Trace Elliot Commando Boxer 65W - need expert help is it oscillating ?

[chazpope]

Hi

I am kind of stuck with this baby - shematic attached - it is the 65W Boxer
model. I got it in a state where the big caps were wobbly and when soldered
it worked ok for 2-3 min with headphones. I noticed though that the power
stage is cooking like on full power even when there is no input signal at
all. As I was wondering what to do it blew both fuses and died. Next I found
that TR1 was burned - bought and replaced it stupidly - only to have it blow
again this time burning all 4 power transistors + TR5 + Z1 !

So I bought all parts again (already crying) and this time decided to
proceed gently lest I burn it again. I replaced Z1, TR1, TL072, TR8 (all but
TR7 temp limiter) and with the power transistors T1-4 not installed I turned
the thing back on. Nothing burned this time, rail voltage +-54V and +-15.5V
ok - but I am afraid to install T1-4 on because it still behaves funny and I
think they will burn ..

Basically the LED1,2 at the bias point behave weird and light up for no
reason - with everything uplugged. During this the bases on TR5 TR6 go up to
above 3V and of course they start cooking like hell and I turn it off.

So from start and with gain and vol pots turned down it starts with both
LEDSs not lit up but as soon as I turn up the pots both light up and stay
lit. Again there is no input/speaker during this. The T5/T6 bias and LEDs
behaviour is basically erratic - starts with 1.3 volts only to go way off
and start cooking as soon as I finger anything (e.g R18) or turn pots on.

I tried plugging another amp after R20 and is plays ok also treble/bass
adjusts as it should so looks like the pre amp circuit is working ok in the
audible range anyway.

Any ideas ? I have no scope but I think might be oscillation going on an not
sure where to look next.

thanks

[Enzo]

Look up "light bulb limiter" then make one and use it.

DO not connect a speaker load until we know the amp is stable.

Make sure R35 over by the speaker is not open.

If your outputs burnt, replace the drivers as well, that is TR5 AND TR6. And verify ALL the resistors in that section. Open resistors are just as bad as shorted transistors. That is to say every resistor to the right of the four diodes D1,2,3,4.

SO you powered up without the output transistors and it holds? OK, look at those four diodes, D1,D2,D3,D4. Measure the voltage across each. DO they all four read about the same, roughly half a volt?

LEDs are on? They are driven by IC1-7, so does turning the volume control RV4 to zero affect that? Basically grounding pin 5 of IC1.

[chazpope]

I did replace drivers, the IC and TR8. I spent a lot of time looking at voltages around D1-D4 - they seem to be ok - across all of them the voltage stays at 2.6v however with regards to earth it tends to shift randomly as if driven by the output from R20 and then the LEDS light up. When I turn RV4 this is what happens: The LED comes on and then sometimes stays on even when I turn RV4 down - during this the voltage of TR5 base goes to 3V with the base of TR6 following at 2.6V below that, or it goes the other way respectively. This effect is made stonger if RV5 is turned up as well. Or, when I start the amp and no LEDS are on - then I touch R18 with my finger and one or both the LEDS come up and it starts cooking one or both drivers and it stays in that state untill I turn it off.

At one point I disconnected R20 and the LEDS and then voltage was steady +-1.3 on both driver bases (now I think even then they operate too hot maybe by design; temporarily I have doubled the emitter resistors of the drivers - lest they will burn while I am testing)

Tonight I will recheck R35 and other resistors, and yes I will only try it with power transistors with the bulb limiter in.

[chazpope]

well resistors seem to be ok.

I found another thing though that I had not noticed before: the 15.5 V collapses (variably down to 7v) when the LEDS come on ! respectively on whichever side or both sides this happens. After a while (switched off) it goes back to 15v. The Z diodes are OK so the only reason for this is a sudden increase in load ! Basically it starts ok - then I (electrically) touch R18 - and it goes into LED on state - weird.

also there seems to be lots of AC (2-3V) on R20 - I only have a digital meter.

[Enzo]

Of course ther is AC on R20 if both LEDs light.

The four diodes are your bias, they keep the bases of TR5 and TR6 a steady space apart - your 2.6v. With R20 gone you had a stable 1.3v on each base, so they are working. I am not sure why that is too hot, in that stable state, you can easily measure current through TR5,6 by voltage drops across the 47 ohm resistor. The voltage at the base doesnlt matter, what matters is the relationship between base and emitter at each transistor.

With R20 in place you mention the voltage to ground at the diodes moves about as if driven by R20. Well, yes, that is exactly how the amp works. The outputs follw whatever that string of diodes does, and R20 is what sends the signal from IC1b to them.

COllapsing 15v? That needs to be fixed before anything. Even if it is oscillating, the op amp shouldn't load down the supply. How did you determine the zener was blameless? Does 15v remain across it while the circuit drops? Or if the zenerr has only 7v across it, how can we know it is not at fault, other than replacing it. Then ther is the 2k2 R22 that serves the zener. How about it or its solder?


You need to have at least one pair of outputs in there to complete the circuit. Or put in the load without them. Your op amp as it sits with neither outputs or load has an open DC circuit to pin 6.

[chazpope]

I measured the 15V collapse on both zeners (resp with whichever LED comes on as I turn the gain/vol pots). I replaced both zeners recently and I assumed they work because they do show 15.5v 'before stuff happens'. I work out that (54v-15v)/2.2k=18ma - 5 ma for zener itself = 13ma left to draw before it will start to collapse - but it beats me what causes that. I will try it again tonight.

re drivers too hot to touch yes I worked out what power they do based on emitter current and it was way too high - I will repeat that test to be sure. Feels like the rail voltage is too high (54V) why the hell did they not write what it should be

I also thought about the open output and tried it with headphones plugged in - it made no difference - I will try again with 8 Ohm resistor

[chazpope]

another 2-3 hours spent only to confirm the above... in the end I concluded that 'erratic cooking of drivers' is caused by the collapsing 15v swinging their bias points way off. So I disconnected R2 and R6 for now to save the drivers

I cannot figure out why the 15.5V collapses - when I play my bass through the amp it collapses along with me playing. I put a 5 ohm load for speaker, disconnected the point between R20 and C1-C2, this still happens. But what about the LEDS they must be drawing an additional load ? I have another Commando different 100W model but with similar LEDs and there they do not really light up as all the time when you play. What are the LEDs for anyway - limiter ? Could it be that they are not designed to shine and when they do they overload the supply ?

In desperation I tried halving R22 and R31 - and then the 15V held firm - however the zeners got quite hot and it started creeping closer to 16V and I quit the experiment. So I figure something else is wrong that causes all this behaviour. Again I tried plugging an amp at R20 - it plays but with lots of noise, even when input is off there is some humming 50hz noise - not sure but kind of more than normal despite my exposed wires. Checked rail voltage again - steady 55.5V. I think there is some 'inherent hum' in the preamp that gets amplified and causes all this crap - should I replace small elecrtolytic caps. However it worked ok with them before it blew up so not sure if it would make a difference.

I hope tomorrow brings some fresh ideas ..

[Enzo]

Well, speaking or R31 and R22. When this collapsing happens, what goes on at the hot ends of those two. Do the main rails stay up, or do they fall a lot as well.

[chazpope]

no the rail voltages stay firm and like I said when I reduce R22 and R31 the 15V also holds - so doesn't this mean it's just overloaded for some reason.

actually as soon as I switch off - the collapsed 15V goes back up to where it should be and then gradually goes down as the big caps discharge

it behaves as if the IC is a power amp with the LEDs as its load, and there is some hum generated at TR6 which is amplified to inordinate levels and that causes overload of the 15V. This I think because the weird behaviour depends on RV5 - even when the input socked is grounded. I wish there was some more DC decoupling in this circuit - now I m not sure which point to ground to locate the problem.

[Steve Conner]

Yes, the LEDs are a limiter of some sort. If the power stage were working, and the feedback loop doing its thing, they shouldn't light.

What you're seeing seems quite normal if the power transistors are missing, or the wrong part (not Darlingtons?) or indeed if the feedback loop around the output stage is broken in any other way.

The +/-15V rails collapse because they can't supply the amount of current that the TL072 is capable of pushing through the LEDs. It tries to push this high current because it has no feedback to tell it not to. In normal use, with the feedback intact, the opamp only delivers as much current as needed to drive the power stage.

I guess there is also a possibility that the opamp got cooked by the power stage fault. Might be worth replacing it.

[chazpope]

great but what bothers me in that logic is that this happens when there is no input signal whatsoever and if I put the darlingtons they will burn, bulb limiter or not - I still haven't found why when the amp worked for a while before they were cooking with no input. the 2 4A fuses burned after the power transistors - so I am doing all I can to prevent this this time - these bloody Darlingtons are expensive.

Is there any way I can stop the input to the IC to check if then its output will be stable. I did replace the IC earlier it made no difference so I assume it's ok.

I guess I in the end I will try it with a pair of Darlingtons on and cry

[Steve Conner]

I can't think of any way to "stop the input" to the opamp that would tell you anything useful. Maybe you could temporarily take the feedback from the op amp's output instead of the speaker terminal.

In this circuit, I'd be worried about bias stability. I think your replacement transistors cooked simply because they were biased too hot and went into thermal runaway. I don't see a bias adjust pot, so maybe Trace changed some resistors to suit each batch of output devices? Anyway, in the absence of an adjustment, you might like to try shorting out one or two of the 4 diodes in the bias chain. I don't think it matters which one, as the op amp should readjust for DC balance.

It's well known that transistors sacrifice themselves to save the fuse.

[chazpope]

ok thanks I wil try that toninght. These Darlingtons really make me nervous - also the IC shifts the drivers bias by over 2V easily.

Just to make sure the preamp is not injecting high leveles of hum in the circuit I also plan to connect an amp after R19 and listen will I get a crisp manageable signal there or not

[chazpope]

well I put a bulb in, shorted 2 diodes, soldered a TR1/3 power pair in, crossed myself some and turned on - it did not blow!

so Enzo, Steve you were spot on - the lack of feedback was causing the LED shenanigans and it was just me being paranoid and not turning it on. It played ok with headphones with some crossover distortion. I proceeded to restore the rest of the circuit (less T2/T4and it started to play ok and with a 5 Ohm load the lamp started to come on.

However at one point (with no load or input) the bulb started to shine, T1/T3 got hot - I switched it off ! All power stage was cooking and about to run out. Now at this point I had not fixed T1/T3 to the heat sink - but as far as I was concerned there should have been no big action as there was no input. Next I put T1/3 onto heat sink and covered them together with bias diodes in paste. Turn back on, it was stable and playing well but then again - the drivers are cooking working at 0.75W in quiescent state and they are too hot to touch.. Now isn't this knocking on transistor hell's door !

Now on my other commando I found the drivers to be similarly cooking and untouchable (and though there I could fit 10sqcm heat sinks it made little difference) - the amp worked fine. Is this normal design ?

I mean BF422/3 do not take heat sinks, bias diodes are too far away to know - what's to stop it from blowing

[J M Fahey]

Yes, it's a poor design and unstable to boot.
Teemuk particularly hates this architecture very much (having the output devices supply voltage gain), can't say I don't agree with him !!
And to make it worse , they bias it too hot.
But there is another time bomb ticking there: the DC ground reference for the power amp is through RV4.
Any dust/dirt/crack/bad contact at its wiper will put huge amounts of DC at the speakerterminals, oscillate wildly, you name it.
This might explain some of your earlier symptoms.
Solve it permanently by soldering a 220K resistor from IC1B Pin 5 to the nearest available ground.
Also, to lower output transistor bias and cool the drivers somewhat, replace R7/R11 by 100 ohm ones and R2/R8 by 150 ohm ones. All 1/4W.
Good luck.