Electronics > Repair

Help needed for repair of Efratom FRS-C rubidium oscillator

<< < (6/7) > >>

edpalmer42:

--- Quote from: ejd.pol on May 05, 2024, 10:01:10 am ---* The fan only turns on when the counter is taken out of standby. The remarkable thing is that the fan is cooling the
supply of the Rubidium oscillator, which is on all the time. That power supply needs to work just as hard
when the counter is in stand-by as when it is on. So I don't get it... Is it perhaps the case that when the counter is on,
the fan is supposed to create a more stable temperature environment for the oscillator? (By the way,
that would also explain why the fan is allowed to be operated from a different power supply, which may break...)

--- End quote ---

You should take a close look at the power for the fan.  I think that if the Rb is installed, the fan is supposed to run continuously.  There might be another problem or an option set wrong.  You could remove the Rb oscillator, power supply, and distribution board and put them into a seperate box.  This would allow you to set up a nice passive heatsink with no fan to cool the Rb.


--- Quote ---* When the 10 MHz clock is not there, apparently the display remains empty. But the software is very much alive.
So why is there not some error message displayed? Or at least some kind of sign of life? This boggles my mind.

--- End quote ---

The software isn't alive when the counter is in standby.  What are the power supplies drawing in standby?  The power switch turns the +5V on/off as described in the power supply section of the manual.  The reason the oscillator is necessary is that it acts as the clock for the CPU.  I've also seen that in a counter from Racal.


--- Quote ---Anyway, after installing the replacement supplies, it looks like I will be able to see how accurate the counter is,
after I hook up my GPSDO. I will keep you posted on that part as well.

Again, many thanks for the advice and moral support! EJ

--- End quote ---

You're welcome.  Glad I could help.

Ed

ejd.pol:
 Hello Ed,

The replacement of the power supply is a success. I decided to install a 9-way power connector in order not to molest the main PCB any more than necessary,
even in the case the replacement supplies would not function satisfactorily, and plan B would have to be activated (whatever plan B is...).
The three small boxes happen to fit precisely (and I mean _precisely_) into the cavity of the removable metal shield of the original supply.
In two of the three perpendicular directions, the margin is about 1 mm !!  See the pictures below. 8)

Of course first I performed a dry run, connecting only AC in, not DC out, and measuring the output voltages. Then the DC wires were fixed to the new power connector,
and after triple checking of the connections, the moment came to plug it in, and turn it on. No black smoke, the display showed "NO SIGNAL",
and after a while the unlock light turned off. Ah, great!

The first guinea pig to hook up was a simple and small digital signal generator. The reading on the PM6685R deviated only a few ppm from the setting,
so that was pretty good. Of course I hoped that by far the major part of the deviation as due to the generator, not the counter.
I played around a bit with the frequencies, and found the counter could measure down to 4 Hz on an amplitude of 5 V. So far so good.

Next, I hooked up an OCXO that I had calibrated to my GPSDO roughly three years ago. It is stand-alone, housed a big metal box with a decent toroidal transformer in it,
and linear voltage regulators. When I switched it on, the deviation was significant, around 100 ppm or so too high. But the box started to warm up, the deviation reduced,
and after about two hours the reading had stabilized to less than 100 mHz too low on the 5 MHz target. That is 20 ppb.
Not bad, given both instruments had been last calibrated years ago.

The proof of the pudding was the GPSDO itself. Once I had got it going again (likewise, it was years ago since it last was fired up),
the reading on the counter swung between 10.000,000,00 and 10.000,000,01, roughly evenly, so the deviation is about 0.5 ppb!
That is the best that could be hoped for, I would think. I will not even touch the calibration pot!  :)


--- Quote from: edpalmer42 on May 05, 2024, 09:09:07 pm ---You should take a close look at the power for the fan.  I think that if the Rb is installed, the fan is supposed to run continuously.  There might be another problem or an option set wrong.  You could remove the Rb oscillator, power supply, and distribution board and put them into a seperate box.  This would allow you to set up a nice passive heatsink with no fan to cool the Rb.

--- End quote ---

I did check all jumpers, and they all seem to be set correctly. And the counter seems to operate just fine. Do you have any suggestions what else I can check?

The suggestion of putting the oscillator in a separate box was indeed the plan in case the counter could not be resurrected.
But now seeing it working so nicely, I'd like to keep them together. Unless the fan issue should force it?


--- Quote ---The software isn't alive when the counter is in standby.  What are the power supplies drawing in standby?  The power switch turns the +5V on/off as described in the power supply section of the manual.  The reason the oscillator is necessary is that it acts as the clock for the CPU.  I've also seen that in a counter from Racal.

--- End quote ---

From memory, the supply current in standby is about half of the current during operation, so roughly 300 mA for the 5V rail.
As far as the clock is concerned, there is a 10 MHz crystal on the main pcb, so I had assumed that that would deliver the main clock for the processor.
If not, then what is it there for? If the system requires the Rb clock to be always there, the crystal ought to be superfluous, no?

The last thing to check is the GPIB interface.... I will get to that once I have a hardware driver assembled!  :-DMM

Cheers, EJ

Pictures will follow in the next post (there seems to be a bug in the system: canceled attachments are not deducted from the total attachment size)

ejd.pol:
p10: Inside the replacement power supplies (couldn't resist taking a peek inside)
p11: power connector strip installed
p12: the three modules fit precisely
p13: after closure of the supply compartment
p14: OXCO hooked up
p15: GPSDO hooked up

edpalmer42:
Looks great!  Congratulations!

Regarding the fan, I don't know.  If the Rb is running continuously, I'd expect the fan to do the same.  I don't have the schematic handy, but I seem to remember seeing a fan connector labelled for the Rb option only.

A similar situation may exist for the clock.  Since the Rb option is fitted, the unit is strapped to always use the Rb.  If the Rb option wasn't installed, the unit would be strapped to use the internal oscillator.  I believe there's also a strapping option to use the internal high stability OCXO instead of the basic oscillator.  Some of this could be different with your unit.  I've got the PM6681 which is quite similar to yours, but may not be identical.

Does your model have math functions?  I often set mine to make a measurement and then subtract the expected value from the measurement so the counter displays the error.  This gives me a couple of extra digits of resolution without having to count all those zeros!

Extending the gate time will give you a more stable reading if necessary.  This is particularly helpful when measuring something like the GPSDO which often has short-term wobbles.

When I *REALLY* want to make good measurements I do crazy things like set the gate time for 100 sec., make ten measurements, and have the counter display the average of the ten readings.

Ed

ejd.pol:
Hi Ed, Just to give you an update on the state of things:

-As far as I can tell, the fan is connected to the proper connector as indicated on the main PCB. As remarked earlier, it may be that the Rb oscillator is fine without it in stand-by.
 The oscillator is bolted to quite a large heat sink, being the floor plate of thick solid aluminum that runs completely front-to-back, and had good thermal contact
 with the external metal box. So, if the power consumption of the counter itself is low enough in stand-by, the oscillator may be cool enough.
 When the counter is switched on, the fan will help thermal flow to the external box.
 This also provides motivation why the fan is allowed to run on a supply that is different from the Rb oscillator supply....

-Speaking about thermals, I have disassembled the new supply in order to drill some holes in the three enclosures. It should not be urgent or even necessary,
 but I thought it would help to reduce temp inside the three boxes as much as possible. The metal supply protection box doubles as a heat sink as well.

-Doing precise measurements: I would hope (have not investigated this point yet) that, in stead of making 10 measurements of 100 seconds each,
 it is possible to set up a measurement of 1000 seconds. When doing so, the statistics of the results would improve, I think. Or am I missing something?

Anyway, it has been a great experience doing this repair, and I would like to thank you for all your help and guidance!

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod