[Fixed] IRS20957S based amplifier sound loss

[Rutherberg]

Hello,

I'm repairing an Ampeg bass amplifier, with IRS20957S based mosfet power amplifier. Here is the power amp section schematic :

http://music-electronics-forum.com/attachments/23973d1372015087-svt-7_-main-bd.pdf

IRS20957S datasheet : https://www.infineon.com/dgdl/irs20957spbf.pdf?fileId=5546d462533600a401535676143e2799

The problem is output signal disappear and come back randomly. After troubleshooting, I've found that the voltage at the CSD pins (shutdown pin) fluctuates a lot, so the power amp is turned on and off randomly.

I isolated the CSD pins, first removing R39, then C174, then both. The voltage at CSD pins continued to fluctuate in each case.
VDD and VCC voltages at the IRS20957S are stable.

I replaced both IRS20957S, without success. I also reflowed all related solder joints.

I really don't understand what is happening, could you help me with this one please ?

Thanks !

[bktemp]

The internal current into CSD is controlled by the overcurrent protection circuit.
Check the highside overcurrent protection (CSH pin) and lowside current setpoint (OCSET pin).

[Rutherberg]

Thank you, I'll check that.

By the way, here is the correct schematic (page 2) : http://music-electronics-forum.com/attachments/23973d1372015087-svt-7_-main-bd.pdf
I edited the first post.

[bktemp]

It looks like highside overcurrent protection is not used (R112 not populated).
That leaves the low side overcurrent protection and the high and low side undervoltage protection.
So verfify if Vcc to -55V is >10V and also VB to VS.

[Rutherberg]

My measurements so far, for each side :

When power amp shut down, CSH goes from 2,4V to 0V
OCSET = -53V, this looks not right. I dessoldered R26 and R38 and measured it, the values are good.
+-55V is at +-58V and -43V is at -47V.

By this point the amp worked correctly again, no more shutdown, so next values are in normal condition :
Vcc to -55V = 11,5V
VB to VS = 11,2V

I try to reproduce the problem. It looks like a bad contact somewhere...

[bktemp]

OCSET is connected to the low side mosfet, so its voltage is relative to the negative rail. -53V and -58V gives a voltage of 5V for OCSET.
Based on the resistor values from the schematic OCSET is pulled to slightly above 5V. That is out of specs according to the datasheet.

[Rutherberg]

I reproduced the problem. So to summarize :

                     Nomal condition                  Fault condition
                     -----------------                 ----------------
OCSET =              6.2V                                   4.8V
Vcc - COM =        11.5V                                 8.5-9V (not stable)
VB - VS =            11.2V                                 15.5V
Vdd =                  10.3V                                 10.3V 
+/-B =              +/-58.8V                            +/-59.3V
"-43V" =              -47V                                 -50.5V

So, -47V is dropping to -50.5V, thus Vcc drop also under the minimum 10V recommended. Tell me if I'm wrong, but it seems to be an undervoltage detection in the low side !

On the first page of the document, there is the power supply schematic. There is a 7812 (U1) to "transform" -58V to -47V. Temperature approx. 55°C.
Under normal and faulty condition, I have -44V at the input of the regulator, so transformer, D23 and C78 seems to be good.
At the output, -47V under normal condition and -50.3V under faulty condition. So I assume it could be the 7812.

That said, there is some components connected between +/-B and -47V. Is it possible that one of these components occasionaly "leaks" to the -47V rail ?

Thanks for your help, it is greatly appreciated

[rg58]

Those amps based on IRS20957S are horrible to troubleshoot. I gave up on 3 plate amps from RCF at that time, one being ART-310A MKIII. Sound cut out and came back randomly on the base channel, the PWM signal looked wrong. Like yourself, I changed the IRS and some caps with no luck. All transistors tested good as well, even all SMD transistors and diodes tested good in-circuit. Did lots of testing too with the scope, and eventually blamed it on the DSP or firmware issue (those beasts are complex!). But don't let this discourage you, there must be a way to fix it *if* you are willing to put a lot of time into it. Sorry that I couldn't contribute more. Good luck!

[Rutherberg]

But don't let this discourage you, there must be a way to fix it *if* you are willing to put a lot of time into it. Sorry that I couldn't contribute more. Good luck!

Thanks, I don't want to put more time in it but I'm frustrated and I want to repair it ! 

I consider to remplace the 7812 and C26, plus some parts between +/-B and -47V (D31, D41 and C75)
Any suggestions before this last foolish action ? 

Thanks

[bktemp]

Since you already have identified the problem (the 12V supply voltage of the mosfet driver driver stage is dropping), you need to find what is causing this problem.
55°C for the 7812 sounds rather high, since it is only powering the mosfet drivers (I would expect 100mA at most), so maybe something is drawing more current than usual, causing the 7812 to overheat and shutdown.
But it could be also the 7812 being the actual problem.
If you measure those voltages, use the -55V as the reference, that makes it easier because the 12V is referenced to the -55V rail.
You said the 7812 "transform" -58V to -47V. Does that mean at the input of the 7812 is at +23V (above -55V)? If so, that would explain why the 7812 is getting hot.
23V sounds rather high since the capacitor is only rated for 25V. And since they use a regulated SMPS, why waste so much power in a linear regulator? I would have expected +15-18V or so. The +/15V rails seem to be used for the input opamps without any linear regulator.
Check all voltage rails from that SMPS, maybe there is some other problem causing the high output voltage.

And why is there a +300V rail? Is this voltage used somewhere?

[Rutherberg]

At the input of the 7812, I have approx. -44V stable even in fault condition, so about 14.5V referenced to the -58.8V.
Typical voltage drop of a 7812 is 2 to 2.5V so it could also be the problem. Maybe try with a LDO 12V regulator ?

There is a preamp tube in the amp, the +300V is used to supply it.

Thanks for the advice

[rg58]

Haha, I know that feeling very well. It's the challange that keeps us going and going and going - and uppps, another day has passed 

Anyway, as nobody else is jumping in having more experience with the IRS20957S, I took a bit time to look at the schematic and your problem. The fact that...

Vcc - COM = 11.5V    8.5-9V (not stable)

...made me think as well - maybe it's not even an issue with the IRS. As you said, it might be related to the power supply. It should be 12V (stable) between those 2 pins (-55V + 12V = -43V). So I would change U1 (7812) and if no success D23 as well (could fail randomly too!), then see what happens. If still no 12V behind U1, then it could be that there is too much current being drawn on the -43V rail. And -44V on the input of the regulator sounds a little low as well, could be a little more reserve there (45-48V maybe). Again, could be an indication of too high current draw.

And if the -43V rail has -47V by default, there's definately something fishy going on (=8V). This again points to U1 or the transformer winding on the input is shorting randomly (or D23 has a problem). Many possibilities. If still no success, I would test the power supply separately by unplugging it from the amp circuit if that is possible. Remember to put load resistors accross the outputs when testing a SPS. I would put 2 resistors from -55/+55V to ground so that it draws maybe 100-500mA on each side (mind the wattage of the resistors!). That would make troubleshooting easier as you could narrow down further and rule out too high current draw on the -43V rail by the amp circuit or for other reasons we have not considered yet. Lots of ideas in my head...

[bktemp]

The typical voltage drop of a LM78xx regulator should be <2V at 1A and <1.5V for a couple of 100mA. So 14.5V should be fine. Even going from the worst case specs, it can not explain why the voltage drops to 8V.
So it is either ther 7812 overheating ((55°C-25°C)/50°C/W = 0.6W, at 2.5V that gives a current of 240mA, rather high for just two mosfet drivers) or shutting down due to overcurrent or it is faulty.

Using an external current limited 12V power supply could help finding the culprit.

[Rutherberg]

So it is either ther 7812 overheating ((55°C-25°C)/50°C/W = 0.6W, at 2.5V that gives a current of 240mA, rather high for just two mosfet drivers) or shutting down due to overcurrent or it is faulty.

I re-measured the input of the 7812, it is a steady 14.89V in normal condition and 15.3V in fault condition.

The +/-B fluctuates between 58 and 60.5V, it is really high compared to the 55V announced.
The input of the 7812 fluctuates in the same extent, so input voltage is constant in normal condition.

I will change the 7812 anyway, it's cheap and quickly done.

Using an external current limited 12V power supply could help finding the culprit.

So if replacing 7812 doesn't work, replacing it with an external current limited PSU. Negative output to the -58V rail and positive output to the -47V rail ?
To detect an enventual overcurrent when the -47V rail collapse I assume ?

[bktemp]

The B+/- voltage is unregulated, so it depends on your mains voltage and output load. 60V instead of 55V should be fine.

[Rutherberg]

So, replaced 7812, with a 1 ohm resistor in series with the output to monitor current. I try to reproduce the problem.
So far I have an output current between 103.5 and 104mA in normal condition, temperature still about 50°C after 15 minutes powered on.

So wait and see. Thanks again 

[Rutherberg]

Well, it is more than 2 hours in idle, no sound loss, 7812 apparently was the culprit.
I'll play with it this weekend before confirming it is repaired.

Thanks for all the help, what a relieving feeling ! 

[Rutherberg]

3 days powered on, no problems : fixed ! 

[bktemp]

Does the 7812 still get hot or does it get less hot than before?

[Rutherberg]

Where do you take temperature on a TO-92 package ? It varies depending on where I take it of course.
It seems to stay at about 40-45°C on the back (heatsink surface), it is better but still hot. Can it be caused by the high frequency ripple at the input ?
There is an extracting fan, so temperature will probably slightly drop when the top lid is closed.

[bktemp]

Wait, the 7812 U1 is actually a 78L12 in TO92 package?
If so, the old 78L12 must not have been faulty, the whole design was just marginal and depending on the temperature it went into protection mode, reducing the output voltage.

[Rutherberg]

Oh sorry, TO-220 ! It was a long weekend 

[Zucca]

It seems to stay at about 40-45°C on the back (heatsink surface), it is better but still hot.

I hear a bomb ticking... bigger heat sink? A big power R or a DC/DC step converter before the 78L12 (hope not to introduce some buzzz noise in the audio system....)?
Anyway congrats for the repair... well done bro!

[Rutherberg]

It was a mistake, it is a regular 7812, TO-220 package. The original design is without heatsink.
I don't find 40-45°C shocking for a 7812.