subject:"\[time\-nuts\] Newbie questions"

Bill,

Thanks. Your reply was very informative. I understand the reason for using
decibels for the applications you mention. However, I did not consider the
output level of a non-RF signal to be in that category.

You are right, I have never seen a DMM reading in decimals. I know that
digital scopes can do so (and understand how they work), but again I have
never seen or used one. Given the zero point (finally figured that one out, I
was missing the assumed impedance factor which was varying) I can convert
levels - not quite in my head, since I can't do logs in my head - but close
and can get a pretty good guestimate that way.


Mike

P.S. I also used punch cards. I kept my source library first on punch cards
and then mag tape. I eventually bought my own IBM disk drive for the 360 at
work. I started on an RCA 70/46 using punch tape in a time sharing
environment. I had a trash-80 before S-100s. I even had a full size IBM tape
drive and four hard disk drives. Never could do anything with them, I just
couldn't get 3-phase 440 in an apartment complex and IBM used some very
unusual logic levels. I eventually gave the hard disk drives away and
abandoned the tape drive. I had the original CP/M source code from Gary
Kildall written in PL/M. Threw all of that stuff out about 10 years ago and
even more a bit more about a year and a half ago. But I still have the MASM
Assembler manuals from Microsoft. Those were last published around 1995 - and
I needed them just a few months ago.


> Hi, MIke. I used the university CDC6400/6600 supercomputer while in
> engineering school with punch cards or Teletypes and was familiar with S-100
> vintage equipment. Somewhere I may still have a MIcrosoft BASIC pre MS-DOS
> (HDOS or CP/M) looseleaf manual. I haven't retired yet, but have been an
> Application Engineer at Tektronix for nearly 30 years. So I can appreciate
> your mindset. I'm going to only answer two of your questions:

> 1. What is the zero value for voltage and watts using logarithmic
>scaling   (at least as used here)? Is there actually a consistent
>underlying value   across all applications?

> 2. Why use it for specifying voltage or power in a limited range? Why
>not   just say that the output is 1.0v rms or 0.7v, or that it uses
>50mW? There   does not appear to be any actual advantage to using a
>logarithmic scale   for a small range of values - and 1mV to 1kV IS a
>small range.   Especially when you have to convert the logarithmic
>value to a "real"   value to actually do anything with it.

> RF and audio (including telephone) types have used logarithmic (dB)
> units for many decades. There is often a need to discuss thermal noise
> levels and transmitter power levels in the same circuit, which can lead
> to very large voltage and power ratios. It's common to need to relate
> voltages over a 10^10 range and powers over a 10^20 range
> (1:0.0001). Engineers and scientists like to use numeric
> values which are easier to work with. You wouldn't want to specify a
> hard drive size as 1,000,000,000,000 bytes but as 1 TB (ignoring the
> power of 2 vs power of 10 issue). So we use "engineering units" (powers
> of 1,000) for frequency (kHz, MHz, GHz, THz), voltage (pV, nV, uV, mV,
> V, kV, MV, GV).

> RF applications are more naturally dealt with in terms of power. The
> noise generated in a resistor due to thermal agitation (Johnson-Nyquist
> noise) is P = kTB, where P = power in watts, k = Boltzmanns constant
> (1.28 x 10^-23 J/K), T = absolute temperature in Kelvins, and B is the
> measurement bandwidth in Hertz. Many RF components are rated by power
> dissipation. Historically it's been much easier to measure RF signals
> levels by measuring thermal changes due to signal power.

> The very large dynamic range required for characterizing sound and
> telephone line levels and relating them to human perceived level change
> led to the definition of the Bel (power ratio of 10, named for Alexander
> Graham Bell). The decibel (1/10 Bel) is the logarithmic unit which is
> used in practice. The noise delivered by a resistor to a matched load at
> room temperature and normalized to a 1 Hz measurement bandwidth is about
> -174 dBm/Hz.

> The "m" in "dBm" stands for 1 mW (milliwatt). So 0 dBm = 1 mW. You
> should read "dBm" as "decibels relative to 1 milliwatt". Since most
> power levels in RF equipment tend to be within a couple of orders of
> magnitude below or above a milliwatt, dBm is the main unit used for RF
> equipment which can fit on your lab bench. In some cases it's convenient
> to use logarithmic voltage units, and the common units are dBuV
> (decibels relative to 1 microvolt), dBmV (decibels relative to 1
> millivolt), or dBV (decibels relative to 1 volt). But except for cable
> television and a few other applications (including noise levels at low
> audio frequencies), dBm rules the RF world.

> You are correct that linear non-logarithmic units work well when a small

Re: [time-nuts] Newbie questions

Jeremy,

While you are right, "real" mathematicians don't have anything to do with
numbers! Its all functions, sets, maps, groups, rings and formal logic (very
tongue in cheek here, of course). All very abstract.

I also understand about the "chain of elements", but my assumption was that it
was a significant part of this arena. Certainly due to my inexperience in the
area.


Mike


> That hits the nail on the head. Mathematicians (the OP mentioned he was
> one) learned long ago that it's easier to add and subtract than to multiply
> and divide.

> Jeremy


> On Wednesday, January 27, 2016, Jim Harman  wrote:

>> Another benefit of using dB vs Watts or Volts is that systems often consist
>> of a chain of elements with gains and losses. Working with gains and losses
>> in dB lets you calculate the signal level at any point along the way and
>> the system gain by adding and subtracting rather than multiplying and
>> dividing.
>>
>> On Wed, Jan 27, 2016 at 3:02 AM, wb6bnq >
>> wrote:
>>
>> > Hi Mike,
>> >
>> > The element that you are missing is the impedance.  When you look at the
>> > common formula it refers to a ratio of power or voltage and the impedance
>> > is left out with the understanding that the impedance is equal for each
>> > power or voltage in the ratio.  The actual formula (for power) is DB =
>> 10 X
>> > log10 ( E1^2 / R ) / ( E2^2 / R ).  Hint   Power P = E^2 / R.
>> >
>> > In the RF world that impedance is 50 Ohms and ZERO DBm(illiwatts) is ONE
>> > milliwatt into 50 Ohms which is 0.223606797749979 Vrms.
>> >
>> > In the audio world the reference impedance is 600 Ohms.  So ZERO DBu is
>> > One milliwatt into 600 Ohms which is 0.7745966692414834 Vrms.  More
>> > commonly referred to as 0.775 Vrms.
>> >
>> > As for your GPS questions, I will leave that up to others to answer.
>> >
>> > BillWB6BNQ
>> >
>> >
>> > time...@metachaos.net  wrote:
>> >
>> > Hi,
>> >>
>> >> I am a newbie to this list. I have downloaded the archives and read
>> about
>> >> 5,000 of the past messages. I plan on building my own GPSDO, probably
>> >> using a
>> >> LEA-6T (but LEA-7T or LEA-M8T would be good if I can find one
>> >> affordably). I
>> >> have a MTI 260 on order (although it could wind up being a 261 since
>> they
>> >> all
>> >> appear to ship one or the other randomly).
>> >>
>> >> Currently, my resources include a DMM (well, a couple) and soldering /
>> >> desoldering stations and quite a few tools. I also have an oscilloscope
>> >> that I
>> >> am currently repairing - a 400Mhz Tektronix 2465BCT analog scope. I am
>> >> waiting
>> >> on the final parts from Mouser. Once that is done I need to get it
>> >> calibrated.
>> >> All of that will probably take me another month. I also need to finish
>> >> fixing
>> >> my cassette deck - and then to finish writing a special recording
>> program
>> >> to
>> >> use raw device drivers to get around the fact that Windows is not real
>> >> time. I
>> >> interrupted that project to work on the scope.
>> >>
>> >> In the meantime, I am reading the time-nuts messages (and lots of other
>> >> things) to gather information and ideas about how I am going to do this
>> >> and
>> >> generally to learn more.
>> >>
>> >> So, I have some questions. Let me tell you a bit about me, so that you
>> >> know
>> >> the context and my limitations. I am a retired programmer. I wrote just
>> >> about
>> >> everything including device drivers, operating systems, utilities,
>> >> various AI
>> >> programs, telephone systems, compilers, encryption, web applications and
>> >> much
>> >> more. If I need to throw 50,000 LOC at a project, no problem. I have
>> used
>> >> many
>> >> languages including quite a few different assembly languages (I have
>> also
>> >> written an assembler). I consider myself a mathematician / programmer,
>> >> although I haven't really needed Calculus or Differential Equations for
>> >> decades, so I am pretty rusty in that area. I do more work in formal
>> logic
>> >> than higher mathematics. But, I THINK like a mathematician. Formalism
>> and
>> >> abstraction come naturally to me.
>> >>
>> >> During my career I also helped to debug hardware during S-100 days. I
>> have
>> >> sporadically messed with electronics off and on, informally, with no
>> >> education
>> >> in the area. Now that I am retired (and have more time, but less money -
>> >> it IS
>> >> a zero sum game!), I am trying to learn more about electronics and start
>> >> doing
>> >> hardware projects. I have never been into model building or anything
>> >> similar,
>> >> so my construction skills are lacking. I understand a lot of things in
>> >> theory,
>> >> but practice still eludes me. For example, knowing a part exists or
>> >> determining which of 10,000 apparently identical parts is the "right"
>> >> choice.
>> >> It can hours or even days to find the "right" connector. In many cases,
>> >> the
>> >> names or descriptions are

Re: [time-nuts] Newbie questions

2016-01-27 Thread Attila Kinali

Moin,

As no-one seems to want to answer the GPS related questions

On Tue, 26 Jan 2016 19:04:10 -0500
time...@metachaos.net wrote:

> 
> Paradoxically, I have no interest in time. As in time of day, day of week,
> etc.. I have never had a job where I got to work on time. My philosophy has
> always been "go to bed when sleepy, get up when not". I was notorious in high
> school for only showing up on test day. But, I am interested in being able to
> timestamp events accurately and in measuring time (and other things). I am
> also interested in how a very accurate frequency source can be used in
> other applications and test instruments. That brings me to my desire to build
> a GPSDO and my questions.

I have wirtten a couple of mails with references to popular GPSDO's
a couple of times in the past. You might want to have a look at them
and the discussions they were in:

https://www.febo.com/pipermail/time-nuts/2012-March/064873.html
https://www.febo.com/pipermail/time-nuts/2013-June/077368.html

> I have also been researching GPS antennas. From what I can see there are two
> basic types - the flat puck and the helical. 

There are actually many more, but the cross dipole, the spiral antenna
and the two you mentioned are by far the most popular ones.

> I have not seen anything to
> distinguish the two types based on performance or usage or to indicate that
> one or the other might be better for GPS timing.

What makes good antennas and what not is actually not that easy to define.
There was a discussion on this last spring. look for
"Important parameters for a GPS/GNSS antenna" in the archives.

> However, I have seen "GPS
> Timing Reference Antennas" advertised. Most or all of those appear to be
> helical. But, I have not seen anything that specifies the difference between 
> an active GPS antenna and an active GPS Timing Reference Antenna.

>1. What is the difference between a "normal" GPS antenna and a GPS Timing
>   Reference antenna? What features are of interest?

The timing reference antennas are usually build by manufacturers with the
intended use for high precision instruments. Either timing or geodetic survey.
Their parameters are more strictly controlled, they have (usually) a more
stable phase center and some of them are also dual band (L1/L2).

Unless you need sub-ns stability and accuracy or have a dual band receiver,
go for a standard antenna.

>2. Is there anything extra needed besides a GPS antenna to enable the use
>   of WAAS or other services? Apparently the ubolt receivers can make use
>   of some of that, but it is not clear what is needed to provide that
>   information to them, or if they just pick it up automatically using a
>   standard GPS antenna.

I am not sure what the LEA modules need to active WAAS/EGNOS/etc, it's
probably listed under one of the commands in the protocol reference.

You don't need any special antenna for those, as these augmentation
services use the same frequency as L1 GPS.

> Also, from what I have read, using carrier phase for timing is potentially
> more accurate by a couple orders of magnitude. Are there any GPS timing
> receivers available that use carrier phase? 

Yes, but they are darn expensive.

> Or use both L1 and L2 for increased accuracy? 

Yes, but again, expensive.

> I see that the ubolt receivers can report some carrier
> phase information, but that doesn't appear to translate to increased accuracy.

If anything, it would lead to higher precision ;-)

The reason why carrier phase tracking does not lead to large improvements
is because most receivers are single frequency receivers which cannot
directly calculate the ionospheric corrections. Thus they have to rely
on predictions send out by the GPS satellites (and by WAAS/EGNOS).
But even with those predictions, the ionospheric terms will dominate
the error. Thus there is very little gain from using carrier phase
tracking over code phase only tracking for an L1 only receiver.

> And the LEA M8T use dual channels, but don't appear to mix GPS and GLASNOS to
> improve accuracy. Do any receivers do that?

I did not have a look at the M8T yet, so i cannot answer this.
I am pretty sure that there are dual system timing receivers out there.

>  I suspect that building a GPS
> receiver is probably more complex than can be easily handled by an amateur so
> I am most likely restricted by available receivers.

Not really. I know of three people on this mailinglist who are
doing exactly this. It is just a lot of work and takes a lot of time.
The basics of a GPS receiver can be taught in an afternoon and would
enable you to build your own. But, doing it right and and time-nut
grade, with support for more than just L1 C/A requires quite a bit
more understanding and even more time.

> I have also read, more than once, statements in this forum that something or
> another could be had for some low, low price so why build it yourself? I think
> that there are several

Re: [time-nuts] Newbie questions

Bob, et. al.,

Thanks for the advice and information. That has helped a lot in closing some
holes and gives me a lot to consider. I am continuing to research and learn,
this is not a short term project as in building it this week or month. I need
to learn and work up to the full project. I will probably start small with
pieces and gather what equipment I can. Like many of my projects, this one
will probably have multiple sub-projects. Especially if I manage to snag test
equipment that needs repair.

Regrettably, the 260 was ordered from China, there were none from the US when
I ordered. I did know that it would be a crap shoot, but for $26 it was
something I could get now. I might even be lucky!

I am planning on including sawtooth correction in hardware. But, I have a good
bit to learn for that. It appears to be fairly simple given your processor
reads the sawtooth correction from the GPS receiver and sets up a programmable
delay for each cycle.

I have a dozen pieces floating in my head, I am certain I will be back for
more advice.

Thanks again.


Mike

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

2016-01-27 Thread wb6bnq


Hi Mike,

The element that you are missing is the impedance.  When you look at the 
common formula it refers to a ratio of power or voltage and the 
impedance is left out with the understanding that the impedance is equal 
for each power or voltage in the ratio.  The actual formula (for power) 
is DB = 10 X log10 ( E1^2 / R ) / ( E2^2 / R ).  Hint   Power P = E^2 / R.


In the RF world that impedance is 50 Ohms and ZERO DBm(illiwatts) is ONE 
milliwatt into 50 Ohms which is 0.223606797749979 Vrms.


In the audio world the reference impedance is 600 Ohms.  So ZERO DBu is 
One milliwatt into 600 Ohms which is 0.7745966692414834 Vrms.  More 
commonly referred to as 0.775 Vrms.


As for your GPS questions, I will leave that up to others to answer.

BillWB6BNQ


time...@metachaos.net wrote:


Hi,

I am a newbie to this list. I have downloaded the archives and read about
5,000 of the past messages. I plan on building my own GPSDO, probably using a
LEA-6T (but LEA-7T or LEA-M8T would be good if I can find one affordably). I
have a MTI 260 on order (although it could wind up being a 261 since they all
appear to ship one or the other randomly).

Currently, my resources include a DMM (well, a couple) and soldering /
desoldering stations and quite a few tools. I also have an oscilloscope that I
am currently repairing - a 400Mhz Tektronix 2465BCT analog scope. I am waiting
on the final parts from Mouser. Once that is done I need to get it calibrated.
All of that will probably take me another month. I also need to finish fixing
my cassette deck - and then to finish writing a special recording program to
use raw device drivers to get around the fact that Windows is not real time. I
interrupted that project to work on the scope.

In the meantime, I am reading the time-nuts messages (and lots of other
things) to gather information and ideas about how I am going to do this and
generally to learn more.

So, I have some questions. Let me tell you a bit about me, so that you know
the context and my limitations. I am a retired programmer. I wrote just about
everything including device drivers, operating systems, utilities, various AI
programs, telephone systems, compilers, encryption, web applications and much
more. If I need to throw 50,000 LOC at a project, no problem. I have used many
languages including quite a few different assembly languages (I have also
written an assembler). I consider myself a mathematician / programmer,
although I haven't really needed Calculus or Differential Equations for
decades, so I am pretty rusty in that area. I do more work in formal logic
than higher mathematics. But, I THINK like a mathematician. Formalism and
abstraction come naturally to me.

During my career I also helped to debug hardware during S-100 days. I have
sporadically messed with electronics off and on, informally, with no education
in the area. Now that I am retired (and have more time, but less money - it IS
a zero sum game!), I am trying to learn more about electronics and start doing
hardware projects. I have never been into model building or anything similar,
so my construction skills are lacking. I understand a lot of things in theory,
but practice still eludes me. For example, knowing a part exists or
determining which of 10,000 apparently identical parts is the "right" choice.
It can hours or even days to find the "right" connector. In many cases, the
names or descriptions are completely meaningless. That all appears to be an
experience related issue, so I will (hopefully) overcome that in time.

I have no problem with soldering / desoldering, but I haven't designed or
built my own PCB yet. I have designed / redesigned some minor circuits,
especially on the power supply side. I can follow schematics reasonably well,
but I am not comfortable with Eagle or other PCB layout programs. Every time
I have tried one of those programs, half of the parts I needed were not
available. I have started using TinyCAD which is much easier to use. So, I
have a lot to learn. But, that is basically what I do, all day, every day. I'm
the type of person that gets bored easily and quickly. As #5 said "more input,
more input"!   6.02059991327962

Paradoxically, I have no interest in time. As in time of day, day of week,
etc.. I have never had a job where I got to work on time. My philosophy has
always been "go to bed when sleepy, get up when not". I was notorious in high
school for only showing up on test day. But, I am interested in being able to
timestamp events accurately and in measuring time (and other things). I am
also interested in how a very accurate frequency source can be used in
other applications and test instruments. That brings me to my desire to build
a GPSDO and my questions.

I understand the logarithmic scaling used for voltage and power. I even
understand why voltage uses a multiplier of 20 and power a multiplier of 10.
It makes sense when working with a wide range of values.

[time-nuts] Newbie questions

Ha!

I finally answered one of my questions that has been bugging me!

I was taking dBm as the definition of voltage - which it is, but only
round-about. The hidden variable is the assumed impedance of 50 ohms where 1mW
is the zero point for power. That is also why I wasn't getting consistent
answers - different application areas I am interest in assumed different
values for the impedance and so my calculations were all over the place. I
found a reference in the PRS10 reference that tied it all together for me.


Mike

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

2016-01-27 Thread Bill Byrom

Hi, MIke. I used the university CDC6400/6600 supercomputer while in engineering 
school with punch cards or Teletypes and was familiar with S-100 vintage 
equipment. Somewhere I may still have a MIcrosoft BASIC pre MS-DOS (HDOS or 
CP/M) looseleaf manual. I haven't retired yet, but have been an Application 
Engineer at Tektronix for nearly 30 years. So I can appreciate your mindset. 
I'm going to only answer two of your questions:

1. What is the zero value for voltage and watts using logarithmic
   scaling   (at least as used here)? Is there actually a consistent
   underlying value   across all applications?

2. Why use it for specifying voltage or power in a limited range? Why
   not   just say that the output is 1.0v rms or 0.7v, or that it uses
   50mW? There   does not appear to be any actual advantage to using a
   logarithmic scale   for a small range of values - and 1mV to 1kV IS a
   small range.   Especially when you have to convert the logarithmic
   value to a "real"   value to actually do anything with it.

RF and audio (including telephone) types have used logarithmic (dB)
units for many decades. There is often a need to discuss thermal noise
levels and transmitter power levels in the same circuit, which can lead
to very large voltage and power ratios. It's common to need to relate
voltages over a 10^10 range and powers over a 10^20 range
(1:0.0001). Engineers and scientists like to use numeric
values which are easier to work with. You wouldn't want to specify a
hard drive size as 1,000,000,000,000 bytes but as 1 TB (ignoring the
power of 2 vs power of 10 issue). So we use "engineering units" (powers
of 1,000) for frequency (kHz, MHz, GHz, THz), voltage (pV, nV, uV, mV,
V, kV, MV, GV).

RF applications are more naturally dealt with in terms of power. The
noise generated in a resistor due to thermal agitation (Johnson-Nyquist
noise) is P = kTB, where P = power in watts, k = Boltzmanns constant
(1.28 x 10^-23 J/K), T = absolute temperature in Kelvins, and B is the
measurement bandwidth in Hertz. Many RF components are rated by power
dissipation. Historically it's been much easier to measure RF signals
levels by measuring thermal changes due to signal power.

The very large dynamic range required for characterizing sound and
telephone line levels and relating them to human perceived level change
led to the definition of the Bel (power ratio of 10, named for Alexander
Graham Bell). The decibel (1/10 Bel) is the logarithmic unit which is
used in practice. The noise delivered by a resistor to a matched load at
room temperature and normalized to a 1 Hz measurement bandwidth is about
-174 dBm/Hz.

The "m" in "dBm" stands for 1 mW (milliwatt). So 0 dBm = 1 mW. You
should read "dBm" as "decibels relative to 1 milliwatt". Since most
power levels in RF equipment tend to be within a couple of orders of
magnitude below or above a milliwatt, dBm is the main unit used for RF
equipment which can fit on your lab bench. In some cases it's convenient
to use logarithmic voltage units, and the common units are dBuV
(decibels relative to 1 microvolt), dBmV (decibels relative to 1
millivolt), or dBV (decibels relative to 1 volt). But except for cable
television and a few other applications (including noise levels at low
audio frequencies), dBm rules the RF world.

You are correct that linear non-logarithmic units work well when a small
range of values are being used. Amateur Radio handheld transmitters have
power ratings usually given in linear watts (100 mW, 1 W, 3 W, 5 W,
etc.). A HF (high frequency 3-30 MHz) transmitter may have an output
power meter marked in linear units of Watts. But the received signal
strength meter is marked in logarithmic units, since with an automatic
gain control the receiver dynamic range is many orders of magnitude too
large to be shown with linear units.

You happen to have test equipment which measures linear units. Many
voltmeters designed to measure audio levels are calibrated in dBmV. Most
RF signal level measurement instruments (power meters and spectrum
analyzers) are usually used with dB or dBm scaling. Since the ratio of
RF signals is often the main measurement of interest (such as harmonic
or intermodulation level), spectrum analyzers are usually set up with a
a variable full scale value (reference level) and a dB ratio vertical
scale. The horizontal scale of a spectrum analyzer is usually linear
frequency, but in many cases can be changed to logarithmic frequency.

Modern oscilloscopes (those made in the past 20 years) digitize the
voltage waveform, and they can easily show RMS voltage and even power
levels. For many RF measurements the oscilloscope uses an FFT to create
a spectrum display scaled in dBm. A DMM is not used to measure RF
levels, but RF power meters directly display dBm power, as also used by
spectrum analyzers.

So the reason that your test equipment doesn't produce measurements in
the same units commonly used by RF engineers is that you

Re: [time-nuts] Newbie questions

2016-01-27 Thread Bob Camp

Hi

Ok, so let me answer the questions you *should* have asked:

(They are in no particular order. Number 3 probably should come first)

1) Is the gear I have enough to do this project? 

No, you will need some sort of frequency  / time standard. An atomic
clock of some sort is pretty much a minimum. You probably also need
a working GPSDO (or set of them) for comparison as well. You will also 
need a working / modern precision counter that will give you data down 
in the < 100 ps range. 

2) How will this ultimately be built? 

At the very least, you will be building this with surface mount devices. 
If it’s a scratch build, you will be dealing with fine pitch parts. That gets
you into a whole bunch of gear. It also gets you into a very real “is this 
fun or not” sort of question. 

3) What *is* the goal? 

"I’m going to make dinner” is the start of a process. It’s not enough of a 
goal to accomplish the task. Starting the task with a general objective is fine.
It does need to be refined a bit before you go much further.  

Is this what most of us would call a GPSDO (self contained box) or is it
something with a PC in the middle of it? 

Is this an OCXO based “precision” device or is it something more simple?

Is a pure software solution good enough?

Each of those decisions (and that’s by no means a full list) will send you off 
in a very different direction. 

4) How long is this likely to take?

Best guess based on the others who have done the same thing - several 
years. 

5) How much is this likely to cost?

If done the way others have done it, several thousand dollars up to 
quite a bit more than that. 

6) How much research is involved?

Quite a bit. The information you need is scattered all over the place. Figure
that you likely will read at least several hundred papers. There is a whole 
statistical language that is unique to these gizmos. This is *not* a follow a 
set recipe sort of project. 

Lots of fun !!!

Bob

> On Jan 26, 2016, at 7:04 PM, time...@metachaos.net wrote:
> 
> Hi,
> 
> I am a newbie to this list. I have downloaded the archives and read about
> 5,000 of the past messages. I plan on building my own GPSDO, probably using a
> LEA-6T (but LEA-7T or LEA-M8T would be good if I can find one affordably). I
> have a MTI 260 on order (although it could wind up being a 261 since they all
> appear to ship one or the other randomly).
> 
> Currently, my resources include a DMM (well, a couple) and soldering /
> desoldering stations and quite a few tools. I also have an oscilloscope that I
> am currently repairing - a 400Mhz Tektronix 2465BCT analog scope. I am waiting
> on the final parts from Mouser. Once that is done I need to get it calibrated.
> All of that will probably take me another month. I also need to finish fixing
> my cassette deck - and then to finish writing a special recording program to
> use raw device drivers to get around the fact that Windows is not real time. I
> interrupted that project to work on the scope.
> 
> In the meantime, I am reading the time-nuts messages (and lots of other
> things) to gather information and ideas about how I am going to do this and
> generally to learn more.
> 
> So, I have some questions. Let me tell you a bit about me, so that you know
> the context and my limitations. I am a retired programmer. I wrote just about
> everything including device drivers, operating systems, utilities, various AI
> programs, telephone systems, compilers, encryption, web applications and much
> more. If I need to throw 50,000 LOC at a project, no problem. I have used many
> languages including quite a few different assembly languages (I have also
> written an assembler). I consider myself a mathematician / programmer,
> although I haven't really needed Calculus or Differential Equations for
> decades, so I am pretty rusty in that area. I do more work in formal logic
> than higher mathematics. But, I THINK like a mathematician. Formalism and
> abstraction come naturally to me.
> 
> During my career I also helped to debug hardware during S-100 days. I have
> sporadically messed with electronics off and on, informally, with no education
> in the area. Now that I am retired (and have more time, but less money - it IS
> a zero sum game!), I am trying to learn more about electronics and start doing
> hardware projects. I have never been into model building or anything similar,
> so my construction skills are lacking. I understand a lot of things in theory,
> but practice still eludes me. For example, knowing a part exists or
> determining which of 10,000 apparently identical parts is the "right" choice.
> It can hours or even days to find the "right" connector. In many cases, the
> names or descriptions are completely meaningless. That all appears to be an
> experience related issue, so I will (hopefully) overcome that in time.
> 
> I have no problem with soldering / desoldering, but I haven't designed or
> built my own PCB yet. I have designed / redesigned some

Re: [time-nuts] Newbie questions

2016-01-27 Thread Jim Harman

Another benefit of using dB vs Watts or Volts is that systems often consist
of a chain of elements with gains and losses. Working with gains and losses
in dB lets you calculate the signal level at any point along the way and
the system gain by adding and subtracting rather than multiplying and
dividing.

On Wed, Jan 27, 2016 at 3:02 AM, wb6bnq  wrote:

> Hi Mike,
>
> The element that you are missing is the impedance.  When you look at the
> common formula it refers to a ratio of power or voltage and the impedance
> is left out with the understanding that the impedance is equal for each
> power or voltage in the ratio.  The actual formula (for power) is DB = 10 X
> log10 ( E1^2 / R ) / ( E2^2 / R ).  Hint   Power P = E^2 / R.
>
> In the RF world that impedance is 50 Ohms and ZERO DBm(illiwatts) is ONE
> milliwatt into 50 Ohms which is 0.223606797749979 Vrms.
>
> In the audio world the reference impedance is 600 Ohms.  So ZERO DBu is
> One milliwatt into 600 Ohms which is 0.7745966692414834 Vrms.  More
> commonly referred to as 0.775 Vrms.
>
> As for your GPS questions, I will leave that up to others to answer.
>
> BillWB6BNQ
>
>
> time...@metachaos.net wrote:
>
> Hi,
>>
>> I am a newbie to this list. I have downloaded the archives and read about
>> 5,000 of the past messages. I plan on building my own GPSDO, probably
>> using a
>> LEA-6T (but LEA-7T or LEA-M8T would be good if I can find one
>> affordably). I
>> have a MTI 260 on order (although it could wind up being a 261 since they
>> all
>> appear to ship one or the other randomly).
>>
>> Currently, my resources include a DMM (well, a couple) and soldering /
>> desoldering stations and quite a few tools. I also have an oscilloscope
>> that I
>> am currently repairing - a 400Mhz Tektronix 2465BCT analog scope. I am
>> waiting
>> on the final parts from Mouser. Once that is done I need to get it
>> calibrated.
>> All of that will probably take me another month. I also need to finish
>> fixing
>> my cassette deck - and then to finish writing a special recording program
>> to
>> use raw device drivers to get around the fact that Windows is not real
>> time. I
>> interrupted that project to work on the scope.
>>
>> In the meantime, I am reading the time-nuts messages (and lots of other
>> things) to gather information and ideas about how I am going to do this
>> and
>> generally to learn more.
>>
>> So, I have some questions. Let me tell you a bit about me, so that you
>> know
>> the context and my limitations. I am a retired programmer. I wrote just
>> about
>> everything including device drivers, operating systems, utilities,
>> various AI
>> programs, telephone systems, compilers, encryption, web applications and
>> much
>> more. If I need to throw 50,000 LOC at a project, no problem. I have used
>> many
>> languages including quite a few different assembly languages (I have also
>> written an assembler). I consider myself a mathematician / programmer,
>> although I haven't really needed Calculus or Differential Equations for
>> decades, so I am pretty rusty in that area. I do more work in formal logic
>> than higher mathematics. But, I THINK like a mathematician. Formalism and
>> abstraction come naturally to me.
>>
>> During my career I also helped to debug hardware during S-100 days. I have
>> sporadically messed with electronics off and on, informally, with no
>> education
>> in the area. Now that I am retired (and have more time, but less money -
>> it IS
>> a zero sum game!), I am trying to learn more about electronics and start
>> doing
>> hardware projects. I have never been into model building or anything
>> similar,
>> so my construction skills are lacking. I understand a lot of things in
>> theory,
>> but practice still eludes me. For example, knowing a part exists or
>> determining which of 10,000 apparently identical parts is the "right"
>> choice.
>> It can hours or even days to find the "right" connector. In many cases,
>> the
>> names or descriptions are completely meaningless. That all appears to be
>> an
>> experience related issue, so I will (hopefully) overcome that in time.
>>
>> I have no problem with soldering / desoldering, but I haven't designed or
>> built my own PCB yet. I have designed / redesigned some minor circuits,
>> especially on the power supply side. I can follow schematics reasonably
>> well,
>> but I am not comfortable with Eagle or other PCB layout programs. Every
>> time
>> I have tried one of those programs, half of the parts I needed were not
>> available. I have started using TinyCAD which is much easier to use. So, I
>> have a lot to learn. But, that is basically what I do, all day, every
>> day. I'm
>> the type of person that gets bored easily and quickly. As #5 said "more
>> input,
>> more input"!   6.02059991327962
>>
>> Paradoxically, I have no interest in time. As in time of day, day of week,
>> etc.. I have never had a job where I got to work on

[time-nuts] Newbie questions

2016-01-26 Thread timenut

Hi,

I am a newbie to this list. I have downloaded the archives and read about
5,000 of the past messages. I plan on building my own GPSDO, probably using a
LEA-6T (but LEA-7T or LEA-M8T would be good if I can find one affordably). I
have a MTI 260 on order (although it could wind up being a 261 since they all
appear to ship one or the other randomly).

Currently, my resources include a DMM (well, a couple) and soldering /
desoldering stations and quite a few tools. I also have an oscilloscope that I
am currently repairing - a 400Mhz Tektronix 2465BCT analog scope. I am waiting
on the final parts from Mouser. Once that is done I need to get it calibrated.
All of that will probably take me another month. I also need to finish fixing
my cassette deck - and then to finish writing a special recording program to
use raw device drivers to get around the fact that Windows is not real time. I
interrupted that project to work on the scope.

In the meantime, I am reading the time-nuts messages (and lots of other
things) to gather information and ideas about how I am going to do this and
generally to learn more.

So, I have some questions. Let me tell you a bit about me, so that you know
the context and my limitations. I am a retired programmer. I wrote just about
everything including device drivers, operating systems, utilities, various AI
programs, telephone systems, compilers, encryption, web applications and much
more. If I need to throw 50,000 LOC at a project, no problem. I have used many
languages including quite a few different assembly languages (I have also
written an assembler). I consider myself a mathematician / programmer,
although I haven't really needed Calculus or Differential Equations for
decades, so I am pretty rusty in that area. I do more work in formal logic
than higher mathematics. But, I THINK like a mathematician. Formalism and
abstraction come naturally to me.

During my career I also helped to debug hardware during S-100 days. I have
sporadically messed with electronics off and on, informally, with no education
in the area. Now that I am retired (and have more time, but less money - it IS
a zero sum game!), I am trying to learn more about electronics and start doing
hardware projects. I have never been into model building or anything similar,
so my construction skills are lacking. I understand a lot of things in theory,
but practice still eludes me. For example, knowing a part exists or
determining which of 10,000 apparently identical parts is the "right" choice.
It can hours or even days to find the "right" connector. In many cases, the
names or descriptions are completely meaningless. That all appears to be an
experience related issue, so I will (hopefully) overcome that in time.

I have no problem with soldering / desoldering, but I haven't designed or
built my own PCB yet. I have designed / redesigned some minor circuits,
especially on the power supply side. I can follow schematics reasonably well,
but I am not comfortable with Eagle or other PCB layout programs. Every time
I have tried one of those programs, half of the parts I needed were not
available. I have started using TinyCAD which is much easier to use. So, I
have a lot to learn. But, that is basically what I do, all day, every day. I'm
the type of person that gets bored easily and quickly. As #5 said "more input,
more input"!   6.02059991327962

Paradoxically, I have no interest in time. As in time of day, day of week,
etc.. I have never had a job where I got to work on time. My philosophy has
always been "go to bed when sleepy, get up when not". I was notorious in high
school for only showing up on test day. But, I am interested in being able to
timestamp events accurately and in measuring time (and other things). I am
also interested in how a very accurate frequency source can be used in
other applications and test instruments. That brings me to my desire to build
a GPSDO and my questions.

I understand the logarithmic scaling used for voltage and power. I even
understand why voltage uses a multiplier of 20 and power a multiplier of 10.
It makes sense when working with a wide range of values. However, my DMM, my
scope and generally schematics work directly with current, voltage and watts.
So, I am constantly seeing statements like an output is 7 dBm or 13 dBm. If I
knew the actual value for 0 dBm then the basic equations would resolve the
values. However, I have not found a consistent answer for that. When I have
attempted to work values backwards from various statements, again I don't get
a consistent value (probably because those statements were approximations and
not exact values). I always see statements that an increase of 6dBm doubles
the value. It is used so often that most people forget it is an approximation.
It is 6.02059991... and sometimes, it may make a difference. Worse, the zero
value appears to be different for different applications. In some it appears
to be

Re: [time-nuts] Newbie questions

2010-01-06 Thread Bruce Griffiths


But the result disagrees with the manufacturer's specs for RG59.
There's an error somewhere in your measurement setup.
For example, it could be a time offset error (or even differing trigger 
levels with a sinewave input) between the time interval counter start 
and stop channels.


Bruce

Tom Duckworth wrote:

Magnus,

We've made this measurement using a 20 ps time interval counter and a 
GPS disciplined Rubidium frequency standard as the time base; making 
many concurrent measurements with no dead time between. The resultant 
measurement was very close to the 1 ns/ft benchmark with RG-59 (BNC 
connectors), 10 MHz source. So we felt ok with using the 1 ns/ft 
estimate.


Tom
Tom Duckworth
tomd...@comcast.net
- Original Message - From: Magnus Danielson 
mag...@rubidium.dyndns.org
To: Discussion of precise time and frequency measurement 
time-nuts@febo.com

Sent: Tuesday, January 05, 2010 7:26 PM
Subject: Re: [time-nuts] Newbie questions



Tom Duckworth wrote:

Jim,

We use a benchmark 1 ns per foot of coax (RG-59).


This sounds fast. The normal taxiometer is at 66% of speed of ligth 
in vaccum, which for 1 ns is about 3 dm so for the RG-59 that would 
be about 2 dm.


Some cables reach 78%, but RG-58 and RG-59 is down at normal 66%.

You could measure the delay by using a resistive splitter (50 ohms) 
and two cables (say a 2 foot and a three foot, each terminated at 
the far end with a 50 ohm pass through terminator). Drive the 
splitter with your 10 MHz signal and measure, at the far end, using 
an appropriate 2-channel scope or counter with the necessary 
resolution, the difference in time delay between the two, which will 
give you a pretty accurate delay per foot. Both cables should be the 
same coax type.


Being a time-nut, using time-interval counters or TDR would be my 
choice, but these tools/toys outnumbers the scopes...


Cheers,
Magnus

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to 
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts

and follow the instructions there.




___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to 
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts

and follow the instructions there.





___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

2010-01-06 Thread David C. Partridge

Except the 1nS/ft figure is a good approximation to C (speed of light in
vacuo).  In RG58, you would expect to see 0.66C or about that.

So 1nS for 8 of cable is a good ROT.   Or put the other way round, about
1.5nS per foot.

Dave 

-Original Message-
From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On
Behalf Of Tom Duckworth
Sent: 06 January 2010 02:23
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Newbie questions

Jim,

We use a benchmark 1 ns per foot of coax (RG-59).

You could measure the delay by using a resistive splitter (50 ohms) and two
cables (say a 2 foot and a three foot, each terminated at the far end with a
50 ohm pass through terminator). Drive the splitter with your 10 MHz signal
and measure, at the far end, using an appropriate 2-channel scope or counter
with the necessary resolution, the difference in time delay between the two,
which will give you a pretty accurate delay per foot. Both cables should be
the same coax type.

Tom
Tom Duckworth
tomd...@comcast.net
- Original Message -
From: Jim Mandaville z...@dakotacom.net
To: Discussion of precise time and frequency measurement 
time-nuts@febo.com
Sent: Tuesday, January 05, 2010 11:17 AM
Subject: [time-nuts] Newbie questions


I am new to the list (although lurking now a while) and also new to the 
more precise species of  frequency and time measurement.  I have recently 
powered up an LPro and a Thunderbolt, both of which appear to be working by

the book.  Connecting the TBolt to my scope external sync and the LPro as 
an unknown I see the pattern moving one division (cm) to the left in 295 
seconds with a 0.05 us-per-division setting on the scope (the fastest 
setting available).  This, if my newly-learned calculations are correct, 
indicates a difference of 1.7 X10-10 (0.0017 Hz).  This appears to be 
confirmed by my HP 5335A counter, which shows the LPRO 1or 2 thousandths of

a Hz low, using the TBolt as an external time source. An HP manual I have 
indicates that a low unknown pattern should be moving to the right, not the

left, on the scope, so this sort of puzzles me.

 I have a few questions that I'm hoping some of you more experienced hands 
 can help with:

 1.  Can someone tell me the meaning and significance of the Timing 
 Outputs numbers in the lower left corner of the TBolt monitor window? 
 (Mine right now is showing plus 3.75 ns and plus 0.01 ppb). The TBolt 
 manual does not describe these, although on one page it lists them as 
 estimates of UTC/GPS offsets.  Do these numbers show the difference 
 between my receiver outputs and the time being kept by my present 
 satellites?  Or is it the difference between my receiver outputs and 
 master gps time (somewhere)?  Neither of these?  The use of two decimal 
 places on nanoseconds implies great accuracy.  Is this obtained in 
 practice?  My ppb on 10 MHz usually lies between plus 0.1 and minus 0.1, 
 often hanging around 0.01 or 0.02.  I have not so far put in any 
 compensation for cable delay.

 If the TBolt knows what these differences are, why doesn't it just 
 factor them into its outputs?  Or does it?

 2.  What is a reasonable expectation of TBolt accuracy (at any given time 
 that I use it for measurement) for the 10 MHz relative to NIS? How 
 accurate would it be, say, 90 percent of the time? (Looking for just an 
 experienced guesstimate here).

 3.  What format do I use to put in pps nanoseconds compensation for cable 
 delay (I use about 19 feet of RG-58U).  I understand this should be a 
 negative number.

 4.  Does anyone know a way to force the 5335A counter to display another 
 decimal place in frequency measurements?  I am getting to 0.001 Hz by 
 using the mean of 100 counts function on the counter, but I think the 
 counter has at least one more digit available which I would like to use 
 when accuracy justifies it (e.g. when using the TBolt as an external time 
 source).

 Any comments and suggestions appreciated
 Jim, KF7A


 ___
 time-nuts mailing list -- time-nuts@febo.com
 To unsubscribe, go to 
 https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
 and follow the instructions there.
 


___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.


___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

Hi

The 5335 is a pretty nice counter, you can use the math functions to get just 
about anything you want displayed. One simple example is to just subtract 10 
MHz from the count and display the difference. 

The problem you run into is overflow in the counter chip. Somewhere around 100 
seconds you run out of storage with a 10 MHz input. 

It's been *years* since I ran one, I could be off on the 100 seconds 

Most of the ones you see for sale have at at least one input channel blown 
out.. They are easy to fry and the front end chips aren't anything you can get 
off the shelf. 

Bob


On Jan 6, 2010, at 2:55 AM, Hal Murray wrote:

 
 z...@dakotacom.net said:
 4.  Does anyone know a way to force the 5335A counter to display
 another decimal place in frequency measurements?  I am getting to
 0.001 Hz by using the mean of 100 counts function on the counter,
 but I think the counter has at least one more digit available which I
 would like to use when accuracy justifies it (e.g. when using the
 TBolt as an external time source). 
 
 I've never used a 5335 but I have a 5334.
 
 How many digits are there on the display?  How many of them are you getting?
 
 With the 5334, you can get a few more digits by running for 100 seconds but 
 you can only see them via GPIB.
 
 In case you haven't noticed, many people on this list are happy with the 
 Prologix GPIB/USB gadget.
 
 
 -- 
 These are my opinions, not necessarily my employer's.  I hate spam.
 
 
 
 
 ___
 time-nuts mailing list -- time-nuts@febo.com
 To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
 and follow the instructions there.
 


___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions


Tom Duckworth wrote:

Magnus,

We've made this measurement using a 20 ps time interval counter and a 
GPS disciplined Rubidium frequency standard as the time base; making 
many concurrent measurements with no dead time between. The resultant 
measurement was very close to the 1 ns/ft benchmark with RG-59 (BNC 
connectors), 10 MHz source. So we felt ok with using the 1 ns/ft estimate.


Well, I think you should reconsider. Theory says that

v = 1/sqrt(my*epsilon) = 1/sqrt(my_r * epsilon_r) * c

For a coaxial cable, we have the magnetic properties about the same as 
vacuum, so my_r = 1 is a fair approximation. RG-58 and RG-59 use solid 
polyethylene having epsilon_r = 2,25 and that cranks out as v = 0,66*c,

in agreement with tabulated values:

http://hyperphysics.phy-astr.gsu.edu/Hbase/tables/diel.html
http://www.epanorama.net/documents/wiring/coaxcable.html

Notice how foam-PE rates at 0,78*c rather than 0,66*c. This is due to 
the lower dielectric constant (about 1,64) of foam PE.


I think it this relationship is wellfounded. I use 2 dm/ns for coax and 
fiber myself for my reality check calculations and found good 
correlation with reality whenever I tried it.


Oh, did you use sine as waveform? A few different frequencies and/or 
amplitude would be good to ensure that biases could be canceled out. I 
do that for reality check myself. Swapping the cables is another.


You have to excuse me, but the value you have measured does not really 
correlate well with my experience or view of theory.


Cheers,
Magnus

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

Bob Camp wrote:

The 5335 is a pretty nice counter, you can use the math functions to get just about anything you want displayed. One simple example is to just subtract 10 MHz from the count and display the difference.

Indeed. The 5335 can display 12 digits but does not overflow the display
as some other counters but display the 12 most significant digits. The
math function allows you to remove the most significant digits in order
to display more digits. Good math-functions is necessary for good
counters, I regularly use them. Correctly used you may often crank out
the numbers you are interested in fairly quickly.

The problem you run into is overflow in the counter chip. Somewhere around 100 seconds you run out of storage with a 10 MHz input.

The Gate Time is specified from 100 ns to 10 Ms. I think the CPU handles
the overflow flag in the MRC and clears it.

The Gate Time can be set up to 4 seconds on the front (I really love the
directness of the Gate Adjust knob, miss that in many other counters)
and about 1 second using GPIB GA command. There are three ways to get
longer gate time, Manual Trigg on the front panel, External Arm input at
the back and using the GO (Gate Open) and GC (Gate Close) GPIB commands.

The single shot resolution is about 1 ns, each of the interpolators is
good for about 500 ps, but since they always combine the result is 1 ns.

While increased time-base will give increased frequency resolution, the
stability of the reference time-base becomes an issue. To measure
stability the way we like (Allan Deviation and friends) they vary with
the gate-time (which we call tau in time-nut-speech), so a counters
resolution can be expressed as resolution divided by tau as a single
figure of merit (which is an oversimplification). Another figure of
merit is the single-shot resolution, the time resolution for a single
trigger event. There is actually the hardware resolution and that which
includes trigger-jitter (which is more usefull). The SR-620 is at about
25 ps, but the hardware counting is in units of 4 ps. These numbers
correlate to some degree with the Allan Deviation of the instrument for
shorter taus, but as taus is allowed to increase various limits kicks in.

It's been *years* since I ran one, I could be off on the 100 seconds

Most of the ones you see for sale have at at least one input channel blown out.. They are easy to fry and the front end chips aren't anything you can get off the shelf.

It's a nice bench-counter thought and is fairly flexible.

The HP5334A is the economy version of the HP5335A but has an additional
feature which is usefull for time-nuts... the binary output dumps the
unprocessed MRC values and can do that in a steady stream, allowing for
time-stamp records to be recorded. Together with picket fence it allows
a higher rate recording than the HP5335A.

Hmm, I have a HP5335A to repair. Probably the PSU.

Cheers,
Magnus

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

Hi

Ahhh, that jogs a few of the tired old brain cells

You can run the gate off of the GPIB input to get all kinds of silly long
gates. Then you start to get into the cpu overflow issues. Since it's a MSB
overflow you can usually clean it up in software *if* you know it's
happening. 

Since the GPIB is simply telling the counter start about now and finish
it up now the accuracy of the GPIB timing does not get into the result. The
counter still uses it's time base as the standard of comparison.

Bob

-Original Message-
From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On
Behalf Of Magnus Danielson
Sent: Wednesday, January 06, 2010 9:28 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Newbie questions

Bob Camp wrote:
 Hi
 
 The 5335 is a pretty nice counter, you can use the math functions to get
just about anything you want displayed. One simple example is to just
subtract 10 MHz from the count and display the difference. 

Indeed. The 5335 can display 12 digits but does not overflow the display 
as some other counters but display the 12 most significant digits. The 
math function allows you to remove the most significant digits in order 
to display more digits. Good math-functions is necessary for good 
counters, I regularly use them. Correctly used you may often crank out 
the numbers you are interested in fairly quickly.

 The problem you run into is overflow in the counter chip. Somewhere around
100 seconds you run out of storage with a 10 MHz input. 

The Gate Time is specified from 100 ns to 10 Ms. I think the CPU handles 
the overflow flag in the MRC and clears it.

The Gate Time can be set up to 4 seconds on the front (I really love the 
directness of the Gate Adjust knob, miss that in many other counters) 
and about 1 second using GPIB GA command. There are three ways to get 
longer gate time, Manual Trigg on the front panel, External Arm input at 
the back and using the GO (Gate Open) and GC (Gate Close) GPIB commands.

The single shot resolution is about 1 ns, each of the interpolators is 
good for about 500 ps, but since they always combine the result is 1 ns.

While increased time-base will give increased frequency resolution, the 
stability of the reference time-base becomes an issue. To measure 
stability the way we like (Allan Deviation and friends) they vary with 
the gate-time (which we call tau in time-nut-speech), so a counters 
resolution can be expressed as resolution divided by tau as a single 
figure of merit (which is an oversimplification). Another figure of 
merit is the single-shot resolution, the time resolution for a single 
trigger event. There is actually the hardware resolution and that which 
includes trigger-jitter (which is more usefull). The SR-620 is at about 
25 ps, but the hardware counting is in units of 4 ps. These numbers 
correlate to some degree with the Allan Deviation of the instrument for 
shorter taus, but as taus is allowed to increase various limits kicks in.

 It's been *years* since I ran one, I could be off on the 100 seconds 
 
 Most of the ones you see for sale have at at least one input channel blown
out.. They are easy to fry and the front end chips aren't anything you can
get off the shelf. 

It's a nice bench-counter thought and is fairly flexible.

The HP5334A is the economy version of the HP5335A but has an additional 
feature which is usefull for time-nuts... the binary output dumps the 
unprocessed MRC values and can do that in a steady stream, allowing for 
time-stamp records to be recorded. Together with picket fence it allows 
a higher rate recording than the HP5335A.

Hmm, I have a HP5335A to repair. Probably the PSU.

Cheers,
Magnus

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.




___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

Hi

The thing that always gets me is when I pick up a piece of air dielectric
coax (like a line stretcher). I have to recalibrate all over again

Bob

-Original Message-
From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On
Behalf Of Magnus Danielson
Sent: Wednesday, January 06, 2010 8:17 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Newbie questions

Tom Duckworth wrote:
 Magnus,
 
 We've made this measurement using a 20 ps time interval counter and a 
 GPS disciplined Rubidium frequency standard as the time base; making 
 many concurrent measurements with no dead time between. The resultant 
 measurement was very close to the 1 ns/ft benchmark with RG-59 (BNC 
 connectors), 10 MHz source. So we felt ok with using the 1 ns/ft estimate.

Well, I think you should reconsider. Theory says that

v = 1/sqrt(my*epsilon) = 1/sqrt(my_r * epsilon_r) * c

For a coaxial cable, we have the magnetic properties about the same as 
vacuum, so my_r = 1 is a fair approximation. RG-58 and RG-59 use solid 
polyethylene having epsilon_r = 2,25 and that cranks out as v = 0,66*c,
in agreement with tabulated values:

http://hyperphysics.phy-astr.gsu.edu/Hbase/tables/diel.html
http://www.epanorama.net/documents/wiring/coaxcable.html

Notice how foam-PE rates at 0,78*c rather than 0,66*c. This is due to 
the lower dielectric constant (about 1,64) of foam PE.

I think it this relationship is wellfounded. I use 2 dm/ns for coax and 
fiber myself for my reality check calculations and found good 
correlation with reality whenever I tried it.

Oh, did you use sine as waveform? A few different frequencies and/or 
amplitude would be good to ensure that biases could be canceled out. I 
do that for reality check myself. Swapping the cables is another.

You have to excuse me, but the value you have measured does not really 
correlate well with my experience or view of theory.

Cheers,
Magnus

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.




___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions


Hi!

Bob Camp wrote:

Hi

Ahhh, that jogs a few of the tired old brain cells

You can run the gate off of the GPIB input to get all kinds of silly long
gates. Then you start to get into the cpu overflow issues. Since it's a MSB
overflow you can usually clean it up in software *if* you know it's
happening. 


Notice that it is specified to support gate-times up to 10 Ms (10^7 s). 
I would suspect that both the event and time counters is CPU extended.


Maybe I should pull the PROM and do a readout.


Since the GPIB is simply telling the counter start about now and finish
it up now the accuracy of the GPIB timing does not get into the result. The
counter still uses it's time base as the standard of comparison.


Well, it will cause a variation in gate-time (about 2 ms can be assumed, 
since that is what programmed gate using GA allows) , but since this is 
used for longer gate times it is not significant. If better precission 
is needed, the external arm input shall be used.


Cheers,
Magnus

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

Hi

I agree with you on the spec sheet and the fact that the cpu *should* extend
the data. My observation was that it didn't do it if the input frequency was
high enough. I tried it on a couple dozen counters built over a period of
several years.

Bob

-Original Message-
From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On
Behalf Of Magnus Danielson
Sent: Wednesday, January 06, 2010 12:10 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Newbie questions

Hi!

Bob Camp wrote:
 Hi
 
 Ahhh, that jogs a few of the tired old brain cells
 
 You can run the gate off of the GPIB input to get all kinds of silly long
 gates. Then you start to get into the cpu overflow issues. Since it's a
MSB
 overflow you can usually clean it up in software *if* you know it's
 happening. 

Notice that it is specified to support gate-times up to 10 Ms (10^7 s). 
I would suspect that both the event and time counters is CPU extended.

Maybe I should pull the PROM and do a readout.

 Since the GPIB is simply telling the counter start about now and finish
 it up now the accuracy of the GPIB timing does not get into the result.
The
 counter still uses it's time base as the standard of comparison.

Well, it will cause a variation in gate-time (about 2 ms can be assumed, 
since that is what programmed gate using GA allows) , but since this is 
used for longer gate times it is not significant. If better precission 
is needed, the external arm input shall be used.

Cheers,
Magnus

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.




___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions


Bob Camp wrote:

Hi

I agree with you on the spec sheet and the fact that the cpu *should* extend
the data. My observation was that it didn't do it if the input frequency was
high enough. I tried it on a couple dozen counters built over a period of
several years.


That would imply that only the time counter is CPU extended, but not the 
event counter. This is kind of fair, since it allows the lower 
frequencies to gather equalent amount of events as higher frequencies 
and thus similar precission achieved.


Should be a simple exercise to establish the event overflow properties.

I have not seen any detailed description of the MRC chip, but I know it 
is being used in HP5315A, HP5316A, HP5334A and HP5335A. The later two 
are bigger counters which includes the interpolators for 1 ns 
resolution versus the 100 ns time resolution of the simpler units when 
using the 10 MHz directly. The MRC chip tolerates 100 MHz directly, so 
for the 200 MHz (1/2) and 1,3 GHz (1/20) responses prescalers is used.


Thus, the time counter ticks at 10 MHz while the event counter ticks at 
up to 100 MHz for normal frequency measurement. For frequency 
comparision, the time counter is being used for the B channel, so it can 
tick at 100 MHz then.


Cheers,
Magnus

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

Hi

I have a pile of 5334's in the shed. I'm getting the itch to pull them out
and see what they can do. I've always looked down on them a bit, since
5335's were always available for what I needed to do.

Bob

-Original Message-
From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On
Behalf Of Magnus Danielson
Sent: Wednesday, January 06, 2010 12:41 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Newbie questions

Bob Camp wrote:
 Hi
 
 I agree with you on the spec sheet and the fact that the cpu *should*
extend
 the data. My observation was that it didn't do it if the input frequency
was
 high enough. I tried it on a couple dozen counters built over a period of
 several years.

That would imply that only the time counter is CPU extended, but not the 
event counter. This is kind of fair, since it allows the lower 
frequencies to gather equalent amount of events as higher frequencies 
and thus similar precission achieved.

Should be a simple exercise to establish the event overflow properties.

I have not seen any detailed description of the MRC chip, but I know it 
is being used in HP5315A, HP5316A, HP5334A and HP5335A. The later two 
are bigger counters which includes the interpolators for 1 ns 
resolution versus the 100 ns time resolution of the simpler units when 
using the 10 MHz directly. The MRC chip tolerates 100 MHz directly, so 
for the 200 MHz (1/2) and 1,3 GHz (1/20) responses prescalers is used.

Thus, the time counter ticks at 10 MHz while the event counter ticks at 
up to 100 MHz for normal frequency measurement. For frequency 
comparision, the time counter is being used for the B channel, so it can 
tick at 100 MHz then.

Cheers,
Magnus

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.




___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

2010-01-06 Thread SAIDJACK

On the 5334, you can get 11 digits resolution on the LED display  by 
subtracting 10MHz from the measurement (-10MHz offset), and setting the  
mesurement intervall to 99 seconds. Both sources need to be 10MHz of course for 
 this 
to work.
 
On the 5335 I seem to get one digit less with the same offset procedure, I  
could never get it to display 1E-011 digits..
 
bye,
Said
 
 
In a message dated 1/5/2010 23:55:56 Pacific Standard Time,  
hmur...@megapathdsl.net writes:


I've  never used a 5335 but I have a 5334.

How many digits are there on the  display?  How many of them are you 
getting?

With the 5334, you can  get a few more digits by running for 100 seconds 
but 
you can only see them  via GPIB.

In case you haven't noticed, many people on this list are  happy with the 
Prologix GPIB/USB  gadget.


___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions

2010-01-06 Thread John Ackermann N8UR


Bob Camp wrote:

Hi

I have a pile of 5334's in the shed. I'm getting the itch to pull them out
and see what they can do. I've always looked down on them a bit, since
5335's were always available for what I needed to do.


I've always liked the 5334s for both general bench use, and as TICs 
where 2ns resolution will do.  One advantage is that they don't have a 
fan to make yet more noise in the lab.


John

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Newbie questions