Author Topic: Economical option for precision frequency reference?  (Read 155275 times)

0 Members and 1 Guest are viewing this topic.

Offline N8AUM

  • Regular Contributor
  • *
  • Posts: 132
  • Country: us
Re: Economical option for precision frequency reference?
« Reply #125 on: July 29, 2015, 07:30:51 am »
The 15Mhz  doesnt look too bad from what I can see and I didnt realise the output was so hefty! Used 50db of attenuation when I made this sweep. I  think I will buffer the 5Mhz osc and mix them together ? 
Wish I could make some phase noise measurements but my HP 8591E isn't much better  :(
As far as distribution goes, rather than building one I might try a video distribution amp.
 

Offline N8AUM

  • Regular Contributor
  • *
  • Posts: 132
  • Country: us
Re: Economical option for precision frequency reference?
« Reply #126 on: July 29, 2015, 07:43:52 am »
Glad you got it all working. I made a proper DB9 cable for mine today, so no more terminal strip with wires hanging off it :)

I need to figure out how the DB9 power connection locks in. I read somewhere that the little pins (instead of screws) are called "slidelock". Have to see if I can get a connector with those.
[/quote]

Why not just replace those with regular hardware ?
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #127 on: July 29, 2015, 11:07:06 am »
I need to figure out how the DB9 power connection locks in. I read somewhere that the little pins (instead of screws) are called "slidelock". Have to see if I can get a connector with those.

Why not just replace those with regular hardware ?

Yeah, well, because then it's not in "pristine" condition :)
 

Offline SoundTech-LG

  • Frequent Contributor
  • **
  • Posts: 788
  • Country: us
Re: Economical option for precision frequency reference?
« Reply #128 on: July 29, 2015, 03:28:14 pm »
I'm assuming most of these posts are referring to the

LUCENT/SYMMETRICOM/HP Z3810AS, KS24361 L101 REF0 / L102 REF1, GPSDO TIMING SYSTEM

as shown. Maybe the thread name can reflect that?

Great stuff by the way...
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #129 on: July 29, 2015, 03:35:27 pm »
I'm assuming most of these posts are referring to the

LUCENT/SYMMETRICOM/HP Z3810AS, KS24361 L101 REF0 / L102 REF1, GPSDO TIMING SYSTEM

as shown. Maybe the thread name can reflect that?

Great stuff by the way...

Yes, most, but there are plenty of posts about other options. I didn't want to change the thread title because When I have done that in the past, people end up dropping off the thread because they can't find it.
 

Offline edpalmer42

  • Super Contributor
  • ***
  • Posts: 2260
  • Country: ca
Re: Economical option for precision frequency reference?
« Reply #130 on: July 30, 2015, 06:23:50 am »
evb, you ask a lot of questions!  But they're good questions, so here goes.

(a)  Common failure modes.  None that I've heard of.  Since these things are NOS, there might be some infant mortality, but I've heard that the seller is quite good about replacements.

(b)  Delicate equipment?  Hell, no!  They were made by HP for Lucent.  It doesn't get much better than that.  I have an older member of the family.  I've had it for a few years and it's been monitored continuously since I got it.  I suspect that most of the units in service with hobbyists are the same.  There was a question about whether the Z38xx program had a subtle bug that could wear out the EEPROM in the Z3801A and, by extension, other members of the family.  The author patched the program to remove that danger.

(c)  Dying for no reason.  No, that was a somewhat off-topic comment about a totally unrelated unit (Rockwell Jupiter GPS receiver).

(d)  Data collection.  Start with the program called Z38xx written by the late Ulrich Bangert.  Although it needs a hack to recognize these units, it's a minor thing.  If that program isn't to your taste, roll your own.  It's just a serial port.

(e)  Antennas.  Get a timing grade antenna.  Typically, they're bullet shaped rather than a puck.  The timing grade antennas have better filtering and typically higher gain than other types.  Make sure it will tolerate 5V.  Newer ones might be 3V3 only.  More definitive recommendations are difficult because local conditions like cable length and local signal obstructions will have an effect.  Earlier in the thread motocoder and I posted graphs and info on our setups.  That's as good a guide as any.

(f) and (g)  Initial vs. burned-in performance.  That's not clear.  I'm beginning to wonder just what their ultimate performance will be.  As we speak, I'm playing with yet another GPSDO from Trimble.  The performance is quite similar to the Lucent boxes we're discussing here.  But when I removed the OCXO from the board and measured it seperately, its short-term performance was much better than its short-term performance while on the board.  It seems that the board is reducing the OCXOs short-term performance while improving its long term performance.  There's no way to tell if that was intentional or irrelevent to the designers.  So until we hear from someone who has had these units online for a few months - and there should be some since they first came on the market late last year - we won't know what the ultimate performance will be.

But remember, I'm coming at it from the point of view of a slightly obsessed, PITA, time-nut who bitches and whines about a GPS receiver that has a 10 MHz output that's off by 0.00015 Hz!  For sane, normal people these units will perform perfectly well out of the box.

(h)  Run one unit only?  There was discussion on that, but since I got these as spares, I haven't investigated.  You should be able to find the info on time-nuts.

(i)  Reaction to moving a unit?  Anytime you move the antenna, you introduce some error into the system.  The error can show up as more jitter in the output or as jumps when a satellite joins or leaves the list of tracked birds.  Whether the error is significant will depend on your application.  In general, when you move the antenna, you should rerun the position survey.  I think there's a command for that, but I'm not sure.

 :phew:

Ed
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #131 on: July 30, 2015, 07:56:03 am »
(h)  Run one unit only?  There was discussion on that, but since I got these as spares, I haven't investigated.  You should be able to find the info on time-nuts.

I found one person on time-nuts that described how this was done. It involved building a small circuit to toggle the values on inputs on J5 in response to certain outputs. His approach involved wiring into the signal on the Fault LED, so it wasn't an external-only solution.

(i)  Reaction to moving a unit?  Anytime you move the antenna, you introduce some error into the system.  The error can show up as more jitter in the output or as jumps when a satellite joins or leaves the list of tracked birds.  Whether the error is significant will depend on your application.  In general, when you move the antenna, you should rerun the position survey.  I think there's a command for that, but I'm not sure.

I believe you can start a survey with this command:

:PTIMe:GPSystem:POSition:SURVey:STATe ONCE

By default, the units are set up to do a survey on start-up. You can manually trigger a reset/start-up sequence with this command:

:SYSTem:PON

If you don't like the survey on start behavior, you can disable that with this command:

:GPSystem:POSition:SURVey:STATe:POWerup 0

More commands can be found here:
http://do-nyc.bodosom.net/
 

Offline edpalmer42

  • Super Contributor
  • ***
  • Posts: 2260
  • Country: ca
Re: Economical option for precision frequency reference?
« Reply #132 on: July 30, 2015, 01:11:31 pm »
(h)  Run one unit only?  There was discussion on that, but since I got these as spares, I haven't investigated.  You should be able to find the info on time-nuts.

I found one person on time-nuts that described how this was done. It involved building a small circuit to toggle the values on inputs on J5 in response to certain outputs. His approach involved wiring into the signal on the Fault LED, so it wasn't an external-only solution.

I had vague memories that there was a 'version 2' of this mod so I did some digging.  Take a look at:

http://www.febo.com/pipermail/time-nuts/2014-November/087793.html plus a half dozen or so messages after that.

Apparently you can just set up a jumper plug.

Somebody should set up a wiki for all the information on these boxes.

Ed
 

Offline Shock

  • Super Contributor
  • ***
  • Posts: 4200
  • Country: au
Re: Economical option for precision frequency reference?
« Reply #133 on: July 30, 2015, 01:19:35 pm »
Run one unit only?  There was discussion on that, but since I got these as spares, I haven't investigated.  You should be able to find the info on time-nuts.
I found one person on time-nuts that described how this was done. It involved building a small circuit to toggle the values on inputs on J5 in response to certain outputs. His approach involved wiring into the signal on the Fault LED, so it wasn't an external-only solution.

Or even better than that, read what is discussed here:
http://www.febo.com/pipermail/time-nuts/2014-November/087972.html

Which indicates that the GPS unit J5 Pin 2 requires a 470ohm pull down resistor to ground (pin 8 or pin 13) and pin 3 requires wiring directly to ground. As far as I'm aware the numbering is right to left looking at the unit front on. Please double check this yourself.
Soldering/Rework: Pace ADS200, Pace MBT350
Multimeters: Fluke 189, 87V, 117, 112   >>> WANTED STUFF <<<
Oszilloskopen: Lecroy 9314, Phillips PM3065, Tektronix 2215a, 314
 

Offline Shock

  • Super Contributor
  • ***
  • Posts: 4200
  • Country: au
Re: Economical option for precision frequency reference?
« Reply #134 on: July 30, 2015, 01:42:01 pm »
As for a PCB for the 5MHz processing doubler and distribution amp, I could make some gerber files that would be compatible with OSHPark, Seeedstudio, elecrow, dirtypcbs, et. al. publically available in the near future if there is continued interest and I can get the CAD data on the particular parts involved in the publication.

I would like to see a 10Mhz out for the Z3811A or a cleaner 10MHz signal for the Z3812A, I have 16 port distribution amp already so I'm a little against the need for adding a large distribution modification to the Z3811A.

As there is so far a rear power supply mod, standalone Z3811A mod, Z3811A GPS backup battery mod, I think a single 10Mhz output (with 1pps if required) on the rear of the Z3811A seems the best idea.

Drilling multiple holes in the face seems to be over engineering it and butchering what could otherwise be a very discrete modification.

If someone comes out with a minimalist approach I'm interested in adding the PCB and modding mine.
Soldering/Rework: Pace ADS200, Pace MBT350
Multimeters: Fluke 189, 87V, 117, 112   >>> WANTED STUFF <<<
Oszilloskopen: Lecroy 9314, Phillips PM3065, Tektronix 2215a, 314
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #135 on: July 30, 2015, 01:46:08 pm »
(h)  Run one unit only?  There was discussion on that, but since I got these as spares, I haven't investigated.  You should be able to find the info on time-nuts.

I found one person on time-nuts that described how this was done. It involved building a small circuit to toggle the values on inputs on J5 in response to certain outputs. His approach involved wiring into the signal on the Fault LED, so it wasn't an external-only solution.

I had vague memories that there was a 'version 2' of this mod so I did some digging.  Take a look at:

http://www.febo.com/pipermail/time-nuts/2014-November/087793.html plus a half dozen or so messages after that.

Apparently you can just set up a jumper plug.

Somebody should set up a wiki for all the information on these boxes.

Ed

Maybe we could just add a page on Wikipedia?
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #136 on: July 30, 2015, 01:50:23 pm »
Run one unit only?  There was discussion on that, but since I got these as spares, I haven't investigated.  You should be able to find the info on time-nuts.
I found one person on time-nuts that described how this was done. It involved building a small circuit to toggle the values on inputs on J5 in response to certain outputs. His approach involved wiring into the signal on the Fault LED, so it wasn't an external-only solution.

Or even better than that, read what is discussed here:
http://www.febo.com/pipermail/time-nuts/2014-November/087972.html

Which indicates that the GPS unit J5 Pin 2 requires a 470ohm pull down resistor to ground (pin 8 or pin 13) and pin 3 requires wiring directly to ground. As far as I'm aware the numbering is right to left looking at the unit front on. Please double check this yourself.

Ah, I saw this post before. Near the end of the message, he decides the connection to Pin 2 isn't needed:

Quote
...all that's needed to enable a Ref-1 unit stand alone is to link together J5 pins 2, 10, 12,  and 15, and to ground pin 3 to pin 8, and then just hang around for hours and  hours on end with yer fingers crossed:-)

And I THINK in a later reply he decides the connection to pin 3 should be a pull down resistor instead of a direct short to ground. I'll have to double-check that as I know you have to be careful about having both pull-ups and pull-downs attached to a CMOS circuit.

What does he mean when he says this: "When pin 5 and pin 11 are  observed together, the usual GPS sawtooth pattern is  evident."

« Last Edit: July 30, 2015, 02:08:38 pm by motocoder »
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #137 on: July 30, 2015, 01:52:14 pm »
As for a PCB for the 5MHz processing doubler and distribution amp, I could make some gerber files that would be compatible with OSHPark, Seeedstudio, elecrow, dirtypcbs, et. al. publically available in the near future if there is continued interest and I can get the CAD data on the particular parts involved in the publication.

I would like to see a 10Mhz out for the Z3811A or a cleaner 10MHz signal for the Z3812A, I have 16 port distribution amp already so I'm a little against the need for adding a large distribution modification to the Z3811A.

As there is so far a rear power supply mod, standalone Z3811A mod, Z3811A GPS backup battery mod, I think a single 10Mhz output (with 1pps if required) on the rear of the Z3811A seems the best idea.

Drilling multiple holes in the face seems to be over engineering it and butchering what could otherwise be a very discrete modification.

If someone comes out with a minimalist approach I'm interested in adding the PCB and modding mine.

I think you make a good point. How about just modding it so the 15 MHz SMA connector becomes a 10 MHz output?
 

Offline SoundTech-LG

  • Frequent Contributor
  • **
  • Posts: 788
  • Country: us
Re: Economical option for precision frequency reference?
« Reply #138 on: July 30, 2015, 03:42:16 pm »
As of June 19, eLoran is on the air in the United States. The low-frequency signal emanates from a single station, a former U.S. Coast Guard Loran Unit in Wildwood, N.J., which sports a 625-foot signal mast that has been out of action for five years. The signal is receivable at distances of up to 1,000 miles.

If you already own a receiver,(in that area) well, could be economical...
http://gpsworld.com/eloran-progresses-toward-gps-back-up-role-in-u-s-europe/


 

Offline SoundTech-LG

  • Frequent Contributor
  • **
  • Posts: 788
  • Country: us
Re: Economical option for precision frequency reference?
« Reply #139 on: July 30, 2015, 03:53:06 pm »


Nigel, and I worked on this project to get the Stanford Research FS-700 manual complete with schematics up on K04bb.
I provided high resolution copies, and he organized, and collected it all into one pdf. I then uploaded onto K04bb.
He mentions good prices on the FS-700, but they recently have gone through the roof! Not sure why yet...

https://www.febo.com/pipermail/time-nuts/2015-July/092812.html
 

Offline N8AUM

  • Regular Contributor
  • *
  • Posts: 132
  • Country: us
Re: Economical option for precision frequency reference?
« Reply #140 on: July 31, 2015, 08:32:37 am »
I wanted to see how close the Lucent RFTG unit is compared to Nortel Trimble NTBW50AA Thunderbird unit. Both units using the same antenna over a 3 hour time period.  Triggering on Ch 1 (Lucent) yellow trace and Ch 2 is the 10 Mhz TP output from the RFTG unit with the scopes persistence set to infinity.
Forgive the crappy pic, am tiered and cant see straight lol

 

Offline Shock

  • Super Contributor
  • ***
  • Posts: 4200
  • Country: au
Re: Economical option for precision frequency reference?
« Reply #141 on: July 31, 2015, 12:50:50 pm »
I would like to see a 10Mhz out for the Z3811A or a cleaner 10MHz signal for the Z3812A, I have 16 port distribution amp already so I'm a little against the need for adding a large distribution modification to the Z3811A.
As there is so far a rear power supply mod, standalone Z3811A mod, Z3811A GPS backup battery mod, I think a single 10Mhz output (with 1pps if required) on the rear of the Z3811A seems the best idea.
Drilling multiple holes in the face seems to be over engineering it and butchering what could otherwise be a very discrete modification.
If someone comes out with a minimalist approach I'm interested in adding the PCB and modding mine.

I think you make a good point. How about just modding it so the 15 MHz SMA connector becomes a 10 MHz output?

All depends how much modification it requires, am still reading through the time-nut posts.

I prefer everything out the back and mount just the GPS unit in an enclosure rather than a rack.
Soldering/Rework: Pace ADS200, Pace MBT350
Multimeters: Fluke 189, 87V, 117, 112   >>> WANTED STUFF <<<
Oszilloskopen: Lecroy 9314, Phillips PM3065, Tektronix 2215a, 314
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #142 on: July 31, 2015, 01:43:47 pm »
I would like to see a 10Mhz out for the Z3811A or a cleaner 10MHz signal for the Z3812A, I have 16 port distribution amp already so I'm a little against the need for adding a large distribution modification to the Z3811A.
As there is so far a rear power supply mod, standalone Z3811A mod, Z3811A GPS backup battery mod, I think a single 10Mhz output (with 1pps if required) on the rear of the Z3811A seems the best idea.
Drilling multiple holes in the face seems to be over engineering it and butchering what could otherwise be a very discrete modification.
If someone comes out with a minimalist approach I'm interested in adding the PCB and modding mine.

I think you make a good point. How about just modding it so the 15 MHz SMA connector becomes a 10 MHz output?

All depends how much modification it requires, am still reading through the time-nut posts.

I prefer everything out the back and mount just the GPS unit in an enclosure rather than a rack.

If you keep reading the time-nuts thread on standalone operation of REF-1, eventually you find this post:

https://www.febo.com/pipermail/time-nuts/2014-November/088043.html

Which describes a mod that can be done, entirely via an adapter plugged into J5, to allow REF-1 to operate in standalone mode. Since REF-1 does not have a 10 MHz output, you will obviously also have to do something there. The adapter is:

      470R
2 ---/\/\/\---8/13
3 ----------- 8/13


I.e. short pin 3 directly to ground (available on either pin 8 or 13), pull down pin 2 to ground via a 470 ohm resistor.I haven't tried it to verify, but I believe they are saying that it can take an hour for REF-1 to come up in "ON" mode with the 15 MHz and 1PPS outputs enabled when using this adapter.

 

Offline edpalmer42

  • Super Contributor
  • ***
  • Posts: 2260
  • Country: ca
Re: Economical option for precision frequency reference?
« Reply #143 on: July 31, 2015, 03:54:53 pm »
I wanted to see how close the Lucent RFTG unit is compared to Nortel Trimble NTBW50AA Thunderbird unit. Both units using the same antenna over a 3 hour time period.  Triggering on Ch 1 (Lucent) yellow trace and Ch 2 is the 10 Mhz TP output from the RFTG unit with the scopes persistence set to infinity.
Forgive the crappy pic, am tiered and cant see straight lol

Okay, but what does that mean?  Every GPSDO has jitter.  You'd see something similar if you compared two identical units.  Also, your measurement technique makes it look like the Nortel unit is perfect and all the jitter is on the Lucent unit when they're actually both jittering around.

In the past I measured the jitter of the 1 PPS output of a few bare GPS receivers and GPSDOs.  Here are the results:

GPS Devices -- Performance of 1 PPS Output

Device ............... Std Dev (ns).... Range (max-min)(ns) ... Device Type

Navsync CW12 ......... 4 - 5 .......... 20 - 25 ............... GPS Rcvr (1) see below
Motorola UT+ ......... 40 - 55 ........ 95 - 110 .............. GPS Rcvr (2)
Rockwell Jupiter ..... 10 ............  50 ...................  GPS Rcvr (3)
Motorola M12M ........ 10 - 15 ........ 40 - 60 ............... GPS Rcvr

Trimble Thunderbolt .. 0.4 - 0.5 ...... 2 - 4 ................. GPSDO
HP Z3801A ............ 0.1 - 0.2 ...... < 1 ................... GPSDO   
HP Z3817A / CW12 ..... < 0.1 .......... < 1 ................... GPSDO    (4)
Jackson Labs GPSTCXO . 0.3 - 0.4 ...... 2 - 3 ................. GPSDO


Results are based on multiple runs of ~ 1000 measurements each.
Sawtooth correction has not been used for any of the GPS receivers.  Where supported, it would reduce the numbers substantially.
GPS Rcvr measurements made with HP 5372A.  GPSDO measurements made with HP 5370B.
All units were connected to the same antenna system.

1.  Sawtooth correction not supported.
2.  Most 'range' results were in this group, but there were a few at 20 - 30.
3.  Only one test.
4.  Requires external 1 PPS input.  Equipped with E1938 oscillator.



I haven't done these tests on the RFTG-u.  I'll have to add it to the list.

Ed
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #144 on: August 01, 2015, 12:30:17 am »
I haven't looked at the details fully but that photograph looks like there was quite a lot of phase variation relative to the entire 10MHz cycle, which surprised me.  I would have expected the jitter to be visible only on one zoomed in edge of the waveforms and to account for a small fraction of a cycle's phase over a relatively short (hour) interval with there being a bit of phase noise with an underlying phase drift progression due to the frequency discord.

I also wonder about whether the clock outputs wouldn't get cleaner wrt. jitter during GPS LOS periods since there is no further input for the corrections.  Of course I suspect the devices create a mathematical model of a given OCXO's 'inherent' frequency error and temperature sensitivity and may try to keep correcting for the predicted error according to the trained model data (and perhaps also temperature measurement though probably not if it is assumed to simply be stable in the oven's feedback).  But eventually I suppose the 'remembered' feedback due to the frequency error modem might 'age out' and then one would be left with basically the OCXO's intrinsic output given the temperature / power supply / noise related fluctuations.
Or I suppose if you cold started a unit and could get it to output the 15MHz or 10MHz or whatever without having the GPS related corrections enabled due to insufficient data to model the OCXO error at which point you'd be left with less control system jitter, though it may simply not output the high frequency at such a time.  I recall seeing something about a fault described in the RFTG's SW report descriptions as 'flywheel failed, duration > 8 hr' so I wasn't sure what it'd do wrt. HF output after 8hr or 24hr in flywheel.

I suppose it might also be possible to trick the non-GPS RFTG unit into giving 'pure' OCXO output by doing something like feeding its own 15MHz output divided by 15M back into its PPS reference input as if that was coming from the GPS disciplining unit, I suppose one might need to do some phase adjustment to get it to accept that there was a 0 static phase error in that process.

The microprocessor in the unit is making EFC corrections periodically to keep the thing in sync with the GPS signal. I can see in the reported PPS stats that the PPS value is varying between about +/- 40ns of the GPS time, and for others here, whose units arent' quite as well behaved, it's more like +/-90ns. The o-scope capture posted earlier was I assume taken by turning on infinite persistence on the scope and letting it run for 3 hours. So I would entirely expect this sort of display as the EFC value gets updated many times over that 3 hour span.
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #145 on: August 01, 2015, 06:48:29 am »
Interesting.  That is quite a lot of jitter for some of the units (40-90ns).  Particularly out of a 100ns period 10MHz cycle anything more than 1ns of jitter is relatively speaking a lot (3.6 degrees / ns @ 10MHz).

BTW I looked at the data from KO4BB's site but I still haven't seen anything like an output specification for the 15MHz or PPS signals, is there a published value for what the frequency accuracy and jitter and such are supposed to be when the unit is operating normally?

As I understand it the PPS signals for timing oriented equipment are actually supposed to be synchronous with the transition of the GPS referenced second's transition time, generally down to nanoseconds levels of precision.  I'm not sure that the RFTG units are oriented toward "timekeeping" so much as an accurate long term stable frequency reference at 1Hz and 15MHz though I suppose they probably also keep the clock edge of the PPS relatively aligned to the actual transition of the second though it sounds like that is only to the level of ~40ns to ~90ns which really sounds like it is about a half cycle of the 5MHz crystal clock.
Though it is odd if it can flywheel for 24hrs with microseconds of drift accuracy and still get large fractions (90ns) of that as short term corrections.

It's not 40 - 90ns of jitter on a waveform at 10 MHz (100ns). It's 40-90ns delta over the GPS reference over a 1 second interval. The hardware has no mechanism to measure jitter over a individual periods of the 10Mhz clock. What it does it count the number of 10 MHz (probably 5 Mhz in this case - but let's use 10MHz for simplicity of discussion) OCXO cycles that occur from one GPS PPS rising edge to the next GPS PPS rising edge. If the count is 9,999,996, it calculates that it's -40ns relative to GPS, and it uses this value as an error signal into the PID loop that is controlling the EFC voltage. Similarly, if the count is 10,000,009, it calculates that it's 90ns relative to the GPS, feeds that into the PID loop, and adjusts the EFC accordingly.

Regarding the stability over the course of the day, I suspect it estimates this based on some calculation involving the rate of change of EFC and the standard deviation or variance of frequency when in closed-loop mode (GPS tracking active). The theory being that if it goes into an open-loop mode (no GPS), it will leave the EFC voltage at a constant value, and the drift in frequency will be a function of how much it was having to change EFC previously and how much the frequency was varying even when EFC was tightly controlled.
« Last Edit: August 01, 2015, 07:45:53 am by motocoder »
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #146 on: August 01, 2015, 02:44:48 pm »
I have uncovered a couple of interesting data points on my Lucent unit.

1) I noticed that the PPS fluctuations seemed to increase every time the A/C turned on here. So I put a small 12V fan behind the unit blowing air through at a very low CFM rate. Notably, this seemed to cause the fluctuations to get worse. I observed this over many hours, and as soon as I removed the fan, it returned to the previous pattern.
2) After removing the fan, I put a  piece of cardboard around the A/C vent so that the air would not circulate past the Lucent unit as much. This settled things way down, and the PPS value seems to mostly be +/- 10ns with occasional excursions to +/- 20ns and a much rarer spike to +/-30 ns.

During #2, the HUD value started to go down for awhile, but then started increasing again. It appears strongly correlated with the rate of change of the EFC value, which was fairly steady for a short time (while HUD was decreasing), but then started to rise again.

So in conclusion, the unit is sensitive to the environmental air temperature and possibly the EMI generated by that fan. The HUD value seems to be strongly correlated with the rate of change of EFC. The more constant EFC is when the unit is locked to GPS, the lower the HUD value, regardless of how much "jitter" there is of PPS.
 

Offline edpalmer42

  • Super Contributor
  • ***
  • Posts: 2260
  • Country: ca
Re: Economical option for precision frequency reference?
« Reply #147 on: August 01, 2015, 03:17:07 pm »
I have uncovered a couple of interesting data points on my Lucent unit.

1) I noticed that the PPS fluctuations seemed to increase every time the A/C turned on here. So I put a small 12V fan behind the unit blowing air through at a very low CFM rate. Notably, this seemed to cause the fluctuations to get worse. I observed this over many hours, and as soon as I removed the fan, it returned to the previous pattern.
2) After removing the fan, I put a  piece of cardboard around the A/C vent so that the air would not circulate past the Lucent unit as much. This settled things way down, and the PPS value seems to mostly be +/- 10ns with occasional excursions to +/- 20ns and a much rarer spike to +/-30 ns.

During #2, the HUD value started to go down for awhile, but then started increasing again. It appears strongly correlated with the rate of change of the EFC value, which was fairly steady for a short time (while HUD was decreasing), but then started to rise again.

So in conclusion, the unit is sensitive to the environmental air temperature and possibly the EMI generated by that fan. The HUD value seems to be strongly correlated with the rate of change of EFC. The more constant EFC is when the unit is locked to GPS, the lower the HUD value, regardless of how much "jitter" there is of PPS.

I noticed that mine reacted strongly when a stray sunbeam fell on it for a few minutes.  I'm surprised that a double-oven OCXO would be that sensitive to external thermal influences.  The high number of holes in the case suggests that when these units were in service there was a fan tray or something similar that flooded the entire bay with a continuous stream of air to maintain a constant temperature.  Closing off those holes would have to be done cautiously to make sure that nothing was going to overheat, but reducing drafts on the oscillator would likely improve performance.

By the way, don't add extra insulation around the OCXO!  The internal heat must be allowed to escape.  If you prevent that, the oven controller won't be able to control the temperature properly.  Worst case, the internal temperature starts to rise and the unit cooks itself.
 

Offline motocoderTopic starter

  • Supporter
  • ****
  • Posts: 769
  • Country: us
  • Electrical Engineer
Re: Economical option for precision frequency reference?
« Reply #148 on: August 01, 2015, 05:01:55 pm »
I noticed that mine reacted strongly when a stray sunbeam fell on it for a few minutes.  I'm surprised that a double-oven OCXO would be that sensitive to external thermal influences.  The high number of holes in the case suggests that when these units were in service there was a fan tray or something similar that flooded the entire bay with a continuous stream of air to maintain a constant temperature.  Closing off those holes would have to be done cautiously to make sure that nothing was going to overheat, but reducing drafts on the oscillator would likely improve performance.

Yes, my theory on this is that we're talking about such minute changes in frequency, that even a tiny outside temperature change can have an effect. I wish I had some other unit to compare against - not sure if this is normal. Do you think the stray sunbeam was impacting the outside temperature of the oven, or is there perhaps some chip-scale IC inside these units?

BTW, I found an interesting paper from Milliren on crystal oscillator aging: http://www.mti-milliren.com/MTIPapers/Ext_Aging_Perf_Results.pdf

By the way, don't add extra insulation around the OCXO!  The internal heat must be allowed to escape.  If you prevent that, the oven controller won't be able to control the temperature properly.  Worst case, the internal temperature starts to rise and the unit cooks itself.

Noted. Thanks
« Last Edit: August 01, 2015, 05:09:29 pm by motocoder »
 

Offline edpalmer42

  • Super Contributor
  • ***
  • Posts: 2260
  • Country: ca
Re: Economical option for precision frequency reference?
« Reply #149 on: August 01, 2015, 05:23:05 pm »
Yes, my theory on this is that we're talking about such minute changes in frequency, that even a tiny outside temperature change can have an effect. I wish I had some other unit to compare against - not sure if this is normal. Do you think the stray sunbeam was impacting the outside temperature of the oven, or is there perhaps some chip-scale IC inside these units?

You mean like the RPi a few months back?  I think these units are too old for that.  If I remember correctly, it took the unit several minutes to recover after I blocked the sunbeam so I think it was a thermal effect.

I have a Z3801A which is equipped with an HP double-oven oscillator.  The oscillator is surprisingly crude for an HP unit.  They basically took an HP 10811 oscillator (really good reputation, by the way), made a few internal changes to increase the electrical tuning range, and then wrapped another heater and insulation around it.  There are autopsy photos online.  Once you dig out the oscillator, it even has the original label on it.  In spite of how crude it is, it works quite well.  The entire unit is in a solid steel box with a reasonable number of slots for cooling.  Thermal changes don't really bother it - even at these tiny levels.

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf