On 5/25/2013 12:43 AM, Mark C. Stephens wrote:
Hey Ed, Just knocked up a temperature sensor with a Dallas single wire temperature sensor thingy, a pic and an LCD display. So, I will know for sure I am hitting the xtal turning point. I figured you had been right about everything so perhaps my temperature gun was out. It's probably a handy bit of kit to have, because no doubt I have a couple of ovens here that are playing up ad some sort of accurate temperature measurement and logging would be a handy bit of kit to have. It spits out the time and temperature reading once per second via a serial interface as well as displays in on the LCD.. I had already sort of written the code 10 years ago, just hadn't gotten around to finishing the hardware. :) A few changes here and there for the pic pin out and it worked! Just going through the warm-up cycle now, I expect an hour should do it. Seems to be heating that crystal very slowly. I think there still may be a problem around that wee transistor that limits the current to the heater transistor. Or perhaps by putting the right value 6K81 resistor and a new op-amp changed the dynamics and I may need to revisit R8. I have checked the PSU, its 17.3v spot on. I checked the temperature after an hour and it was only 58 degrees ): I stuck a resistance box across R8 and adjusted until I got ~79 on my temperature meter. The resistance is 680R which seems very low to me, but not outside the circuit specifications.
As you change R8, watch the frequency of the oscillator as it changes. When your temperature meter shows 79C, is the frequency at its lowest point? Although the frequency changes with age, I've never seen anything to suggest that the turning point changes with age. Anyone reading this, please correct me if I'm wrong. Regardless of the actual temperature, you want to adjust the crystal temperature so that you hit the turning point. For an AT crystal, that will give you the lowest frequency. Once you get to the turning point you will see if you will be able to get it working by changing the capacitors. For test purposes, if the crystal frequency is only slightly out of range, change the temperature with R8 until you can get to 10 MHz.
But appear to have dug myself into a hole.
It looks like the hole was already there. But you kind of jumped in with both feet! :)
I disconnected the EFC as per the manual and hooked up a 10K 'Pot' across the 17.3v supply and the wiper to the crystal boards E9 point (EFC) At zero volts the 10 MHz is very high at around 10000500 and at 17.3 volts it is around 10000650. So, there is either a problem, with the crystal or the surrounding circuit.
If the crystal really has drifted that far you might have trouble pulling it back. Check the temperature as shown above. Note that your frequency is too high and that's just what you would expect if you were away from the crystal turning point.
Looking at the circuit C11 and C12 are selectable components. C11 selects the centre frequency and C12 selects the adjust range. The range in the circuit is specified as 4.7 to 100 pf, but my schematic is a bit blurry so it could be 4.7pf to 180pf. So I went from 10pf -> 100pf for C11. At 100 Pf the 10Mhz is closest at ~10000300 (with 0v EFC) There is no way I will hit 10Mhz by adjusting C11.
My copy of the schematic isn't any clearer than yours. But, who cares what the manual says? Load on more capacitance and see what happens. If it works, it works. Make sure that any capacitor you use there is NPO/C0G.
I guess my next step is to pull the board out again and check components, especially C3/C4/C5.
The manual only talks about C10, C11, and C12 as affecting the frequency. C3 looks like it might resonate with T1 to couple the 10 MHz to the next stage. A fault there would affect the output amplitude, but not the frequency. It wouldn't hurt to check C4 and C5. Running high in frequency suggests that a capacitor has dropped in value which is a likely failure mode if, for example, it cracked.
No hope of lock when the EFC can't pull the Xtal into the correct frequency is there.
Nope. The oscillator has to hit 10 MHz and it should have room to spare on the EFC to accomodate shifts in temperature, voltage, aging, etc.
I am wondering if we getting off subject for the list?
If a repair saga of a relatively common Rb standard is off-topic, the list is in a really sad state.
Final point: SLOW DOWN!!! You're changing too much, too fast. You're taking the thing apart and putting it together multiple times. This circuit is over 20 years old and has spent it's entire working life running hot. The boards are getting brittle and the glue holding the traces onto the board is fragile. There are parts that are completely irreplaceable. You're going to ruin it if you keep going the way you are. I know you're enthusiastic, but less speed often results in more progress.
Ed
marki -----Original Message----- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of Ed Palmer Sent: Friday, 24 May 2013 4:03 PM To: time-nuts@febo.com Subject: Re: [time-nuts] 9390 GPS RX Hold on mate! You're not quite done yet. :) You said you tweaked the cap to get the Control Voltage to 8 volts and saw the proper frequency. You should be able to move the cap, and therefore the control voltage, anywhere in the 2 to 12 volt range and still see the proper frequency. Does that work? The next area is the Rb lamp itself. What's the voltage on pin 7 (Rb Lamp)? If it's much below 7V you could have trouble with the unit dropping out of lock. While you've got the unit on the bench, you should also check the temperature of the lamp since it's easy to do. Don't try to measure it while it's powered. Open up the access port to the lamp, power down, and measure the temperature. The tolerance on the temperature is not stated, but since it's running at 115C, too hot will burn out the poor transistor that's being used as a heater while too cold will give a weak light that could cause locking problems. Finally, power the unit on from cold and make sure that the frequency sweep still works properly and after a few minutes the sweep stops, the crystal control voltage stabilizes, the frequency looks good, and pin 5 (Resonance Lock) goes from open to ground. If everything looks good, you can consider your repairs complete. The manual lists other things that you could check, but in my limited experience, if the above tests are good, it's not worth doing any more. Ed On 5/23/2013 7:31 PM, Mark C. Stephens wrote:You were spot on about the negative efc thing. Turns out I was using a bad negative point. I used the supply negative and I saw the EFC was sitting at about 3.5Volts. So I tweaked the Cap to 8 Volts and what do you know, my counter says 10 000 000 07. Dead on 10 Mhz, we have done it Ed, you and I! :) I want to do some checks for spectral purity and drift before I install the 9390's Ball back in. Good stuff mate :) -marki -----Original Message----- From: Mark C. Stephens Sent: Friday, 24 May 2013 10:37 AM To: time-nuts@febo.com Subject: RE: [time-nuts] 9390 GPS RX Ed, I nailed it, R7 2M ohm, a crappy old carbon composition resistor had gone high to 3 Meg in the positive feedback loop to the 741. Once replaced The temperature still wasn't hitting 79 degrees I checked all the other resistors and found R12 6K81 was a little low at 6K79 so I went through my metal film boxes and found one that was spot on 6.81K. Replaced that' but still not the right temperature. The only thing left was the thermistor and the 741. The thermistor appeared okay at 18K cold so that left the 741. Luckily, I had some new 741HM metal can jobs in stock. Once it was replaced all is now hunky dory. Temperature is a steady 79 degrees. Multiple component failures are a worry to me, I can explain the 2M resistor but the 741 failing is not good. Also the back of the board is lacquered. When I used my PCB cleaner for clean the solder flux off, it melted the lacquer leaving a sticky white coloured goo on the back of the board. I had to get a tooth brush and scrub it all off and apply a new coat of lacquer. Didn't see that one coming ;) Well I am happy to say that The output isn't sweeping anymore, however its locked to the wrong frequency! The counter I trust implicitly, I have 6 house standards that all read 10 000 000 07 on it so I know exactly what 10 Mhz is on it. For some reason the FRK has locked to 10 000 117 54 and there it stays. According to the troubleshooting flowchart it's the A21 Crystal oscillator board. They mention one should set the trimmer cap to midway and disconnect the wire off E9 and apply +6V there. Form there you have to pick 2 values for 2 caps to set the centre frequency and adjust range (C11 and C12) But with my luck its probably not that easy.. I really should get some sleep... -marki -----Original Message----- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of Ed Palmer Sent: Friday, 24 May 2013 5:02 AM To: time-nuts@febo.com Subject: Re: [time-nuts] 9390 GPS RX I applaud your enthusiasm, but Slow Down! Don't go randomly trying things. The more you touch, the more you break! If you damage the Rb lamp, you will not be able to find a replacement. You shouldn't have touched it without a good reason. 1. Did you confirm that the frequency sweep isn't getting to 10 MHz? What are the frequency limits on the sweep? Are you sure about the calibration of your counter? Since you've already adjusted the trimmer you don't know what the values were before, but can you adjust the trimmer so that the sweep includes 10 MHz exactly? If your counter isn't up to the challenge, put your house standard on one channel of your scope and put the FRK output on the other. Trigger off the house standard and if the FRK is centered around 10 MHz you should see the FRK as a blur that eventually slows down so you can see it scrolling left or right, stops, and then starts scrolling the other direction until it's a blur again. If it's just a continuous blur, it isn't reaching 10 MHz, but it's tough to tell whether it's high or low in frequency. Hopefully, your counter can handle that. 2. The temperature shift on the crystal seems high. I don't trust non-contact temperature guns. The emissivity of the target has too much effect on the reading. Do you have a thermocouple or solid-state probe that you can use? Before changing resistors, check for bad solder joints in the temperature sensing circuit and check the transistor that's bolted to the crystal enclosure. Is the bolt snug? It might be worth the effort to carefully remove the bolt, clean off the old heat sink grease, and apply fresh grease. Take special notice of the odd washer under the bolt. It should be cone-shaped and the outside of the cone should be against the transistor. I have one FRK where a similar washer was installed upside-down. The transistor eventually burned out. Maybe coincidence, maybe not. 3. The crystal control voltage should be moving up and down with the frequency, but it should be sweeping from ~ 1 - 12 volts. It can't go negative because there's no negative supply. Check the ground point of your meter. There are other things that need to be checked, but that's enough for now. Ed On 5/23/2013 10:38 AM, Mark C. Stephens wrote:Okay, I pulled the black box cover off and was greeted by a picture of how things should be made. After some considerable time staring at the veritable work of art, I thought I saw the crystal hiding behind the foam. So I gently removed the foam and marked on the xtal is 79 degrees C. Cool, Well actually that's hot! So I have powered it up for 30 minutes and measured the temperature with a very accurate temperature gun I use for measuring preheating on rework jobs. Huh, its sitting dead on 60 degree's. I think therein, lies our problem, the crystal never reaches turning point. According to the fine manual, Increasing R8 lowers the temperature. Therefore, If I decrease R8 It should theoretically, increase the temperature? The conundrum is, Shall I just change R8 value or perhaps there is something else wrong. What do you think Guys, R8 or something else? I am not going further without some of the abundant collective wisdom and experience contained in the time-nuts members! -marki -----Original Message----- From: Mark C. Stephens Sent: Friday, 24 May 2013 2:01 AM To: Discussion of precise time and frequency measurement Subject: RE: [time-nuts] 9390 GPS RX Ed, Okay, Lying there I couldn't stop thinking about its lone Ball. So I setup the jig, Powered it up and yep, after half an hour the frequency is doing the up/down swing thing. Interesting, I just happened to come across that page that reckons you can rejuvenate a lamp with a heat gun. So I thought, hmm that's easy, let's try that, so I did, and the frequency is not swinging so quickly now. Could be an symptom, but I think coincidence. Anyway, next I connected a DVM to the xtal volts output monitor pin as per the manual. Uhuh, its swinging in tandem with the frequency swing of the 10 Mhz output swing. Gets to 2V and then swings down to -2V Nowhere near the 8V defined in the manual. So I tried adjusting the trimmer as per the manual. Nope won't go above 2V at the fully clockwise limit of the trimmer. So that is where I am at. Looks like I will have to pull it apart and fix it then. Does it sound/look like the crystal has gone out of adjustable range? -marki -----Original Message----- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of Mark C. Stephens Sent: Friday, 24 May 2013 12:22 AM To: Discussion of precise time and frequency measurement Subject: Re: [time-nuts] 9390 GPS RX Hey Ed, Well I have its Ball (just one? ;) Efratom out on the bench. Seems they built this stuff to service, huh? Nice. I managed to make a mating connector out of 2.54mm pitch header strip as per your suggestion. Whew, that was my biggest concern! Anyway, its after midnight and I have a full day ahead, Looking forward to getting this old gal working again. I can't help giving the 9390 a reassuring pat every time I walk past it ;) its sitting there merrily running off the house standard although PDOP hasn't gone below 4 which has me a tad worried. It seems, I have become quite attached to this 9390 :p -marki -----Original Message----- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of Ed Palmer Sent: Thursday, 23 May 2013 1:16 AM To: time-nuts@febo.com Subject: Re: [time-nuts] 9390 GPS RX On 5/22/2013 4:58 AM, Mark C. Stephens wrote:It's a 9390-55024 I have plugged my counter into the Efratom rubidium oscillator thing and disconnected the EFC. It is actually wobbling ~ +/-650Hz, peaking as much as +/- 1KHz. So, hazarding a guess, something is very wrong inside the black box thing.Is it really wobbling around randomly, or is it sweeping up and down. Your counter could be fooling you by taking 'snapshots' at intervals dependent on the gate time setting. If the frequency is wobbling randomly that would likely mean an actual fault in the FRK rather than something that needs adjusting. If it's sweeping up and down, that means that it's searching for the rubidium signal, but not finding it. Every rubidium will do that, but it should find the signal within 5 - 10 minutes. That could mean an adjustment or a fault depending on the actual frequency range that it's sweeping over. Obviously, you want the frequency range to be 10 MHz +- something.My biggest fear is, wherever will I get the 'Winchester' connector used on this oscillator?That's easy. Improvise! Go into your box of junk connectors and find a female connector that uses pins of approximately the same diameter. Remove the sockets from the connector body, solder wires onto them, insulate them with tape or heat shrink and slide them over the pins on the FRK connector. For example, RS-232 sockets are a bit big, but could be squashed down. Floppy disk connectors are usually really easy to remove from the body. I'm not sure if they'd work, but you get the idea. It doesn't have to be neat or elegant. It's just temporary for a test. EdWhat I mean is, for a proper bench job, according to the manual, I'd have to remove the whole rubidium from the 9390 and put it on the clean bench for disassembly. Then I can connect various voltmeters to monitor the various signals, and if needed, replace parts and re-align. However, I am unwilling to lop off the connector in the 9390, to only have to put it back on/in after Mr. Balls Efratom is back to normal. The other way, I suppose is to bodge up wires from the inside of the connector to some sort of temporary Jig for the service job. Reading the Most Interesting FRK.PDF, it sounds like the crystal oscillator assembly has issues, whether the crystal oven is broken or similar. That is a really interesting document, worth the read, even just for the heck of it ;) -marki
_______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
