Behringer GMX110 - 20vDC at Speaker Output

[NickH]

Hi,

I have a Behringer V-Tone GMX110 here that is driving me crazy, I am trying to fix it for a friend (I realise it’s probably not worth the time or parts, but it’s become a point of principle now! + I am learning things in the process...)

When he gave it to me it blew the fuse shortly after being turned on every time, I disconnected the lead from the transformer to the amplification board and it stopped bowing the fuses. I quickly checked all the resisters in circuit and they all seemed to be ok ish, the caps seemed visually ok, (didn’t fancy desoldering them all to test at that stage) but the amp IC (TDA2050V) was shoring across a few of the pins so I ordered a new on with some new 7815 & 7915 voltage regulators at the same time...

With the new IC chips in, it powered on with the Amp board connected but the TDA2050V got very, very hot very quickly and it was pushing 20vDC to the Speaker output.

Shortly after that the new TDA2050V exploded...

I took some measurements on the TDA2050V mounting point with the TDA2050V removed and got the following:-

Pin1 (Non Inverting Input) 20V (steadily dropping)
Pin2 (Inverting Input)
Pin3 (-Vs) -34V
Pin4 (output) 0V
Pin5 (+Vs) -10V

I also took some measurements on the voltage regulators, I can’t honestly remember whether it was with the TDA2050V installed or not (sorry it was a week or so ago, lesson learned on note taking..)(Most likely not installed though...)

7915
Output -25V
Input -39V
Ground -8V (dropping)

7815
Output +6.9
Ground -6V
Input 17V

I then gave in and started unsoldering all the Caps & Diodes to test out of circuit, and this is what is doing my head in... I can’t find a big fault with any of it...

Some of the very,very small caps on the underside of the board I didn’t remove to test, but I had removed just about everything else by this stage and they gave sensible readings on the multimeter cap test in circuit, the only one of the tiny ones I did remove (because it wouldn’t give a reading in circuit) was C42 which gave 208nF when removed. (I can’t read the printing on the board to see what that one should be, but I took any reading on that one as a semi positive sign...

All the other caps are within tolerance as far as I can tell, but C7, C8 & C9 seemed way high...

C7 = 133 nF
C8 = 133 nF
C9 = 135 nF
According to the board they should be 100 nF and they were marked as +-2% tolerance

Could they be causing the problem? Seems a bit subtle to me....

The 5402 Diodes (D1 to D4) seemed ok with the following readings on the voltage drop:-

0.5644v
0.5823v
0.5294v
0.5814v

Resistor R1 is also a little high, reading 4.63 ohms (should be 4.2 ) which is out of its +/-5% tolerance, again could this be the problem?


C4, C5,C12 & C13 are little rectangular caps marked 100nJ100 they read 101.30, 101.69, 103.39 & 102.35. I don’t know what the tolerance for these should be as I can’t find a datasheet, but they don’t seem too bad...

Any Ideas folks? This is driving me utterly crazy, I spent a few hours browsing past threads on here to glean some clues, but I couldn’t find anything quite like this, but I did work out that you seem an amazingly knowlagable bunch & if anyone might know it would be you guys...

Many, Many thanks for taking the time to read this..



Nick

[Enzo]

Is there a burnt open ground trace somewhere? And what are you using for your meter ground? You don't have proper voltages at the power IC, and the lack of grounds at the Vregs is also suspicious.

[NickH]

Is there a burnt open ground trace somewhere? And what are you using for your meter ground? You don't have proper voltages at the power IC, and the lack of grounds at the Vregs is also suspicious.

Thanks for the reply Enzo, you are a star!

I will check all of the ground trace extremely carefully with the multi meter and then solder the components back in the board and take fresh readings, I was using the negative post of the speaker output as my meter ground, maybe I screwed up and accidently slipped it onto the +

Would I be better to solder a temporary lead to clip to off the back off the PCB where the transformer negative comes in?

Thanks again...



N

[Enzo]

I'd steer away from the speaker leads for that. You have jacks all over the thing, use the sheell of one of the tape jacks, or plug a guitar cord into the input and clip your meter ground to the shell of that. Or the effects send jack. All those things go to ground. One side of the speaker SHOULD be ground, but let's not count on that yet.

Until you measure +/-15VDC for the preamp and +/-20 or 25 or whatever it is for the power amp, nothing else matters.

Was the old 2050 insulated from the heat sink? Mica or silicone insulator with a plastic washer on the mounting screw? Is that stuff still there? Is the 2050 tab shorted to ground? That's important.

Forget tolerances, tolerances don't blow fuses and burn up TDA2050s. They might have a subtle effect on tone or EQ or something, but that's about all.

Perfectly fine to leave the 2050 off the board until you find all the voltages and a reasonable ground. In this case, I recommend it.

[NickH]

I'd steer away from the speaker leads for that. You have jacks all over the thing, use the sheell of one of the tape jacks, or plug a guitar cord into the input and clip your meter ground to the shell of that. Or the effects send jack. All those things go to ground. One side of the speaker SHOULD be ground, but let's not count on that yet.

Ok, cool! I have the rest of the amp disconnected from that amp board at the moment & the board hanging out of the case away from everything else, the only thing connected to it is the transformer lead. It did this to isolate the problem, was this a bad plan? I am I better to connect all the other stuff back in again? Thinking about it, at the moment that board is not receiving a ground from the rest of the amp at all, again is this a bad thing?

Until you measure +/-15VDC for the preamp and +/-20 or 25 or whatever it is for the power amp, nothing else matters.

Gotcha, I am now focused!

Was the old 2050 insulated from the heat sink? Mica or silicone insulator with a plastic washer on the mounting screw? Is that stuff still there?

I haven't actually got the heatsink on at the moment, I have been running it in short bursts without it to test, it did have a small sliver of clear plastic that is exactly the same size as the tab on ether the 7815, 7915 or the 2050, but because I am an idiot, I didn't see which one it was on before it fell off and I noticed it for the first time, because all three connect to the same heatsink, and because the 2050 & & 7815 have -DC at the tab, I assumed it should go on the 7915 as that has +DC at the tab, again have I screwed up?

Is the 2050 tab shorted to ground? That's important.

No at the moment it's not connected to anything because of the heatsink not being there, should it be?

Forget tolerances, tolerances don't blow fuses and burn up TDA2050s. They might have a subtle effect on tone or EQ or something, but that's about all.

Very interesting & extremely useful information, I have logged it in my long term memory!

Perfectly fine to leave the 2050 off the board until you find all the voltages and a reasonable ground. In this case, I recommend it.

Cool ok..

Thanks Again!



N

[Enzo]

The 7815 has a grounded tab - well a tab that can be grounded, but in some circuits is not. In this case it is grounded. The others 2050 and 7915 need to be insulated from ground. I was unclear. It was important that the 2050 tab NOT be shorted to ground.

Having the board hang by the transformer wires may well leave ground connections unmade. I usually work with everything conected. Sometimes when I raise the board for service, I have to add clip wires from various points back down to the chassis ground.

[NickH]

Thanks for the clarification Enzo

I checked the PCB carefully for a burnt open ground trace and did continuity tests with the meter and couldn't find a problem.

I then soldered all the components back onto the board & bar the 2050 and measured the voltages again using the tab on the 7815 for my meter ground.

This time I got:-

7915
Output = -14.7
Input = -22.7
Ground = 0


7815
Output = +14.9
Ground = 0
Input = +22

At the 2050 socket I got :-

Pin 1 = 0 (Non Inverting Input)
Pin 2 = 0 (Inverting Input)
Pin 3 = -22 (-vs)
Pin 4 = 0 (Output)
Pin 5 = +22 (+vs)

I then soldered the 2050 chip in, made some insulators cut from a clear plastic CD wallet for the 2050 & 7915 added thermal paste to the tabs & the heat sink & put everything back together.

The great news is the amp powers up nicely, the 2050 doesn't seem to be overheating and the 20vDC is gone from the speaker output! Yay! Thanks Enzo!

The amp sounds fine on headphones, but very rattley and has an odd distortion round the edges of the tone (I can take it down the studio & do you a recording if it would help...) when running through the speaker...

My first thought was the DV voltage hosed the speaker, but I ran a 1k (sinewave) test tone through the amp clean channel from my signal generator, hooked the o-scope up & took the following readings at the 2050 pins:-

Pin 1 = (Non Inverting Input)


Pin 2 = (Inverting Input)


Pin 4 = (Output)

Any Ideas?

I was thinking of seeing if any of the 4 X 12 cabs we have in the studio will run in 4 ohm to test it with another speaker, but the o-scope readings have me wondering if something else is messed up here, I was expecting a pretty clean sinewave, do they always look like that when put through an amp?

Many Thanks Again!



Nick

[Enzo]

OK, it now functions. You insulated the 2050 tab with plastic, but you really should get a mica insulator or a silicone one. Look up transistor insulator or find most any old piece of electronics with parts mounted to a heat sink and steal one. Got a dead cassette deck? I bet there are a couple regulators in that on a heat sink. Your plastic can melt.

Also, I hope you have a plastic spacing washer under the screw through the 2050 tab. Just as there should be with the 7915. If the metal screw touches the metal tab, it will ground it to the heat sink.

What your scope shows is the amp oscillating at high freq. You are seeing the audio moving the HF waveform up and down. Speed your trace up a lot and the wide band of light will resolve into a sine wave or something close.

Pin 4 of the 2050 is its output pin, which ultimately is connected to the speaker jack and then internal speaker. But also from pin 4 you should find C7. C7 and R1 are in series from pin 4 to ground. R1 is 4.7 ohms - make sure it is not open. C7 is 100nf, might as well check it too, but I'd bet on the resistor. C7 and R1 are your stability network.

[NickH]

OK, it now functions. You insulated the 2050 tab with plastic, but you really should get a mica insulator or a silicone one. Look up transistor insulator or find most any old piece of electronics with parts mounted to a heat sink and steal one. Got a dead cassette deck? I bet there are a couple regulators in that on a heat sink. Your plastic can melt.

Ahh ok, I have ordered some from ebay, I didn't have any kicking around... I will put it in as soon as it arrives..

Also, I hope you have a plastic spacing washer under the screw through the 2050 tab. Just as there should be with the 7915. If the metal screw touches the metal tab, it will ground it to the heat sink.

Good to know for the future, but this thing just has one metal bar that holds all three chips against the heat sink (not touching the tabs) no screws go through the IC's.. weird looking arrangement..

What your scope shows is the amp oscillating at high freq. You are seeing the audio moving the HF waveform up and down. Speed your trace up a lot and the wide band of light will resolve into a sine wave or something close.

Pin 4 of the 2050 is its output pin, which ultimately is connected to the speaker jack and then internal speaker. But also from pin 4 you should find C7. C7 and R1 are in series from pin 4 to ground. R1 is 4.7 ohms - make sure it is not open. C7 is 100nf, might as well check it too, but I'd bet on the resistor. C7 and R1 are your stability network.

You are of course exactly right! It was the R1 resistor, Thanks! That problem is now solved! Getting a nice clean sinewave now...

I am getting one other problem now, the amp is fine at low levels, but as soon as you turn it up and play loud chords the amp starts cutting out a bit like a tremolo gate effect, I did try the amp with an external 4ohm 4x12 and it has the same problem on that, although there is no horrible rattle on the 4x12 so that speaker is definitely hosed, no drama I will pick up another for it, but the 'tremolo fault' is more of a worry, every time it does it my 'Poor mans light bulb voltage limiter' lights up like crazy in time with the pulsing..

It also glows brighter when you strum a hard chord but I am guessing that is normal...

Cheers



N

[Enzo]

Oh, OK, no screw on the clamp type mounting, so forget the plastic washer.

When you played through that 4x12, did you use the wires that would have gone to the internal speaker or did you just use the external speaker jack? The external jack puts the 4x12 in series with the internal speaker, so any electrical problems with the internal speaker can still affect the external one. Disconnect the internal speaker wires from the internal speaker and use clip wires to conect them to your 4x12 or other cab. NOW does it stil act funny?

I'd suspect that internal speaker for sure. Especially if it was sitting there with the 20v from the shorted 2050 trying to melt it.

Try this, disconnect the speaker and listen through headphones. Does it still cutout/tremolo through the phones?

[NickH]

When you played through that 4x12, did you use the wires that would have gone to the internal speaker or did you just use the external speaker jack? The external jack puts the 4x12 in series with the internal speaker, so any electrical problems with the internal speaker can still affect the external one. Disconnect the internal speaker wires from the internal speaker and use clip wires to conect them to your 4x12 or other cab. NOW does it stil act funny?

I did use the external jack, but the internal speaker was completely disconnected...

Try this, disconnect the speaker and listen through headphones. Does it still cutout/tremolo through the phones?

I will check that out & get back to you...

Thanks again!



N

[NickH]

Try this, disconnect the speaker and listen through headphones. Does it still cutout/tremolo through the phones?

Hi Enzo,

Just tried that now, it works fine through the phones, with or without the speaker attached...

Cheers



N

[Steve Conner]

If you haven't got the proper thermal washers yet, your TDA2050 will be overheating, and cutting out from its on-chip thermal protection when you play loud. I bet that is the explanation for your "tremolo sound".

CD wallet plastic won't cut it, to conduct that heat away you need either mica with thermal grease, or Silpads. The place where you bought your TDA2050 should sell them. The ones with a hole in will work fine even if there are no mounting screws, just make sure the hole in the washer is lined up roughly with the hole in the transistor tab, so as not to waste surface area for heat transfer.

You've probably spent 10 times the cost of the amp in labor by now, but it's worth it for the learning experience

[Enzo]

It may be thermal related, but I don't think the thermal protection can cycle that fast, certainly the heat in the IC/heatsink can't.

[NickH]

The chip is quite cool to the touch even 5 seconds after the fault...

Not sure if that helps at all?



Thanks everyone..



N