[neonixie-l] Re: How close together do a controller and crystal need to be?

[chuckrr]

A handy way I use, to remember the approximate speed of light, which is 
also the approximate

speed at which an electrical signal travels in a wire is just to think 
of it

as 1 nanosecond per foot.   Approximately.
 

 Original Message 
From: "gregebert" 
Sent: 12/27/2020 12:32:41 AM
To: "neonixie-l" 
Subject: Re: [neonixie-l] How close together do a controller and 
crystal need to be?
 

I'm assuming you are routing the output signal of an oscillator, not 
the crystal signals themselves.

 

The rise- & fall-times of the clock signal will determine how long the 
trace can be without termination. Faster edge-rates, say in the 2-3nsec range, 
will limit your trace to around 1 inch.

Signals propagate around 150psec/inch, and if the rise/fall times are 
about 10x (or larger) longer than the flight-time, then reflections should not 
have sufficient amplitude to cause false clocking.

 

In the example above, 1 inch of trace has a round-trip flight-time of 
300psec. If the rise and fall delays are 3nsec or larger, you can safely use 1 
inch of trace without using termination networks or controlled-impedance traces.

 

SPICE simulations are very helpful when deciding how to design clock 
lines when you cant satisfy the above rule.


On Saturday, December 26, 2020 at 4:06:26 PM UTC-8 Bill van 
Dijk wrote:




As long as there is not something very 
noisy on the other side of the board you’ll be just fine.

 

Bill





 

 

From: neoni...@googlegroups.com 
[mailto:neoni...@googlegroups.com] On Behalf Of Erick Anderson
Sent: Saturday, December 26, 2020 6:53 
PM
To: neonixie-l 

Subject: [neonixie-l] How close 
together do a controller and crystal need to be?

 


I designed a board for the 6-digit All 
Spectrum controller, which uses the Dallas TCXO chip. That's what goes in the 
DIP-14 socket in the picture. Right now they're as close to each other as 
possible. I'm thinking about redesigning the board to be a bit shorter, and 
moving the socket into the empty space at the right of the board would help. 
This would make the clock signal trace much longer, but is that actually a 
problem?





--
You received this message because you 
are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop 
receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, 
visit 
https://groups.google.com/d/msgid/neonixie-l/1bbc978e-4883-4b0e-920f-d5f3ff4c4c2cn%40googlegroups.com.





--
You received this message because you are subscribed to the Google 
Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send 
an email to neonixie-l+unsubscr...@googlegroups.com.
To view this discussion on the web, visit 
https://groups.google.com/d/msgid/neonixie-l/bbe8ae0a-bb3c-4f4a-98f6-9f56bc1e1805n%40googlegroups.com.

-- 
You received this message because you are subscribed to the Google Groups 
"neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to neonixie-l+unsubscr...@googlegroups.com.
To view this discussion on the web, visit 
https://groups.google.com/d/msgid/neonixie-l/202012271120827.SM185808%40ADT-IM1.

Re: [neonixie-l] How close together do a controller and crystal need to be?

[David Forbes]

Don't worry about it.
This is a CMOS 5V signal, I presume. It only has one load, at the far end
of the trace. You do not say what the frequency is, so I will guess that
it's on the order of 5-20 MHz.
The signal integrity is important, to prevent ringing that would
double-trigger the load circuit. A non-terminated transmission line will
have a reflection, whose size is a function of the sharpness of the rising
or falling edge, and the length of the trace.
The output transition time of the oscillator (the rise and fall time) is
likely greater than 5 nanoseconds. As long as the trace length is a lot
shorter than the transition time, you don't have to worry about reflection
from the end of the line causing ringing.
FR4 has a propagation delay of 2ns/foot, based on its dielectric
coefficient of ~4.5. So a trace up to one foot long will 'swallow' the edge
rather than causing ringing.
I apologize for the slightly hand-waving analysis, but this isn't a
critical case, so a detailed analysis isn't needed.
I recently designed a clock driver circuit board for a 4 GHz spectrometer,
if you're wondering what I do in my day job that makes me think about clock
traces.



On Sat, Dec 26, 2020, 4:53 PM Erick Anderson  wrote:

>
> I designed a board for the 6-digit All Spectrum controller, which uses the
> Dallas TCXO chip. That's what goes in the DIP-14 socket in the picture.
> Right now they're as close to each other as possible. I'm thinking about
> redesigning the board to be a bit shorter, and moving the socket into the
> empty space at the right of the board would help. This would make the clock
> signal trace much longer, but is that actually a problem?
>
> --
> You received this message because you are subscribed to the Google Groups
> "neonixie-l" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to neonixie-l+unsubscr...@googlegroups.com.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/neonixie-l/1bbc978e-4883-4b0e-920f-d5f3ff4c4c2cn%40googlegroups.com
> 
> .
>

-- 
You received this message because you are subscribed to the Google Groups 
"neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to neonixie-l+unsubscr...@googlegroups.com.
To view this discussion on the web, visit 
https://groups.google.com/d/msgid/neonixie-l/CAPbqtvfgTaKkiAWEZvtOHNJ1sEwwy_5pOtrg18-JKkfHy2uPUw%40mail.gmail.com.

Re: [neonixie-l] Re: How close together do a controller and crystal need to be?

[Charles MacDonald]

On 2020-12-27 6:27 p.m., chuckrr wrote:
Well even thoughe electricity flowing through a wire is somewhat slower 
than the speed of light, exactly how much slower


is something I do not know about.   this install

order called for the use of a 100 foot cable.   The result was that the 
newly installed processor complex was unable to synch


up with the rest of the switch.   Nobody knew why.   I calculated that 
at the speed of light, the 12.352 mHz clock signal would be


late at the new processor by around half of a cycle.  


it is legendary that Rear Admiral Grace Hopper, (the Computer genius who 
is credited with the early computer language COBAL) used to had out 
little pieces of wire, cut to the length that would take one Nanosecond 
to travel.  "This Gentlemen is a Nanosecond".


she was of course attempting to get computer designers to make the 
interconnects shorter, as that would allow faster computers. as well as 
explaining lag in satellite communications.


https://en.wikipedia.org/wiki/Grace_Hopper#Anecdotes

--
Charles MacDonald  VA3CPY   Stittsville Ontario
cm...@zeusprune.ca  Just Beyond the Fringe
No Microsoft Products were used in sending this e-mail.

--
You received this message because you are subscribed to the Google Groups 
"neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to neonixie-l+unsubscr...@googlegroups.com.
To view this discussion on the web, visit 
https://groups.google.com/d/msgid/neonixie-l/2f17c8d8-ab45-77d8-9b7c-dd9027fd660f%40zeusprune.ca.

Re: [neonixie-l] Re: How close together do a controller and crystal need to be?

[Adrian Godwin]

https://en.wikipedia.org/wiki/Velocity_factor

On Sun, Dec 27, 2020 at 10:05 PM Paul Andrews  wrote:

> Hmm. My recollection from high school physics was that the speed of
> propagation along a wave guide was around 90% the speed of light -
> presumably limited by the dielectric. Signals along a plain old wire, on
> the other hand, were more like 1/3 the speed of light. Now I'm going to
> have to double-check that for the first time in over 40 years!
>
> On Sunday, December 27, 2020 at 2:20:06 PM UTC-5 Chuck wrote:
>
>> A handy way I use, to remember the approximate speed of light, which is
>> also the approximate
>>
>> speed at which an electrical signal travels in a wire is just to think of
>> it
>>
>> as 1 nanosecond per foot.   Approximately.
>>
>>
>>  Original Message 
>> From: "gregebert" 
>> Sent: 12/27/2020 12:32:41 AM
>> To: "neonixie-l" 
>> Subject: Re: [neonixie-l] How close together do a controller and crystal
>> need to be?
>>
>> I'm assuming you are routing the output signal of an oscillator, not the
>> crystal signals themselves.
>>
>> The rise- & fall-times of the clock signal will determine how long the
>> trace can be without termination. Faster edge-rates, say in the 2-3nsec
>> range, will limit your trace to around 1 inch.
>> Signals propagate around 150psec/inch, and if the rise/fall times are
>> about 10x (or larger) longer than the flight-time, then reflections should
>> not have sufficient amplitude to cause false clocking.
>>
>> In the example above, 1 inch of trace has a round-trip flight-time of
>> 300psec. If the rise and fall delays are 3nsec or larger, you can safely
>> use 1 inch of trace without using termination networks or
>> controlled-impedance traces.
>>
>> SPICE simulations are very helpful when deciding how to design clock
>> lines when you cant satisfy the above rule.
>> On Saturday, December 26, 2020 at 4:06:26 PM UTC-8 Bill van Dijk wrote:
>>
>>> As long as there is not something very noisy on the other side of the
>>> board you’ll be just fine.
>>>
>>>
>>>
>>> Bill
>>>
>>>
>>>
>>>
>>>
>>> *From:* neoni...@googlegroups.com [mailto:neoni...@googlegroups.com] *On
>>> Behalf Of *Erick Anderson
>>> *Sent:* Saturday, December 26, 2020 6:53 PM
>>> *To:* neonixie-l 
>>> *Subject:* [neonixie-l] How close together do a controller and crystal
>>> need to be?
>>>
>>>
>>>
>>>
>>> I designed a board for the 6-digit All Spectrum controller, which uses
>>> the Dallas TCXO chip. That's what goes in the DIP-14 socket in the picture.
>>> Right now they're as close to each other as possible. I'm thinking about
>>> redesigning the board to be a bit shorter, and moving the socket into the
>>> empty space at the right of the board would help. This would make the clock
>>> signal trace much longer, but is that actually a problem?
>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "neonixie-l" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to neonixie-l+...@googlegroups.com.
>>> To view this discussion on the web, visit
>>> https://groups.google.com/d/msgid/neonixie-l/1bbc978e-4883-4b0e-920f-d5f3ff4c4c2cn%40googlegroups.com
>>> 
>>> .
>>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "neonixie-l" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to neonixie-l+...@googlegroups.com.
>>
>> To view this discussion on the web, visit
>> https://groups.google.com/d/msgid/neonixie-l/bbe8ae0a-bb3c-4f4a-98f6-9f56bc1e1805n%40googlegroups.com
>> 
>> .
>>
> --
> You received this message because you are subscribed to the Google Groups
> "neonixie-l" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to neonixie-l+unsubscr...@googlegroups.com.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/neonixie-l/480dcaf7-ce2b-45be-b8cc-bb1ee79bc5a2n%40googlegroups.com
> 
> .
>

-- 
You received this message because you are subscribed to the Google Groups 
"neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to neonixie-l+unsubscr...@googlegroups.com.
To view this discussion on the web, visit 
https://groups.google.com/d/msgid/neonixie-l/CALiMYruP%2BG1MwK5ivp-_SXgFPXO6AfnFPRAXNYpc68uFohPHmg%40mail.gmail.com.

Re: [neonixie-l] Re: How close together do a controller and crystal need to be?

[chuckrr]

Well even thoughe electricity flowing through a wire is somewhat slower 
than the speed of light, exactly how much slower

is something I do not know about.   What I do know is that the 
approximation of it being "around" the speed of light has served me

very well over the years.   One time on an an installation job at a 
telephone switching office where all previoisly installed

processor complexes had been installed using 50 foot coax cables to get 
the 12.352 mHzmaster clock to them all, this install

order called for the use of a 100 foot cable.   The result was that the 
newly installed processor complex was unable to synch

up with the rest of the switch.   Nobody knew why.   I calculated that 
at the speed of light, the 12.352 mHz clock signal would be

late at the new processor by around half of a cycle.     We talked this 
over with network operations and they confirmed my calcualtions

and then instructed the crew to change out ALL the clock cables in the 
entire exchange to 100 footers to match the new one.

There was lost of slack to stow, but it all came up just fine after 
that.  The install crew wanted to know how I managed to

know what the problem was.  I said, "simple gentlemen, it's the speed 
of light and our new intsall was getting the message rather late"
 

 Original Message 
From: "Adrian Godwin" 
Sent: 12/27/2020 5:42:02 PM
To: neonixie-l@googlegroups.com
Subject: Re: [neonixie-l] Re: How close together do a controller and 
crystal need to be?
 

https://en.wikipedia.org/wiki/Velocity_factor



On Sun, Dec 27, 2020 at 10:05 PM Paul Andrews  
wrote:

Hmm. My recollection from high school physics was that the 
speed of propagation along a wave guide was around 90% the speed of light - 
presumably limited by the dielectric. Signals along a plain old wire, on the 
other hand, were more like 1/3 the speed of light. Now I'm going to have to 
double-check that for the first time in over 40 years!



On Sunday, December 27, 2020 at 2:20:06 PM 
UTC-5 Chuck wrote:


A handy way I use, to remember the 
approximate speed of light, which is also the approximate

speed at which an electrical signal 
travels in a wire is just to think of it

as 1 nanosecond per foot.   
Approximately.
 

 Original Message 
From: "gregebert" 
Sent: 12/27/2020 12:32:41 AM
To: "neonixie-l" 

Subject: Re: [neonixie-l] How close 
together do a controller and crystal need to be?
 

I'm assuming you are routing the output 
signal of an oscillator, not the crystal signals themselves.

 

The rise- & fall-times of the clock 
signal will determine how long the trace can be without termination. Faster 
edge-rates, say in the 2-3nsec range, will limit your trace to around 1 inch.

Signals propagate around 150psec/inch, 
and if the rise/fall times are about 10x (or larger) longer than the 
flight-time, then reflections should not have sufficient amplitude to cause 
false clocking.

 

In the example above, 1 inch of trace 
has a round-trip flight-time of 300psec. If the rise and fall delays are 3nsec 
or larger, you can safely use 1 inch of trace without using termination 
networks or controlled-impedance traces.

 

SPICE simulations are very helpful when 
deciding how to design clock lines when you cant satisfy the above rule.


On Saturday, December 26, 2020 
at 4:06:26 PM UTC-8 Bill van Dijk wrote:


   

[neonixie-l] Re: How close together do a controller and crystal need to be?

[Paul Andrews]

Hmm. My recollection from high school physics was that the speed of 
propagation along a wave guide was around 90% the speed of light - 
presumably limited by the dielectric. Signals along a plain old wire, on 
the other hand, were more like 1/3 the speed of light. Now I'm going to 
have to double-check that for the first time in over 40 years!

On Sunday, December 27, 2020 at 2:20:06 PM UTC-5 Chuck wrote:

> A handy way I use, to remember the approximate speed of light, which is 
> also the approximate
>
> speed at which an electrical signal travels in a wire is just to think of 
> it
>
> as 1 nanosecond per foot.   Approximately.
>  
>
>  Original Message 
> From: "gregebert" 
> Sent: 12/27/2020 12:32:41 AM
> To: "neonixie-l" 
> Subject: Re: [neonixie-l] How close together do a controller and crystal 
> need to be?
>  
> I'm assuming you are routing the output signal of an oscillator, not the 
> crystal signals themselves.
>  
> The rise- & fall-times of the clock signal will determine how long the 
> trace can be without termination. Faster edge-rates, say in the 2-3nsec 
> range, will limit your trace to around 1 inch.
> Signals propagate around 150psec/inch, and if the rise/fall times are 
> about 10x (or larger) longer than the flight-time, then reflections should 
> not have sufficient amplitude to cause false clocking.
>  
> In the example above, 1 inch of trace has a round-trip flight-time of 
> 300psec. If the rise and fall delays are 3nsec or larger, you can safely 
> use 1 inch of trace without using termination networks or 
> controlled-impedance traces.
>  
> SPICE simulations are very helpful when deciding how to design clock lines 
> when you cant satisfy the above rule.
> On Saturday, December 26, 2020 at 4:06:26 PM UTC-8 Bill van Dijk wrote:
>
>> As long as there is not something very noisy on the other side of the 
>> board you’ll be just fine.
>>
>>  
>>
>> Bill
>>
>>  
>>
>>  
>>
>> *From:* neoni...@googlegroups.com [mailto:neoni...@googlegroups.com] *On 
>> Behalf Of *Erick Anderson
>> *Sent:* Saturday, December 26, 2020 6:53 PM
>> *To:* neonixie-l 
>> *Subject:* [neonixie-l] How close together do a controller and crystal 
>> need to be?
>>
>>  
>>
>>
>> I designed a board for the 6-digit All Spectrum controller, which uses 
>> the Dallas TCXO chip. That's what goes in the DIP-14 socket in the picture. 
>> Right now they're as close to each other as possible. I'm thinking about 
>> redesigning the board to be a bit shorter, and moving the socket into the 
>> empty space at the right of the board would help. This would make the clock 
>> signal trace much longer, but is that actually a problem?
>>
>> --
>> You received this message because you are subscribed to the Google Groups 
>> "neonixie-l" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to neonixie-l+...@googlegroups.com.
>> To view this discussion on the web, visit 
>> https://groups.google.com/d/msgid/neonixie-l/1bbc978e-4883-4b0e-920f-d5f3ff4c4c2cn%40googlegroups.com
>>  
>> 
>> .
>>
> --
> You received this message because you are subscribed to the Google Groups 
> "neonixie-l" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to neonixie-l+...@googlegroups.com.
>
> To view this discussion on the web, visit 
> https://groups.google.com/d/msgid/neonixie-l/bbe8ae0a-bb3c-4f4a-98f6-9f56bc1e1805n%40googlegroups.com
>  
> 
> .
>

-- 
You received this message because you are subscribed to the Google Groups 
"neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to neonixie-l+unsubscr...@googlegroups.com.
To view this discussion on the web, visit 
https://groups.google.com/d/msgid/neonixie-l/480dcaf7-ce2b-45be-b8cc-bb1ee79bc5a2n%40googlegroups.com.

[neonixie-l] Nixie Entertainment!

[martin martin]

I thought you guys would like this YouTube fellow!

https://www.youtube.com/watch?v=Us6rxEsLrfE=28s

Enjoy..

-- 
You received this message because you are subscribed to the Google Groups 
"neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to neonixie-l+unsubscr...@googlegroups.com.
To view this discussion on the web, visit 
https://groups.google.com/d/msgid/neonixie-l/30b00be0-9267-4b12-ac95-0da201e1456an%40googlegroups.com.