Author Topic: OCXO tuning screws not working on old equipment  (Read 4546 times)

0 Members and 1 Guest are viewing this topic.

Offline tverbeureTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: us
    • tomverbeure.github.io
OCXO tuning screws not working on old equipment
« on: June 26, 2023, 07:56:26 pm »
I just received a GPSDO. Last night, I tried to calibrate my HP 5334A and Racal-Dana 1992 frequency counters as well as a stand-alone OCXO reference clock generator, by putting the GPSDO 10MHz and the reference output on a scope and turning the tuning screws until the clock don't drift anymore.

This worked fine with the stand-alone reference clock generator (it's an old commercial product, but really nothing more than an OCXO with power supply, 2 outputs, and a calibration screw hole.)

However, with both the 5334A and the 1992, turning the calibration screws doesn't anything. I can turn them from all the way CW or CCW yet the signal drifts just the same.

It's such an odd thing to fail on two of the 3 different devices on which I've tried this.

Is this a common failure point? I could obviously use the REF-IN to sync these devices to a calibrated clock or to the GPSDO itself, but I'd prefer them to be operating stand-alone.
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2439
  • Country: de
Re: OCXO tuning screws not working on old equipment
« Reply #1 on: June 26, 2023, 08:35:24 pm »
The Lissajou method is absolutely inappropriate for comparing frequencies on a 10-8 .. 10-10 level.
A pity that you mis-adjusted the OCXOs of both your counters.

Use Time Interval methods (accumulated phase shift measurement with ns resolution, between GPSDO and OCXO) instead, or simply measure the frequency of the GPSDO on these interpolating counters. They resolve 9..10 digits, when you use 10..100sec Gate Time.

Give TimeLab by John Miles: https://www.miles.io/timelab/beta.htm a try.

I assume that your counters are not supported directly, but check out if other interface options might work.

Frank
« Last Edit: June 26, 2023, 08:37:20 pm by Dr. Frank »
 

Offline J-R

  • Super Contributor
  • ***
  • Posts: 1319
  • Country: us
Re: OCXO tuning screws not working on old equipment
« Reply #2 on: June 26, 2023, 09:40:19 pm »
I think this is possible especially on the HP5334A since it is easily externally accessible. Which oscillator is installed in the 1992?  Maybe pull one of the trimmers and test it.  And of course I'll be captain obvious and ask if you've double-checked your connections & setup?

Not sure what the Dr. is upset about, as using a scope in this manner is pretty common and I've done it many times.  Of course if you're going for a high-quality adjustment you need to monitor the scope for a little bit.  When I first got my GPSDO, I immediately adjusted half a dozen devices and in all cases drift or trim pots were blatantly the weak points, not the scope.  Specifically I remember on my HP5334B with option 10 having to literally tap-tap-tap the trimmer screw to get it where I wanted it.  A bit further than necessary, but a year later it's still perfection.
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2439
  • Country: de
Re: OCXO tuning screws not working on old equipment
« Reply #3 on: June 26, 2023, 10:49:23 pm »


Not sure what the Dr. is upset about, as using a scope in this manner is pretty common and I've done it many times.  Of course if you're going for a high-quality adjustment you need to monitor the scope for a little bit.  When I first got my GPSDO, I immediately adjusted half a dozen devices and in all cases drift or trim pots were blatantly the weak points, not the scope.  Specifically I remember on my HP5334B with option 10 having to literally tap-tap-tap the trimmer screw to get it where I wanted it.  A bit further than necessary, but a year later it's still perfection.

Oscilloscope / Lissajou method might be common for trimming ordinary XTALS to 10-6..10-7, where a 360° phase shift might happen within a few seconds.
OCXOs need to be trimmed to 1000 times less deviation, i.e. that phase shift happens on a 1000 times longer time scale, so that's a quite stupid method, as that takes 15min at least, staring monotonously on the scope screen. The counters themselves offer a very comfortable method by T.I. phase measurement, and TimeLab or a simple PYTHON script would allow sampling over several hours. Stable32 then could calculate Allen Deviation and longterm drift,to be trimmed to near zero.
I really don't understand why people are fascinated by those outdated methods from way back then, when it is also much easier to measure frequency directly and trimm to 10-9 within 1 or 10 sec, respectively.

Then it would be interesting, which OCXO models are inside each counter. Is it a 10811 in the 5334A?
The 1992  had several different OCXO options, up to a high stability model.
I can't imagine, that both trimmers of both OCXO are defect. As well as there is no information, how fast the phase change appears to be, or if no change is visible.
In latter case, I would search for the error on operators side.
If the phase slips very fast, on the other side, then there must be a very high frequency deviation, that could also mean that the GPSDO does not work (yet) properly, i.e. not locked, no antenna signal, not disciplined yet, etc.

Did it really sound if I was upset? In fact, I dislike if no or rarely any information is given when asking for help, so one can only make an educated guess.

Frank
« Last Edit: June 26, 2023, 10:56:09 pm by Dr. Frank »
 

Offline aeg

  • Regular Contributor
  • *
  • Posts: 130
  • Country: us
Re: OCXO tuning screws not working on old equipment
« Reply #4 on: June 26, 2023, 11:13:42 pm »
The oscilloscope method is by adjusting for no drift when the DUT oscillator waveform is displayed with the sweep synchronized to the reference oscillator. It is, in effect, a time interval measurement. It does not involve a Lissajous display.

 

Offline J-R

  • Super Contributor
  • ***
  • Posts: 1319
  • Country: us
Re: OCXO tuning screws not working on old equipment
« Reply #5 on: June 27, 2023, 02:44:57 am »
As previously mentioned, in my experience drift and the trim pots on these and similar class devices are by far the weak points and they simply do not have the precision to exceed what you can see on the scope within about 10-15 seconds.  On a large-screen DSO, zoomed in to the waveforms, the tiniest change is easily observed and the smallest of tweaks to the pot sends the waveform off one way or another pretty much immediately.
 

Offline tverbeureTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: us
    • tomverbeure.github.io
Re: OCXO tuning screws not working on old equipment
« Reply #6 on: June 27, 2023, 03:46:42 am »
The Lissajou method is absolutely inappropriate for comparing frequencies on a 10-8 .. 10-10 level.
A pity that you mis-adjusted the OCXOs of both your counters.
I'm not using the Lissajou method, but simply displaying the 2 waveforms on top of each other and checking the drift against each other. It allows for zooming in much more than with the Lissajou mode.

And I wish I misadjusted those two counters! :) Because it would mean that their OXCOs can be trimmed. But unfortunately, as I said, I can put the trim screws in any position, and there is zero difference before and after, even with a 10s gate time.

Furthermore, when I now use the GPSDO as external reference clock and the successfully calibrated stand-alone OCXO as input, the Racal with a 10s gate time toggles between 9,999,999.998Hz and 9,999,999.99Hz, waaaay better than before. Similarly, the 5334A now toggles between 9,999,999.99MHz and 10,000,000.0MHz. Yes, it took a lot of finessing that little screw, and the 300MHz scope was zoomed in to the max. I frankly stunned that it worked so well.

Quote
...or simply measure the frequency of the GPSDO on these interpolating counters.
I did that too. Same result: turning the trimming screws makes zero difference on both counters.

FWIW: the drift-on-the-scope calibration method is significantly faster. You don't need to wait 10s per measurement, and as proven with the stand-alone OCXO, the result is equally good.

All that said, my question was not whether or not I used the right calibration technique, but whether or not it's common for these kind of trim screws to go bad to the point where they don't work at all.

The oscilloscope method is by adjusting for no drift when the DUT oscillator waveform is displayed with the sweep synchronized to the reference oscillator. It is, in effect, a time interval measurement. It does not involve a Lissajous display.
Yes, this is the method that I used.



« Last Edit: June 27, 2023, 04:01:46 am by tverbeure »
 

Offline tverbeureTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: us
    • tomverbeure.github.io
Re: OCXO tuning screws not working on old equipment
« Reply #7 on: June 27, 2023, 04:00:41 am »
I think this is possible especially on the HP5334A since it is easily externally accessible. Which oscillator is installed in the 1992?  Maybe pull one of the trimmers and test it.  And of course I'll be captain obvious and ask if you've double-checked your connections & setup?
I think opening up the two counters is unavoidable. Another fun project to spend time on. The 1992 has the 04E option. I'm pretty sure that the connections were right.

Quote
When I first got my GPSDO, I immediately adjusted half a dozen devices and in all cases drift or trim pots were blatantly the weak points, not the scope.  Specifically I remember on my HP5334B with option 10 having to literally tap-tap-tap the trimmer screw to get it where I wanted it.  A bit further than necessary, but a year later it's still perfection.
Of course, all of this is just for the sake of having a correct measurement. It's not as if I have any project that requires this kind of accuracy!  ;D
« Last Edit: June 27, 2023, 04:03:16 am by tverbeure »
 

Offline tverbeureTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: us
    • tomverbeure.github.io
Re: OCXO tuning screws not working on old equipment
« Reply #8 on: June 27, 2023, 04:34:55 am »
Oscilloscope / Lissajou method might be common for trimming ordinary XTALS to 10-6..10-7, where a 360° phase shift might happen within a few seconds.
But why would anyone need a 360 degree phase shift? All you need to see is movement: 0.5 degree phase shift in a few seconds works just the same.

So if you say that a 360 phase shift in a few seconds is good to trim a 10-7 XTAL, and if I can work with a 0.5 degree phase shift over a few seconds, then that's 3 orders of magnitude right there. Your 10-7 becomes my 10-10.

I'm not being facetious here. 10MHz is a 100ns period. 0.5 degrees of that is 0.14ns. When your scope is set at 1ns/div, it's trivial to see movement at that scale.

Quote
OCXOs need to be trimmed to 1000 times less deviation, i.e. that phase shift happens on a 1000 times longer time scale, so that's a quite stupid method, as that takes 15min at least, staring monotonously on the scope screen.
Have you actually tried? It took me maybe 15 tries to calibrate my standalone OCXO to 2x10-10. Give or take 5min tops? It takes longer to properly screw in a GPIB cable. (I kid, I kid, mostly.)

Quote
I really don't understand why people are fascinated by those outdated methods from way back then, when it is also much easier to measure frequency directly and trimm to 10-9 within 1 or 10 sec, respectively.
People use them because they work very well. And it's much faster to do. Do you have any idea how much work it would be to create the setup that you propose? I mean, TimeLab doesn't even work on Linux, I'd have to setup a Windows PC first. ;D I'm well versed in controlling my equipment over Python (I wrote a full Racal 1992 driver for pymeasure), but it's completely overkill for something as simple a calibrating an OCXO.

The issue is not the measurement. It's the fickleness of the trim screws.

Quote
Then it would be interesting, which OCXO models are inside each counter. Is it a 10811 in the 5334A?
I haven't opened it up, but it's the high stability option 010. According to the schematics, that's a 10811.
(BTW: I bought a pristine service manual on eBay: I've scanned them with high quality so that you don't need to buy them anymore.)

Quote
The 1992 had several different OCXO options, up to a high stability model.
Yup. I have the 04E high stability option.

Quote
I can't imagine, that both trimmers of both OCXO are defect. As well as there is no information, how fast the phase change appears to be, or if no change is visible.
In latter case, I would search for the error on operators side.
It's really hard to screw up turning a little screw. Other than making sure that the INT/EXT switch on the 5334A in the back is in the right position, there's really not much that can go wrong: connect the REF-OUT to the scope and observe the signal against the GPSDO.

Quote
If the phase slips very fast, on the other side, then there must be a very high frequency deviation, that could also mean that the GPSDO does not work (yet) properly, i.e. not locked, no antenna signal, not disciplined yet, etc.
The phase slip is a few seconds for a 360 phase. Nothing too bad. It's not that the counters were horribly wrong to begin with. Just that I'm pedantic in wanting to get the things as accurate as possible.
The GPSDO has been running for a couple of days. It's also stuck in a drawer to reduce temperature changes even more. It seems stable against the calibrated stand-alone OCXO, against the REF-OUT of my R&S SMHU signal generator and against the REF-OUT of my Advantest R3273, both have a high-stability OCXO as well.

The 5334A was a flea market buy and the 1992 was a gift, so it's not completely unthinkable that they both have issues. (You can seem them here in action: https://tomverbeure.github.io/2023/06/16/Frequency-Counting-with-Linear-Regression.html) I love the red LED look.  :)
« Last Edit: June 27, 2023, 04:58:26 am by tverbeure »
 
The following users thanked this post: rf-loop, Dr. Frank

Offline tverbeureTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: us
    • tomverbeure.github.io
Re: OCXO tuning screws not working on old equipment
« Reply #9 on: June 27, 2023, 06:42:39 am »
Unlike the 1992, the trimmer of the 5334A seems to be outside of the OCXO itself. So that seems to be first step: open it up and check the potmeter itself.

« Last Edit: June 27, 2023, 06:44:46 am by tverbeure »
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: OCXO tuning screws not working on old equipment
« Reply #10 on: June 27, 2023, 06:51:10 am »
The oscilloscope method is by adjusting for no drift when the DUT oscillator waveform is displayed with the sweep synchronized to the reference oscillator. It is, in effect, a time interval measurement. It does not involve a Lissajous display.



I don't know why people keep talking about Lissajou, the OP said nothing about it.

I have used this method of triggering on a reference oscillator while displaying the signal I want to check and it is extremely sensitive, gently blowing on the outside casing of a double oven OCXO is enough to cause the waveform to start moving for a while before settling down.
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2439
  • Country: de
Re: OCXO tuning screws not working on old equipment
« Reply #11 on: June 27, 2023, 08:36:59 am »


I don't know why people keep talking about Lissajou, the OP said nothing about it.

I have used this method of triggering on a reference oscillator while displaying the signal I want to check and it is extremely sensitive, gently blowing on the outside casing of a double oven OCXO is enough to cause the waveform to start moving for a while before settling down.

Very easy: Because the OP initially also said nothing specific about the setup:

'...by putting the GPSDO 10MHz and the reference output on a scope and turning the tuning screws until the clock don't drift anymore.'

From such scarce information, you can only guess, how it's done, either Lissajou / XY mode, or you could display both channels and trigger on first channel, but you could as well only display one signal and only use the 2nd on the trigger channel. Depends on your scope.
And it was not evident at all, how fast the scope / timebase used was.. 10ns? 1ns? Less than that? If you have a slow scope, the method described in the hp manual
gets really tedious, especially when you measure even more stable atomic standards, like a Rb clock vs. GPSDO, latter having a lot of jitter.

That's always what annoys me, that no details about the methods are given.
So if I made a wrong guess, it's now even more annoying if some know-it-all try to tell me, that I could have known it correctly upfront.  :-//


Frank
« Last Edit: June 27, 2023, 08:41:31 am by Dr. Frank »
 
The following users thanked this post: tverbeure

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2439
  • Country: de
Re: OCXO tuning screws not working on old equipment
« Reply #12 on: June 27, 2023, 09:40:20 am »
Unlike the 1992, the trimmer of the 5334A seems to be outside of the OCXO itself. So that seems to be first step: open it up and check the potmeter itself.

(Attachment Link)

Hello Tom,
I have to admit that the scope method is usable for trimming the OCXOs.
The T.I. and direct freq. methods are much easier, anyway, also from point of cabling.
Please check, what happens if you use 10..100sec Gate Time on your 5334A, whether the display always rounds to 9 digits, or displays the rightmost digits with OF.. That could give additional resolution.
Identical procedure like your scope method applies, if you connect GPSDO signal to channel A of your 5334A, and the 1992 output to channel B and then choose T.I. A->B on the 5334A (mode which allows +/- T.I.?).. it should resolve 1..2ns, maybe additional averaging is possible, which you can either track visually over several minutes, or by a PYTHON script.. The effort is the same, but to my experience much more comfortable. TimeLab would directly plot this curve and freq. deviation.
I last used the scope method (XY and 2 channel) when I was a teen, building my first TTL grave counter with an OCXO, comparing to 200 kHz Droitwich, so I don't have to redo this experiment again.

Btw.: you have a really nice frequency blog, so you could now try and add these different phase measurement methods, which were regularly used by those 'time-nuts', but also for the Allen Deviation measurements. Maybe you could reveal, which GPSDO you have acquired. Some of them, especially when there is no disciplined OCXO inside, show a very bad short term jitter on the order of 10-8.

Concerning your counters, it's getting really interesting. The 5334A/B uses the external EFC trimming, if a 10811 is assembled inside.
That's not the case with its bigger brother HP5335A, which I own. I have to open the lid and trim the capacitor directly inside the 10811.
For my 5370B, I drilled a hole into the lid, to access this said capacitor trimmer in the 10811.

It might turn out that in your 5334A, a different OCXO is assembled, or that this feature had been disabled for some reason.

I'm not so familiar with the 1992, I repaired one with that high stability option 27 years ago .. didn't that have two trimmers, coarse and fine?
Maybe you only have access to the fine trimmer from outside, so you won't see any difference, as the scope method is not sensitive enough for very small changes?
Same might occur for the 5334A.

It would be great, if you would provide some Photographs from the interior of the 5334A, please.

Frank
« Last Edit: June 27, 2023, 09:57:59 am by Dr. Frank »
 

Offline Wrenches of Death

  • Regular Contributor
  • *
  • Posts: 84
  • Country: us
Re: OCXO tuning screws not working on old equipment
« Reply #13 on: June 27, 2023, 12:53:20 pm »

Specifically I remember on my HP5334B with option 10 having to literally tap-tap-tap the trimmer screw to get it where I wanted it.  A bit further than necessary, but a year later it's still perfection.

I'm guessing that you are using a standard insulated trimmer tool. I've found that a simple mod will make them much easier to use for precise adjustments. I slip a couple of inches of snug fitting clear plastic hose over them to increase the diameter. Now the tool diameter is doubled and it is MUCH easier to make tiny adjustments.

WoD


 
The following users thanked this post: tverbeure

Offline tverbeureTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: us
    • tomverbeure.github.io
Re: OCXO tuning screws not working on old equipment
« Reply #14 on: June 27, 2023, 06:16:22 pm »
Very easy: Because the OP initially also said nothing specific about the setup:
To be honest, I have never heard about the Lissajou method for trimming an OCXO and none of the OCXO tutorials that I ran into use it. To borrow your terminology, it seems like quite a stupid method, more like a party trick to show during physics class, because it's much a harder to zoom into an XY signal and get the required precision.

But also, my question wasn't about the trimming method to begin with, but about how common it is for these trimming screws to fail...

Quote
So if I made a wrong guess, it's now even more annoying if some know-it-all try to tell me, that I could have known it correctly upfront.  :-//
Alright, so far I let it slide, but since you're doubling down, here's some well-meaning advice: others commented about you sounding upset. It wasn't that, it came over as condescending, the way some doctors talk down to a nurse. "a pity you misadjusted both of your counters", "quite a stupid method", incorrectly assuming that observing Lissajou is normal way of observing drift, claiming you need see 360 degrees shifts, and then calling *others* annoying know-it-alls when they don't make that assumption. It may be a cultural or language thing, and totally acceptable where you live, but you'll just have to accept that this is an international forum where that kind of stuff will rub people the wrong way.

Anyway, let's move on...

Quote
It might turn out that in your 5334A, a different OCXO is assembled, or that this feature had been disabled for some reason.
I'll open it up tonight and report back.

Quote
didn't that have two trimmers, coarse and fine?
It does. Both screws are accessible from the outside. Both do nothing.

Quote
Maybe you only have access to the fine trimmer from outside, so you won't see any difference, as the scope method is not sensitive enough for very small changes?
Leaving aside that the scope method is actually sensitive enough to observe 10-10 changes (seriously, try it!), I've done all the regular measurements with the 1992 using a 10s gate time and 10 digits precision. The last digit of the 1992 doesn't move no matter which position I put the fine trimmer in.



 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: OCXO tuning screws not working on old equipment
« Reply #15 on: June 27, 2023, 07:08:49 pm »
Very easy: Because the OP initially also said nothing specific about the setup:

'...by putting the GPSDO 10MHz and the reference output on a scope and turning the tuning screws until the clock don't drift anymore.'

I knew immediately what he was talking about, you display the two signals side by side and when they are not exactly the same frequency one of them will appear to drift relative to the signal that the scope is triggering from. Yes he could have been a little more specific but it's a common technique that is the only sensible way to use a scope to adjust something to a precision reference.
 
The following users thanked this post: egonotto

Offline BillyO

  • Super Contributor
  • ***
  • Posts: 1848
  • Country: ca
Re: OCXO tuning screws not working on old equipment
« Reply #16 on: June 27, 2023, 09:57:34 pm »
There is nothing wrong with the drift method and with a modern DSO it's a snap to tune a 10MHz reference to 10^-10 in a few seconds.

I'd also be willing to bet that 99.999% of folks on this forum don't need any better than one part in 100 million.  You can do that without even having zoom on your scope in just a few seconds.  You can do that on an old Tek 465.
Bill  (Currently a Siglent fanboy)
--------------------------------------------------
 
The following users thanked this post: egonotto, tverbeure

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3516
  • Country: us
Re: OCXO tuning screws not working on old equipment
« Reply #17 on: June 28, 2023, 03:33:10 am »
The voltage required to set an OXCO to frequency varies as it ages.  As a consequence it can drift outside the range of the divider.   I bought 10 CTI OSC5A2B02 OXCOs for $30.  There's on the order  of a volt between the highest and lowest voltage required  to set them to 10 MHz.

When the divider is in the can it's difficult to fix.  The $20 PCBs that use the CTI OXCOs are remarkably good.  Better than the OXCO in my 5386A which is very hard to shift a small amount.  I have 2 that I was able to trim to ~1e-10 out of 4.  The other 2 will have to have the divider resistors changed.

If you put them in a temperature controlled enclosure you should be able to limit temperature effects.  The best precision, high resolution trimmer I know of is a $25 Vishay part with a 50 ppm/C tempco.  That means  that the trimmer can only be a small part of the divider with the rest being lower tempco fixed parts and should be paired with a resistor with the opposite.  The stability of the reference voltage is critical. Hence the value of a double oven.

Reg

 
The following users thanked this post: tverbeure

Offline tverbeureTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: us
    • tomverbeure.github.io
Re: OCXO tuning screws not working on old equipment
« Reply #18 on: June 28, 2023, 04:11:05 am »
Here's what the drift looks like for one 10MHz and one 9,999,999.97MHz signal. Deviation: 3x10-9. The scope is set to a pedestrian 10ns/div. The counter is set to 1s gate time.



The drift is obviously trivial to see and you see the changes of turning the trim screw instantly. Very quick to calibrate!

And here's what it looks like for one 10MHz and one 9,999,999.999MHz signal. Deviation: 1x10-10. (Whether or not the GPSDO is that accurate is a different story!)



The scope is now set to 1ns/div. The counter is set to 10s gate time. The drift is now much harder to see, but they still drive 1 div in 20 to 25 seconds. If all you need to see is whether or not the signal is where it needs to be, you don't need to wait a full diff: it's obvious after a few seconds that there's drift.

If you'd calibrate with the counter alone, it would have to wait up to 20 seconds after each trim.

Calibrating with a scope is just more convenient and faster.


« Last Edit: June 28, 2023, 05:49:56 am by tverbeure »
 
The following users thanked this post: egonotto

Offline tverbeureTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: us
    • tomverbeure.github.io
Re: OCXO tuning screws not working on old equipment
« Reply #19 on: June 28, 2023, 04:30:49 am »
The voltage required to set an OXCO to frequency varies as it ages. As a consequence it can drift outside the range of the divider.
Oh, definitely. Of course, even if it's out of range, you'd still expect the screw to at least do something!

Quote
I bought 10 CTI OSC5A2B02 OXCOs for $30.  There's on the order  of a volt between the highest and lowest voltage required  to set them to 10 MHz.
I see that I'm not the only one with a bunch of OSC5A2B02 and other OCXOs in my part box.  :)

Quote
If you put them in a temperature controlled enclosure you should be able to limit temperature effects.  The best precision, high resolution trimmer I know of is a $25 Vishay part with a 50 ppm/C tempco.  That means  that the trimmer can only be a small part of the divider with the rest being lower tempco fixed parts and should be paired with a resistor with the opposite.  The stability of the reference voltage is critical. Hence the value of a double oven.
I've seen others add temperature control to their enclosures. DIY, it's probably cheaper than a DOCXO.

I've been wondering why people don't use a precision voltage reference with PWM and 5 cent microcontroller or a DAC on their stand-alone OCXOs to avoid the hassle with trimpots. It's too costly for a OSC5A2B02, but it starts to make sense for more expensive ones like my eBay 8663. Maybe GPSDOs are just to cheap to justify it.

I also have an HP 5384A that was suspiciously stable for not having the high stability option. Here's what I found when I opened it up:



Note the AD780 voltage reference!

I bought that 5384A from the lead designer of the HP 5334A.
« Last Edit: June 28, 2023, 04:34:20 am by tverbeure »
 
The following users thanked this post: egonotto

Offline tverbeureTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: us
    • tomverbeure.github.io
Re: OCXO tuning screws not working on old equipment
« Reply #20 on: June 28, 2023, 05:33:48 am »
The HP 5334A mystery is solved!

I tried to trim the OCXO with the trim pot that's available in the back. Well, it turns out, it's not a trim pot, but a trim cap. And while its purpose is indeed to adjust the oscillator, it's there to adjust the XTAL oscillator... which is only present when you don't have the OCXO model!

Here's the schematic, the trim cap is C1.



And here's trim cap C1 on the PCB:



Right below (and under the white cable) is the socket for XTAL Y1... which isn't populated. Was it really *that* hard to also remove C1 to avoid confusion?

My unit has indeed the 10811 OCXO, with its own frequency adjust screw. When turning that, I'm able to fine tune the output frequency as expected.



It's a bit tricker than other trim pots: the act of inserting a metal screwdriver in the trim hole is sometimes (but not always) enough to influence the output frequency.

Still, the 5334A is now calibrated well enough for ~10-9 precision.

One final mystery was the EFC (electronic frequency control) input of the 10811. It turns that there's an additional trim pot on the 10811 assembly PCB, and that you're supposed to calibrate the OCXO that way. This second trim input is accessible through the back panel, but it was covered with a calibration sticker.  ::) And the other problem is: you need a really long and thin screw driver to reach it.



« Last Edit: June 28, 2023, 06:04:39 am by tverbeure »
 
The following users thanked this post: Dr. Frank, egonotto

Offline Wrenches of Death

  • Regular Contributor
  • *
  • Posts: 84
  • Country: us
Re: OCXO tuning screws not working on old equipment
« Reply #21 on: June 28, 2023, 02:55:27 pm »

It's a bit tricker than other trim pots: the act of inserting a metal screwdriver in the trim hole is sometimes (but not always) enough to influence the output frequency.


It's not a good idea to use a metal screwdriver. Not only can it detune the circuit while it's in the hole, it can short the oscillator circuitry to ground. A number of the HP canned oscillators are specifically labeled "USE INSULATED TUNING TOOL ONLY".

WoD

 
The following users thanked this post: egonotto, tverbeure

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3516
  • Country: us
Re: OCXO tuning screws not working on old equipment
« Reply #22 on: June 29, 2023, 02:14:29 am »
A wooden dowel with a 16 gauge aluminum blade epoxied in a slot in the end is easy to make and does a great job.
 

Offline BillyO

  • Super Contributor
  • ***
  • Posts: 1848
  • Country: ca
Re: OCXO tuning screws not working on old equipment
« Reply #23 on: June 29, 2023, 02:21:12 am »
Yes, yes.  A wooden stake through it's heart, that will liven it up! :-DD
Bill  (Currently a Siglent fanboy)
--------------------------------------------------
 
The following users thanked this post: egonotto

Offline trobbins

  • Frequent Contributor
  • **
  • Posts: 826
  • Country: au
Re: OCXO tuning screws not working on old equipment
« Reply #24 on: June 29, 2023, 04:17:14 am »
Some of the earlier HP OCXO modules (such as the HP 00105-6013) used a fibre/poly rod to interface between a users adjustment screwdriver and the actual oven assembly - and over time that rod can easily snap, leaving no easy way to provide the standard coarse adjustment of frequency.
 
The following users thanked this post: tverbeure


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf