HP 5065A Rubidium Standard Repair Log

[5065AGuru]

Tim,

If you adjust the 5Mhz output (in open loop) for 5.00000025 MHz that should give you a usable level of the 137Hz to see on the scope at A7 TP2. Just use a scope at 100mv/Cm.

Once you are able to see that signal (after adjustments) then you can adjust the frequency to zero offset (5.00000000Mhz) and observe the 2nd harmonic signal. You would like to be able to get at least 500mv P-P.

Intent is to get max 2nd harmonic level at A7TP2 when at frequency.

If nothing is discernible run the TED at 1 AMP for a few days while monitoring A7TP2. Hopefully you will see the 2nd harmonic slowly increasing in amplitude.

On the A3 turn the top pots R3 and R11 max CW and the side bias pot R40 and matching caps C59 and C61 for a max level at A7TP2. Start with R40 max CCW and turn it CW while looking for a signal. This usually will occur within the first half of the pots range. Also measure A3 TP2 for around 5VDC.

Cheers,

Corby

[trobbins]

Hi Corby.  I applied the deflooding current to the TEC for 3 days.  In loop open, the A7 TP2 spectrum signal was then tweaked for a null in 137Hz using the oscillator fine tune adjust and the 274Hz then tweaked for max level using the various A3 trimmers you indicated.  Some of the A3 trimmers are quite sensitive, although the REW spectrum analyser makes it easy to confirm trim peaks (rather than use the Meter 2nd harmonic level).  The maximum 274Hz signal level achieved was about 80mVrms (290mVpp), which relates to a 2nd Harmonic meter reading of 24, and the 137Hz signal was at least -50dB below that, and the 4th harmonic about 19dB down. 

Switching to Operate mode, and resetting the logic gives a Continuous Operation light, and the control and error levels remain at zero reading.  Interestingly the A7 TP2 spectrum shows the 137Hz level increases to only a few dB below the 274Hz level.  The oscillator fine tune adjust continues to allow the Control level to be moved either side of 0.  All the meter readings appear nominal although the lamp oven reading is still swinging slowly, and the cell oven reading is swinging very slowly, both of which you indicate would be related to the physical separation of the new heaters to their respective thermistors.

I have noticed the 2nd Harmonic meter level glitches higher every now and again.  I tried to identify if that was a coax connection issue, but that wasn't consistent.  A glitch doesn't align with any perturbation of the A7 TP2 signal level.  The owner did recall having to replace some leaking tants in a shielded assembly, so I will remove A1, A3, A7, and A13 assemblies and inspect internal parts, as the thought is it may be a supply rail bypass tant failing.

The green indicator is also turning off after some time, but can then be reset.  The instant of turn off seems to align with the lamp oven meter reading swing to its maximum (which is nearly 50) - given the A14 logic circuitry is 'instantaneous' for an out of range oven signal voltage, that would seem to imply that is a good place to focus on.  Looking at the A11 heater control circuitry, I'd anticipate that either the bridge circuit or the doubler rectifier circuit could be places to modify the open-loop response to try and attenuate the peak-peak swing level without hopefully changing the average level of control voltage that controls the heater.  At the moment the open loop meter swing period is about 10 seconds, so one thought could be to introduce a new RC after R26 (between Q6 base and the doubler positive rail) of say 10k and 220uF (2.2 sec time constant) with the risk that the 10k would shift the Q6 dc level and hence the lamp temp.  One hassle is the optical unit would need to be removed to allow a temp probe to assess if the lamp temp was shifted.  Any thoughts on this?

Ciao, Tim

[5065AGuru]

Tim,

You can turn the rear pot on A7 to set a new reference level for the 2nd harmonic meter position. Usually between 35 and 40.
Once set you can monitor the 2nd harmonic meter for another couple days. If the reading is stable you can reconnect the TED to its normal spot. If still increasing wait till it stops before reconnecting.

The older A7 actually nulls the 2nd harmonic meter reading at the center frequency and rises if you go either way off frequency. This is because it does not have a bandpass circuit in the 2nd harmonic channel.
I'd suspect the A7 for the meter glitches but since your 2nd harmonic level at TP2 is 290mVpp it could just be noise getting in. If you use a scope at A7TP2 what 2nd harmonic level do you see?

The operate lamp circuit will indeed turn the lamp off if the either oven is at max. This is normal.
One thing you can try to eliminate the oven oscillations is to connect a decade box (#1) in parallel across the pins on A11 for the lamp thermistor. At the same time pull A11, open one end of the lamp temp setting resistor. Connect two wires to bring it out to another decade box (#2) set to the original resistor value and reinstall A11. On the other decade box try starting at 2K and work down to maybe 820 Ohms. You will have to insert a thermocouple probe into the lamp end to monitor the lamp temperature and as you make changes to decade box #2 readjust decade box #1 to maintain the correct temperature. You want to be around 92 degrees C.

You can set the optical unit as shown in the pictures so you can work on the lamp end. Extenders are used for the Db9 connection, the lamp power, the cable to A7 and the TED wires. You can't use an extender for the cable to A3 but the original wire will reach.

Cheers,

Corby

[trobbins]

I found a heat damaged R48 (51 ohm) on the Y output of A7 when inspecting inside the A7 assembly.  Although the series connected 60uF tant C24 measured ok on my LCR, there was leakage going on when the assembly was powered on the bench and the dc voltage on output of IC2 was measured and Y was grounded.  Otherwise A7 voltages were nominal, so those two parts have been replaced.

I have restarted de-flooding - initial duration was 3 days, so as suggested I will monitor 2nd H meter level and see if it stabilises over coming few days - it is nearly 30 now (Continuous Operation).  So far there has been no 'glitching' behaviour so here's hoping that aspect has been solved.

As before, the heater reading on the meter is swinging and nearly reaching 50, but hasn't stopped continuous operation yet.  There may be a causal link to local ambient and that the top and bottom covers have been off all the time.  Once deflooding is finished, I'll put the covers back on and see if that suppresses the meter swing.

One additional thought is that if the heater meter swing is not adversely affecting stability performance, then the signal level to the A14 logic board could be a simpler target to fool by introducing an RC time constant that suppresses the peak-to-peak swing of the heater signal reaching the logic, but otherwise allows a continuous fault low or fault hi to annunciate itself.  That path doesn't then alter the pre-existing thermistor setting and regulation control.

[trobbins]

I continued the de-flooding process for a few more days and was able to get the A7 TP2 level of 2nd H up to 150mVrms (ie. circa 420mVpp).

I then returned the TEC wiring to nominal and put the 5065A in continuous operation - for about a day now.  After half a day I reset the 2nd harmonic meter level (A7 R19) to about 25, as it had slowly subsided to the point of the logic tripping out of Continuous Operation.  I've just checked the 2nd H level and it has reduced from 150mVrms to about 65mVrms, and that was once the various A3 trimpots had been optimised again for max 2nd H.  Perhaps that indicates that the de-flooding process still needs to continue?

With the 5065A in open loop, and the OCXO fine tune set for zero error and a null of the 137Hz signal in the spectrum on A7 TP2 (137Hz <-45dB below 273Hz), a comparison of the 5065A 5MHz output to my GPSDO shows negligible roll of down below 240 secs per 200ns full-cycle sine, so circa 200n/240 = 0.8 ppb, and below the 1ppb that the GPSDO is indicating it is within.

Corby, what I don't understand is that when I switch to continuous operation, the A7 TP2 spectrum shows that 137Hz signal rises to within a few dB of 273Hz, and control moves to -14 on the meter and the 5MHz shifts to about 0.01Hz over the GPSDO (one 200ns cycle roll in about 25 secs).  Any thoughts on how I can check why the control is pushing the OCXO off frequency ?

Ciao, Tim

[5065AGuru]

Tim,

If you de-flood too long you can overshoot!
If so once you hook the TED back to normal the signal  level will drop slowly till it stabilizes at the new level.
It does sound like you have an offset. Check the zeroing adjustments for A9.
Have you checked the phasing of the 137Hz? Try checking using this procedure.

Cheers,

Corby
 

[trobbins]

Corby, thanks for the procedure doc - much appreciated.

In open-loop I set the OCXO fine adjust to just get 273Hz on A7 TP2, although the signal level was only 100mVpp.  I went through the A3 adjustment procedure a few times but can only achieve a max A7 TP2 ac voltage of about 140mVpp, so nowhere near 500mVpp.

With the A8 sync adjustment, I had gone through that before but had not observed the simple half-sine waveform, but rather a higher frequency and noisier signal, but still showing the 'splitting'.  With respect to your notes, the fine freq adjust does vary the level of splitting and that splitting can be effectively nulled (although the phase pot is at full CW), but the 'signal' is now just noise (and A7 TP2 is effectively the same 273Hz amplitude of about 100 to 120mVpp).

With the loop gain setting procedure, I can only reach about -14 on meter error reading when fine tune is 50 units CW of centre frequency.  My OCXO doesn't have coarse adjust capability, so the fine tune setting can't be easily made '250'.  With the loop closed, the control voltage at the rear bnc is 0V at centre frequency, and goes to +0.8V at 50 units CW fine tune, and to -0.7V at 50 units CCW, so not quite +1 to -1V.  The error meter reading swings from -14 to +14 respectively.  With the fine tune back at 50 units CW, and loop operate switched on, the error falls to 0.

The next adjustment of meter 2nd harmonic level is ok, although the meter needle has some fine jitter of about +/-1 about 35 setting, and A7 TP1 is still about 100mVpp.

Control is now at +10 meter and +0.8Vdc rear bnc, and the 5065A 5Mhz output is now within 1ppb of the GPSDO.  So no 0.01Hz difference that was observed a few days ago, and on review that seems to be a result of not moving the OCXO fine tune to 50 units CW of centre frequency. 

Any thoughts on why the 2nd H level from A7 TP2 is relatively low?  Perhaps I should wait a week or two and confirm if that level changes. 

I don't have access to a frequency standard but will now try and see if the scope phase shift of 5065A to GPSDO can be minimised using the magnetic field adjustment 10 turn pot, given the op manual indicates that adjustment has a nominal 2ppb span (increasing the pot increases the 5065A frequency).

Ciao, Tim

[trobbins]

Well I'll be damned !  Now at 400mVpp 273Hz on A7 TP2 - I even had to reset the 2nd H meter level as it was over-ranging. 

A possible catalyst was that I had to turn the unit off to repair the magnetic coil current supply (failed A15 Q6 dual npn) for a few hours.  So with a goodly level of 2nd H I have just been setting the magnetic 'clock' for negligible drift against the GPSDO - down to a level where the GPSDO tracking jitter is circa 0.1ppb (https://www.eevblog.com/forum/projects/budget-gpsdo-a-work-in-progress/50/. 

So I think I am pretty much done with my repair log - touch wood.  Thanks so much to Corby for the tireless assistance, and to Stanley for helping with the heater wire supply.

Ciao, Tim

[HighPrecision]

A possible catalyst was that I had to turn the unit off to repair the magnetic coil current supply (failed A15 Q6 dual npn) for a few hours.

Same failure fixed on my unit:
https://www.eevblog.com/forum/metrology/inspecting-aligning-and-testing-an-hp-5065a-rubidium/msg3271248/#msg3271248

Best, Stephano

[trobbins]

Ta Stephano - I breezed through that thread at the start of the journey, but one then gets blinkered by the immediate faults being faced and forgets to sweep through all the threads again - and yes it makes sense that the correct line wasn't being used given the abnormal scope waveform I had when checking phasing.  In this case the repairs were being done iteratively as functions were being checked, and the magnetic field repair was the last function to check and repair.  And as the ops manual points out, the loss of magnetic field is not monitored by the Continuous operation indicator and logic.  I really should go back and redo all the setup procedure again, but that can be for another day by the owner to enjoy 

[Whaleback]

Hi folks. There was a recent group of 5065As that surfaced on the auction site all from the same seller.
I gambled and got the first one listed with option 3 in "parts or not working condition". It arrived and I started meticulously cataloging its condition.

The serial number is 0956A00250. It looks like it was reconditioned at some point, as it has a 10811 OCXO, the digital clock/PPS option, and what appears to be a later era RVFR.

The first problem is that it is missing the A13 buffer assembly. So I have no way to close the servo loop if the physics package is still ok. Is that something I can jumper over (A3J3 direct to A1J2) temporarily?

I removed all the circuit cards (they're all there). Unbelievably, there's no corrosive cap leakage on any of them. I will be replacing all the recommended devices anyway (electrolytic caps, hot diodes, etc).
The A12 RVFR looks like it is a replacement. The housing is unpainted, and it carries an 05065-6071 part number as well as a 1220A02147 serial number.
I ohm'ed out the DB9 and and the heaters were both 50 ohm, the thermistors both 2k, and the C-field coil was 11 ohms. No heater short the the chassis.
I opened it up to find someone's greasy thumbprint on the filter cell that I'll clean up with an ethanol prep pad.
I put 20V across the feedthru and the lamp module chassis, and it drew almost no current.

I was sad.   :'(

Turns out the A12A1 R1 1.33k resistor was open. I snipped that out and soldered a temporary 1.3k across the PCB lands and it lit at nominal start and run current.

So now that I know I have a unit that potentially is recoverable, I thought I'd log the bring up here? Should I start a new thread or append here?

I also have a few items I need to source, primarily that A13 buffer module. I'm hoping some on list or over at Time-Nuts has one in a dead chassis I can buy or beg?
Also looking for a complete set of side panels and handle panels, it's missing all 6. They're PNs 5000-0738, -0739, -0766.
Does anyone know the part numbers for the spring metal clips that hold top and bottom cover fasteners on? I need 8 of those and the pozidriv screws. I can't seem to find a master HW part list for old HP gear.

Finally , I really want to say thanks to Corby Dawson, Ed Palmer, maat, Timeok, and everyone who's dropped 5065A restoration knowledge here.

[KE5FX]

Very cool.  I saw that listing -- guessing it was surplussed by Boeing given the seller's location and the other stuff he's listing.  Nice to hear it has an updated RVFR, 10811, and a good home.  I had just bought one from another seller for beaucoup bucks so I didn't feel like jumping into the fray for this one. 

If anyone is still looking for a unit in decent shape, this one is listed by the same seller I just bought from.  It's pretty spendy.  No guarantees, obviously, but the only things that were actually wrong with the one I bought were a missing indicator lamp and a pair of open 20 uF caps on the RVFR heater controller board.  Fortunately the RVFR heaters were OK.  I think it has the best 10811-60109 I've ever seen; it does about 2.5E-13 @ t=1s in open-loop mode.

[Whaleback]

Hi John - thanks for responding. Yeah it was this one. Same seller.
I did find a donor chassis from a 8616A.
Does anyone know the difference between the -6001 and the -6071 physics packages? - I see - from the manual it's a replacement unit.

[Whaleback]

Lamp photo - temporary 2k6||2k6 installed at R1.

[Whaleback]

Oven controller and PSU .
PSU has some oxidation on the bridge diodes. One of the diodes is suspect as didn't measure correctly on the meter.

I'm looking for recommendations on the wet tants across all the cards. I'm thinking about just replacing all of them in every card/assy.
Do I use 'modern' dry pill tants? Nichicon/Chemicon radial electrolytics? All sources of high quality axial al el caps appear to have been retired.

I don't plan on ever returning the battery back up to this unit. I'm planning on chopping out all the components on the charger board except for the ext/int DC or-ing diodes.

[trobbins]

FWIW, the early version of 5065A I repaired had only a few caps identified as faulty and replaced, and I did check a few given that they were plausible fault initiators.  In summary:
05065-6024 RVFR temp controller A11: C23 and C24 had both gone bad (low uF and high R).
05065A-6010 Signal amp A7: C24 (60uF tant) appears to be leaky when grounding Y.  C18 had already been replaced.

So in general, I wouldn't immediately replace all tants unless the first few you suspect due to a known fault, end up measuring bad.  But that is just my view from a single sample of equipment.

[coppercone2]

can the physics package of one of these be improved by modern insulators like aerogel ?

[KE5FX]

Oven controller and PSU .
PSU has some oxidation on the bridge diodes. One of the diodes is suspect as didn't measure correctly on the meter.

I'm looking for recommendations on the wet tants across all the cards. I'm thinking about just replacing all of them in every card/assy.
Do I use 'modern' dry pill tants? Nichicon/Chemicon radial electrolytics? All sources of high quality axial al el caps appear to have been retired.

I don't plan on ever returning the battery back up to this unit. I'm planning on chopping out all the components on the charger board except for the ext/int DC or-ing diodes.

You should be OK replacing the tantalums with modern 105C aluminum electrolytics.  I don't think it's worthwhile to declare a general jihad against all of the hermetic parts, though -- just do visual and ESR checks and replace any that look suspicious. 

I do replace all of the 10 dry non-hermetic 100 uF parts when overhauling one of these, plus the two 20 uF parts on the oven control board. 

[Whaleback]

Thanks for the useful guidance guys.
So my plan is to replace all the caps that Tim and John have  mentioned, as well as some rectifier doides, and after checking, some other aforementioned devices that are known to be suspect such as C-field Q6, C5A/B, the chassis electrolytics, etc.

My initial steps will be to test the main PSU alone on the bench, fix any deficiencies, install it and run only the 10811 initially while measuring it against my SRS FS740 to check its health. I am aware the adev will be masked by the FS740s internal PRS10.

NOTE: SRS FS740 works with timelab. I have matlab or octave stream data off it to a file, then run timelab in 'Acquire live data file' mode. I'm happy to share the matlab script that does this. Anyone can code this up in python. It was just quickest for me to do in matlab.

I built one of Corby's DMTD boards, haven't had time to integrate it in front of 740. This OCXO check is only looking for 'gross defect' behavior.

Next I'll reinstall the oven controller and the RVFR and test the oven control. I am planning on installing a thermal fuse.

Hopefully soon after that I'll have found an A13 module, and can start looking at closing the phase servo loop and continue down the path to a green continuous operation lamp.

[Whaleback]

Pics of bad caps. Some were ok, many were in the puffs and hundreds of ohms. I found a bad 1N4998 on the A15 PSU card so I pulled them all. I'll be jumpering over some of the diodes there and putting a bridge down on the chassis bottom as I see the newer revisions have. I haven't cracked A1/A3/A7, but bought replacement caps for them with intent.

Couple of questions if anyone can answer:

My RVFR tube is missing the fiber washer(s) I read about here. It sounds like there is only one? What is the intent of that washer? It's not thermal or electrical isolation right? There isn't one on the nut clinching _inside_ the outer shield, just direct metal-to-metal contact there? Can someone outline the fastener stack-up there so I can get it right? Is the fiber washer just that thick red kind-of cardstock material? Cork rubber?Does it need to be anything special? Can I use a delrin washer? Does it just provide drag to allow the nuts to clinch?

Does anyone have a PDF of a manual more recent that the one i see all over for series 1908 in 1979? I bought the Artek copy, but its series 940.


 

[Whaleback]

More debugging notes.

Recapped all the 100u, 20u and a bunch of other caps. Replaced the oring diodes on the batt controller. Stripped the remaining circuits off the batt controller.
Inserted A15 regulator and surprise - _no regulation_.
After some careful analysis and poking around determined n-ch JFET Q4 was bad. Replaced with what I had on hand which was 2N4393. Not an ideal replacement, but it worked.
Now have A15 regulating. Also have 2N4416A fets on order for proper cross replacement for that part.

Installed thermocouple into RVFR. plugged A11 oven controller in as well at P15/J15 of RVFR.
Monitored cell temp first - over appx 40 mins it rose and settled to 67.7C.
Drew thermocouple out of oven until reaching new max temp at lamp oven of 90.1C.
Temp regulation looked tight.

5 and 1 MHz outputs looked pretty good. 100 kHz output looked terrible. High harmonics and subharmonic. Will be ejecting both dividers and all of the clock display HW unit later.
Next steps are review performance of synthesizer, multiplier, and sig amp.
Hopefully after that run, deflood, and adjust locking signals.

[5065AGuru]

The correct stackup for the threaded lamp post is:

Once the first two shield ends are in place install the first threaded nut. I just look across the edge of the outer shield and position the nut so it just protruding. Then install the outer shield. With the shield pressed into place firmly the inner nut should just be contacting the outer shield. The fiber washer is then put on and the outside nut screwed firmly in place. The intent is to make sure the outer shield contacts the inner nut and after tightening against the outer shield functions as a heat sink. The fiber washer is to allow a little play with temperature. Fiber, card stock or nylon is fine.

Cheers,

Corby

[Whaleback]

Thanks for that info Corby!

So my current progress has found & replaced a failed 2nd harmonic buffer NPN - Q9 on the 6010 version A7 board. I replaced that and started the alignment.
Progressed through the alignment, and everything worked out fine down to the phase detector board.
I get a nice rectified sine at A8TP3 testpoint which as I deviate the fine frequency control switches sign after passing through a min amplitude as I cross the on-frequency point. Amplitudes are good down the chain, and phase splitting is symmetric.

The issue I'm having at the moment is after switching into operate.
When I dial the fine to on-frequency (null and noise only at the phase detector), and then switch into servo mode, the integrator swings and pins the meter immediately to the -50 end of its range.
The output at A8TP3 shows the integrator has tuned the OCXO way away from the null.
 
I have not yet started debugging this board, aside from checking diodes and BJTs. Those devices all seem ok. I haven't checked the dual-FET yet, or the ADI op-amp. I haven't checked the coarse or fine tuning of the integrator.
I also haven't checked the neg EFC control on the 10811.

If anyone has thoughts I'd love to hear them.

I guess I should run through 5-29, 5-30, and 5-31 procedures before asking for additional guidance.

[5065AGuru]

In loop open monitor the error meter reading.
On frequency it will be zero.
Move fine freq. CCW  and frequency on a counter will go down and the error meter will move to the + side.
Move fine freq. CW and the frequency on a counter will go up and the error meter will move the - side.

If not make sure the plus and minus EFC jacks are not reversed.

Cheers,

Corby

[bnz]

I have not found the schematics, or at least the manual changes and list of replaceable parts, for the later A3 Multiplier Assay 05065-6106 (Series 2644) in any manual on internet up to now.    Does anybody has seen that information somewhere?