Stumped on Avionics Power Supply Repair

[alexturner]

Hi there!

I'm in the process of repairing a power supply for an old aircraft radio, it's an interesting unit as a fairly old ringing choke converter with a small toroid on the board outputting, 12V, -26V, a regulated 5V for CMOS logic, 4.5V for an external instrument and 189V for a gas discharge display. As it is at the moment the power supply isn't outputting much, the 189V circuit is hovering at about 5V, and I'm barely getting anything on the other circuits.

I started by reviewing the oscillator circuit which is build around a Quad Op-Amp IC which is used as the oscillator (frequency source) as well as a couple of comparators. On review, the frequency source appears to be operating correctly and the power source (an external LDO) is producing a solid 5V for the quad op amp IC -  no issues there. The saw tooth oscillator is producing (albeit slightly off frequency) an appropriate waveform (measured at the input to the variable pulse generator):



Digging a bit further with everything connected, measuring the waveform at the output of the variable pulse generator (I302) I'm seeing an appropriately timed variable pulse - however perhaps the on-time is too short?



With this running as is, measuring at the test point on the collector of the larger NPN switching BJTs (Q303), I'm seeing a waveform quite different to that outlined in the schematic:



At this point I'm just taking components out and testing them one by one; which isn't smart. I'm trying to understand how this particular design works and I'm stuck at this chicken and egg problem where the op-amp by R306 is dependant on the 6.2V from the transformer. Similarly the feedback going into the comparator (with R308 as the negative feedback) is dependent on the 189V supply functioning.

I'm thinking perhaps a passive has failed causing the VPG to vastly shorten the on time which isn't giving Q303 enough time to saturate, thus, not ringing.

One thing I've also noticed which I find unusual is that without any load the board is drawing 1.5W which appears to be dissipated entirely in Q303 as it's getting very hot. I removed the transformer to test resistance across the various taps. Primary was sitting at around 1ohm which seems correct and all others were proportional to their output voltages.

At this point I've
- Replaced the regulator I303
- Replaced the Quad opamp IC
- Replaced all transistors on the board
- Tested all electrolytics
- Tested all passives in the feedback circuit by R302
- Tested all diodes
- Tested all passives on the switching circuit (R319, R322, etc)

Any thoughts on where or what this could be?  Schematic below for reference


Thanks so much for reading
Alex

[Fred Koke]

The quad device is a comparitor with open collector output and B part looks to be an overcurrent inhibit with R323 as the sense. Check that first, then overide the feedback or other parts to force a wider pulse.

[shakalnokturn]

Dead sure about R301 not open and R302 track to brush being good?

I can't agree with +6.2V coming from LDO.
LDO IC is supplied with +12V but is there to regulate +6.2V from transformer down to +5V.
Either because LDO needs more than +6.2V Vcc or as a way of power sequencing.

Looking at amplitude of your 3rd waveform on TP302 makes me ask what voltage you are using as input to the power supply?
I assume it should be 28V?

If it is running on 28V with a hot Q303 and so high Vce in on state I'd expect a secondary over current.

[alexturner]

@Fred, thanks for your reply. You may very well be correct, I've just removed R323 and it's reading 0.6Ohm which is far from it's rated 0.05Ohm however that's also dependant on the accuracy of my Fluke 18B+; easy enough to replace the part.

[alexturner]

Thanks @shakalnokturn
You're right about the 6.3V from the LDO, i misread the circuit - it's from the tap and the LDO is producing a regulated 5V which is not used within the PSU circuit, I've updated my original post.

The input is 13.4V~ (aircraft alternator which is similar to a car)

Dead sure about R301 not open and R302 track to brush being good?

I can't agree with +6.2V coming from LDO.
LDO IC is supplied with +12V but is there to regulate +6.2V from transformer down to +5V.
Either because LDO needs more than +6.2V Vcc or as a way of power sequencing.

Looking at amplitude of your 3rd waveform on TP302 makes me ask what voltage you are using as input to the power supply?
I assume it should be 28V?

If it is running on 28V with a hot Q303 and so high Vce in on state I'd expect a secondary over current.

[shakalnokturn]

I haven't worked on avionics equipment that much I have come across 115V 400Hz and 28V D.C. but wasn't aware of 14V D.C.

I assumed 33V Zener CR301 with R304 to be there to take out supply spikes on a standard 28V input.
There are also the voltage levels given on reference waveform for TP302, I don't see why the off state would be at 23V with a 14V  input.

[T3sl4co1l]

Looks like it's trying to switch a too-heavy load.  (Confirm by measuring R323.)  Shorted transformer?  Sure the diodes and capacitors are okay?  No shorted traces on the board?

Tim

[alexturner]

Good theory. I removed R323 to validate it's resistance - I don't have an accurate enough DMM so I attached it in series with a 1A CC load and measured it as 0.056Ohms, within tolerance.

Placing it back in the board, and measuring the voltage drop across is 0.008V so 0.16A (what the bench power supply is reading). The primary of the transformer read a couple of ohms when I removed it to test so I'm not so sure it's shorted - especially with the low current across it at R323.

The board looks fine, no visually failed components, no shorted traces. I think the next step is to remove the diodes on the secondary side of the coil and see if the load goes away (CR305, CR306, CR307, CR308) but I'm not sure what the effect of removing the feedback circuits on the 189v and 6.2v taps will do. Thoughts?

Looks like it's trying to switch a too-heavy load.  (Confirm by measuring R323.)  Shorted transformer?  Sure the diodes and capacitors are okay?  No shorted traces on the board?

Tim

[floobydust]

If R323 current-sense is OK what about C316?
The Q303 collector scope pic doesn't make sense - where is 0V? Is it 2V or 20V/div and if it goes negative CR304 is open.

[alexturner]

C316 is ok - not shorting and correct capacitance.

I've just reuploaded the scope pic from TP302 for clarity - May have been some user error in the first image.



Ignore the current (it was set to max for max excitement)

If R323 current-sense is OK what about C316?
The Q303 collector scope pic doesn't make sense - where is 0V? Is it 2V or 20V/div and if it goes negative CR304 is open.

[floobydust]

You've tested and replaced so many parts, I wonder if your component testing not thorough enough or something got overlooked.

I see Q303 does not really turn on much, 5V drop although it did not get a lot of time. Q301 should be doing a lot of work.
So I would scope TP301, and the base of Q302, Q303 to see why the switch section is lazy. What did you get testing the CR303 array for Vf?
You might have more than one problem, go after the narrow PW or the lazy switch. This is supposed to be a fun puzzle 

[dietert1]

The switch saturates at 5 V. Can you take a measurement at pin 2 of the primary coil? Maybe the primary coil has a short between 1 (TP302) and 2, so the driver Q302 doesn't work as intended.
Are you using a current limited supply for testing?
Any loads connected? This type of converter needs some load, at least on the output used for voltage control, in your case the 189 V line. The voltage divider around R302 may not be enough.

Regards, Dieter

[alexturner]

Re the parts testing, that's my thought. For your requested scopes:

TP301:


Q302 Base:


Q303 Base:


CR303 Vf is around .6V I believe.

You've tested and replaced so many parts, I wonder if your component testing not thorough enough or something got overlooked.

I see Q303 does not really turn on much, 5V drop although it did not get a lot of time. Q301 should be doing a lot of work.
So I would scope TP301, and the base of Q302, Q303 to see why the switch section is lazy. What did you get testing the CR303 array for Vf?
You might have more than one problem, go after the narrow PW or the lazy switch. This is supposed to be a fun puzzle 

[alexturner]

Thanks Dieter,

I don't have a load connected - 189V is hard to do in the lab, I have a programmable load but it only goes to 150V - do you think I could load up another coil or must it be on the voltage feedback circuit? When I reinstall the power supply into the radio, the symptoms remain identical.

Interesting thinking on the short, I've just removed the transformer and you may very well be right, I'm reading 0.0-0.1Ohm between Pin 1 and 2, and 2 and 3 - but how could it have failed?!
 

The switch saturates at 5 V. Can you take a measurement at pin 2 of the primary coil? Maybe the primary coil has a short between 1 (TP302) and 2, so the driver Q302 doesn't work as intended.
Are you using a current limited supply for testing?
Any loads connected? This type of converter needs some load, at least on the output used for voltage control, in your case the 189 V line. The voltage divider around R302 may not be enough.

Regards, Dieter

[T3sl4co1l]

Placing it back in the board, and measuring the voltage drop across is 0.008V so 0.16A (what the bench power supply is reading). The primary of the transformer read a couple of ohms when I removed it to test so I'm not so sure it's shorted - especially with the low current across it at R323.

Well, read the waveform.  0.16A with that kind of duty cycle has to be some ridiculous peak current.

Intra-winding shorts won't show up on an ohm test, obviously.  You have to use an impedance meter or signal generator.

Tim

[dietert1]

You can hardly detect a transformer winding short with an ohmmeter. The total coil resistance may be a fraction of an ohm and the loss (short) of one winding changes that by a small fraction. A short of one winding can be caused by many reasons, e.g. over temperature that destroys insulation, or arcing.
Unless you test the device with a current limited supply, you will never succeed, because each time you turn on another part may suffer. The current limit needs to match the size of the device. Can you show an image of the device so we get an impression of the size? Strange enough the schematic does not include a fuse.
Output load should be at least some fraction of normal. A large 100 Ohm resistor across the +26V output (transmitter supply?) may be a good match. It will take about 7 W of heat. But you can also use another output - select the one with the most solid rectifier.

Regards, Dieter

[floobydust]

Based on the scope traces, I see trouble at Q302, the base should not idle at 12.6V, recheck Q301, Q302 and the resistors around there.

[fzabkar]

I can't understand why there is a sawtooth on IC302-11. Surely the sawtooth should appear on IC302-10. Pin #11 should be a DC level that varies between 1.5V and 3.1V, thereby generating a variable width pulse.

[alexturner]

Got you - here are a few pics of the measurements of the removed transformer connected to my frequency generator. The 26V output is actually -26V and it's for some strange EEPROM on the microcontroller board - all the rectifiers are as large as each other.



This is between pins 1 and 3 on the transformer with 2 and 2 connected with a jumper


Pin 7 (12V)


Pin 6 (189V)


Pin 5 (5 V)


Pin 4 (-26V)


I'm not sure I see anything unhealthy there. I can put a load on the -26V supply, I have appropriate resistors.

You can hardly detect a transformer winding short with an ohmmeter. The total coil resistance may be a fraction of an ohm and the loss (short) of one winding changes that by a small fraction. A short of one winding can be caused by many reasons, e.g. over temperature that destroys insulation, or arcing.
Unless you test the device with a current limited supply, you will never succeed, because each time you turn on another part may suffer. The current limit needs to match the size of the device. Can you show an image of the device so we get an impression of the size? Strange enough the schematic does not include a fuse.
Output load should be at least some fraction of normal. A large 100 Ohm resistor across the +26V output (transmitter supply?) may be a good match. It will take about 7 W of heat. But you can also use another output - select the one with the most solid rectifier.

Regards, Dieter

[George Edmonds]

Hi

To check this type of transformer you do a "ringing test" monitor one winding with the scope and apply a single short pulse to another winding, you should see a decaying sine wave, if it decays in one or two cycles you have shorted turns, if it takes around six cycles it is OK.

You must load the  +5VDC (+6.2VDC) to obtain correct operation as it is primary voltage that the control feedback comes from.

George G6HIG

[fzabkar]

You must load the  +5VDC (+6.2VDC) to obtain correct operation as it is primary voltage that the control feedback comes from.

The 189V rail provides the feedback, at least after the supply starts regulating.

The inverting input of comparator I302A is the +5V reference for the error amp, so the non-inverting input would need to sit at this same voltage. The voltage at the junction of R331 and R329 is +1.8V (= 6.2V x 10K / 34K), so it cannot influence the feedback once the regulator has reached steady state.

If we set R302 at its midpoint, then the HV supply would be ...

5V x (371.1K / 10.1K) = 184V

If we set R302 at its extremes, then the range of HV is from 167V to 204V.

[fzabkar]

Based on the scope traces, I see trouble at Q302, the base should not idle at 12.6V, recheck Q301, Q302 and the resistors around there.

I agree with this.

[Gangway]

Greetings, just asking is this unit a Bendix/King. I have tried to repair same problem with  a KY-196 VHFComm.
Spent hours trying different ways to make this thing start up, no joy. Suspect transformer.
Seems to be very critical on loading. So now it’s on my list of “ when I get time” jobs,
Good luck

[alexturner]

Hey Gangway, it is! It's an ADF (KR87) and for those wondering it does appear to be a transformer issue. In testing the transformer (running 5v pulse at <100ma) I lost all output on the secondary - not sure if I totally killed it or if it was already on it's way. I went to order a new transformer but being Avionics the NOS part was going to run $300USD+. I managed to find a questionable unit on eBay for $100 which I'm waiting to arrive. TBD if I test the new unit and just use that, or replace the transformer / powersupply into a unit I know is calibrated an works. I'll update here when I get the parts none the less.

[Gangway]

Hi AT, thought I recognised the circuit diagram, very similar to mine. Feel your pain with parts cost. I paid $65 for one transistor ( germanium). I did purchase online svc manual, which was very good, but like most units, experience is the key. Sometimes you have to change a part to break the cycle.
Will have a look next time I’m down the hanger for anything in the dumpster!
Keep us informed of any progress.