Aguilar DB750 overcurrent

[Swampy]

Hi Everyone,

My solid state general knowledge is still a little cold so if anyone can help me troubleshoot this beast it'd be much appreciated.

Attached are the power amp and voltage driver schems with voltage readings made w/ light bulb limiter. Got some dc on the output.



Chur

[Jazz P Bass]

First, you must straighten out the drive board.
There have to be a few bad transistors to get voltage readings like that.
A digital volt meter would be most helpful.
Diode check would be used on all of the transistors, looking for shorts.

[Gtr_tech]

That thing runs 15v rails for teh output section? That can't be right...

[Jazz P Bass]

Yeah, I kinda thought that was odd, what with all the paralleled mosfets.
I have attached the files in pdf. format.
I find it much easier to read pdf.

[52 Bill]

I think that the voltages he listed are with a light bulb limiter on the primary circuit. Something may be pulling enough current to suck the main power supplies down to 15 volts.

[Swampy]

Yes you are correct 52 Bill, with the power and drive section supplies disconnected, and still running the lightbulb limiter, I read about + and - 100 volts on the rails.

I'll hunt for shorts today and report back.

[J M Fahey]

Yes, of course 52 Bill is right.
To begin with, the expensive output MosFets are not shorted, they just follow what the driver board sends them.
Congratulations Swampy on providing many accurate voltages.
Many of them look reasonable, but not around the differential input.
That uncompensated important voltage difference between the +/- diff inputs is abnormal.
Iīm even mistrusting the offset balance trimmer, but let me study it a little more.

ADDED: +2.83V on Q501īs base is too much, even in a low beta/Hfe transistor such as an MPSA92.
Its twin brother, Q502, should have a similar voltage, thatīs what diff pairs do !!! , but although the output rail voltage is *positive* (+ 8.5V) , Q502s base is *less positive* īthan Q501's (which in such a circuit acts as being "negative") ... and yet the circuit canīt compensate the output rail back to zero V.
In fact, even compensating it to + 2.83V , take por give 10 or 20 mV would be fine, since it would just be following what's on Q501's base, but we do not even have that ... so .....
First suspect: that offset trimmer, 1.5V difference between its ends is too much. Might be worn, open or dirty.
Second suspects: Q501/Q502.
There may be others, but I would start with them.
There are other hairy things I would do if it were in my bench, but wonīt suggest them for safety, unless we find us in a real dead end.
Good luck.

[Swampy]

Cheers J M, but I think you misread the output of the diff pair, was sitting at -8.5 v (I think you read + 8.5??) at the time of that reading.

Anyhows, it's the future now and after verifying the impedance of the trimmer in circuit (all good) I've adjusted it to balance the inputs on the diff pair and remeasured the circuit. The base emitter drop of Q507 seems a little off, could a partial short be the cause of the stray -7.18 v on the output node of the diff pair?? My multimeter reads the base to emitter diode drop at approx .3 volts ...

Chur

aguilar_driver_for_web_2.pdf

[J M Fahey]

Hi swampy.
2 details:
1) just for the record: I correctly read +8.5V (your readings) at the *output rail* which is the *speaker* output rail and is whatīs worrying us, was not referring to the diff amp collector voltage.
By sheer chance the first one is +8.5V ; the latter -8.5V which makes the confusion easy.
Yes, itīs the past, the main reason for this correction is to make sure whatīs called "the output rail* .... which still reads bad.
Oh well.
2) now to the present: all voltages read reasonable now, yet the diff pair (brain) is not compensating voltages through Q509 (muscle).
I mistrust somewhat C503 and the 4 diodes in antiparallel with it.
Please lift one end of R508, to turn NFB (all of it, including DC) into 100%.
I think the output rail offset will drop to around 1.5V .
Reading the future in my coffee cup residue (yes, I drink Turkish style coffee), after that I would short R503.
I guess output DC offset voltage will now drop to a few mV.
Or not, in which case I'll have to rethink everything.
But for now, these are 2 tests I would make if I had it in my bench.
Good luck.

PD: just for the record, please add voltages read at R503/C508 as you did before. Voltage *now*, before lifting one end of R508.
Also the voltage at the node "NFB" D508/D509 and also voltages at the emitters of Q512 and Q519.
Yes, I know what they "should" be ... but maybe they are not so, we're trying to find abnormal readings, and here most look good ... yet the diff pair does not seem to be doing its work ..... although I suspect something.
Congrats on having been able to balance it though, by using the trimmer.

[Swampy]

Hi J M,

My apologies, i wsa oncsufed.

Attatched is an update ("2b") to the second image I posted including the voltages you questioned, and a third image (3) of the measurements with R508 lifted. Output rail voltage has dropped to -.06 V dc . Those base to emitter drops on Q507 and 509 are off though right?

Chur

aguilar_driver_for_web_2b.pdf
aguilar_driver_for_web_3.pdf

[J M Fahey]

Looks like we are getting *much* closer.
That amp *does* have 100% DC NFB, thatīs what C503 is there for.
This is negated by diodes D503/4/5/6 which provide a DC path.
In theory DC offset should be way below 1.4V and they should never conduct ... but anything over that and we are in big trouble.
I would reconnect R508 *but* pulling those diodes out.
They can do more harm than good.
If in doubt, test C503 for shorts or just plain replace it with a fresh one.
You *should* be able to zero the output rail offset.
Next step is knowing why this amp pulls so much idle current.
Lowering bias voltage *should* solve that.
If it does not, *maybe* one of the output MOS is, not shorted but lossy (sometimes happens after prolonged overheating) and the poor differential stage, as Enzo already put it (or was it RG or Steve Conner?) is just doing its job, *trying* to balance things out.
If some device on a rail passes more current than it should (because of a defect), the Diff Stage will *force* the other side to do the same (the balancing act), making the amp pass too much idle current, even if Bias does not say so.
As you see, repairing an amp often opens a can of worms.
Oh well.

[Swampy]

Cheers J M

With C503 and D503/4/5/6 lifted and adjusting the bias to minimum I have taken the readings on the attached image. Still way to much idle current flow but...

Pulling the connector to the driver board (supplies driver with + and - , negative feedback etc ) lowers idle current dramatically...am I correct in abstracting this rules out any shorted or lossy output MOSFETS?

aguilar_driver_for_web_4.pdf

[J M Fahey]

What drives me crazy is that from an initial +/- 8.5V rail voltage, low but still functional, we are now working with +/- 1.5V rails, where everything is at saturation voltage and even balancing anything becomes meaningless.
Iīm distrusting the ouput MosFet power pack, since they are the only ones with enough muscle to turn that PSU down through the series lamp limiter.
Although you say you measured no shorts in them, I have my *big* doubts.
Letīs try *just* the Mos pack.
Using this image as reference:

1) leave it connected to the power rails (+Vss and -Vss)
2) disconnect its input from the driver board (hope you have a connector just for this)
3) short together the V+ drive and V- drive pins. We donīt care about crossover or biasing now, just testing the capability of the Mos pack to stand real world voltage.
The multimeter ohms function applies less than 2Vdc.
4) connect a 10K 1W or 22K 1/2W or 47K 1/4W from the V+ drive pin to +Vss and an equal one from V- drive to -Vss
We are applying a balanced voltage (net result about 0Vdc) to the gates, I *hope* to find close to 0Vdc at the "output" rail, and, most important, almost full (I think you said around 100V each) Vss rail voltages, since the Mos pack is unbiased and should pull practically "0" current from the PSU, ergo, through the lamp limiter.
I guess that a Mos which didnīt show anything unusual under the very low voltage applied with the multimeter will now fail under stress.
OK, letīs do it.

[Swampy]

Do I short the drive pins on the drive board, or on the power board (where we have our two resistors balancing at 0vdc)??

[Swampy]

All good, I used my noggin.

Got + and - 23 volts on the rails, -.035 V on the gate resistors (drive side) and -0.60 on the output. So...I guess I start pulling em one at a time?? There must be a cheeky insider approach to this task?