Datum PRS-50 Cesium FRS. A tale of woe.

[Oldtestgear]

Hard to believe that I bought this nearly 18 months ago & am finally writing it up. To be honest it has been a bit of a disaster but if I can get some information it may have a happier ending.

When I received it the PRS-50 had been packed in a large box. No packing material worth mentioning, just a box full of air. That combined with the carrier guaranteed that it would be smashed & it was. The seller really did not care.

The mechanical repairs were bodged to allow testing. No outputs from the PSU as all the electrolytic caps had spewed their insides over the PCB. Tracks & vias were  etched & broken. Replacement caps were fitted & checked for continuity. All seemed OK. At switch on there was hope as the LEDs came on; followed by the PSU self igniting before I could switch it off.  Looks as if there was an internal short in the multi layer PCB.
So, does anyone have access to service information for the power supply?  If I cannot get any information then I am definitely at a loss how to attempt a repair . I do not want to spend many hours trying to trace it out & then work out which voltages are produced & where these go.  I am thinking about fitting an external PSU  as a next step but  need basic information before I try this.

Any ideas very welcome & yes I will consider selling it if someone really wants the challenge.

Thanks

Phil

[Theboel]

may I ask about the passive maser ?

[5065AGuru]

The passive Maser is still a work in progress!

http://www.leapsecond.com/maser/pasivemaserblogd.docx

Cheers,

Corby

[LapTop006]

Just having been off for a year is enough that the tube may well have gone.

My own Symmetricom (now Microsemi, but it's a Datum-derived design) CS4000 cesium was better treated for shipping, but I didn't find out until after I'd bought it that it had been left on the shelf in storage for a year, I still haven't got around to figuring out if it has a dead tube or bad electronics, as the alarms could indicate either.

[edpalmer42]

Just having been off for a year is enough that the tube may well have gone.

My own Symmetricom (now Microsemi, but it's a Datum-derived design) CS4000 cesium was better treated for shipping, but I didn't find out until after I'd bought it that it had been left on the shelf in storage for a year, I still haven't got around to figuring out if it has a dead tube or bad electronics, as the alarms could indicate either.

A year on the shelf won't kill a Cesium tube.  It'll just take a bit longer to pump the tube down when you power it up.  Are you running Monitor3?  What is the Ion Pump reading?

Ed

[LapTop006]

A year on the shelf won't kill a Cesium tube.  It'll just take a bit longer to pump the tube down when you power it up.  Are you running Monitor3?  What is the Ion Pump reading?

Only to a point, here was it after ~3 months of power on. It's just sitting near me, but it takes more time than I have at the second to talk to it.

[edpalmer42]

50K power on hours!  Ouch!!  I don't know what kind of life expectancy a tube like that should have, but that seems like a lot.

The ion pump current is high, but not fatal.  The Clock Signal level isn't right, but again, it's not fatal.  Is the level constant or does it change wildly?  My experience with my Datum 4065A is that the Zeeman Signal level is meaningless until the unit locks.

Since the CS4000 is newer than my 4065A, I'm surprised that it doesn't seem to have a storage mode where the Ion Pump is kept alive while the rest of the unit is powered off.  Since you've got nothing to lose, you might consider unplugging the Cesium tube except for the Ion Pump connection.  This will ensure that you're not using up your Cesium supply while the Ion Pump keeps working away at pumping down the tube.  A cleaner tube might allow a weak tube to lock easier.  There's nothing in the manual to suggest that the processor will shut anything down if it doesn't like the readings, so you should still be able to monitor the pump current.  You said the screen capture was after running for 3 months.  Was the pump current coming down at all?

Ed

[LapTop006]

50K power on hours!  Ouch!!  I don't know what kind of life expectancy a tube like that should have, but that seems like a lot.

It's high enough that the storage might have killed it, but low enough that it still might be resurrected. It's just been a low priority for me, my house reference is a nice rubidium, and I don't really need better, I'd use a GPSDO except I have no usable GPS signal at my apartment.

Here's it after about a month. (It was living in California for the first six months I owned it until I could ship it back to Sydney, and I didn't have the right cables with me on the first visit)

[edpalmer42]

So the Ion Pump current is dropping and the Clock Signal is rising.  Both are good.  Maybe it will live again.

Ed

[Oldtestgear]

Since it seems that there is no helpful information out there I am left with the options to scrap it or sell for parts if anyone is interested.

If anyone wants it then please make me an offer by PM.

Thanks

Phil

[CJay]

Yes.

[rubidium]

I'm tagging onto this thread because my post is related, and I have been exchanging notes recently with the OP.

This is about my recent acquisition of a PRS-50 - and it arrived today. The unit is immaculate inside, and clean outside with just a few scuffs. Operationally, though, "no joy". But there's enough hope to encourage me to move forward with a few fundamental repairs before throwing in the towel. When the unit first arrived, I decided to go for broke and connect 2 independent 48VDC supplies to the rear panel. I had previously set up monitor3 and connected up a working USB-to-RS232 cable to my laptop. Initially, the DC POWER LED on the front panel illuminated - although suspiciously not with what I would call normal brightness - and then after about 10 seconds all of the front panel LEDs (except for the 2 bad fuse indicators) illuminated for another 10 seconds, and then finally the LEDs started to blink on and off in what appeared to be a fairly complicated, but repeating, sequence. Again, all of the LEDs were a bit dim in my opinion when they did illuminate. Meanwhile, there was absolutely no activity on the RS232 ports, no matter what I did. So I suspected a power supply problem and decided to pull the main module bearing the CBT. Removing that module was rather easy, and I gained an appreciation for the layout of the build that apparently had serviceability in mind.

Without any schematic to know what voltages ought to be at which test points, I chose to probe 3 places that were obvious. Two of those places were the inputs to the two HV bricks, for which the input voltages were each marked on the respective brick. The other was an LM340T-12 linear regulator that clearly fed the OCXO. The results confirmed power supply problems:
- Input to HV brick 1: 10.8v (should be 24)
- Input to HV brick 2: 6.65v (should be 10)
- Output of LM340T-12: 5.83v (should be 12), problem due to only 7.37v into the regulator

Further checking how many other test points are wrong was immaterial at this point, since it's obvious that the power supply card has issues. So ... time to take the cover off to the compartment housing the power supply and take a look at the board. After gaining access to the board I, of course, discover all but 2 of the 16 larger electrolytics were badly swollen and about 11 of them were oozing electrolyte. Been down that road plenty! The good news is that electrolyte was simply oozing out of the tops of the caps and absolutely none of it got on the PCB or anywhere else. Once I order some replacements I'll make the repairs and test the voltages at the caps themselves. Somewhere on eevblog I recall that someone posted a "map" of the caps and the voltages on them from his working unit (may not have been a PRS-50 but one of its siblings.). I'll dig that up to guide my testing once I replace the caps.

More after I have a working supply. Perhaps there's some good news here - though it may end up being just be wishful thinking on my part - and that is that the tube actually may still have some life left in it, and that it was the power supply failures that led to the unit's retirement. Perhaps that retirement was fortuitous for the tube.
Jim

[edpalmer42]

Somewhere on eevblog I recall that someone posted a "map" of the caps and the voltages on them from his working unit (may not have been a PRS-50 but one of its siblings.).

That was me.  My Datum 4065A is based on the 5045A Cesium Module.  Your PRS-50 is similar, but the main board looks a lot different.  The RF side of the unit looks very similar, however.  The thread you're looking for is here:

https://www.eevblog.com/forum/metrology/no-luck-on-my-first-attempt-at-acquiring-a-cesium-frequency-standard/

Ed

[rubidium]

Yes, there seem to be many similarities and several differences between our power boards. Yours appears to have 15 electrolytics and mine has 16. Transformers look very similar, if not the same. I ordered replacements for the caps, but didn't get any replacement power MOSFETs (IRF640). Perhaps I should have, since there's no telling at this point if any croaked when the caps gave out. I'm also hoping that none of the control circuitry croaked either. I'm guessing that the microcontroller was browning out with the present power situation, and that may explain the cyclic blinking of the front panel LEDs.

I'm also not sure about communication with monitor3. Certainly at this point, without proper voltages coming out of that power board, I can't expect to get anything out of those RS232 ports. But there are 2 ports (1 front and 1 rear) and potential confusion as to whether monitor3 will work with either or both. There is the so-called TL1 option board at the top of the enclosure above the 4201A module that a is involved with these ports per the user's manual. I'm not sure if it works with  monitor3. But, for now, I need to get proper power first.
Jim

[edpalmer42]

TL1 is an ugly language to use unless you already know it.  Try to avoid that board if possible.

[rubidium]

On the inside of the front hinged panel is an array of DIP switches. I infe from the user's guide that flipping them all on and removing the TL1 board ought to connect the rear RS232 directly to the 4201A module without having the TL1 performing as a communications intermediary and thus imposing its associated language. Hopefully that will allow monitor3 to talk to the module. From a communication I had with the author of monitor3, his code should be able to talk directly to the module. Of course, I have to get the power supply in proper order first.
Jim

Edit: Indeed I discovered that the rear RS232 RXD and TXD lines go directly into the 4201A module if you set the DIP switches 1 and 2 on the back of the front panel to "ON." The TL1 board gets bypassed.

[notfaded1]

Good Morning Jim-

At least the problems may be caps... that's really an easy fix and compared to others with this unit you're lucky it didn't destroy the circuit board.  You may have read in another thread where someone's entire pcb traces were ruined by the caps leaking.  I'd actually be happy that it might be old caps and possibly not something much worse.  My first Cs has all good power... unfortunately on it the tube was unable to ever get up to correct output.  At least, thanks to Corby, posting his scans of FTS 4060 manual gave me the docs I needed to finally understand how the 5030 really works.  Sadly that also showed me where the test points to measure and to adjust the Cs output and no matter how many times I turned to multiturn pot it never got close to what the manual says it's needed to be.  For me this was on the board attached to the Cs tube right next to it on vertical pcb with two blue multi turn pcb pots on top.

Regards,

Bill

[rubidium]

While I'm awaiting parts, if anyone who has been down a similar road before has any advice on removing the old ones I'm open to it. I don't know how many layers are in this board, but I suspect at least 4. Presumably one is a ground plane that will make a nice heat sink that proves to be problematic when de-soldering that particular lead of each cap. If the board didn't sustain any damage from any leaking electrolyte, I certainly don't want to introduce any with my de-soldering efforts. Would hot air from my SMD rework station or iron be best?
Jim

[notfaded1]

Jim-

I usually use a HAKKO FR301-03/P Desoldering Tool but either the air or the iron can work with some desoldering braid to clean up.  I think the main thing is you don't want to overheat anything longer than necessary.  Also with the pcb being older I wouldn't put too much pressure on it making the whole thing flex too much.  You really just don't want to break or lift and traces or heat it so much that it causes more damage than good.  That's the best advice I can give.  Sometimes it can't hurt to inspect the other solder joints as well and reflow any that don't look right.  I've fixed a few things by just reflowing the solder joints on a pcb where they didn't look right.  Also keep in mind... I think there is conformal coating on the boards so you may have to get through it to get to the solder which complicates things a little.

Bill

[rubidium]

I just completed a more in-depth examination of the power board. Fortunately, there is no conformal coating anywhere on the board. There is a spot of glue at the base of each electrolytic however. It is a slightly dark uniform maple syrup color and is translucent/glassy. I'm convinced it's not leaked electrolyte for several reasons: It does not dissolve with IPA on a cotton swab, whereas the dried electrolyte that leaked out of the tops of the caps does. The leaked electrolyte is brown with color gradations and opaque and grainy. There are spots of the same stuff at each screw anchoring power MOSFETs and BJTs to the heat sink (see the photo for a typical example). Slight lateral rocking displacement on a couple of the caps shattered the stuff and it went flying. Others remained intact under slight pressure. I don't want to apply too much pressure without applying perhaps a little focused hot air.

[rubidium]

Yea ... that stuff is glue. I just removed 3 caps and all rubber seals are intact. The glue they used largely captured the cap's shrink wrap and holds it to the board. The combination of residue is like concrete. I'll experiment with a chip of the stuff to see if I have some solvent that will soften it without damaging the board. I have a small chem lab here, and "solvents are us." But the caps more or less easily came out without damaging pads . I just used an iron, rosin flux, and added a bit of solder to the joint to effect good heat conduction without a lengthy dwell.

[notfaded1]

That looks good.  Yeah it's glue they used to keep the caps stable in place.  Since it's old sometimes it'll just crack and break right off or with help of carefully placed semisharp edge it'll pop right off.  Usually you can tell if it's going to come right off.  Obviously don't force it if it's taking a lot of force.

Regards,

Bill

[rubidium]

I tried a lot of solvents on a piece that I chipped off, and it's resistant to just about everything: IPA, Ethanol, Butanol, and even MEK. Hell with it. I'll just chip as much as I can off. If the new caps have to ride a millimeter above the board, I'll live with that.  I'm not planning to go to war with this thing. 

[LazyJack]

I'd guess it may be some kind of resin, like epoxy or polystyrene. If it is, then probably you don't want anything near the board that would dissolve it Instead just try heating it with a hot air gun, and see if it becomes rubbery. Then it should pop off the board with little effort.

[rubidium]

Well ... here we go! I replaced all 16 electrolytics on the power board. This took quite some time, as I wanted to be super careful and not damage the PCB.  Still, my heart was in my throat with about 4 of them, as I was left with barely enough solder pad to get a connection on the new caps. I Applied 48V input to the module on the bench and noticed that it was initially drawing about 510-520mA, and then after about 10 minutes it started drawing about 835-840mA. I also started to feel some warmth from things, like various chips, and the OCXO. So I decided to put the module back in the cabinet, power it up there, and see what happens.

Back in the cabinet, the crazy front panel LED behavior was gone, and I had nice bright illuminated LEDs on for "DC POWER" (green), "INITIALIZATION" (yellow), and "MAJOR ALARM" (red). After about 25 minutes, both the "INITIALIXATION" and the "MAJOR ALARM" LEDs went out and I got the ever-so-elusive "CESIUM LOCK" (green) LED.    I'm ecstatic!!!  But we'll see how long it stays on.

What I'm getting for output is a rock-solid 5MHz. I would expect 10MHz, but I do know that this is a setting that can be changed somewhere. What I don't have is any activity on the rear RS232 port. Monitor3 doesn't detect the unit at all, and so I can't take a look at the various performance parameters. So I need to  play around with that and get it going. But, for now, "CESIUM LOCK" is all my tired brain and body needs.

Jim