EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: BreakingOhmsLaw on May 15, 2020, 01:26:02 pm
-
Well, the moment that i have always dreaded has come. My trusty Efratom FRK-L oscillator stopped locking.
These units are notorious rabbit holes, but fortunately a service manual is available here:
http://ftb.ko4bb.com/manuals/176.198.185.59/Efratom_FRK-L_Service_Manual.pdf (http://ftb.ko4bb.com/manuals/176.198.185.59/Efratom_FRK-L_Service_Manual.pdf)
Thanks to KO4BB for that, thanks mate! :-+
I followed the troubleshooting guide on page 29 to the very last entry - "Check resonator and Lamp temperature".
And - sure enough - the lamp thermostat did not heat.
It took me a while to find the culprit - there is a "jumper" that allows you to select external heating voltage. My unit has internal heating, so that could not be it, right? Wrong.
On the place where i expected to find a jumper, i found nothing. Only when i started scouting the PCB under the Mantis i found this:
[attachimg=1]
In the location where the probe is pointing, there should be what i think was a solder jumper. Instead there's a crater in the PCB. I guess it has evaporated. You can see a little bit of copper trace left.
Anyway, i bridged it to the test point to the right, and indeed, the heating is back.
But.....still no lock. Something else is foul. Doh! |O
Down the rabbit hole i go...
If anyone has experience fixing these, feel free to make suggestions. Currently I am crawling through the servo board. The 127Hz is just 125Hz, bit I don't think it's an issue since it is divided down from the related 254Hz in the lamp regulator loop. I haven't checked the resonator temperature yet, but it *is* heating so for now i don't want to mess with the physics package if not absolutely neccessary.
-
Well, it seems to be locking *once* when the unit is warmed up and power is turned on.
It then locks for 20 seconds...aaaaannd it's gone.
I can see the the control signal for the VCO is still rising once lock has been achieved, and eventually it hits the upper rail and loses lock. It then has to be restarted to lock again.
My suspicion is that something is leaking and the integrator that controls the VCO gets a DC bias it's not supposed to. Next step is replacing the CD40538 triple analog switch that feeds the input signal to the integrator. Have ordered some abroad, so let's see when they arrive. It could be a leaky compensation cap of the integrating opamp as well, I will test that when i get the CD40538 so i can leave the unit assembled for now.
If anyone has a working unit, i would be grateful for a scope screenshot of TP1 from a working unit. There's one in the service manual, but it looks nothing like my signal when under lock.
-
What other tools and toys do you have access to?
One of the first things I always suggest is to check whether the unlocked frequency is sweeping up and down and confirm that it's sweeping through the rated output frequency. That requires something like a GPSDO or a calibrated frequency source and/or counter. The FRK should sweep over a range of ~15 Hz.
The other thing to check is that all the oven temperatures are correct. If any of the temperatures are off, you won't achieve lock. A quick cruise through the manual shows the lamp should be ~115C, the resonance cell should be 75 - 78C. The OCXO wasn't listed, but one unit I've got has 77.1 written on the crystal.
If the OCXO is off temperature, that will shift the frequency. You'll see that when you look at the sweep range. If the temperature is off, fix that first. Then, if necessary, adjust the sweep range.
To measure the lamp, you'll probably have to turn off power and then poke the thermometer through the hole in the heatsink. There's so much RF around the lamp that it tends to scramble the reading from my probe. Don't use an optical thermometer, use either a semiconductor or thermocouple thermometer - basically any kind of contact thermometer.
The resonator temperature is measured as described in the manual. Be careful! You're pressing the probe against a piece of glass. If you break it, your FRK becomes a parts unit.
The lamp voltage is another major thing to check. If it's too low, nothing will happen. The lamp temperature has a huge effect on the lamp voltage so you have to check the temperature first.
Then there are obvious things like power supply voltages, ripple, etc.
I've never had to really dive into the FRK so I don't know a lot about the internals. But I do have one that needs repair so maybe now is the time. :)
-
Hi edpalmer,
The FRK is - or was - my labs time reference. I do have a GPDSO ordered from China as i was thinking along those lines, but it hasn't arrived yet. According to my Siglent 1202X-E, the OCXO Frequency stays exactly at 10MHz during the sweep. But: it does not have enough digits to show a 15Hz difference anyway I haven't measured with my frequency counter, as it normally takes the FRK as external reference. But i will try to see if i can see the OCXO sweep, even if the absolute value is wrong, so thanks for that tip.
For now i have measured the lamp temperature only. What i did was press a PT1000 against the metal casing of the lamp, basically covering the lamp inspection hole and pressing it place with the detachable lamp cover and some foam. The temperature, according to the HMC8012 that the sensor connects to, stops at 88,4 C. However this is a 2-wire measurement, the real value must be lower The FLIR camera pointed at the heater tranny shows just under 100. I did not measure directly on the glass, as the opening is only around 2,5mm and i've no probe that fits. I don't trust either measurement, my gut feeling tells me they are both lower than reality. Have to order a smaller PT100 that fits on the tab of the TO-220 heater tranny, the one i use is way to large, even to strap it to the lamp housing closer to the heater.
I did not measure the resonance chamber and OCXO temperature yet. They both are heating and since the unit worked before, I see no reason why the broken trace of the lamp heater could have offset them, so for now i went at the servo board.
I'm reluctant to mess with the physics package until i have ruled out anything on the boards, but of course you are right to point that out.
If the servo board turns out to be okay, i'm going to do a full alignment of temperatures and OCXO i guess.
Another problem i encountered is, that the service manual does not list the part number for the temperature sensors. Only parts mounted on PCB are listed, but the sensor is inside the lamp unit. So i can't tell if they are off. My unit also has no potmeter for lamp temperature. There is a fixed resistor mounted where the pot should be, which seems to be selected at test like many other resistors.
-
Hi edpalmer,
The FRK is - or was - my labs time reference. I do have a GPDSO ordered from China as i was thinking along those lines, but it hasn't arrived yet. According to my Siglent 1202X-E, the OCXO Frequency stays exactly at 10MHz during the sweep. But: it does not have enough digits to show a 15Hz difference anyway I haven't measured with my frequency counter, as it normally takes the FRK as external reference. But i will try to see if i can see the OCXO sweep, even if the absolute value is wrong, so thanks for that tip.
For now i have measured the lamp temperature only. What i did was press a PT1000 against the metal casing of the lamp, basically covering the lamp inspection hole and pressing it place with the detachable lamp cover and some foam. The temperature, according to the HMC8012 that the sensor connects to, stops at 88,4 C. However this is a 2-wire measurement, the real value must be lower The FLIR camera pointed at the heater tranny shows just under 100. I did not measure directly on the glass, as the opening is only around 2,5mm and i've no probe that fits. I don't trust either measurement, my gut feeling tells me they are both lower than reality. Have to order a smaller PT100 that fits on the tab of the TO-220 heater tranny, the one i use is way to large, even to strap it to the lamp housing closer to the heater.
I did not measure the resonance chamber and OCXO temperature yet. They both are heating and since the unit worked before, I see no reason why the broken trace of the lamp heater could have offset them, so for now i went at the servo board.
But when you think about it, the fact that there's a 'crater' where the trace was suggests that something fairly dramatic happened. Just replacing the trace sounds like a problem just waiting to reoccur. Have you measured the power drain during warmup and after? Is it reasonably close to the specified values?
I don't know if this was intentional or not, but at least some FRK bulbs have a tiny hole in the center of the metal base that allows you to see the purple Rb glow. Does yours? Can you still see the glow? The lamp voltage will also show you if the lamp lights - even if the level doesn't come up as high as it should. You'll see a significant jump in the voltage when the lamp lights. If the lamp lights, that suggests that it's probably not too far from the correct temperature.
I'm reluctant to mess with the physics package until i have ruled out anything on the boards, but of course you are right to point that out.
I understand your reluctance to go near the physics package. It's hard to tell where the fault lies. The physics package is the most irreplaceable part of the system, but it's also the starting point of the system.
If the servo board turns out to be okay, i'm going to do a full alignment of temperatures and OCXO i guess.
Another problem i encountered is, that the service manual does not list the part number for the temperature sensors. Only parts mounted on PCB are listed, but the sensor is inside the lamp unit. So i can't tell if they are off. My unit also has no potmeter for lamp temperature. There is a fixed resistor mounted where the pot should be, which seems to be selected at test like many other resistors.
Yes, that's why you have to measure the temperature directly. I have one very old Rb standard where one thermistor drifted so far that the resonance cell was running ~20C lower than it should have.
-
Yes, my unit has the hole and the lamp actually still ignited without the heater if i remember correctly. I think that it just needs the 77MHz from the Colpitts Oscillator to work, or that also heats the lamp to some extend. The power consumption is bang on, 1,8A on startup, once lamp ignites after 3 or 4 secs the power supply switches from Input voltage to clean 17,3 Volts and consumption drops to 1,5A and decreases gradually to half an amp over the course of 10 minutes. The power rails are all accurate (well, at least as accurate as you're going to get from zener diodes). And I totally agree on the thermistors. Anything that's being barbecued at 113C for years on end is going to suffer. I have new sensors inbound, i will take accurate measurements then.