HP 5065A Rubidium Standard Repair Log

[maat]

Maat,

The schematic you are using for the lamp assy is an old one.
The 130pf cap is correct. No tuning is done.
Tubular ceramics are 3.9 and 11 Pf and I've never seen one go bad.
The tiny cap is a .001uf monolithic radial.
CR1 is a multi-pellet diode and can't be checked with an Ohmmeter. (Never seen a bad one!)
Make sure when you remount it that you hold the transistor heat sink while tightening so as not to twist the transistor leads!
Also once you reassemble (you should have had two big nuts and a fiber insulator on the central lamp post) position the inner nut so that when you tighten the outer nut onto the fiber washer it clamps the outer magnetic shield between them as the outer shield forms a heat sink for t e lamp assy.!
How does the reflector look?

Cheers,

Corby

Thanks a lot for the info. I will just replace the two caps then. The rest seemed fine.

[5065AGuru]

Maat,

I usually use my smallest finger and just push HARD on the glass from the rear and it pops out.

Take a look here for details of an optical filter mod.

http://leapsecond.com/corby/Corby-Super-5065A-Project.pdf

Also the yield when trying the mod is small as everything else has to be in better that spec. condition.

Just remember you will need a Hydrogen MASER to test it!

That goes for trying the Laser excitation also.

A friend has been trying the laser excitation for a few years now and its just not feasible yet.

Why replace a simple lamp that lasts 30 years with a costly Laser system whose reliability can't compete?

If you try it details would be most welcome!

Cheers,

Corby

[maat]

I usually use my smallest finger and just push HARD on the glass from the rear and it pops out.

So yours are glued in as well? I will give it a shot next time I take the lamp appart when installing the caps.

Concerning the lifetime of of the laser diodes is indeed an issue. In my university group, we are mostly using Sharp GH0781JA2C and with optical feedback these last about 2-3 years non-stop before degrading. This number will be higher with less output power, which is a seasonable asumption, because I only need to drive the saturated absorption spectroscopy and less power is better anyway. Cost is another factor. Laser driver, TEC controller, external resonator, the laser filter, Lock-In Amplifier and fibres including splitters will put you down by about 5-10 k$ depending on how much you can do yourself. But this is a nice 'benchtop' experiment one can run in the basement with reasonable cost of ownership while still being close to modern day physics lab experiments.

[edpalmer42]

A friend has been trying the laser excitation for a few years now and its just not feasible yet.

Why replace a simple lamp that lasts 30 years with a costly Laser system whose reliability can't compete?

Corby, do you mean it's not feasible for a retrofit?  The Microsemi Miniature Atomic Clock uses a VCSEL (Vertical Cavity Surface Emitting Laser) instead of a Rb lamp.  The light isn't visible to the eye, but a camera can see it.  I have one that is, unfortunately, dead, but it's fascinating to watch it try to find a combination of parameters that will get a lock.  It spends something like 3 hours trying before it gives up!

SA.3Xm User Guide:  https://www.microsemi.com/document-portal/doc_download/135918-sa-3xm-miniature-atomic-clock-mac-user-guide

Ed

[5065AGuru]

Ed, I know about the Microsemi device.
My point about feasibility, not so much that it could not be done albeit at much expense and fiddling. (Which don't get me wrong, can be fun) However I try to balance modifications to those that improve or at least maintain the existing 5065A performance with a lifetime that is long enough to warrant the effort. So far examples I have seen don't come close. Remember an excellent performing 5065A will have an Allan Deviation approaching or exceeding 1X10-13th at a 100 Sec Tau. The Microsemi units depending on which spec you purchase are 1X10-11th or 8X10-12th at 100 Sec. Of course no modern (commercial) Rubidium comes close to matching a 5065A performance! The drive to smaller, cheaper, low power, directly affected the short term stabilities achieved by modern Rubidium units.

Cheers,

Corby

[edpalmer42]

Ed, I know about the Microsemi device.
My point about feasibility, not so much that it could not be done albeit at much expense and fiddling. (Which don't get me wrong, can be fun) However I try to balance modifications to those that improve or at least maintain the existing 5065A performance with a lifetime that is long enough to warrant the effort. So far examples I have seen don't come close. Remember an excellent performing 5065A will have an Allan Deviation approaching or exceeding 1X10-13th at a 100 Sec Tau. The Microsemi units depending on which spec you purchase are 1X10-11th or 8X10-12th at 100 Sec. Of course no modern (commercial) Rubidium comes close to matching a 5065A performance! The drive to smaller, cheaper, low power, directly affected the short term stabilities achieved by modern Rubidium units.

Cheers,

Corby

I was sure that you were familiar with the Microsemi device, I just didn't quite understand your comment.

Ed

[5065AGuru]

Maat,

It's RTV not cement holding in the glass. It's soft and just to keep it from rattling around!

The bits of original RTV are usually fine when reinstalling. I've never had to use new RTV

Cheers,

Corby

[maat]

Today was bathing time!



From left to right. Distilled water with a dash of Simple Green, Acetic acid mixed with distilled water and finally pure water. There is also a forth bowl with IPA not shown in the picture. I usually start by scrubbing off any ooze, corrosion and minerals that are on the board. When I have a bare copper surface, that is only tarnished from oxidation, I put it in the vinegar bath until the copper is shiny again (a few minutes). Then I wash it in pure water to remove the acid. You may add sodium bicarbonate, but watch out if you use baking soda. Sometimes manufacturers add other stuff to the sodium bicarbonate as well. Check the package first. Then drown it in IPA. Depending on the time it was in the water I usually leave it in there for a while to remove all the moisture. Finally I use compressed air to blow off the IPA. If there are still some stains, reapply IPA, and then immediately wash with distilled water, then soak in IPA again. Usually these are flux residues that get dissolved in the IPA, but rinsing with water while the IPA is still liquid usually takes care of the stuff.

Once I am done with cleaning and the copper looks good again, I apply a layer of solder to protect the copper for all eternity

This is the A15 Board as an example:



I removed the capacitors and the diode in the middle, which looks like this:



after some scrubbing, the green copper sulfite (from a leaking tantalum capacitor probably replaced by the previous owner) the corrosion is gone, but we are left with a layer of oxide. I is impossible to solder to this oxide layer, so we need to remove it first.



Unfortunately I did not take a photo after the acid bath, but only after I was completely done. 



I overdid the solder patch a little in this case. This pad was a GND pad, but the plated holes had no contact, because the copper was already gone around those. So I just poured it with solder then washed it again, leaving a shiny blob of solder

The three tantalum caps looked new (old stock?) to me and their capacitance and ESR (a little more than 1 Ω @ 100 Hz) seemed fine so I left them in for the moment, but I will replace them as soon as Mouser ships my replacement caps (long, long time...)

I also paid that corroded polystere capacitor a visit with the fibreglass brush. Worked like a treat:



Then retinned the lead:


So far so good. I cleaned most of the boards shown in the previous post and I also cleared out the interior of the case so see how far the battery acid had spread:



I think I will try the vinegar again and then some polish to touch it up. We will see how it goes. While life was looking good a new horror crept up on me; the A3 Multiplier Assembly:



I can hear a rattling inside and there is stuff rolling around  I did not yet open it, because the screws have been massively abused and I was running out of time anyway. I guess this could be the board where the previous repair attempt met a gruesome end.

To be continued... 

[5065AGuru]

Maat,

The rattling in the A3 is probably a corroded 100uf tantalum that has come loose.
The A3 has lots of screws and they can be difficult to get loose.
Once you have gotten inside and repaired it you need to make sure the two ceramic trim caps that are next to each other (the output matching stage) are not "stuck". One problem people introduce is to try to turn these and break loose the rotor connection. If they don't rotate with moderate force then you need to apply pressure to the rotor section sideways to break them free. I usually do this while the cover is on by inserting a stiff plastic rod through the tuning hole and pry sideways against the rotor using the leverage against the tuning hole until its free. There are more of those caps and you could break them free while you are at it but the critical ones are the two I mentioned.

Also when you power it up the first time quickly check the plus/minus 20V volts. I also put a scope on the two chassis mounted choke connections. One should be flat DC the other an unfiltered full wave rectified signal.

The only reason I check this is that I once had a unit with a shorted choke. Never seen that since but I still check.

Cheer,

Corby

[maat]

It is time for another update. I took the whole units apart and finally cleaned everything. The upside is, that I am now corrosion free. The aluminium still has some etch marks to show for, but I consider that art  The downside is, that some of the labels printed onto the aluminium are now gone. Oh well, there is always something...



In the meantime my power connector arrived. Or so I thought. I had ordered an MS3106A18-22S connector and at that time I did not bother reading through the full 70 pages of the datasheet of that connector family. As it turned out there are multiple versions of the that connector. Hidden somewhere in the far back of the datasheet there is a section "insert alternate positioning". In total there are five different versions of the connector, differing in the position of the pin insert. Well..., trip to the milling machine and now I am the proud owner of a dual connector 



Just in case someone needs that connector. The correct part number ist MS3106A18-22SW (the W at the end denotes a 70° rotation of the pin insert). Digikey link: https://www.digikey.com/product-detail/en/amphenol-industrial-operations/MS3106A18-22SW/AMS3106A18-22SW-ND/2128451

So finally, I powered up the unit without any of the circuit boards installed to check the transformer output. I do have ~28 Vrms (around 40 V after the rectifier) between between pins 5 and 6. Looked a little high as the manual said 24-32 V, but then again that is unloaded, so it might be OK. Nothing blowing up so far, I decided to install A15 (power supply) and A2 (battery charger). The battery charger is required, because it contains the OR-ing circuit to either connect the battery, the DC supply or mains. Units without option 002 or 003 will contain the A18 jumper board to do the job.

Powering up again things looked nice as well.



I have got nice and stable 20.01 V on the 20 V rail, but still those 40 V on the input side to the A15 board, again that is mostly unloaded. . But seconds after I took that picture there was a bright flash and the circuit breaker in the lab triggered. Bummer....Checking the source, I noticed an awful smell and lots of ash around the power input. Nothing on the circuit boards. Phew. Taking a closer look, I found the charred remains of what looked like a capacitor. Guess what, acute RIFAlitis. There are two capacitors C5A and C5B connected from both live and neutral to earth. These are Y2 caps. Supposed to fail open!  Anyway, I pulled both capacitors and the surviving one was already showing cracks. The dead one literally exploded.



For now I am living without them, because I need to order new ones. While I was at it, I also decided to put shrinking tube on all accessible solder joints that carry mains voltage. HP..., hello?!

So a word of caution to everyone buying a used 5065A, you might want to check the caps or RIFAlitis.

Next up to install is the physics package to see if the heaters are working, but I am still waiting for those mica caps. It will probably be another week before they arrive.

[Oldtestgear]

 "Devonport Management Limited". Port? Do I hear sea port? Salt water?
Devonport management Ltd. took over the old Plymouth Dockyard when the MOD started sub contracting core activities to private companies.  So definitely sea air & salt water,

Phil

[trobbins]

Just to extend this log, although for another 5065A, I have been able to reinstate a 00105-6013 crystal oven assembly after it had multiple part faults including shorting heaters to ground and 'cooked' internal insulation foam.  The internal can needed to be exhumed from its solid foam coffin and the heater winding and insulation inspected - the AC preheat heater was not needed (and was a problem for any mains earth leakage cb) and carbonised deposits were carefully scraped away to achieve a reasonably ok insulation resistance from the dc heater to can - and the can then re-centred in the assembly using neoprene foam strips.  Failed parts were the two voltage doubler diodes for temperature feedback, and the following npn, and the heater temp adjust trimpot at the other end pcb, and the entire ribbon cable was charred and replaced.  The coarse freq adjust pole had also suffered and was broken deep inside the can, with no ability for adjustment, so the heater temperature trimpot was used to trim to 5MHz to within 10ppb with the fine freq adjust in centred position - the heater trimpot adjustment gave about +/- 1.2Hz span, so although using that control may affect some performance specs, it was the only way forward for this assembly (given I didn't want to crack open the can for a deep-dive).

Now I'm just about to start excavating the lamp oven heater from within the RVFR can, as the filament has gone open and is shorting to the heater mount ground.  At least the first-aid docs indicate that the lamp heater can be repaired (although no photos available), so I will take some photos along this voyage of discovery.

[5065AGuru]

T Robbins, Have you looked over this post about the lamp oven repair? If you have and plan to follow it I have a couple new bits of info. if you want.

https://www.eevblog.com/forum/metrology/hp-5065a-lamp-oven-repair-details!/

Cheers,

Corby

[trobbins]

Hi Corby, thanks for chiming in - much appreciated.  I tried to access your lamp oven repair link a few days ago, but that link is broken afaik (?).

I have been in touch with syau (Stanley) - his thread photos show an early version of insulation and construction of the lamp heater region.  My partly deconstructed region is shown below - the heater filament had fused, and the foam insulation had baked next to the aluminium cylinder but was able to be gouged out with care.  Although part of the magnetic field coil has been exposed, I aim to constrain that and not uncover any more.  My concern is how much of the original heater I can uncover without disrupting or stressing the DB9 cable's connection to the lamp heater thermistor, and to the cell oven wiring.  I don't really want to disturb that wiring if I don't have to, and so doubt that removing the aluminium cylinder (as Stanley did) is a practical path to take.  I would appreciate any advice you can offer.

I asked Stanley about his spare Pelican wire (as you also did), and also am on the lookout for a local alternative.

Ciao, Tim

[bingo600]

Hi Corby, thanks for chiming in - much appreciated.  I tried to access your lamp oven repair link a few days ago, but that link is broken afaik (?).

You need to make sure the !/ in the end is also copied to the url in the browser

https://www.eevblog.com/forum/metrology/hp-5065a-lamp-oven-repair-details!/

[5065AGuru]

Don't remove any more potting until you read the post referenced. You can make the job orders of magnitude harder if you mess up the wiring hidden in the potting!!!

https://www.eevblog.com/forum/metrology/hp-5065a-lamp-oven-repair-details!/

Cheers,

Corby

[trobbins]

Thanks Corby - agreed. 

From the photos in that linked thread it is unclear how the DB9 cable extends down into the foam, and then how the lamp heater and thermistor wiring exits that DB9 cable.  It appears that the DB9 cable continues down through the insulation (and located between the magnetic field coil and the heater winding) to the cell oven and doesn't get located in to the machined groove in the lamp oven aluminium cylinder - do you recall ?

I am just mulling over my options, as I am not keen on completely removing the aluminium cylinder if I don't have to.  One possibility is to excavate the outer wall of the aluminium cylinder down to where the groove starts, especially if that doesn't physically impinge on the existing thermistor.  I may then be able to insert the new heating turns such that heat is transferred through the upper-half of the aluminium cylinder, and use the cylinder to then even out the temperature for the lamp assembly.  That does assume that the remnant of the existing lamp oven heater wiring (the common +22V end) can be isolated from touching the cylinder wall.

Ciao, Tim

Ciao, Tim

[5065AGuru]

Tim,

Either way you will lose the original connections to the Db9 and have to route the new lamp oven wires out to the Db9 on the end of the cable. If you remove the cylinder you also lose the original thermistor wires and they would need rerouting also.
The problem with digging down into the "exclusion" zone is that there is no guaranty that you can avoid damage to the cable as it could be very near the top!
As far as removing the cylinder I just finished rewinding one and rather than machining complete grooves I just made the 9 grooves in just the exclusion zone area so that the new windings could be flush. Only took a few minutes with a hacksaw and a dremel to widen the slots a bit.

The existing +22volt end can be disconnected inside the Db9 shell on the cable end.

The original machined slot was for the thermistor only, it does not involve the cable.

If you proceeded like you were going to remove the cylinder and removed the potting as in the post you might be able to make a heater "insert" that would slide down into the opened up area?

Cheers, Corby

[trobbins]

Yes my first option was to see if it made sense to prepare a slip on 'voice-coil' type cylindrical assembly with just the heater.  Stanley has kindly just sent me some of his wire, so I will explore what can be done with that later next week.  I do have 5 ohm/ft Advance resistance wire but it isn't insulated, which makes that path difficult.  I have thin Nomex sheet/paper that I could use as either an inner or outer form.  Preferably the bifilar twisted insulated resistance wire has margin between turns, like with your groove setup, so I need to assess what margin I can achieve between turns versus the depth of excavation (and hence risk of upsetting the thermistor).  A portion of the existing heater may still remain, so I need to see if that can be isolated, as one end will be connected to the +22V that is also connecting to the cell oven heater.

One unknown is the maximum temperature reached by the resistance wire, either as an initial peak during heat-up, and as a continuous level, given that the aluminium cylinder is controlled to 90-92degC.  There will be some nominal thermal resistance between the wire and the cylinder which will mainly dictate the continuous operating temp of the wire. Nearby materials like silicon rubber (engineering grade or RTV) and Nomex are rated to a bit of 200C. I may even have room for a thin layer of neoprene or aerogel to further buffer the heat flow, although just an air cavity is simplest as you have used.

In the interim I am going to assess if the exposed filament exhibits damage just towards the 'top' portion, or if the enamel (I assume) coating has degraded all the way along the cylinder.

Ciao, Tim

[5065AGuru]

Tim,

As I said the +22 volt wire can be disconnected in the connector shell so that is not an issue.

The thermistor is buried in the slot about midway down so you should not encounter it when you "dig"

If you can safely excavate enough material around the top I'd just coil up the 9 or so turns of the  twisted wire, stuff it in and RTV it in place.

Note I use ASI 388 electronics grade silicone RTV. Make sure your RTV is NON-corrosive!

It's a little more work to get the cylinder out as per my post but it is way better than messing up the other wiring!

Cheers,

Corby

[trobbins]

Corby - thanks, I had forgotten that two heater wires join at DB9 pin 4 

Agreed about the type of silicone rubber - at the moment I only have an engineering grade (selleys 401) tube open, and it is acetic cure.

Ciao, Tim

[trobbins]

Touch wood I have jumped the lamp oven repair hurdle.  Photos and a description in https://www.dalmura.com.au/static/HP%2005065-6001%20lamp%20heater%20repair.pdf.  Thanks to Corby and Stanley 

Soon after the three ovens started to regulate I noted a burning aroma and saw that the supply meter reading had dropped, and the A15 board parts were getting very hot, with one inductor cracked/burnt, so now on to the next repair hurdle (but hopefully just simpler electronic fault finding).

Photos from my first hurdle - repair of the 00105-6013 crystal oven oscillator are in https://www.dalmura.com.au/static/HP%2000105-6013%20internal%20wiring%20info.pdf.

[5065AGuru]

Tim,

I'd check for a short somewhere. Unplug the A11 until your A15 is working. Then I'd monitor the +20V current with and without the A11.
You may be able to Ohm out the problem.
Which choke was burnt?

Cheers,

Corby

[TheSteve]

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

[edpalmer42]

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?