traces in an open source design
package called Fritzing. Certainly a big help with wiring up the analyser.
These drawings are the only docs I have however... One thing I'd REALLY like
is a memory map but maybe the original designer of this machine will come up
trumps soon.

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[dwight]

I once worked with a device that came close to telling you

that U15 was failing.

It was called a signature analyzer.

It was good as a first pass production tester.

It was not something I'd expect a hobbyist to use.

It needed a pin bed to match the board under test.

It wasn't good enough to catch the bypass capacitor

tied to a buss feedthru.

Dwight



From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Charles Anthony 
<charles.unix@gmail.com>
Sent: Friday, February 3, 2017 12:51:32 PM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Logic Analysers

On Fri, Feb 3, 2017 at 12:43 PM, Eric Smith <space...@gmail.com> wrote:

> On Fri, Feb 3, 2017 at 1:25 PM, Fred Cisin <ci...@xenosoft.com> wrote:
>
>
> > What would the "System Requirements" be?
> > Would it also advise me that both my hardware and my OS are out-of-date,
> > and need to be upgraded?
> >
>
> I'll add that in a future upgrade (for more $).
>
> I'm also considering a space-rated version, which will be able to detect
> faults in Alpha Echo Three Five units.
>

It can only be attributable to human error.

Re: Logic Analysers

[Charles Anthony]

On Fri, Feb 3, 2017 at 12:43 PM, Eric Smith  wrote:

> On Fri, Feb 3, 2017 at 1:25 PM, Fred Cisin  wrote:
>
>
> > What would the "System Requirements" be?
> > Would it also advise me that both my hardware and my OS are out-of-date,
> > and need to be upgraded?
> >
>
> I'll add that in a future upgrade (for more $).
>
> I'm also considering a space-rated version, which will be able to detect
> faults in Alpha Echo Three Five units.
>

It can only be attributable to human error.

Re: Logic Analysers

[Eric Smith]

On Fri, Feb 3, 2017 at 1:25 PM, Fred Cisin  wrote:

> On Fri, 3 Feb 2017, Eric Smith wrote:
>
>> Hmmm...
>> I think I'll make a box that you plug into a computer and it tells you 'U5
>> is faulty'.
>> It won't be magic, though...
>> :-)
>>
>
> Other than plugging in to the computer (USB?), the rest of it could
> probably be done in software.
>

I'd prefer having a physical piece of hardware. That way I feel like I've
gotten something for my money.


> What would the "System Requirements" be?
> Would it also advise me that both my hardware and my OS are out-of-date,
> and need to be upgraded?
>

I'll add that in a future upgrade (for more $).

I'm also considering a space-rated version, which will be able to detect
faults in Alpha Echo Three Five units.

Re: Logic Analysers

[Tony Duell]

>
> ...And if you don't have a schematic, you ring out the connections and draw
> your own...

Given the number of times I've done that, I half-feel like mentioning
a grandmother
and sucking eggs ;-)

More seriously, to draw out a useful schematic -- not just one that
shows what is
connected to what, but also groups related sections together, also
requres a good
understanding of the device, of what is likely to be going on, etc. In
other words that
brain comes in handy...

-tony

Re: Logic Analysers

[Paul Berger]

On 2017-02-03 3:47 PM, Tony Duell wrote:

On Fri, Feb 3, 2017 at 7:04 PM, dwight  wrote:

I think Tony's statement about the key thing to know about

trouble shouting is to know what it should be doing.

If you don't know that, no scope or logic analyzer with help much.

Yes. I once explained faultfinding in this way. The technical manual,
schematics, microcode listings, etc should tell you what the device
should be doing. The instruments tell you what it is doing, you need
test instruments becuase you can't directly 'see' electrical signals.
You then have to compare the two and work out what could cause
the problem.

There is no magic box that you plug into a computer and it tells
you 'U5 is faulty'. At least not in general.

As I have said before, the most important piece of test gear is a
brain.

-tony
...And if you don't have a schematic, you ring out the connections and 
draw your own...


Paul.

Re: Logic Analysers

[Fred Cisin]

On Fri, 3 Feb 2017, Eric Smith wrote:

Hmmm...
I think I'll make a box that you plug into a computer and it tells you 'U5
is faulty'.
It won't be magic, though...
:-)


Other than plugging in to the computer (USB?), the rest of it could 
probably be done in software.



What would the "System Requirements" be?
Would it also advise me that both my hardware and my OS are out-of-date, 
and need to be upgraded?


--
Grumpy Ol' Fred ci...@xenosoft.com

Re: Logic Analysers

[Tony Duell]

On Fri, Feb 3, 2017 at 8:08 PM, jim stephens  wrote:
>
>
> On 2/3/2017 11:58 AM, Eric Smith wrote:
>>
>> On Fri, Feb 3, 2017 at 12:47 PM, Tony Duell  wrote:
>>
>>> There is no magic box that you plug into a computer and it tells
>>> you 'U5 is faulty'. At least not in general.
>>>
>> Hmmm...
>>
>> I think I'll make a box that you plug into a computer and it tells you 'U5
>> is faulty'.
>>
>> It won't be magic, though...
>> :-)
>
> One of those round ones with an 8 on top with a clear window on the bottom
> to tell you the fault?  Those work very well.

I am not sure those are common this side of the Pond. I assume it's some
kind of fortune telling device that perhaps contains a polyhedron with
a different
fortune on each face or something?

-tony

Re: Logic Analysers

[Tony Duell]

On Fri, Feb 3, 2017 at 7:58 PM, Eric Smith  wrote:
> On Fri, Feb 3, 2017 at 12:47 PM, Tony Duell  wrote:
>
>> There is no magic box that you plug into a computer and it tells
>> you 'U5 is faulty'. At least not in general.
>>
>
> Hmmm...
>
> I think I'll make a box that you plug into a computer and it tells you 'U5
> is faulty'.
>
> It won't be magic, though...
> :-)

Yes, OK, I am sure the HP calculator sitting alongside me could be
programmed to display that message :-). It won't help in actually finding
which IC _is_ faulty though.

-tony

Re: Logic Analysers

[jim stephens]

On 2/3/2017 11:58 AM, Eric Smith wrote:

On Fri, Feb 3, 2017 at 12:47 PM, Tony Duell  wrote:


There is no magic box that you plug into a computer and it tells
you 'U5 is faulty'. At least not in general.


Hmmm...

I think I'll make a box that you plug into a computer and it tells you 'U5
is faulty'.

It won't be magic, though...
:-)
One of those round ones with an 8 on top with a clear window on the 
bottom to tell you the fault?  Those work very well.


Thanks
Jim

Re: Logic Analysers

[Eric Smith]

On Fri, Feb 3, 2017 at 12:47 PM, Tony Duell  wrote:

> There is no magic box that you plug into a computer and it tells
> you 'U5 is faulty'. At least not in general.
>

Hmmm...

I think I'll make a box that you plug into a computer and it tells you 'U5
is faulty'.

It won't be magic, though...
:-)

Re: Logic Analysers

[Paul Koning]

> On Feb 3, 2017, at 2:43 PM, Tony Duell  wrote:
> 
> On Fri, Feb 3, 2017 at 7:35 PM, Mouse  wrote:
 the propagation delay as the signal gets to each pin (remember a
 foot is about a nanosecond.  [...])
>> 
>> Not really.  A foot is about a light-nanosecond, yes, but
>> high-frequency signals in copper travel by skin effect, moving
>> significantly more slowly - somewhere around .6c, I think it is.
> 
> It's not really the skin effect that matters here. It's the dielectric
> medium that surrounds the conductors that effectively slows the
> fields down.

Yes.  Consider open wire transmission line, which has a velocity factor around 
98%.  Or air dielectric coax, similarly high value.  The smaller numbers. like 
66%, are found in traditional solid-dielectric (not foam) coax cable.

paul

Re: Logic Analysers

[Tony Duell]

On Fri, Feb 3, 2017 at 7:04 PM, dwight  wrote:
> I think Tony's statement about the key thing to know about
>
> trouble shouting is to know what it should be doing.
>
> If you don't know that, no scope or logic analyzer with help much.

Yes. I once explained faultfinding in this way. The technical manual,
schematics, microcode listings, etc should tell you what the device
should be doing. The instruments tell you what it is doing, you need
test instruments becuase you can't directly 'see' electrical signals.
You then have to compare the two and work out what could cause
the problem.

There is no magic box that you plug into a computer and it tells
you 'U5 is faulty'. At least not in general.

As I have said before, the most important piece of test gear is a
brain.

-tony

Re: Logic Analysers

[Tony Duell]

On Fri, Feb 3, 2017 at 7:35 PM, Mouse  wrote:
>>> the propagation delay as the signal gets to each pin (remember a
>>> foot is about a nanosecond.  [...])
>
> Not really.  A foot is about a light-nanosecond, yes, but
> high-frequency signals in copper travel by skin effect, moving
> significantly more slowly - somewhere around .6c, I think it is.

It's not really the skin effect that matters here. It's the dielectric
medium that surrounds the conductors that effectively slows the
fields down.

But that's why I said 'about'. I am doing order-of-magnitude calculations,
not trying to design a delay line. I would estimate that between adjacent
ICs on the same board you'd get a delay measured in 10's or 100's of
picoseconds. That sort of order. So a 25MHz logic analyser, with an
effective time resolution of 40ns (if that) is not going to show it.

There is no way you're going to get delays of 40ns between adjacent
ICs on any reasonable PCB.

>
> It's still on the general order of c, mind you; for the purposes of
> this discussion, c and .5c - even .1c - are much the same.

Exactly.


-tony

Re: Logic Analysers

[Mouse]

>> the propagation delay as the signal gets to each pin (remember a
>> foot is about a nanosecond.  [...])

Not really.  A foot is about a light-nanosecond, yes, but
high-frequency signals in copper travel by skin effect, moving
significantly more slowly - somewhere around .6c, I think it is.

It's still on the general order of c, mind you; for the purposes of
this discussion, c and .5c - even .1c - are much the same.

/~\ The ASCII Mouse
\ / Ribbon Campaign
 X  Against HTMLmo...@rodents-montreal.org
/ \ Email!   7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B

Re: Logic Analysers

[dwight]

I think Tony's statement about the key thing to know about

trouble shouting is to know what it should be doing.

If you don't know that, no scope or logic analyzer with help much.

Dwight



From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Tony Duell 
<ard.p850...@gmail.com>
Sent: Friday, February 3, 2017 9:06:34 AM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Logic Analysers

On Fri, Feb 3, 2017 at 4:46 PM, dwight <dkel...@hotmail.com> wrote:
> Different strokes for different folks.

Yes. It depends a lot on what you work on, what you are trying to do, and
how you think.

This is a problem with mailing lists. There are many knowledgeable people
here, but each has their own way of doing things. All are right. But a person
trying to learn is going to get conflicting advice. Not because anyone is
being unkind, but because what they say is what they do, it works for them.

There is no one 'right way' to do this. Any way that finds the problem (and
that you know has found the problem!) is OK. Any instrument is just a way
of finding out what the device under test is actually doing. Faultfinding should
then consist of comparing that to what the device should be doing and
working out what could cause the differences.

Needless to say I would not want an LA if I was repairing an SMPSU. I'd use
a 'scope. But a lot of what I work on involves investigate a processor or a
complex interface controller (possibly microcoded, so in a sense a special-
purpose processor) at gate level. Believe me, you do not want to try to
debug an HP9800 (bit serial, microcoded, downright odd in places) with
a 'scope...

-tony

Re: Logic Analysers

[Adrian Graham]

On 03/02/2017 16:41, "Jon Elson"  wrote:

> On 02/03/2017 02:55 AM, Adrian Graham wrote:
>> Ah yes, sorry, I'm aware of that. What I meant in this
>> specific case is that with 4 2764s right next to each
>> other with a direct signal path between adjacent address
>> and data pins that has a resistance of 0.5 ohms pin to pin
>> surely I should be able to put a clip on each (for
>> example) A4 address line and see the same pulse at all
>> four channels? 
> Well, if the two logic analyzers were synched together, or
> you were sampling at 100 MHz or above, then yes.

Neither of them can go that fast but I didn't think that was necessary since
the system clock on this machine is 6MHz so sampling at 25+ should be
sufficient. If I reduce the number of channels to 6 I can drive one of them
at 50MHz but that didn't seem to make a difference.

> totally regular square waves.  If not, then the LA may not
> be sampling at a regular rate, or might have gaps while
> sending data to the PC.  I'm just suspicious of these units,
> given the results you report.

So am I :) I mean, the most expensive one was ukp40 direct from China so if
it's not fully accurate I can't really be surprised. I have a Zeroplus
coming next week from another collector who used it on Apple][/PETs as well
as car ECUs with good results. In the meanwhile the external clock signal
from the 8085 on this machine is accessible in 3 locations so I'll try 3
channels and see if it's properly square.
 
> (On my $130,000 Tektronix analyzer, I don't have to worry
> about such stupid stuff, I know they got it right.  I paid
> less than $750 for it, it will do 100 MHz on 288
> synchronized channels, with a 128K record length.  But, it
> is bigger than a big kitchen microwave, and much noisier, too.)

I'm looking at a lower-end HP/Agilent for around ukp200-250 which should be
enough since I doubt I'll ever work on anything with a clock speed of more
than 8MHz.

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Paul Berger]

On 2017-02-03 1:23 PM, jim stephens wrote:



On 2/3/2017 9:09 AM, Paul Berger wrote:
I also have a 16700A, 16600A, and a 16500C but they are rarely if 
ever used these days.


Paul. 
the 16600A has one slot.  We had one with a scope card installed, very 
nice compact setup if the builtin channels were sufficient.

thanks
jim

Yep that is why I got it, its hard to image 192 channel +  12 clocks not 
being enough it would be better is they where deep memory like the cards 
in my 16700B but you can't have everything.  Since it is the same CPU 
card I moved the option 3 card from the 16700A into it when I took the 
16700A out of service.  I also had an extra 10?100 card from a 16700B I 
think I may have stuck it in there too.  The 10/100 card will work in an 
A but you need to remove the RJ45 connector that is on the CPU card.


Paul.

Re: Logic Analysers

[Adrian Graham]

On 03/02/2017 15:27, "Tony Duell" <ard.p850...@gmail.com> wrote:

> On Fri, Feb 3, 2017 at 8:55 AM, Adrian Graham
> <wit...@binarydinosaurs.co.uk> wrote:
> 
>> Ah yes, sorry, I'm aware of that. What I meant in this specific case is that
>> with 4 2764s right next to each other with a direct signal path between
>> adjacent address and data pins that has a resistance of 0.5 ohms pin to pin
>> surely I should be able to put a clip on each (for example) A4 address line
>> and see the same pulse at all four channels?
> 
> Yes, subject to the following unlikely cases :
> 
> 1) There is a standing wave developed between the pins. Technically that trace
> is a transmission line. I have never heard it happen between ICs next to each
> other at 8-bit micro speeds though.
> 
> 2) There is a bad connection (IC socket?) on one of the pins

That was my first thought so there's nice new turned pin sockets on there
now. I did find one connection on the data bus that was held together by
solder and luck so fixed that with a small piece of jumper wire.
 
> 3) If you have a very fast logic analyser you might be able to see the
> propagation delay as the signal gets to each pin (remember a foot is about
> a nanosecond. So you are talking 10s of picoseconds delay). You will not
> see that with the sort of analyser you or I have :-)

I wondered about that but the fastest I can go is 25Mhz, also the
missing/extra signals are across (in this case) 6 channels and don't go in
channel order, so there'll be a missing pulse on channels 1+2 or channel 4
or an extra one on 2+5 etc. I'm watching the output pins on an LS373
flip-flop (new socket+chip) and the corresponding ROMs (all new
sockets+chips) along with an LS21N in the decoding circuit.

> If you try your test with 4 of your logic analyser channels on the A4 pins of
> the EPROMs, I assume you get different traces for each channel -- that is
> what you are commenting on. What happens if you swap the logic analyser
> channels round?

I thought of that the other day, also swapped the grabber ends since they're
not the sturdiest of things. I haven't tried a PC though, my host is an
iMac.

Hopefully next week I'll have a Zeroplus to try which while still being USB
attach is 16CH+external clock and onboard RAM for storage. Could be useful.
 
> Incidentally, I'd better comment on the Logic Analysers I use. I use them
> a lot more than a 'scope, but that's because of what I generally need to do.
> 

 Ah, someone else with an HP16xx. This is making me think I should
join the club. 

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Paul Berger]

On 2017-02-03 12:41 PM, Jon Elson wrote:

On 02/03/2017 02:55 AM, Adrian Graham wrote:
Ah yes, sorry, I'm aware of that. What I meant in this specific case 
is that with 4 2764s right next to each other with a direct signal 
path between adjacent address and data pins that has a resistance of 
0.5 ohms pin to pin surely I should be able to put a clip on each 
(for example) A4 address line and see the same pulse at all four 
channels? 
Well, if the two logic analyzers were synched together, or you were 
sampling at 100 MHz or above, then yes.
But, if the logic analyzers are running too slow, sampling irregularly 
(I have no trust in Chinese gizmos until PROVEN that they do it right) 
you could get very different results.  Is there a clock on the 
microprocessor that you can check?  Maybe something like a baud rate 
clock or something that is at a few MHz.  See if that shows up as 
totally regular square waves.  If not, then the LA may not be sampling 
at a regular rate, or might have gaps while sending data to the PC.  
I'm just suspicious of these units, given the results you report.


(On my $130,000 Tektronix analyzer, I don't have to worry about such 
stupid stuff, I know they got it right.  I paid less than $750 for it, 
it will do 100 MHz on 288 synchronized channels, with a 128K record 
length.  But, it is bigger than a big kitchen microwave, and much 
noisier, too.)


Jon
Same with my Agilent 16700B, however I current only have 192 channels 
available because I have two card slots occupied by digital scope 
cards.  The analyzer cards I have can do 110 MHz state or 500 MHz timing 
which more than meets my needs for the 30-40 year old computers I use it 
with.  One big advantage of a setup like this is one instrument can be 
used to trigger the other, for instance if I want to see what a signal 
really looks at at some point, I can use the logic analyzer to trigger 
the scope cards.  I also really like the external monitor on the 16700, 
its nice to work on a 19" display with a regular keyboard and mouse.  I 
have the 16700 set up with the back facing my work bench which gives me 
extra reach with the cable.  The only thing I need access to the front 
for is the power switch.   I do also have a analogue scope but I find 
myself using it less these days especially for digital circuits.
I also have a 16700A, 16600A, and a 16500C but they are rarely if ever 
used these days.


Paul.

Re: Logic Analysers

[Tony Duell]

On Fri, Feb 3, 2017 at 4:46 PM, dwight  wrote:
> Different strokes for different folks.

Yes. It depends a lot on what you work on, what you are trying to do, and
how you think.

This is a problem with mailing lists. There are many knowledgeable people
here, but each has their own way of doing things. All are right. But a person
trying to learn is going to get conflicting advice. Not because anyone is
being unkind, but because what they say is what they do, it works for them.

There is no one 'right way' to do this. Any way that finds the problem (and
that you know has found the problem!) is OK. Any instrument is just a way
of finding out what the device under test is actually doing. Faultfinding should
then consist of comparing that to what the device should be doing and
working out what could cause the differences.

Needless to say I would not want an LA if I was repairing an SMPSU. I'd use
a 'scope. But a lot of what I work on involves investigate a processor or a
complex interface controller (possibly microcoded, so in a sense a special-
purpose processor) at gate level. Believe me, you do not want to try to
debug an HP9800 (bit serial, microcoded, downright odd in places) with
a 'scope...

-tony

Re: Logic Analysers

[dwight]

Different strokes for different folks.

I've only used a logic analyzer once and even for that I found

it cumbersome and inadequate. I needed it to solve a sequential

problem that had a lot of time sequential actions.

Things like is does this, then this, then that. Ignore it and restart

if it does this and then something else.

I find that I can work faster with a 'scope. If I have issues with

something not of the bus or processor, most things have EPROMs

and I write test code.

Most logic analyzers are not real good at showing voltages.

Contention on a bus may be missed.

Dwight



From: cctalk <cctalk-boun...@classiccmp.org> on behalf of Tony Duell 
<ard.p850...@gmail.com>
Sent: Friday, February 3, 2017 7:27:07 AM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Logic Analysers

On Fri, Feb 3, 2017 at 8:55 AM, Adrian Graham
<wit...@binarydinosaurs.co.uk> wrote:

> Ah yes, sorry, I'm aware of that. What I meant in this specific case is that
> with 4 2764s right next to each other with a direct signal path between
> adjacent address and data pins that has a resistance of 0.5 ohms pin to pin
> surely I should be able to put a clip on each (for example) A4 address line
> and see the same pulse at all four channels?

Yes, subject to the following unlikely cases :

1) There is a standing wave developed between the pins. Technically that trace
is a transmission line. I have never heard it happen between ICs next to each
other at 8-bit micro speeds though.

2) There is a bad connection (IC socket?) on one of the pins

3) If you have a very fast logic analyser you might be able to see the
propagation delay as the signal gets to each pin (remember a foot is about
a nanosecond. So you are talking 10s of picoseconds delay). You will not
see that with the sort of analyser you or I have :-)

If you try your test with 4 of your logic analyser channels on the A4 pins of
the EPROMs, I assume you get different traces for each channel -- that is
what you are commenting on. What happens if you swap the logic analyser
channels round?

Incidentally, I'd better comment on the Logic Analysers I use. I use them
a lot more than a 'scope, but that's because of what I generally need to do.

1) (Is is an LA?) The HP LogicDart. 3 Channels, 100MHz. No external clock
facility. But it is pocket sized. HP called it the 'advanced logic
proble' and that's
really what it is. A better version of the blinking-light probe I used
to use. Great for
checking clocks, power supply voltages (it has a voltmeter function),
serial data
streams, etc. Normally the first instrument I grab for an unknown logic problem
just to eliminate the 'sillies'

2) An old Gould-Biomation K100D. 16 channels 100MHz. with external clocking. I
do have the 32 channel adapter for it which can only work with an
external clock.
This was my first LA and I still have a soft spot for it.

3) An HP1630. I forget which one, probably a 1630G. It does all I
want. I was also
AFAIK the last HP LA to have a proper component-level service manual. It's also
a classic computer in its own right (6809 + 6829 MMU). Oddly the CRT is scanned
vertically, I have no idea why.

-tony

Re: Logic Analysers

[Jon Elson]

On 02/03/2017 02:55 AM, Adrian Graham wrote:
Ah yes, sorry, I'm aware of that. What I meant in this 
specific case is that with 4 2764s right next to each 
other with a direct signal path between adjacent address 
and data pins that has a resistance of 0.5 ohms pin to pin 
surely I should be able to put a clip on each (for 
example) A4 address line and see the same pulse at all 
four channels? 
Well, if the two logic analyzers were synched together, or 
you were sampling at 100 MHz or above, then yes.
But, if the logic analyzers are running too slow, sampling 
irregularly (I have no trust in Chinese gizmos until PROVEN 
that they do it right) you could get very different 
results.  Is there a clock on the microprocessor that you 
can check?  Maybe something like a baud rate clock or 
something that is at a few MHz.  See if that shows up as 
totally regular square waves.  If not, then the LA may not 
be sampling at a regular rate, or might have gaps while 
sending data to the PC.  I'm just suspicious of these units, 
given the results you report.


(On my $130,000 Tektronix analyzer, I don't have to worry 
about such stupid stuff, I know they got it right.  I paid 
less than $750 for it, it will do 100 MHz on 288 
synchronized channels, with a 128K record length.  But, it 
is bigger than a big kitchen microwave, and much noisier, too.)


Jon

Re: Logic Analysers

[Tony Duell]

On Fri, Feb 3, 2017 at 8:55 AM, Adrian Graham
<wit...@binarydinosaurs.co.uk> wrote:

> Ah yes, sorry, I'm aware of that. What I meant in this specific case is that
> with 4 2764s right next to each other with a direct signal path between
> adjacent address and data pins that has a resistance of 0.5 ohms pin to pin
> surely I should be able to put a clip on each (for example) A4 address line
> and see the same pulse at all four channels?

Yes, subject to the following unlikely cases :

1) There is a standing wave developed between the pins. Technically that trace
is a transmission line. I have never heard it happen between ICs next to each
other at 8-bit micro speeds though.

2) There is a bad connection (IC socket?) on one of the pins

3) If you have a very fast logic analyser you might be able to see the
propagation delay as the signal gets to each pin (remember a foot is about
a nanosecond. So you are talking 10s of picoseconds delay). You will not
see that with the sort of analyser you or I have :-)

If you try your test with 4 of your logic analyser channels on the A4 pins of
the EPROMs, I assume you get different traces for each channel -- that is
what you are commenting on. What happens if you swap the logic analyser
channels round?

Incidentally, I'd better comment on the Logic Analysers I use. I use them
a lot more than a 'scope, but that's because of what I generally need to do.

1) (Is is an LA?) The HP LogicDart. 3 Channels, 100MHz. No external clock
facility. But it is pocket sized. HP called it the 'advanced logic
proble' and that's
really what it is. A better version of the blinking-light probe I used
to use. Great for
checking clocks, power supply voltages (it has a voltmeter function),
serial data
streams, etc. Normally the first instrument I grab for an unknown logic problem
just to eliminate the 'sillies'

2) An old Gould-Biomation K100D. 16 channels 100MHz. with external clocking. I
do have the 32 channel adapter for it which can only work with an
external clock.
This was my first LA and I still have a soft spot for it.

3) An HP1630. I forget which one, probably a 1630G. It does all I
want. I was also
AFAIK the last HP LA to have a proper component-level service manual. It's also
a classic computer in its own right (6809 + 6829 MMU). Oddly the CRT is scanned
vertically, I have no idea why.

-tony

Re: Logic Analysers

[Adrian Graham]

On 03/02/2017 08:01, "Christian Corti" 
wrote:

> On Thu, 2 Feb 2017, Adrian Graham wrote:
>> is fixed 5v. Also you'd expect that sampling at four times the clock speed
>> (they'll both do 25Mhz with 6 channels) then every pulse would be picked up.
> 
> No, because the pulse length may be far inferiour to the sample clock
> rate. You may also need to capture signal transitions instead of signal
> levels (i.e. you record a "pulse" if there was a transition between two
> sample clock pulses).

Ah yes, sorry, I'm aware of that. What I meant in this specific case is that
with 4 2764s right next to each other with a direct signal path between
adjacent address and data pins that has a resistance of 0.5 ohms pin to pin
surely I should be able to put a clip on each (for example) A4 address line
and see the same pulse at all four channels?

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Christian Corti]

On Thu, 2 Feb 2017, Adrian Graham wrote:

is fixed 5v. Also you'd expect that sampling at four times the clock speed
(they'll both do 25Mhz with 6 channels) then every pulse would be picked up.


No, because the pulse length may be far inferiour to the sample clock 
rate. You may also need to capture signal transitions instead of signal 
levels (i.e. you record a "pulse" if there was a transition between two 
sample clock pulses).


Christian

Re: Logic Analysers

[ben]

On 2/2/2017 7:28 PM, Chuck Guzis wrote:

On 02/02/2017 05:59 PM, Ian S. King wrote:

OK, I'm going to give the minimalist/cheap-bastard perspective.  I've
done some solid troubleshooting with my HP 1630G. Sure, it only has a
1K event memory, but the triggering options are pretty flexible.
With some creativity, you can focus on the behavior you need to
observe.  It can self-clock or externally clock, with three separate
clock inputs.  I've used it to find some pretty obscure race
conditions (one-shots are evil!) and glitches.


No, you're a really cheap bastard if you use a PC printer port as your
logic analyzer.  Believe it or not, there's software out there for that.

--Chuck


I say if it meets your needs, go for it.
Ben.

Re: Logic Analysers

[Chuck Guzis]

On 02/02/2017 05:59 PM, Ian S. King wrote:
> OK, I'm going to give the minimalist/cheap-bastard perspective.  I've
> done some solid troubleshooting with my HP 1630G. Sure, it only has a
> 1K event memory, but the triggering options are pretty flexible.
> With some creativity, you can focus on the behavior you need to
> observe.  It can self-clock or externally clock, with three separate
> clock inputs.  I've used it to find some pretty obscure race
> conditions (one-shots are evil!) and glitches.

No, you're a really cheap bastard if you use a PC printer port as your
logic analyzer.  Believe it or not, there's software out there for that.

--Chuck

Re: Logic Analysers

[Ian S. King]

OK, I'm going to give the minimalist/cheap-bastard perspective.  I've done
some solid troubleshooting with my HP 1630G. Sure, it only has a 1K event
memory, but the triggering options are pretty flexible.  With some
creativity, you can focus on the behavior you need to observe.  It can
self-clock or externally clock, with three separate clock inputs.  I've
used it to find some pretty obscure race conditions (one-shots are evil!)
and glitches.

Full disclosure: I do also have a really nice 16700-series I got from Glen
Slick, but for some things the easy setup of the 1630G wins.  -- Ian

-- 
Ian S. King, MSIS, MSCS, Ph.D. Candidate
The Information School 
Dissertation: "Why the Conversation Mattered: Constructing a Sociotechnical
Narrative Through a Design Lens

Archivist, Voices From the Rwanda Tribunal 
Value Sensitive Design Research Lab 

University of Washington

There is an old Vulcan saying: "Only Nixon could go to China."

Re: Logic Analysers

[Adrian Graham]

On 02/02/2017 23:49, "Jim Brain"  wrote:

> On 2/2/2017 5:21 PM, Chris Hanson wrote:
>> On Feb 2, 2017, at 10:43 AM, Fritz Mueller  wrote:
>>> I bought a used 1662 off eBay for cheap, and it has been indispensable for
>>> the work I've been doing on a PDP-11.  I think the external clock is often a
>>> pretty critical feature in being able to sensibly interpret traces.
>>> Sophisticated triggering is also very useful for catching a suspected
>>> misbehaving chip in the act.
>>> 
>>> I had considered some of the more modern USB options at the time because it
>>> seemed they would be convenient, but in the end I opted for a used
>>> old-school tool because it had the features I needed at less actual cost.  I
>>> have not regretted the decision -- super useful tool!
>> My HP 1660CS with Ethernet has ensured I¹m not even remotely interested in
>> one of the USB options, between having 136 channels and way higher bandwidth.
>> 
>> Even one of the models without Ethernet would be worthwhile since they¹re
>> straightforward to interface with via serial or GPIB.
>> 
>>-- Chris
>> 
> 
> Hmm, I'm not sure I'd say that.  I have a 1650, but the memory depth is
> not enough to quickly grab issues.  A Logic or Logic 16 can dump a ton
> of data into the PC, where it is easy to look at.

My chinese clone is of a Logic 16, I couldn't afford a genuine one since
they're ukp500 at least. Obviously caveat emptor, but that's out of my
league as a hobbyist.

Currently looking at HP1670G's in the US since they're reasonable assuming
the shipping and import charges are as quoted. Also a fellow collector is
loaning me a zeroplus 16 channel to try so that's another avenue to explore,
and there's the DSLogic Pro too which was supposed to be open hardware/open
source but development on that one stalled in 2015 and the current lot on
e*ay seem to be clones too.

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Jim Brain]

On 2/2/2017 5:21 PM, Chris Hanson wrote:

On Feb 2, 2017, at 10:43 AM, Fritz Mueller  wrote:

I bought a used 1662 off eBay for cheap, and it has been indispensable for the 
work I've been doing on a PDP-11.  I think the external clock is often a pretty 
critical feature in being able to sensibly interpret traces.  Sophisticated 
triggering is also very useful for catching a suspected  misbehaving chip in 
the act.

I had considered some of the more modern USB options at the time because it 
seemed they would be convenient, but in the end I opted for a used old-school 
tool because it had the features I needed at less actual cost.  I have not 
regretted the decision -- super useful tool!

My HP 1660CS with Ethernet has ensured I’m not even remotely interested in one 
of the USB options, between having 136 channels and way higher bandwidth.

Even one of the models without Ethernet would be worthwhile since they’re 
straightforward to interface with via serial or GPIB.

   -- Chris



Hmm, I'm not sure I'd say that.  I have a 1650, but the memory depth is 
not enough to quickly grab issues.  A Logic or Logic 16 can dump a ton 
of data into the PC, where it is easy to look at.


Jim


--
Jim Brain
br...@jbrain.com
www.jbrain.com

Re: Logic Analysers

[Chris Hanson]

On Feb 2, 2017, at 10:43 AM, Fritz Mueller  wrote:
> 
> I bought a used 1662 off eBay for cheap, and it has been indispensable for 
> the work I've been doing on a PDP-11.  I think the external clock is often a 
> pretty critical feature in being able to sensibly interpret traces.  
> Sophisticated triggering is also very useful for catching a suspected  
> misbehaving chip in the act.
> 
> I had considered some of the more modern USB options at the time because it 
> seemed they would be convenient, but in the end I opted for a used old-school 
> tool because it had the features I needed at less actual cost.  I have not 
> regretted the decision -- super useful tool!

My HP 1660CS with Ethernet has ensured I’m not even remotely interested in one 
of the USB options, between having 136 channels and way higher bandwidth.

Even one of the models without Ethernet would be worthwhile since they’re 
straightforward to interface with via serial or GPIB.

  -- Chris

Re: Logic Analysers

[Adrian Graham]

On 02/02/2017 18:43, "Fritz Mueller"  wrote:

> I bought a used 1662 off eBay for cheap, and it has been indispensable for the
> work I've been doing on a PDP-11.  I think the external clock is often a
> pretty critical feature in being able to sensibly interpret traces.
> Sophisticated triggering is also very useful for catching a suspected
> misbehaving chip in the act.
> 
> I had considered some of the more modern USB options at the time because it
> seemed they would be convenient, but in the end I opted for a used old-school
> tool because it had the features I needed at less actual cost.  I have not
> regretted the decision -- super useful tool!

Cheers Fritz, given that I'll only ever be working on this era kit I too
think the old-skool approach might be better.

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Adrian Graham]

On 02/02/2017 06:50, "Chuck Guzis"  wrote:

>> An HP 1660 or 1670 series self-contained portable logic analyzer
>> might be nice to pick up if you can get a decent deal on one. The
>> main limitation of the 1660 series is that the sample depth is only
>> 4K samples per channel. The 1670 series offers deeper sample depths,
>> and they may sell at a higher price as a result. And of course a
>> sub-model with more channels is better than one with fewer channels.
> 
> The problem with used logic analyzers is that the prices can vary
> widely.  My 1663 was purchased from eBay for $47 shipped, was clearly
> from a rental service, as it was complete; nothing missing.

Nice! It seems that there's only one seller one e*ay uk selling ex-MOD
units. There's actually 3 different IDs but the writeups look similar and
the location's the same so...
 
> I use an LA less than a 'scope, so the price is right for me.  Mostly,
> I'm interested in looking at peripheral device timing, so it's perfect
> for the occasional head-scratcher.

My Tek 2225 scope is getting a lot of use on this project, but because I
have zero documentation there's an awful lot of unknowns so following a
signal around the board should be easier with an LA.
 
> I don't know where you're located, but if you're in a region with an
> Electro Rent depot, you might inquire to see if they turn their old gear
> over to an auctioneer.

I'm in the increasingly-isolationist and possibly soon to be ringfenced UK.

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Adrian Graham]

On 02/02/2017 04:26, "Glen Slick"  wrote:

>> 
>> Whilst looking for better quality units I came across a couple of 'proper'
>> HP/Agilent analysers, a 1663A 34 channel and 1661A 102 channel which seem
>> complete apart from the chip leg grabbers. Am I right to assume some of you
>> might have experience of these beasts?
>> 
> 
> For debugging something like an 8085 CPU system you could connect up
> the analyzer to the CPU and if you capture traces in state mode
> (address, data, and status signals clocked in by RD, WR, or INTA
> edges) the 8085 inverse assembler software running on the analyzer can
> decode the bus transactions into the 8085 instruction stream.

That would be excellent to see, all I can do at the moment is fake an
external clock by using the RD signal and duplicate one of the higher
address lines (A12 currently) to make up for losing one channel since it's
only a 16 channel analyser. Then I run that through a 'simple parallel'
decoder which eventually gives me a set of addresses which happen to match
my disassembled code.

I guess for a sanity check I could watch the CPU on a PET or Apple][ since I
do have the full code listings for those.
 
> Too bad you're not local. I could make you a good deal on a bigger
> 16500B system. I have more of those than I need. Unfortunately the
> cost to ship one is close to or may even exceed their current market
> value, just within the US.

Indeed, looking at e*ay there's an almost-complete 1670G in the US with
working hard drive that appears to 'just' be missing the pods/grabbers. If
the shipping and import costs are correct the whole thing would cost me
ukp250. For ukp100 less than that I can get a complete 1663A in the UK.

I'm amazed I don't appear to know anyone in the UK with a spare one for sale
:)

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Fritz Mueller]

I bought a used 1662 off eBay for cheap, and it has been indispensable for the 
work I've been doing on a PDP-11.  I think the external clock is often a pretty 
critical feature in being able to sensibly interpret traces.  Sophisticated 
triggering is also very useful for catching a suspected  misbehaving chip in 
the act.

I had considered some of the more modern USB options at the time because it 
seemed they would be convenient, but in the end I opted for a used old-school 
tool because it had the features I needed at less actual cost.  I have not 
regretted the decision -- super useful tool!

  --FritzM.

Re: Logic Analysers

[Adrian Graham]

On 02/02/2017 03:13, "Jon Elson" <el...@pico-systems.com> wrote:

> On 02/01/2017 05:18 PM, Adrian Graham wrote:
>> Evening folks,
>> 
>> I have two so-called Logic Analysers, both cheap Chinese clones of other
>> more expensive units that hook up to the host via USB2 and stream readings
>> direct to software, in one case the open source Sigrok and in the other
>> genuine Saleae Logic.
>> 
>> I'm getting different and inconsistent readings out of both of them whilst
>> sampling at 25MHz which should be more than enough for this 6MHz Executel
>> I'm working on. Both of them are good for spotting dead or stuck outputs but
>> I still can't get a good set of readings from eg all points on a single
>> address line. Tonight I replaced all four ROM sockets and ROM chips, tested
>> each individual line for resistance (0.5 ohms on all apart from an
>> occasional 0.4 and 0.6) but still get ghost readings.
>> 
>> Is it me or the cheap clones?
> Well, these lines likely have pulses, and depending on when
> they are sampled, you can get a high, low, or indeterminate
> state.  Also, the LA may have fixed threshold voltages that
> may not correspond to the actual thresholds of the logic you
> are testing.

True, but one is adjustable for either 3.3v or 5v threshold while the other
is fixed 5v. Also you'd expect that sampling at four times the clock speed
(they'll both do 25Mhz with 6 channels) then every pulse would be picked up.
 
> Generally, you would clock the LA off some master clock in
> the unit under test.

Yeah, neither of them will do that. Units with external clock capability are
over the ukp100 mark which initially was far more than I was comfortable
paying when I wasn't sure about how it all worked. For troubleshooting CBM
PETs the cheapy one worked fine.

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Logic Analysers

[Chuck Guzis]

> On Wed, Feb 1, 2017 at 3:18 PM, Adrian Graham 
>  wrote:
>> 
>> Whilst looking for better quality units I came across a couple of
>> 'proper' HP/Agilent analysers, a 1663A 34 channel and 1661A 102
>> channel which seem complete apart from the chip leg grabbers. Am I
>> right to assume some of you might have experience of these beasts?
>> 
> 
> An HP 1660 or 1670 series self-contained portable logic analyzer
> might be nice to pick up if you can get a decent deal on one. The
> main limitation of the 1660 series is that the sample depth is only
> 4K samples per channel. The 1670 series offers deeper sample depths,
> and they may sell at a higher price as a result. And of course a
> sub-model with more channels is better than one with fewer channels.

The problem with used logic analyzers is that the prices can vary
widely.  My 1663 was purchased from eBay for $47 shipped, was clearly
from a rental service, as it was complete; nothing missing.

I use an LA less than a 'scope, so the price is right for me.  Mostly,
I'm interested in looking at peripheral device timing, so it's perfect
for the occasional head-scratcher.

I remember when we got in the then-new HP 1615 logic analyzer--I was
blown away by it.  We were using a Tek lab scope plug-in which was
pretty limited.  The idea of a piece of test equipment with a keyboard
and its own display was pure genius.  Biomation, IIRC, was also big in
the business, but not nearly as polished as the HP units.

I don't know where you're located, but if you're in a region with an
Electro Rent depot, you might inquire to see if they turn their old gear
over to an auctioneer.

--Chuck

Re: Logic Analysers

[Glen Slick]

On Wed, Feb 1, 2017 at 3:18 PM, Adrian Graham
 wrote:
>
> Whilst looking for better quality units I came across a couple of 'proper'
> HP/Agilent analysers, a 1663A 34 channel and 1661A 102 channel which seem
> complete apart from the chip leg grabbers. Am I right to assume some of you
> might have experience of these beasts?
>

An HP 1660 or 1670 series self-contained portable logic analyzer might
be nice to pick up if you can get a decent deal on one. The main
limitation of the 1660 series is that the sample depth is only 4K
samples per channel. The 1670 series offers deeper sample depths, and
they may sell at a higher price as a result. And of course a sub-model
with more channels is better than one with fewer channels.

For debugging something like an 8085 CPU system you could connect up
the analyzer to the CPU and if you capture traces in state mode
(address, data, and status signals clocked in by RD, WR, or INTA
edges) the 8085 inverse assembler software running on the analyzer can
decode the bus transactions into the 8085 instruction stream.

1660A - Mono CRT. 4K sample depth. Floppy only. No LAN.
1660C - Mono CRT. 4K sample depth. Hard drive. LAN optional.
1660E - Color LCD. 4K sample depth. Hard drive. LAN.

1670A - Mono CRT. 64K (optional 500K) sample depth. Hard drive. LAN.
1670D - Mono CRT. 64K (optional 1M) sample depth. Hard drive. LAN.
1670E - Color LCD. 1M sample depth. Hard drive. LAN.
1670G - Color LCD. 64K (optional 256K or 2M) sample depth. Hard drive. LAN.
(Information manually gathered from datasheets, hope it's all correct)

There are also options for built in dual channel oscilloscopes and
pattern generators.

Too bad you're not local. I could make you a good deal on a bigger
16500B system. I have more of those than I need. Unfortunately the
cost to ship one is close to or may even exceed their current market
value, just within the US.

Re: Logic Analysers

[Jon Elson]

On 02/01/2017 05:18 PM, Adrian Graham wrote:

Evening folks,

I have two so-called Logic Analysers, both cheap Chinese clones of other
more expensive units that hook up to the host via USB2 and stream readings
direct to software, in one case the open source Sigrok and in the other
genuine Saleae Logic.

I'm getting different and inconsistent readings out of both of them whilst
sampling at 25MHz which should be more than enough for this 6MHz Executel
I'm working on. Both of them are good for spotting dead or stuck outputs but
I still can't get a good set of readings from eg all points on a single
address line. Tonight I replaced all four ROM sockets and ROM chips, tested
each individual line for resistance (0.5 ohms on all apart from an
occasional 0.4 and 0.6) but still get ghost readings.

Is it me or the cheap clones?
Well, these lines likely have pulses, and depending on when 
they are sampled, you can get a high, low, or indeterminate 
state.  Also, the LA may have fixed threshold voltages that 
may not correspond to the actual thresholds of the logic you 
are testing.


Generally, you would clock the LA off some master clock in 
the unit under test.


Jon

Logic Analysers

[Adrian Graham]

Evening folks,

I have two so-called Logic Analysers, both cheap Chinese clones of other
more expensive units that hook up to the host via USB2 and stream readings
direct to software, in one case the open source Sigrok and in the other
genuine Saleae Logic.

I'm getting different and inconsistent readings out of both of them whilst
sampling at 25MHz which should be more than enough for this 6MHz Executel
I'm working on. Both of them are good for spotting dead or stuck outputs but
I still can't get a good set of readings from eg all points on a single
address line. Tonight I replaced all four ROM sockets and ROM chips, tested
each individual line for resistance (0.5 ohms on all apart from an
occasional 0.4 and 0.6) but still get ghost readings.

Is it me or the cheap clones?

Whilst looking for better quality units I came across a couple of 'proper'
HP/Agilent analysers, a 1663A 34 channel and 1661A 102 channel which seem
complete apart from the chip leg grabbers. Am I right to assume some of you
might have experience of these beasts?

Forums seem to mostly think the streaming USB units aren't worth anything
for more than a few channels but I'm still a relative beginner to all of
this. I really need to watch all 16 lines of an address bus and externally
clock it as Tony has suggested.

Any insights appreciated!

-- 
Adrian/Witchy
Binary Dinosaurs creator/curator
Www.binarydinosaurs.co.uk - the UK's biggest private home computer
collection?

Re: Looking for the Tek 465 of Logic Analysers

[Sean Caron]

Hi Tony,

I don't mean to throw shade on your HP 1630 or your K100 ... just in my own
personal opinion, when I see a HP 1630 on eBay for $100 or a HP 1660 on
eBay for $100 ... and you see this all the time ... I think going with the
1660 is the better deal ... you do make a good point about schematics and
repairability ... and hack-ability ... there are plenty of very exotic
looking components ... especially on the acquisition cards ... on the newer
machines that I'm sure are basically impossible to repair.

I just try to manage that risk by keeping as many spare cards on hand as I
can get for the 16500 series, and I've always got the 1662 standing by in
case something happened to it ... honestly, when you can get a HP 166x for
$100, unless you enjoy the sport of troubleshooting and repair in and of
itself (and I do so sometimes myself) is it really worth sinking much time
and effort into repairing a 1630 series, or a K100?

I just want folks to be aware ... people might not necessarily realize,
what their money will get them; I'd hate for someone to shell out for a
1630 when they could have gotten a 166x or 167x for the same price... That
almost happened to me when I was shopping around and I would have regretted
it. I'm glad I did the a little bit more background research beforehand...
No intrinsic dislike for the classic instruments; if you got em, use em! :O

Best,

Sean




On Fri, Jun 12, 2015 at 4:52 PM, tony duell a...@p850ug1.demon.co.uk wrote:


  * The HP 16500B and 16500C use a different '030 based logic board and
 also
 [...]
  * Generally anything the C can do, the B can do and vice versa. The
 big
  difference between the two is that they switched from HP-HIL peripherals
 on
  the B to standard PS/2 peripherals on the C, if you want to connect
 an
  external keyboard and mouse to use in lieu of or in conjunction with the
  touch screen; depending on what you have in your stash ... I happened to
  have a bunch of HIL mice and a HIL keyboard so HP-HIL on the B doesn't
  bug me.

 I don't know if it works on these instruments, but there was an interface
 to
 use a quadrature mouse on an HP-HIL system.

  * IMO, the HP 1630 series and the Biomation K100 are pretty obsolescent
 and
  ... unless you already have one, or someone local to you is giving you
  one... I don't think they're really worth the cost of shipping unless you
  have some nostalgic attachment to a particular instrument.

 As somebody who has both and uses them (and has no later analysers), I have
 to disagree with you there. After all the OP was asking for the 'Tek 465
 of logic
 analysers' which suggests not the latest instruments. The HP1630 and Gould
 K100D
 are easily fast enough for classic computer work (I've never had
 problems). They
 also seem to be a lot better documented than later analysers (can you get
 schematics of later HPs? I seem to remember they are not in the manuals).
 The
 probe interface is also documented and quite simple. While I recomend only
 getting a LA with the original pods, the ability to make custom pods has
 helped
 me on several occasions. The inputs to both those LAs are simple
 differential
 ECL signals, easy to hack about with.

 -tony

RE: Looking for the Tek 465 of Logic Analysers

[tony duell]

 * The HP 16500B and 16500C use a different '030 based logic board and also
[...]
 * Generally anything the C can do, the B can do and vice versa. The big
 difference between the two is that they switched from HP-HIL peripherals on
 the B to standard PS/2 peripherals on the C, if you want to connect an
 external keyboard and mouse to use in lieu of or in conjunction with the
 touch screen; depending on what you have in your stash ... I happened to
 have a bunch of HIL mice and a HIL keyboard so HP-HIL on the B doesn't
 bug me.

I don't know if it works on these instruments, but there was an interface to
use a quadrature mouse on an HP-HIL system. 

 * IMO, the HP 1630 series and the Biomation K100 are pretty obsolescent and
 ... unless you already have one, or someone local to you is giving you
 one... I don't think they're really worth the cost of shipping unless you
 have some nostalgic attachment to a particular instrument.

As somebody who has both and uses them (and has no later analysers), I have
to disagree with you there. After all the OP was asking for the 'Tek 465 of 
logic
analysers' which suggests not the latest instruments. The HP1630 and Gould K100D
are easily fast enough for classic computer work (I've never had problems). They
also seem to be a lot better documented than later analysers (can you get 
schematics of later HPs? I seem to remember they are not in the manuals). The 
probe interface is also documented and quite simple. While I recomend only
getting a LA with the original pods, the ability to make custom pods has helped
me on several occasions. The inputs to both those LAs are simple differential 
ECL signals, easy to hack about with.

-tony

Re: Looking for the Tek 465 of Logic Analysers

[Ian S. King]

Yes, the documentation of the HP 1630 series is phenomenal.

On Fri, Jun 12, 2015 at 1:52 PM, tony duell a...@p850ug1.demon.co.uk wrote:


  * The HP 16500B and 16500C use a different '030 based logic board and
 also
 [...]
  * Generally anything the C can do, the B can do and vice versa. The
 big
  difference between the two is that they switched from HP-HIL peripherals
 on
  the B to standard PS/2 peripherals on the C, if you want to connect
 an
  external keyboard and mouse to use in lieu of or in conjunction with the
  touch screen; depending on what you have in your stash ... I happened to
  have a bunch of HIL mice and a HIL keyboard so HP-HIL on the B doesn't
  bug me.

 I don't know if it works on these instruments, but there was an interface
 to
 use a quadrature mouse on an HP-HIL system.

  * IMO, the HP 1630 series and the Biomation K100 are pretty obsolescent
 and
  ... unless you already have one, or someone local to you is giving you
  one... I don't think they're really worth the cost of shipping unless you
  have some nostalgic attachment to a particular instrument.

 As somebody who has both and uses them (and has no later analysers), I have
 to disagree with you there. After all the OP was asking for the 'Tek 465
 of logic
 analysers' which suggests not the latest instruments. The HP1630 and Gould
 K100D
 are easily fast enough for classic computer work (I've never had
 problems). They
 also seem to be a lot better documented than later analysers (can you get
 schematics of later HPs? I seem to remember they are not in the manuals).
 The
 probe interface is also documented and quite simple. While I recomend only
 getting a LA with the original pods, the ability to make custom pods has
 helped
 me on several occasions. The inputs to both those LAs are simple
 differential
 ECL signals, easy to hack about with.

 -tony




-- 
Ian S. King, MSIS, MSCS, Ph.D. Candidate
The Information School http://ischool.uw.edu

Archivist, Voices From the Rwanda Tribunal http://tribunalvoices.org
Value Sensitive Design Research Lab http://vsdesign.org

University of Washington

There is an old Vulcan saying: Only Nixon could go to China.

Re: Looking for the Tek 465 of Logic Analysers

[Ken Seefried]

Thanks for the input everyone.  In summary, I got recommendations for:

- HP 16500C (lesser a 16500B, but not a 16500A) (mainframe)
- HP 16700/16900/17500 (mainframe, bigger-faster-stronger, still pretty
expensive, can use 16500 cards)
- HP 166x or 167x (portable, modern, look for hard drive)
- HP 165x (portable, older, only buy cheap, look for hard drive)
- HP 1630D, 1630G or 1631 (old, HPIB and HPIL interfaces, no special
software needed, but you apparently need be able to control it with an
external HPIB controller to get the most out of it)
- Tek 1240
- Gould Biomation K100D (software in rom, HPIB interface, assuming you need
HPIB literacy to use effectively)
- Tek DAS9200 (be careful to get the right pods (several variations), etc)
- Tek TLA710/TLA720 (PC attached, Linux capable)

General advice offered:

- Look for hard drives or no special software required for longest life.
Floppies required == bad.
- Make really sure you get probes, and the right probes, for your kit.
They're often not available separately for non-st00pid money.
- There's a crazy array of ancillary features to consider: o-scope
functions, processor assemblers/disassemblers, pattern generators,
specialized inputs/outputs, timebases, embedded Unix machines, etc.  Know
what you might need.

So I've got a lot of manuals to read, but the 16500C sure looks like a ton
of bang per US$.  You can even get an expansion chassis (10 additional
slots) if you need to go to ludicrous speed on the number of
probes/features.

KJ

Re: Looking for the Tek 465 of Logic Analysers

[Noel Chiappa]

 From: Jon Elson

 On 05/28/2015 09:53 PM, Ken Seefried wrote:

 Ease of finding complete kit; nothing worse than dropping a dime on
 what looks like a good deal only to find you're missing the unobtanium
 cable

 The Tek 1240 should work. 

I can second that. I recently bought a flock of them (for spares/parts for
the first one I bought), because they were so cheap on eBay - several I got
for $25 + shipping. Most of the ones I bought came without probes, etc, but I
managed to round up a very complete set of stuff without spending too much
money. Tek documentation is incredibly thorough, and easy to obtain; and they
are very easy to work on (in terms of accessability, etc). Etc, etc, etc.

The speeds/etc you are looking for are within their range. When filled with 4
16-channel data acquisition cards, you get 64 channels. They seem to have
quite powerful triggering/etc capabilities, and they are easy to use/control.

The only possible issue (for some people) is that the memories aren't large
(although you can chain identical DACs together to get slightly longer
memories). And I'll echo Tony's comments - ECL is pretty much de rigeur, and
as for making your own probes, fuhgeddaboutit; the 1240's probe pods (there
are two kinds, TTL-only, and 'pick a voltage') contain giant custom chips.

Noel

RE: Looking for the Tek 465 of Logic Analysers

[Ken Seefried]

From: tony duell a...@p850ug1.demon.co.uk
 - No weird technologies in the design (all TTL/CMOS logic)

That is going to be a problem. AFAIK no 'serious' logic analyser was all
TTL or (high speed) CMOS. If you are looking for one that is mostly/all
standard logic, I think you have to consider ECL here.

I meant I'm not using any weird technology in *my* designs.  Sorry if that
wasn't clear.  I'm not particularly concerned about what the LA is built
from.

KJ

RE: Looking for the Tek 465 of Logic Analysers

[tony duell]

 
 I'm quite fond of my HP 1630G.  It's quite fast enough for the sort of
 machines I'm logic-analyzing.  :-)

Ditto. Well, I can't remember which model I was given, it's the one that's 
maxed out with 
'state' channels, but only the basic 'timing' channels. 

The manual is excellent (and available, AFAIK, from whoever Agilent became). 
The only schematic
you don't get is that of the video monitor -- and be warned that is a bit 
unconvenitonal as the scan 
lines are vertical (across the short dimension of the CRT). But you do get 
schematics of the CPU board,
state and timing boards, PSU, etc.

One practical thing I like is that the pod cables are 'captive' (held on by a 
screw-on metal cover plate) so
the pods are less likely to go walkabout (you've not seen my workshop :-))

As I mentioned earlier, it has both HPIB and HPIL interfaces. In theory you can 
control this thing from
an HP71B handheld (or an HP41 calculator?)

An HP1631 would be fun (it has a DSO board at the bottom as well as the LA 
functions).

-tony

Re: Looking for the Tek 465 of Logic Analysers

[Ian S. King]

I'm quite fond of my HP 1630G.  It's quite fast enough for the sort of
machines I'm logic-analyzing.  :-)

On Fri, May 29, 2015 at 5:57 PM, Ken Seefried seefr...@gmail.com wrote:

 From: tony duell a...@p850ug1.demon.co.uk
  - No weird technologies in the design (all TTL/CMOS logic)
 
 That is going to be a problem. AFAIK no 'serious' logic analyser was all
 TTL or (high speed) CMOS. If you are looking for one that is mostly/all
 standard logic, I think you have to consider ECL here.

 I meant I'm not using any weird technology in *my* designs.  Sorry if that
 wasn't clear.  I'm not particularly concerned about what the LA is built
 from.

 KJ




-- 
Ian S. King, MSIS, MSCS, Ph.D. Candidate
The Information School http://ischool.uw.edu

Archivist, Voices From the Rwanda Tribunal http://tribunalvoices.org
Value Sensitive Design Research Lab http://vsdesign.org

University of Washington

There is an old Vulcan saying: Only Nixon could go to China.

RE: Looking for the Tek 465 of Logic Analysers

[tony duell]

 That's good to hear.
 2 weeks ago I got an HP1630D and last week a 1631.
 Both came with pods, but the first one did not have the plug
 with the test leads, and the second one did. So I bought the second

The actual pods, which are plugged into the back of the instrument are
quite complex and have a ceramic hybrid circuit inside. Not the sort of thing
to attempt to replicate. The input to the pod is a pin header with an 
odd spacing. The lead from that to the grabber clips is just wires, so something
could be kludged up

 one too for the plugs with leads. The 1631 looks quite identical to
 the 1630, but the 1631 has additionally 2 analog inputs. Not sure
 whether I will need them.

They are much the same instrument -- I think the same manual covers both, Most 
of the
boards are the same (there may well be different firmware on the CPU board, but 
I think
that is the only change). The 1631 has an extra PCB at the bottom which 
contains the custom
input and ADC circuitry to make it act as a digital storage 'scope too. 

-tony

Re: OT: Looking for the Tek 465 of Logic Analysers

[wulfman]

i own   a 16500B software is still on HPs site
i got it off ebay for 200$ fully loaded with a FULL set of probes
i am very happy with it

go for the hp


On 5/29/2015 1:37 PM, Marc Verdiell wrote:
 I'd suggest to go for the king of the hill at the time, and get an HP 16xx 
 (163x, 165x, 166x, 167x) for all-in-ones or the 16500 if you like to 
 modularize yourself, although tis latter one is much harder to put together 
 since you have to get the frame, the plug ins, the software, etc... 
 On the 16yx, the higher the y number, the better the machine (i.e faster, 
 better screens, more memory). They are widely available, starting at below 
 $100 for the earlier machines, and up to $300-$400 for the later ones. I'd 
 recommend to have at least one that has a hard disk, so you don't have to 
 find or make an old LIF floppy to boot from.
 As always the rub is the pods/probes. I had to complete my set separately, 
 but they are also widely available.
 This is a relatively small amount to pay for what these machines actually 
 are. The later ones have the inverse assembler for the 68000.
 I got the luxury one, a 1670G with the pattern generator (which I haven't 
 found an excuse to use yet). Here it is at work:
 http://youtu.be/X_6limxLZ_k


 Sent from my iPad

 On May 29, 2015, at 12:00 PM, cctalk-requ...@classiccmp.org wrote:

 Message: 20
 Date: Thu, 28 May 2015 22:53:07 -0400
 From: Ken Seefried seefr...@gmail.com
 To: cctalk@classiccmp.org
 Subject: OT: Looking for the Tek 465 of Logic Analysers
 Message-ID:
caorcwjx8hkjddgclyvzsbrotk8qmgbsf2kvnzbmsjpqrzzf...@mail.gmail.com
 Content-Type: text/plain; charset=UTF-8

 Maybe only semi-OT.  I'm working on a couple of classiccmp-ish projects
 (6303, 6309 and 68030) and I find the trusty old Tek 465 o-scope is no
 longer compensating for my lack of design skill (or I'm getting better at
 hiding bugs in my designs, depending how you look at it).  I'm looking for
 a recommendation for a logic analyzer.  Considering my very modest design
 constraints, I'm thinking:

 - Suitable for 50MHz designs (really more like 16MHz, but you never know)
 - 32 channels would be nice, ~128 probably perfect, less...you know...do
 what you gotta do...
 - No weird technologies in the design (all TTL/CMOS logic)
 - I'm willing to spend a few $$ to get decent kit, but need to spend closer
 to 465 money than TLA7012 money
 - Decent analytics, hopefully more than here's your traces...good luck
 - Ease of finding complete kit; nothing worse than dropping a dime on what
 looks like a good deal only to find you're missing the unobtanium cable, or
 the software disk that the vendor will be more than happy to provide you
 only under a cripplingly expensive support contract.


-- 
The contents of this e-mail and any attachments are intended solely for the use 
of the named
addressee(s) and may contain confidential and/or privileged information. Any 
unauthorized use,
copying, disclosure, or distribution of the contents of this e-mail is strictly 
prohibited by
the sender and may be unlawful. If you are not the intended recipient, please 
notify the sender
immediately and delete this e-mail.