HP 5065A Rubidium Standard Repair Log

[TheSteve]

Every time I see an update to a 5065A thread I want one just a little bit more...

The last one that sold on fleabay went for C$6500.  How deep are your pockets? 

I saw that. I'd be more interested in a broken unit to rebuild - hopefully a little cheaper. I do have spare 3458A's, maybe someone will want to trade some day.

[edpalmer42]

I've always considered an Efratom FRK / M100 or Stanford Research PRS-10 to be a good second choice to the 5065A.

[trobbins]

Arrrgh, testing on my instrument now shows two new issues. 

Firstly the cell oven heater circuit now appears to have a fault -  exhibiting itself as a zero meter reading when cell oven was selected and power applied to the instrument - this wasn't displayed as faulty during earlier testing.  Further testing showed the heater path measures 52 ohm cold, but effective resistance falls when voltage across it is increased (using a separate supply feeding just that heater).  This appears to be a more onerous fault to fix, as there are no photos or posts on that process, and a risk of disturbing the microwave cavity (?) and then needing to retune that (from post #14 above).

The second fault appeared once the lamp oven had reached regulation temperature.  The lamp circuit +20V feed passes through inductor L2 on the supply regulator pcb A15, and on inspection that part had burnt/cracked - once L2 was replaced the A15 board was checked and presented as ok.   The +20V feed current to the lamp circuit (A12A1 spectral lamp oscillator board) was then monitored and a low resistance circuit appeared once oven temp started regulating.  The main 2N3553 Q2 had been previously replaced/repaired as part of an initial repair effort a few years ago by the owner, so I guess that transistor, along with the parts in series with it (that would have also been stressed from a low resistance circuit path through Q2) will need inspection/replacement.

So one step forward, but two steps back 

[trobbins]

Corby, or anyone who has repaired Q2 (2N3553 in TO39 can) in the lamp oscillator assembly, I'd appreciate any comment on how the TO39 fitted into the Thermalloy 1101A heatsink fitting and then to the oven aluminium oven cover plate.  When I opened up my Q2 heatsink, the 2N3553 can wasn't insulated from the 1101A in any manner except for thermal grease, and the 1101A stud wasn't insulated from the aluminium cover plate of the lamp assembly, so I'm presuming the observed low resistance fault was from the TO39/1101A to oven ground due to the previous repair not being properly finished.  The 2n3553 itself tests ok, and even tests ok at up to 95degC (in a thermal chamber), but I was just using a simple uP transistor checker. The other parts test ok, even though they look very heat stressed.

Q2 and 1101A assembly, with plastic spacer used between assembly and pcb.

[5065AGuru]

Tim,

The original fitting was done without any added insulation.

The heat sink anodizing was the insulation.

I have seen later lamps that use an insulating washer that fits into the hole to center the stud on the outside of the lamp similar to washers used on power transistors, some are nylon or fiber and some are thin black anodized aluminum.

You need to grind off the emitter tab to make it smooth and flush or when you tighten down the hex ring the tab will dig in and short out to the stud.

When I have to replace one these days I don't use heat sink grease on the stud. I use a silpad insulating heat sink pad on the inside and the mentioned insulating washer on the outside.

Make sure you tighten down the stud and make sure the leads engage the board properly first before soldering the transistor in. Tightening the stud with the transistor soldered in can stress the leads!

If the1.3K resistor is original you should replace it as they will eventually fail. I use a 1 or 2W MOX 1.3K 2%.

Cheers,

Corby

Cheers,

Corby

[trobbins]

Thanks for response Corby. 

Yes the TO39 can emitter tab was intact, so likely was the presumed point of electrical interconnection arising from thermal expansion (given the low resistance path did not exhibit itself initially).  I'll remove that tab, and install a typical power device stud mount insulation scheme, and swap out that resistor, refit the lamp assembly and confirm all is then good at that end of the optical assembly.

Do you have any comment on what to expect at the other end of the assembly if I remove the outer metal cover cap?  I can anticipate that accessing the cell end of the assembly is fraught with risk, and even a technical inability to recover the cell performance needed for adequate operation, but I have hit a brick wall with this restoration otherwise.

Ciao, Tim

[5065AGuru]

Tim,

Accessing the cell end is not a big deal.

However I'd wait until you get the lamp running. With the lamp working and the cell and lamp ovens up to temp hopefully you will be able to see on a scope (A7 TP2) some 2nd harmonic signal when "on frequency". If you can see even a tiny bit then you can try optimizing the A3. Turn the top two pots max CW and then see if you can get more by tweaking the two matching capacitors and the varactor bias pot. on the side. If you see 100Mv P-P or more on TP2 then turn the rear pot on A7 to give you 30 on the 2nd harmonic meter reading. Then log the reading daily for a week and see if it's slowly increasing. If so let it run until it stabilizes. At that point if you have >700-800Mv  at TP2 then do the rest of the alignments. The cell end is good enough at that point. If not I can give details on opening the other end and tuning the cell.

Make sure the 2 matching capacitors are not frozen. If frozen and you turn them you can break the rotor loose from its connection. I usually lever a plastic shaft between the hole in the cover and the top edge of the capacitor. Apply pressure at a couple spots around the rim until you hear the rotor snap loose.

Cheers,

Corby
 

[trobbins]

Thanks again Corby.  My concern is not appreciating what to expect from the open-circuit cell oven, and whether repairing that cell oven heater could cause me grief with the cell operation just by way of repairing the heater.  Ciao, Tim

[5065AGuru]

Tim,

I thought the cell oven was good and you repaired the lamp oven?
If both heaters are good now you should be able to power up ,assuming you located the cause that loaded the power supply down.

You can pull the 3 caps off the cell end and look in but there is no need yet if the ovens are both good.

One thing to do if you haven't is to Ohm out the TED wires, the tightly twisted black and red wires from the cell end. It should read close to a short.

Cheers,

Corby

[trobbins]

Corby, yes the cell oven was initially working but appears to have now stopped (open-circuit heater) during the latest test.  I had confirmed the lamp oven was working and was starting to confirm normal operating conditions for other circuitry, and sadly then observed both the cell oven fault and the cooking power supply inductor feed to the lamp oscillator (which I have repaired and replaced the 1k33 part and will get a chance to test tomorrow). Tim

[5065AGuru]

Tim,

How unfortunate! If the C-field winding is still good and the cell thermistor is also good I'd consider removing the lamp cylinder per the instructions and seeing if the cable in the exclusion zone was damaged. The first couple times I removed the lamp cylinder I did damage the cable but was able with much care get the break spliced. Since then I have done several without any damage by observing the exclusion zone.
If that doesn't work you would need to consider rebuilding the cell oven also.
Here is a PIX of an optical unit that I had to rebuilt both ovens and rewind the C-field.
If it's possible to get the cell oven removed without damaging the C-field that would save some effort.

Cheers,

Corby

[trobbins]

Thanks for advice and photo Corby - much appreciated.  That gives me a better appreciation of how the heater wiring, thermistor wiring, and C-field wiring span out from the DB9 cable in the zone between the two oven cylinders.  That in-between zone of foam is now hard baked foam, and is presumably where the DB9 cable wiring splays out.

The cell heater now shows correct continuity, so I will aim to check that the lamp circuitry works, and monitor the cell heater operation to see if and when it becomes open-circuit again (presumably related to thermally induced movement).  If it goes open circuit then I may be able to garner some insight on where the break is occurring from capacitance readings between cell heater cylinder and heater wiring ends.

Ciao, Tim

[5065AGuru]

Tim,

Check the wiring to the cell heater inside the Db9 connector, if that's OK monitor the continuity while flexing the cable.
I'm hoping it's external (outside the potting)!

Cheers,

Corby

[5065AGuru]

Tim,

Also check which wire you disconnected from  pin 4 of the Db9!

Originally there should have been a Yellow and a White wire.

The Yellow goes to the cell and should be connected. The White goes to the lamp and should be disconnected as you have a new lamp oven.

Pin 5 (Gray) the other lead to the lamp oven should also be disconnected.

Cheers,

Corby

[trobbins]

Ta Corby, yes those were the wire colours on my DB9, and it was convenient to have them as separate wire feeds from DB9/4 - both for retrofit of a new heater (as the white and grey can be snipped a bit shorter but not risk connecting to anything else either in the DB9 or in the loom/oven), and for faultfinding.  Just trying to progress another restoration and then back to the 5065A 
Ciao, Tim

[trobbins]

Hi Corby, I'd appreciate your insight to the following please.

I took some time to confirm and fault-find operation of the cell heater, given the preceding lamp heater repair and lamp assembly repair.  With the optical assembly out, both heaters tested fine by themselves and regulated fine when connected using the DB9, and also when the repaired lamp assembly was refitted.  But when the optical assembly was fully installed (with connections from the cell end reconnected) there was a cell heater fault.  I confirmed the related circuits were ok (cell heater controller on A11 and its main chassis mount transistor; the logic circuit A14; and the metering path via A17 terminal board).  So I then backtracked and removed the cell end connections and revealed that the cell heater (at the controlled end) is shorting to the ground shield of the CR1 harmonic diode in A12 RVFR, and so shorts to ground when that coax feed connects to J4 on A3 Multiplier assembly.  By resistance I can identify the short is not at the +22-28V end of the cell heater.

Removing the three end covers at the cell end, I can see that the CR1 connection shield is mounted to the cell metalwork, and the CR2 photo diode appears to be isolated from the metalwork (which concurs with resistance measurement of its shield), and TED also appears to be an isolated circuit.  So it does seem that one end of the heater coil is shorting to the cell oven aluminium cylinder, and hence is shorting to equipment ground via the CR1 harmonic generator cable shield.

From your photo in post #60, the cell oven cylinder is quite long.  The thermal insulation around that cylinder at the cell end seems to be rock hard, but I guess it could just be that the surface face is solid and thick and so presents as being quite hard.  I'm now assuming that I could do a similar style of heater repair (as to what I did at the other end for the lamp), and gouge out a channel around the aluminium cylinder but only go to a similar depth of 27mm and so leave the cylinder hopefully undisturbed (along with the cell etc) and still solidly locked in place by the remaining insulation.  The heater leads would have to route to the DB9 outside the whole assembly, and the main risk would then be any disturbance to the cell heater thermistor. 
Ciao, Tim

[5065AGuru]

Tim,

That's a bummer!
I've never seen a cell end short out that way!
I think you will be safe going down to the 27mm depth without encountering any other wiring.
Stay close to the cylinder and don't mess with the potting around the outside edge (red) as the C-field coil is there!
The first 1/4 inch or so is usually hard, as you go down it gets a bit easier to get out.
You could take drill and drill out the majority staying right up against the cylinder, then get the remaining stuff out with a screwdriver.

Cheers,

Corby

[trobbins]

Thanks for looking Corby.  Stanley is sending me the same insulated heater wire again, so I'll start digging when that arrives 

I added a 200mA in-line fuse in the dc feed to the A12A1 lamp oscillator from A15 as it drew over 1A for the previous fault but doesn't run above about 150mA max.

[TheSteve]

Decent looking unit sold for $2555 this morning before I woke up - did anyone here get it?

Seller did say they let some smoke out...

[5065AGuru]

If that was the one on eBay with NASA stickers it was a very old SN prefix 968, I would not pay that much without verifying that the lamp oven has not melted down! It was in nice physical condition. Had the older Quartz oscillator but good that it had the series 6080 A7 module.

Cheers,

Corby

[TheSteve]

That is the one. Perhaps good I slept through it then. Finding one will likely be a long term project.

[trobbins]

Thanks to Corby and Stanley the 5065A now has functional optical assembly heaters.  The unit has been on for less than a day so far, so no 2nd harmonic meter signal yet.  I'll give it 2 days and then start the TED process for a flooded cell if no 2nd harmonic signal.

In the interim I'll check the performance of the A7 signal amp module that amplifies then frequency selects the 137Hz and 274Hz signals.  My A7 module is the 6010 version (series 916) which is not the same as the Manual's 6080 version (series 820) that I have been using, but I tracked down the 6010 info in the on-line hp5065a_part2.pdf, as well as Corby's comment that the 6010 was the first version and quickly upgraded by HP as it has some performance limitations. 

I'll use a battery powered low level audio sinewave source via 10Meg into the input J1, and use my REW/soundcard frequency spectrum analyser to check the output of the first stage amplifier at TP2, and then the frequency filtered outputs at Y and WBO outputs.  I can then confirm gain from TP1 to Y output is 10-40x for ok loop gain.  I can also confirm Meter response for 2nd Harmonic by adjusting sinewave signal.  I think I can then check that the Error signal (via A8 detector) functions with 137Hz level, and check the CONTROL meter level for both Loop Open and Operate conditions, and possibly confirm the output of the OCXO shifts with EFC signal (although I only have a 1Hz resolution on my frequency meter so would have to set up the GPSDO and look for phase shift on a scope).

Just quickly using a DMM I note that TP2 has a +4.3Vdc unbalance, which seems a titch excessive.  The manual has a procedure for measuring residual DCV at J1 (which is meant to be less than +/- 5mV), as well as a procedure for measurement of A7Q1 base DCV (which is J1 input as well) and tweaking that to within 0.5mVdc.  I measure 3.4mVdc at J1, so perhaps no immediate need to adjust the internal R3 trimpot (apart from the 4.3V offset) given that would require removal of the module as R3 is not easy to get to, but I guess I will do that tweak.

One issue I notice with the heaters is that the Lamp oven meter reading is still swinging from 10 to 42 with a 7 second period, and hasn't stabilised like the other two ovens.  I'd guess that is due to a degraded part on the A11 temp controller pcb, rather than due to a changed thermal delay in the heater and thermistor physical coupling - any thoughts, as I'll have to try and fault-find that 

Ciao, Tim

[trobbins]

I've just used REW software and an EMU0404 soundcard for signal generator and spectrum analyser assessment of some aspects of the 5065A, with signal injection to the A7 AC amp J1 input (in lieu of the optical cell) via 10M and ground clip.

A7 AC Amp module J1 input stage to TP2 shows low harmonics for 0.5Vpp in, with +5.4dB gain.  TP1 shows 0.14% H2 with +0.7dB gain from input, and clearly shows the 274Hz notch in the noise floor response. Y output clips for signal levels above ~0.19Vpp at J1, and non-clipped output has low HDs.

As Corby indicates, the 6010 version of A7 presents a 2nd H meter level even for 137Hz signal, and meter level varies with 234 and/or 137Hz magnitude.

With 137Hz signal input, the following A8 Phase detector module generates an Error meter reading that cycles between -50 and +50, and the reference 137Hz oscillator shows sync with external REW signal at 137.035Hz so is ok.

Switching from Open Loop to Operate mode shows the Control meter reading swings and the Integrator Limit indicator lights at <-50, and >+50, and the A9 Integrator output to OCXO EFC- control line cycles from -14.4V to +7.8Vdc, and my SD6250A frequency counter just recognises a change in the 5MHz frequency (which is below the 1Hz resolution of the meter, but identified by toggling of the last digit).

I was also able to confirm the 1MHz and 100kHz outputs are ok, and the A3 multiplier module has 5.4Vdc on TP2, so that aspect is ok.

The Lamp Oven meter swing has subsided a bit to between 15 and 38, so moving in the right direction.

Ciao, Tim

[5065AGuru]

Tim,

Since the original lamp thermistor is being used your cycling is most likely the poor coupling between the thermistor and the oven. This as your modified oven heats only at the top and the thermistor is at the midpoint height wise. I would not worry about it yet as the instrument should operate in that configuration.
For cell flooding I use a 5VDC fixed supply through a 5 Ohm 10watt WW resistor to get the 1Amp. While watching A7 TP2 with a .25Hz offset in open loop adjust the A3 varactor bias pot and the two matching caps to see if you can find a signal. IF not hook up to de=flood the cell, let it run a week and try again.

Cheers,

Corby

[trobbins]

Thanks again Corby.  Could you please elaborate on the A7 TP2 checking process with respect to the following. 

I have a 6008 version of A3, so I'm not sure how aligned that is with the 6078 version module - for the 6078 version schematic I can identify varactors CR3 to CR6, and it looks like they have a bias pot R11 "Mod Level Adj" applied to the incoming 137Hz signal from A8 phase detector.  I can also identify trimmer caps C61 and C59 that would match the signal output going to the RVFR via J4.  Are those the adjustments you were referring to?

I can monitor A7 TP2 using my spectrum analyser to identify any 137Hz and 274Hz signal levels.  Is that the intent, and if so then is the aim to get a max level of 137Hz at TP2, or just to see some signal above the noise floor ?

I am not sure what your reference to 0.25Hz offset means ?

Overnight the Meter reading for 2nd Harmonic moved off zero and is now about 5.  The unit has been on for 48hrs now.

Ta, Tim