Poweramp Behringer HCA2400 zobel network problem

[binito]

I have a Behringer HCA2400 power amplifier which has a bad channel A output. It sounds less and with the oscilloscope I see that it has a high frequency signal on the speaker output which is not audible.
The resistor R62 and the capacitor C74 are bad. I tried to replace them but R62 starts to smoke after a few seconds.
I understand that the Zobel Network burns out because there is something in the circuit that generates a high frequency signal, but I can't find where it is being generated.
Where could the problem be?

Attached a partial schematic.
Attached a pic with output channel B (OK) and channel A (NO OK)

Behringer+HCA2400+DPA.pdf

[x-pro]

Check all capacitors around coil L6 for quality and compliance with ratings, and check L6 too.
It wouldn't hurt to also check the details around L2 along the FB line.
Compare the oscillation frequency with the frequency in channel B.

[Mick Bailey]

Take a close look at the solder joints, too - especially L6. Sometimes there's a circular crack around the leg that's difficult to spot.

[Helmholtz]

I understand that the Zobel Network burns out because there is something in the circuit that generates a high frequency signal, but I can't find where it is being generated.
Where could the problem be?

Some general remarks:
This is a class D amplifier. Operation is based on high frequency switching.
The switched HF carrier is pulse-width modulated (PWM) with the audio signal.
HF and PWM are generated by IC5.
To separate and regain the audio signal after the power MOSFETs T22/T25 steep low pass filtering is necessary.
The HF filtering is mainly accomplished by L4, L6, C96/87/127/124/97.
If you see significant HF at the output there's a filter issue. So check the filter components and their connections.
Do not connect a speaker as long as there's a large HF output.

https://en.wikipedia.org/wiki/Class-D_amplifier

[Jon Snell]

If there is up to say 2volts of HF on the speaker out with a speaker connected, that is normal.
You mention the Zobel network. In this design it does nothing as the impedance of the output stage and the oscillator frequency will ignore an HF load through R61/2. In other words it serves no purpose.
If L3/4 were short circuit, that would allow HF to pass but the speech coil inductance and capacitance of the speaker cable will deal with that. It will just run hot.

Your HF is generated on pins 1 & 2 of IC5, that is how Class D switching works.
Decoupling is extremely important;

[binito]

Some general remarks:
This is a class D amplifier. Operation is based on high frequency switching.
The switched HF carrier is pulse-width modulated (PWM) with the audio signal.
HF and PWM are generated by IC5.
To separate and regain the audio signal after the power MOSFETs T22/T25 steep low pass filtering is necessary.
The HF filtering is mainly accomplished by L4, L6, C96/87/127/124/97.
If you see significant HF at the output there's a filter issue. So check the filter components and their connections.
Do not connect a speaker as long as there's a large HF output.

https://en.wikipedia.org/wiki/Class-D_amplifier

I replaced C97 and C87. After taking them out I measured them and one of them is wrong.
I reworked the welds on C127/ C124/C96/ L4 and L6.
When I turn the amplifier back on, it smells burnt after a few seconds. I see that L6 and C97 get very hot, what could be causing this on L6 and C97?

Edit: Finally I replaced C96 . It was wrong too.
Now output B is correct.

But analysing output A, now it also has some HF on the output. Hahaha

After having analysed this fault, I had to desolder capacitor by capacitor to be able to measure it. Is there any way to do this on the PCB?

[Helmholtz]

I see that L6 and C97 get very hot, what could be causing this on L6 and C97?

I assume that too much HF is reaching those parts.
Most of the filtering should be done by the first filter stage consisting of L4 and C96 (+C127).
Check L4 for inductance (shorted turns) and C96/C127 for capacitance and connections.

You should really scope the HF (regarding frequency and amplitude) at different circuit points and compare to the other channel.
Use a well compensated 10x probe and don't forget to connect the ground clip.
Don't connect a load and be quick. Allow for cooling between tests.

[g1]

But analysing output A, now it also has some HF on the output.

The HF showing in your ch.A photo looks normal to me for this type of amp. The speaker does not reproduce it.