Author Topic: HP5386A and GPSDO - how long to wait?  (Read 6818 times)

0 Members and 1 Guest are viewing this topic.

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
HP5386A and GPSDO - how long to wait?
« on: September 30, 2021, 05:41:07 pm »
I was avoiding the time domain in my home lab for a very long time  ::), happy with the same level of standard accuracy as I have for resistance and voltage - about 10ppm. However now I've crossed the line and bought an old frequency counter - the HP5386A Option 004 and as an afterthought a GPSDO - the "MiniGPS Reference Clock" from Leo Bodnar. The counter has been calibrated by Keysight in May 2017 and most likely was sitting without power for a long time - some months if not years. After I've switched it on and connected it to the Mini Clock it did show a noticeable drift of the OCXO down over period of six hours since midday today. The drift is exponentially slows. So the question is - how long should I wait for the OCXO to stabilise before I attempt to correct it so it matches the GPSDO? A day? A week? A month? The error at the moment is ~1.4Hz from 10MHz or 0.14ppm. It is not that much over 4 years since the last calibration (assuming it was more accurate at that time). Should I bother at all adjusting the OCXO, providing I can use the GPSDO as a reference clock for the counter? Any suggestions on the preventive maintenance for it? I rather like the HP5386A and it is in a very good shape overall.

Cheers

Alex
« Last Edit: September 30, 2021, 05:52:30 pm by Alex Nikitin »
 

Offline bob91343

  • Super Contributor
  • ***
  • Posts: 2675
  • Country: us
Re: HP5386A and GPSDO - how long to wait?
« Reply #1 on: September 30, 2021, 06:19:48 pm »
It depends.  If you plot the error against time it will give you a clue as to the point at which is will be stable enough for your readings to be trusted.  Add to that the aging factor which means it will never stop changing.

If you have a 10 digit readout, it will be within 100 ppb if you are within one count.

I have a rubidium standard and a counter with crystal oven and a signal generator with crystal oven.  The latter two units are alway plugged in so the crystals are always warm.  When making comparisons, the counter is almost spot on when I switch it on (because its master oscillator is always running) but the generator takes about 45 minutes to settle (because its master oscillator only runs when it's switched on).  The generator error runs a few Hz at around 500 MHz, or close enough for most work.  If I need it better, I just wait an hour or so and it's as good as it's going to get.
 
The following users thanked this post: felixd

Offline mendip_discovery

  • Frequent Contributor
  • **
  • Posts: 824
  • Country: gb
Re: HP5386A and GPSDO - how long to wait?
« Reply #2 on: September 30, 2021, 06:51:27 pm »
Quote
Should I bother at all adjusting the OCXO, providing I can use the GPSDO as a reference clock for the counter?

I would say use the GPSDO, its uncertainty is going to be better. I have yet to find out what the Uncertainty is of the GPS standards but from what I have read it is good.
Motorcyclist, Nerd, and I work in a Calibration Lab :-)
--
So everyone is clear, Calibration = Taking Measurement against a known source, Verification = Checking Calibration against Specification, Adjustment = Adjusting the unit to be within specifications.
 

Offline AndrewBCN

  • Frequent Contributor
  • **
  • Posts: 571
  • Country: fr
Re: HP5386A and GPSDO - how long to wait?
« Reply #3 on: October 02, 2021, 04:33:49 pm »
Should I bother at all adjusting the OCXO, providing I can use the GPSDO as a reference clock for the counter?

You have two OCXOs here:
1. The OCXO inside your HP5386A, which you are wondering if you should adjust or not.

2. The OCXO inside the GPSDO, which is continually adjusted "disciplined" by the GPS signal.

If you want to keep your GPSDO continuously powered and connected to your frequency counter, you don't need to adjust the OCXO inside your HP5386A.

Otherwise, if you want to use your HP5386A for high precision frequency measurements without a GPSDO attached to it, you can adjust calibrate the OCXO inside the HP5386A once every three or six months using the GPSDO, it shouldn't take more than 15 minutes.
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #4 on: October 02, 2021, 06:56:17 pm »
Thank you guys, I am learning this stuff as I go along. The MiniGPS Reference Clock I use doesn't have an OCXO, only a TCXO. I have, however, another, rather underused unit with an OCXO, the Marconi 2022C signal generator, which, as I've discovered now, was about 2ppm off (after at least 20 years since the last adjustment/check, possibly more as I've bought it broken and fixed it about that time). I've adjusted it and it appears to be more stable than the OCXO in the counter (at least for now). If I use the GPSDO as a reference for the counter, the drift of the Marconi is below 1ppb in an hour, around 5ppb overnight with a few degrees change in room temperature.

Cheers

Alex

 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: HP5386A and GPSDO - how long to wait?
« Reply #5 on: October 03, 2021, 04:33:19 am »
" how long should I wait ..." 
my personal experience - about 2 weeks till OCXO drift became uncountable.
for hp 5386a 11 digits ,   i would wait at least 3 weeks .... 
the problem , OCXO  should run 100% - when it off for some period of days ,   you need to wait again a few weeks, but ok for 8 digits in few hours ..

GPS with OCXO, approx 2-3 days for -10, around a week for -11 -12 ; with TCXO never use it ... you may try a bit to isolate in some styrofoam, for a day or so ..
(fun stuff , bought one too ,  would be some time when it arrive,  seems 3Ghz input sensitive to damage , no more then 5V )
 

Offline guenthert

  • Frequent Contributor
  • **
  • Posts: 706
  • Country: de
Re: HP5386A and GPSDO - how long to wait?
« Reply #6 on: October 04, 2021, 04:33:09 pm »
[..]
If you have a 10 digit readout, it will be within 100 ppb if you are within one count.
[..]
      Surely you meant 100ppt (the british use the short scale as well, don't they?), i.e. 1e-10 or 0.1ppb.  Otherwise, I think above statement might need some expansion.  ;)
 

Offline bob91343

  • Super Contributor
  • ***
  • Posts: 2675
  • Country: us
Re: HP5386A and GPSDO - how long to wait?
« Reply #7 on: October 04, 2021, 05:09:26 pm »
Sorry if my mental math was flawed.  I didn't get it quite right, I guess.
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #8 on: October 04, 2021, 11:09:31 pm »
The OCXO in the HP5386A is settling down, right now I run it with the GPSDO as a reference, having fun in some unusual ways  :palm: .

Here is a warmup frequency variation of the Marconi 2022C switched on from cold (it defaults to 1MHz).

Cheers

Alex
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #9 on: October 05, 2021, 12:16:04 pm »
The HP5386A measuring the GPSDO Mini Clock 10MHz for 4 hours this morning. I can probably live with the 15ppb offset...

Cheers

Alex

P.S.

the problem , OCXO  should run 100% - when it off for some period of days ,   you need to wait again a few weeks, but ok for 8 digits in few hours ..

The HP5386A runs the OCXO all the time, it doesn't have a power switch, only "Stand By".
« Last Edit: October 05, 2021, 12:21:34 pm by Alex Nikitin »
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #10 on: October 08, 2021, 06:50:59 pm »
Well, I've attempted a re-adjustment of the OCXO 2 days ago, "nudged" it by about 8-10ppb, here is the result over last 48 hours after the unit settled. It looks like the variations are mostly room temperature related with about -2 ... -2.5ppb/C tempco. Is that a reasonable figure for an old OCXO? The manual states +/-100ppb stability for 0C to +50C range, referenced to +25C, or +/- 4ppb/C max.

Cheers

Alex
« Last Edit: October 08, 2021, 07:05:51 pm by Alex Nikitin »
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: HP5386A and GPSDO - how long to wait?
« Reply #11 on: October 14, 2021, 07:02:22 pm »
on back of meter 10M output, you can feed scope ch1 and gps -> Ch2 , you will see phase shifting, adjusting HP you can set it precise,  increasing sweep for more precision. 

im not sure how GPS precise, due to cheapest are just a basic counter in 1sec interval (PPS). 

i have 2 thunderbolt recycling gps with modded, additionally thermally isolated crystal , it suppose to go around -13 after the week of running. but can't measure.

had tp play with cheap ocxo , heavily thermo-shilded with manual adjustment, it extremely stable after a week for a short period (week or so), and you able to see how GPS adjusting frequency back and forward. using a scope method

thermo cycle changing crystal frequency, ideally permanently on state, so you number may be off when off\on alter long period
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #12 on: October 15, 2021, 10:18:04 am »
Thank you. I am now trying to work out what is the main reason for the daily frequency variations of ~10ppb. It might be the GPSDO unit or it might be the OCXO in the HP5386A. How much frequency variations should I expect from the TCXO based GPSDO? I've received today a separate OCXO, specified at 10ppb initial tolerance and +/-3ppb stability over its temperature range. It is a voltage controlled unit though and the open circuit control voltage is not exactly 2.5V . I will put it in a separate enclosure and run against the GPSDO Mini-Clock. It is a slippery slope  :palm: .

Cheers

Alex
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: HP5386A and GPSDO - how long to wait?
« Reply #13 on: October 15, 2021, 09:39:48 pm »
really no idea about TXCO, in my understanding it how well circuit do adjustment,  the meter do phase calculation as well, a reason of higher precision. so jitter of base a paramount for last digits.

i did a basic lt1021-5 DIP8 as for voltage ref. for OCXO and multi-turn for adjustment.   everything, LT and OCXO was in 8cm thick Styrofoam box. - like it , very simple and stable. LT and OXCO case in direct contact for thermal stability
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #14 on: October 18, 2021, 02:19:21 pm »
It looks like the old OCXO in the HP5386A is to blame for the variations. The new OCXO looks very stable, started from about 1ppb/h drift and overnight reduced to <0.2ppb/h, no up/down changes. Perhaps I should upgrade the OCXO in the counter  ;) .

Cheers

Alex
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #15 on: October 21, 2021, 01:58:23 pm »
The new OCXO is slowly settling down, after four days the variations overnight were below 1ppb, here is the latest result. It looks like a reasonable performance from a mid-range OCXO, I might transfer it into the counter (with an externally accessible adjustment). The old OCXO might benefit from an overhaul as its performance is not as good as I expected. This data also confirm that the MiniGPS Referenced Clock has a good long-term stability when locked to GPS however the output varies a bit if I touch the case long enough to affect it's temperature before correcting itself (see the second graph, the coloured band shows the time when MiniGPS Clock is held by hand, dropping the case temperature from 40C to ~35C) .

Cheers

Alex
« Last Edit: October 21, 2021, 04:24:02 pm by Alex Nikitin »
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: HP5386A and GPSDO - how long to wait?
« Reply #16 on: October 21, 2021, 10:10:32 pm »
lowering jitter the cornerstone, for phase precision measurement i would suggest an additional thermal isolation to OCXO and GPS crystal
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #17 on: October 21, 2021, 10:44:35 pm »
I am not so sure a thermal isolation is the answer. It is not very useful for the OCXO as its internal thermal regulation loop would suffer and it is certainly won't be good for the TCXO in the GPS Clock as it is quite warm already and thermal isolation would only increase the case temperature. I was thinking about a substantial heatsink to the GPS Clock, to reduce the speed of temperature variations so the PLL can catch up and it would also be healthy to lower the working temperature of the unit.

Cheers

Alex

P.S. - it is necessary however to reduce any excessive air flow by screening, leaving only a natural convection cooling.
« Last Edit: October 21, 2021, 11:04:52 pm by Alex Nikitin »
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: HP5386A and GPSDO - how long to wait?
« Reply #18 on: October 21, 2021, 11:35:54 pm »
how the thermal control loop acting, for it would be indifferent how isolated the ocxo, goal to reach the desire temp, better isolation, less fluctuation , more precise in goal achievement. you may actively heat the unit and control loop will be off, as goal reached ...

as i said my SC cut OCXO running in 10cm styrofoam box , an excellent result for a short\medium  therm jitter , without box it consumed around 120mA, inside just 25mA at 12V. that basically parasitic heat dissipation.

you may look at gps , a voltage control level ,  usually OCXO control 2-3V, and replace tcxo -> ocxo + insulation.  that would be a maximum the you can do  at this state.
« Last Edit: October 21, 2021, 11:41:09 pm by GigaJoe »
 

Offline MIS42N

  • Frequent Contributor
  • **
  • Posts: 510
  • Country: au
Re: HP5386A and GPSDO - how long to wait?
« Reply #19 on: October 28, 2021, 10:54:10 am »
I just looked at the specification of HP5386A Option 004, it says ageing rate < 3 x 10E-8 per month after 30 days of continuous operation. I think that gives a clue as to how long HP expects you to wait before calibrating. It is assumed the device is never turned off. There is always ageing, the crystal sheds atoms changing its geometry (the same problem as beset the platinum 1kg standard). Ageing usually becomes consistent after a month or three. So depending on the requirement, a calibration schedule is desirable. Alternatively, if the ageing rate is known, and the time since last calibration is known, it is possible to come up with a fudge factor to apply to a reading so delaying the need for another calibration. 

For accurate measurements, the GPSDO should be the preferred reference. Its crystal or TCXO is also ageing but the disciplining compensates. It is not unreasonable to expect frequency to be within 1 x 10E-10 of accurate. This is a drift of 1 cycle over 1000 seconds so a short measurement period like 10 seconds would show less than 4 degrees of phase change which may not be enough for accurate comparison. So long comparison times are desirable.

Most hobby GPSDOs oscillator, the control voltage is derived from a 16 bit DAC. A quality OCXO can have a pulling sensitivity of 1Hz/V for a control voltage of 5V. This allows the control voltage to set the oscillator to better than 10E-11 of the desired frequency. Some oscillators have a higher sensitivity to the pulling voltage so may not be able to go better than 10E-10 in the short term.

Because most GPSDOs use a PLL, the longer they are run the closer the average frequency gets to the satellite provided frequency, which is itself derived from NIST (for the original GPS constellation).
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #20 on: October 28, 2021, 11:09:42 am »
how the thermal control loop acting, for it would be indifferent how isolated the ocxo, goal to reach the desire temp, better isolation, less fluctuation , more precise in goal achievement. you may actively heat the unit and control loop will be off, as goal reached ...

as i said my SC cut OCXO running in 10cm styrofoam box , an excellent result for a short\medium  therm jitter , without box it consumed around 120mA, inside just 25mA at 12V. that basically parasitic heat dissipation.

 :palm:

Do you understand that in the situation you have described ("basically parasitic heat dissipation"), the control over temperature is lost? The temperature might be stable over a period of time for a given ambient, but it is not controlled.

Cheers

Alex

 

Offline mendip_discovery

  • Frequent Contributor
  • **
  • Posts: 824
  • Country: gb
Re: HP5386A and GPSDO - how long to wait?
« Reply #21 on: October 28, 2021, 06:55:36 pm »
Do you understand that in the situation you have described ("basically parasitic heat dissipation"), the control over temperature is lost? The temperature might be stable over a period of time for a given ambient, but it is not controlled.

The OCXO is heated to ~75°C, my basic understanding I heating elements turn on for x wait for y, sense temp, wait for the temp to drop, apply heat for x etc until the temp is close to desired. With better insulation, the temp is easier to achieve and stabilise in the 22°C exterior conditions. Unless the exterior temp is higher than OCXO running temp there might not be much of a need to dissipate heat. The only other issue I could see is if you insulated really well the heat for x might be enough to overheat.
Motorcyclist, Nerd, and I work in a Calibration Lab :-)
--
So everyone is clear, Calibration = Taking Measurement against a known source, Verification = Checking Calibration against Specification, Adjustment = Adjusting the unit to be within specifications.
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: ca
Re: HP5386A and GPSDO - how long to wait?
« Reply #22 on: October 28, 2021, 11:03:59 pm »
control loop for heating linear, basically opamp to compare 2 points, when it reach equilibrium amount of energy to produce equal to dissipate. if you put crystal in ideal thermal chamber - it will overheat, as a heating loop will off completely, but some mW energy consumed by circuit itself will overheat it eventually.   

OCXO in overall a large metal box with a lot of surface to dissipate, and a Styrofoam box, it not an ideal thermal chamber.  as I said it do consume energy to heat itself but X10 less - it called parasitic heat dissipation.

So, better isolation reach 2 targets: less energy waste, and increase precision of temperature, as the loop acting in much much less deviation range of heat production.   (as an example look at old PIEZO , https://i.ebayimg.com/images/g/TEwAAOSwhbdfWfnl/s-l1600.jpg )

so im curious about the reason why: "the control over temperature is lost"  Like it will produce a heat till it meltdown itself ? can we call it thermostat then, or better say it propane grill ?
 
 

Online Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1150
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: HP5386A and GPSDO - how long to wait?
« Reply #23 on: October 28, 2021, 11:20:24 pm »
Just do a very simple experiment: switch the OCXO on from cold and look for how long would it take to reach, say, 10ppb (or even 100ppb) corridor with and without thermal isolation. Something similar to what I did measure for the Marconi generator earlier.

Cheers

Alex
« Last Edit: October 28, 2021, 11:23:41 pm by Alex Nikitin »
 

Offline FriedLogic

  • Regular Contributor
  • *
  • Posts: 115
  • Country: gb
Re: HP5386A and GPSDO - how long to wait?
« Reply #24 on: October 29, 2021, 07:27:36 am »
So, better isolation reach 2 targets: less energy waste, and increase precision of temperature, as the loop acting in much much less deviation range of heat production.   (as an example look at old PIEZO , https://i.ebayimg.com/images/g/TEwAAOSwhbdfWfnl/s-l1600.jpg )

so im curious about the reason why: "the control over temperature is lost"  Like it will produce a heat till it meltdown itself ? can we call it thermostat then, or better say it propane grill ?
 

  An OXCO will draw a certain amount of current even if the heater is at minimum, so with enough insulation this can take it to the required temperature or beyond, and so temperature control can be lost. It does not sound like the OCXO mentioned here has lost temperature control yet, but with the current down from 120mA to 25mA, but there may not be far to go.
  One problem with extra insulation is that the temperature coefficient can be poorer near the temperature limits, so improvements in temperature gradients may not translate as expected into improvements in performance.
  What tends to bother me more about using a lot of insulation like this is what might happen if the temperature controller fails and the OXCO gets really hot.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf