REPAIRING FUNCTION GEN - Damaged my IEC F34 Function Gen - Really dumb mistake

[FlyingHacker]

I was calibrating a DMM, for which is use the function gen, and I managed to do the final check of 120VAC by connecting up mains in parallel... But I forgot to disconnect the function gen! SO I dumped mains into the output of the function gen

There was no smoke or flash, and it took me at least 30 sec to notice.

The function gen is an IEC F34, which is a really nice 3MHz analog generator/sweeper.

The function gen still puts out its waveforms, but I blew something in the output section. I only see waveforms with the output attenuator at full, and they are much weaker than they should be. (~1/10th). The other settings on the attenuator output nothing.

I gotta take this guy apart and see what I fried. Perhaps the output amp transistors, or perhaps just some series resistors. Guess I will find out when I get a chance to look inside.

So don't do what I did! At least I didn't get shocked.

I may be asking for help if I can't figure this out, but IEC provided lots of test points and waveform illustrations. Hopefully nothing is unobtainium.

[Gyro]

Life happens, at least you didn't fry! 

If your output waveforms are undistorted (albeit lower amplitude) then you're probably lucky enough to have not fried anything active. It sounds as if at least some of the attenuator resistors have fried and maybe the output impedance setting resistor.

There might be some odd values in the attenuator but you can probably make those up with series or parallel combinations.

[FlyingHacker]

Thanks for the reply... That is my hope. The AC was at least isolated from the mains by an isolation transformer. So this likely helped a lot.

I have repaired lots of equipment where somebody somehow applied mains to the wrong connector. I used to laugh at them, wondering how you could possibly apply mains to an output by accident. This is one way!

[FlyingHacker]

Been working on troubleshooting this. There was no obvious damage/burning.

Manual and schematic here: https://drive.google.com/file/d/0B6oqWuYTCiclOUI0d1RzSVRROXM/edit?usp=sharing

Recall I get a good waveform (sine, triangle, square, etc.) on the output when the attenuator is off (at the 10Vpp (off) setting the attenuator is effectively bypassed). However, the amplitude is weak and setting any attenuator setting (or using a 50ohm teminator on the scope) flattens the output wave. In other words the wave doesn't have much behind it, and is easily dragged to ground.

I checked the resistors in the output stage. They look good. I check the step attenutator by measuring the resistance from the input to the attentuator (with power off) to the output BNC jack. Essentially I passed my meter through the entire attenuator assembly. The step attenuator works fine, as do the pots.

I found Q19 was getting quite hot. This is labelled as a P00347006 npn BJT. The part itself has a Motorola logo and says 7009 on one line and 312 on the next line. I would really like to find a replacement for this, but I don't seem to be able to find this part. I know the P00347006 number is an IEC (the function gen manufacturer) part number. However, I think it may still be a standard part. I tried replacing Q19 with a 2n2222. It messed up the waveform quite a bit, however the unit still had the same problem. So I don't think Q19 is the main problem, though it may also be faulty. My transistor tester (Chinese ATMega thing) and my hFE meter had trouble reading it. So I do want to replace it.

I then found Q18 was also running hot (should have checked there per the troubleshooting in the manual as well). I pulled it out and it measured as two resistors! I think it is definitely shot to hell. I replaced it temporarily with a 2n2907. This seemed to get things working much better. The waveform is still messed up (could be either or both of the replacement transistors which are not the spec'ed BJTs), but with the replaced Q18 the step attenuator no longer drags the output waveform to flat line. I need to get a 2n3251A to replace Q18 properly.

I am not certain, but I think these two transistors may be the major (or possibly all) the damage caused by inputting 120VAC mains (the mains was from an isolation transformer) to the output jack.

Anyone know what this Q19 is? This is labelled as a P00347006 npn BJT in the BOM. The part itself has a Motorola logo and says 7009 on one line and 312 on the next line.

Anyone? Bueller?

Thanks!

[Kleinstein]

Q19 and Q20 should be about complementary types. So the 2N2222 is not such a bad choice as a replacement for Q19. Because of better heat removal better the TO39 version 2N2219. Just the wrong transistor type should not cause much distortion - if at all it might cause the amplifier to oscillate, upset the bias (not likely in this case) or just break down prematurely if to small.

Similar I would not expect much trouble from having the 2N2907 instead of 2n3251 - might cause some trouble at the high frequency end and maybe a little offset, but no major distortion.

So there is likely something else broken, like Q33 or Q20. I would check the voltage over R123 and the waveform at the base of Q19/Q20.

[FlyingHacker]

Q19 and Q20 should be about complementary types. So the 2N2222 is not such a bad choice as a replacement for Q19. Because of better heat removal better the TO39 version 2N2219. Just the wrong transistor type should not cause much distortion - if at all it might cause the amplifier to oscillate, upset the bias (not likely in this case) or just break down prematurely if to small.

Thanks for the reply. The 2n2219 was going to be my fallback choice because it is the complement to Q20. It is possible I need to adjust R119, the zero adjust for the output stage. I will posts pics of the waveform I am getting. The thing is that the waveform looked fine (just low amplitude) before replacing Q19. I am tempted to try putting the old one back in, even though I cannot get it to test right. Sometimes transistors don't test right anyway.

Similar I would not expect much trouble from having the 2N2907 instead of 2n3251 - might cause some trouble at the high frequency end and maybe a little offset, but no major distortion.

So there is likely something else broken, like Q33 or Q20. I would check the voltage over R123 and the waveform at the base of Q19/Q20.

I will have a look. Thanks again.

[Seekonk]

That reminds me of a salesman that brought me a lot of work. I serviced stereo equipment in college to get extra money.  This salesman told everyone to buy extension cords instead of speaker wire and cut it in the middle.  Then when you had a party on the deck you could move the speakers outside and connect another extension between the two for the extra distance.  Obviously there was drinking involved and someone would always plug the amp output into a wall outlet.

Had a tech that would always connect up things wrong.  When he destroyed a signal generator, I put a lamp in series with the output.  We never had a need for low impedance.

[SaabFAN]

Have you checked the Trim-Potentiometers for continuity?
In the PM 5190 Signal-Gen I repaired a few weeks ago I found one pot that was Open Circuit, causing distortion and strange behavior when I tried to adjust the DC-Offset.
Soldering a resistor across the terminals solved the problem in my device.

My device also had some High Voltage-accident, although I don't know if that also killed the pot. It definitely killed the DC-Set DAC and the CPU, even though they aren't anywhere near the Output.

[FlyingHacker]

That reminds me of a salesman that brought me a lot of work. I serviced stereo equipment in college to get extra money.  This salesman told everyone to buy extension cords instead of speaker wire and cut it in the middle.  Then when you had a party on the deck you could move the speakers outside and connect another extension between the two for the extra distance.  Obviously there was drinking involved and someone would always plug the amp output into a wall outlet.

Had a tech that would always connect up things wrong.  When he destroyed a signal generator, I put a lamp in series with the output.  We never had a need for low impedance.

Haha. Yeah, it was like that, but with test leads connected to banana jacks, stacked on other banana jacks... I should have paused a moment for reflection before connecting AC, like I normally do.

[FlyingHacker]

The trim pots seem OK.

I am now suspecting IC9, which is an obsolete part I have to source off eBay. RCA CA3018A

The base of Q19 has pretty much nothing on it in terms of a signal. Just some noise (presumably from the power supply because it is 120Hz).

Pins 11,12,8,5 of IC also pretty much have no signal. This is why I am thinking IC9. Ping 9 of IC9 has the correct (albeit low amplitude) signal on it.

Q17, which turns out to be a 2n3251A rather than the part number indicated, tests fine. I wonder if I should replace with a 2n2907 with gain matched to Q18.

I have not yet pulled Q23 or Q20, (they are a pain to get to, requiring desoldering a bunch of wires from the attenuator assembly) but the voltage drop across R123 was pretty low, like 0.03V or something. Correct me if I am wrong but that leads me to think Q33 is at least not shorted from emitter to collector.

My current plan is to try a new IC9 then try gain matched 2n2907s for Q17 and Q18. If everything else works, but those have trouble then I will find some 2n3251A's to put in there. I just don't want to have to buy a bunch of those to find a decent match in gain. Do you think that is important here?

Thanks.

[Vgkid]

The ca3018 is a transistor array. You might be able to find an equivalent still in production.

[FlyingHacker]

Thanks Vgkid. I did find some on eBay for a reasonable cost. (Not 2n2222 reasonable as in 3 cents, but a couple dollars USD delivered.)

[Kleinstein]

The low voltage on R123 is indicating something is wrong with Q33 or maybe VR7, R135. Of cause it could also be that Q18 is not delivering a current and thus a to negative voltage at the collector of Q33.

Checking the DC levels might be a good idea. If some transistors are broken they should be way off. This should show if it is IC9 or another transistor.

It's no surprise to have the same transistor for Q17 and Q18, but matching should not be that critical. If not the same a slight offset and maybe more offset drift is possible, but nothing dramatic. So I see no problem with Q17 either a 2N3251 or 2N2907.

Ideally there should be very little (e.g. < 1 mV at low frequencies) signal at pin 9 of the IC. That amplifier is build as an inverting amplifier, having pin 9 as a virtual ground.

[FlyingHacker]

Thanks for the reply, Kleinstein.

I really don't understand this circuit so well... Forgive my hobbyist knowledge level.

I pulled Q33 and it tested as a pair of ~38k resistors... Bad stuff. I do have a replacement, as my local store did have those. How did you know what the voltage drop should be across R123? If you have a moment I would appreciate your thought process. I get the Ohm's Law bit, that it is passing ~59uA.

You mean check DC levels around this part of th circuit, or overall? Overall, at the main test points for power supply voltages they are within spec.

I think Pin 9 of IC9 has on the order of volts in terms of signal. I will remeasure, but the theory of operation section says the waveform switch has 6V peak to peak. I have Q17 out right now or I would remeasure right now. I thought pin 9 (right next to R343 to ground) was the input??

Thanks for the help.

[FlyingHacker]

More info (in addition to the post right above this).

I replaced Q33 now I get a 0.51V drop across R123

R114 has a 21.8V drop across it

The output of the generator now is pure DC at about -12V. The attenuator works on this, scaling it properly.

On IC9 pin9  I get the correct waveform (still at the lower amplitude of 3Vpp)

I get the same waveform on IC9 pin 2. I wonder if IC9 shorted the diode from pin 9 to pin 2 ??

I get no appreciable waveform on any other pins of IC9


Hopefully my new IC9 will show up soon.

Any other thoughts?

Thanks.

[FlyingHacker]

So I am pretty sure with this new symptom of -12ish volts on the output the final culprit will be IC9. I lifted a leg as fully tested VR7, the 6.8V Zener. It was fine, the troubleshooting guide conveniently has a this exact symptom in it and it claims that it is a faulty IC9. New ones should be here by Thursday. 

Now, if I can just keep myself from applying mains to the output again like a total Bobby Doozer I think it should be good! 

[Kleinstein]

The circuit is very much like a classic class B (maybe AB) audio amplifier. IC9 is a long tailed pair of darlingtons, Q17/Q18 are a second differential stage (instead of the more classical emitter type) to do the voltage amplification. Q33 with R123 and VR7 is a constant current source - so it should have more like 6 V over R123.
Different from most audio amps it's used in the inverting configuration - thus the input at PIN9 if IC9 should be close to GND, with little signal (at least a low frequencies).
 
Only 0.51 V over R123 still indicates there is not much current flowing. So there might be still some trouble around Q33 - maybe Q20 broken.
So even if difficult to get there it might be worth checking Q20. The DC voltages at it's pins and the voltage over R127.

Having the same waveform on pin2 as pin 9 is normal, it should be just 0.7 V lower in DC level. Also the 22 V over R114 is about right. As is looks like the amplifier is saturated in one direction, it's normal to not see an AC signal at the other pins - the DC voltage at pins 5/8 and 11/12 should normally be at about +18 V and very similar on both sides. Saturated this should be more like 22 V and 12 V.
If IC 9 is OK it should be the lower voltage at pins 5 and 8.

[FlyingHacker]

Kleinstein, thanks for the great explanation. I had been refamiliarizing myself with some of the building blocks of this circuit yesterday.

I had measured the voltage across R127 and thought it looked OK, but I don't recall why I thought that. I shall have a thorough look again soon, and get those readings around Q20.

Gut feeling would tell me to be very suspect of Q20 because all the other silicon fried! Now you have helped me realize how that could influence the voltage through R123.

I understand what about pins 2 and 9 being a diode drop apart, but the same waveform. Makes sense now.

I will measure the voltages around IC9. I had actually done this for an earlier posting, but the browser crashed and I lost them 

I am a little concerned about the amplitude on pin 9 of IC9. I understand this is inverting. The theory of operation section seems to indicate that the output of the waveform switch S3 should have around 6Vpp, which it is. So wouldn't R108 and R343 make a voltage divider and knock that down to around 3.9Vpp on pin 9 of IC9? Or am I misunderstanding something else?

Thanks so much for all your help. I am learning a lot in the process. I am at the stage where I grasp most of the basic theory, but don't know many building block circuits. I have heard of them, and know what they do as black boxes, but I am now learning to identify more of them.

I will take some more readings and see if I can (possibly with your help) determine if Q20 is indeed bad.

[FlyingHacker]

The negative rail side of R127 is -23.39V. The Q20 side of R127 ( collector of Q20) is -22.82. When I measure directly across R127 I get 0.478V


The Q20 lead of R126 has -12.77V this is the emitter of Q20

The base of Q20 is -22.90V

The base of Q19 is -21.60V

On IC9: DC voltages to ground

Pin 1 : -1.712
Pin 2 : -2.440
Pin 3 : -0.016
Pin 4 : -0.703
Pin 5 : 9.90
Pin 6 : -0.699
Pin 7 : -1.472
Pin 8 : 9.91
Pin 9 : -1.771
Pin 10: -1.674
Pin 11 : 23.20
Pin 12: 23.20 (was wrong earlier)

Frequency set to 3KHz

Output BNC is still a flat DC line at around -12V

[FlyingHacker]

R135 has 21.8V across it.

It measures as 1.3Meg in circuit... That can't be good...

EDIT: False alarm... Must have been reading it poorly, as when I clips leads on it to jump in another resistor it is reading around 4K.

EDIT REDUX:  It was due to residual voltage in the capacitors. Once I short those it reads correct.

[Kleinstein]

The readings for Q20 indicate a problem here - so it's very likely burned: about 10 V from base to emitter is too much. With a working Q20 the output should be more at -21 V.

The readings around IC9 still look reasonable for an intact IC9. No indication for a problem there. However it looks like Q18 is not working correct - maybe it is broken again, possibly because Q33 and Q20 were still bad.


The amplifier works very much like an OP as inverted amplifier - so Pin0 of IC9 should be a virtual ground, with a rather low voltage. The current from the signal source and offset pot is balanced by the feedback bath.
Only hat higher frequencies there should be some measurable signal.

[FlyingHacker]

Thanks for the reply. I will replace Q20, and Q18 and see where we stand. I did get some IC9 ICs in the mail. So I have that too.

I am busy with work (which is not EE stuff as you can tell) for a few days, though. So I won't get to look until next week.

Thanks again.

[FlyingHacker]

So I finally got a chance to work on this guys again (holiday and work...).

I replaced Q18 and Q20 as I said above and it works!

I seem to possibly be having some temperature issues, though. How hot can the case of a TO-92 2n2222A get? The max junction temp is 150C.

The case temp of Q19, with a K-type thermocouple held directly against the flat side (case open).is 92C !! That seems pretty hot.

I am thinking about getting the original metal can transistor parts instead. The originals had these round heat sinks on them.

Thank you so much for your help. I really love this generator for general use. It has a nice big knob for the frequency, and all kinds of nice features. IT may not have the 40MHz of my Tek FG504, but it is still pretty sweet.

The square wave in the scope is the sync/trigger output, and is not terminated correctly.

[Gyro]

Congratulations on the fix!

It nicely demonstrates how wrong my advice can be: 

If your output waveforms are undistorted (albeit lower amplitude) then you're probably lucky enough to have not fried anything active. It sounds as if at least some of the attenuator resistors have fried and maybe the output impedance setting resistor.

It does sound as if that TO-92 is running a bit hot for long term reliability, the thermocouple wont be coupling terribly well to the plastic package so the temperature is probably higher. If the original transistor had a heatsink then you probably need more cooling. You can get little clip-on brass sheet heatsinks for TO92s but you might be as well selecting a different transistor and re-using the original heatsink.

[FlyingHacker]

Thanks, Gyro.

Well, honestly, even given that I am resorting to eBay for the original replacement transistors, I still think the oddball precision resistors in the attenuator section would be harder to find. So in a way it is good you were wrong. The attenuator was my first assumption as well, but it was wrong too 

Kleinstein's assistance was invaluable. Thanks.

I am eager to get the final parts in there. Removing the attenuator a second time to access Q20 is a pain, but at least I know that it works.