request help troubleshooting Amp15 - one channel -29.4V, other channel good

[plim]

Hi all,

I need help, please, with where to start looking for faults with my new build Amp15.

Problem: when muted, the output voltages appear normal (2.6V), but unmuted, one channel can be trimmed to 0V but the other remains stubbornly at -29.4V (measured at the + output).

I've successfully built 2 Amp11s, a Trupath and an Amp32-PS, all working, so I think that I am ok on my soldering skills and general build quality. Having said that, some of the spacings on the Amp15 are a challenge.

Here's more background:

1. I first built the PS section and confirmed it working with a 35V 300VA transformer, i.e., VPP=+49V and VNN=-49V and the VN+10 and V5 all were fine.

2. All the jumpers are correct per the drawings. Only the builk caps have been left out at this point.

3. I have rechecked the soldering with a magnifying glass, resoldered all contacts and cleaned thoroughly with a toothbrush and IPA

4. When powered, both FETs are receiving the VPP and VNN voltages at their Hi and Low pins respectively.

5. Unfortunately, I have soldered the 3020 directly to the board (yes, yes, I know).

Any advice on what to test? Is the -29.4V an indication that there is a problem with the FET, or a problem with the 3020 driving it?

Many thanks in advance for your help.

Patrick

[Jan]

It seems like the high side of the FET is not switching, just the low side.

- Do you have the right gate resistor for the high side?
- Look at the boot cap and diode for the bad side. Do you have 10V over the 47uF cap?

[plim]

Thanks Jan for the fast response!

The gate resistors measure 5.8 Ohms. The diodes all measure good in circuit (0.56V using the MM diode check). BUT, the good side boot cap measures 9.36V and the bad side -0.26V. I'm having a look at the circuit diagram now to figure out what might be causing this.

[plim]

I'm stuck. Both pins of C11 measure about -29V ref to GND. I checked D2, C13, C11 and R9 and they all appear to be correctly installed.

Should VBOOT1 be pulling up the positive side to +10 above the FET Out?

[Jan]

Hi,

The negative side of the 47uF cap (and 0.1uF) is tied to the loudspeaker output, the output side of the FETs, before the output filter. When the output swings to the negative side, the positive side of the cap will charge to 10V from VN10 via a diode. As the negative side is tied to the output those 10V will follow the output, always at 10V above the output, up and down, recharging a little at every negative swing when needed. See this 10V cap as the power supply for the high side driver and the VBOOT pin as the supply input for the high side gate driver.

The 10V cap charges up at power on, at the first negative output switching. There are a few negative cycles done specifically for this by the chip startup sequence, before attempting any positive output. A 220 ohm resistor limits the current at startup, but it will be near 10V after a few cycles. Even if the amp is idle, the outputs will always swing to full positive/negativ voltage (at around 600KHz before the output filter) but the average should of course be zero. After the output filter you can ses the remainders of this switching as a sinusidoal with 600kHz one volt signal.

So if the 10V cap does not charge, there is no chance of the high side FETs turnng on, and the 30V you see is the average of negative and zero output.

As you obviously have negative output, this limits down the possible problems a lot: take a look at:
- the boot diodes D2, that they are not reversed or broken. Double-check the direction. Do you have the VN10 voltage to the diode?
- that the 47uF cap is OK, not shorted / reversed / damaged (less likely)
- that R9/R10 are OK (less likely)
But I would probably rule out the last two as there is the small 0.1uF cap too, which should be enough to charge to at least a few volts.

(10V here is more like 9V on AMP15, as VN10 is via a 9V regulator which is enough to turn those FETs on..)

[plim]

Hi Jan,

I really appreciate your taking the time to explain the details. I missed the fact that the cap was following the pre-filtered FET output. I'll carry on in the morning after a strong cup of coffee.

Thanks again.

Patrick

[plim]

I went through checking D2, C13, C11, R9, R7. Everything measured fine in circuit and I couldn't find any short that would explain why C11 wasn't charging up to VN10.

So, I started removing parts, first C40, C5 and C7 to gain access to the FET pins. I then removed the FET pin 2 and 4 (gateLow and GateHi) and confirmed that there was no short between Out and GateHi. I thought at one point I found some odd resistance readings more likely it was the caffeine running out. When things looked normal and I reassembled everything, I ended up back with the -29.2V output reading.

Anyway, I've done enough damage to the board that I doubt I'd risk ever connecting my good speakers to it. I'm going to put this on the shelf until I move house and then decide what to do.

Lessons learned:

1) Building and testing the power supply side first is a good approach and I'll extend that to taking measurements as I build the rest of the board (on top of just double-checking components . Catching something and fixing it before everything is in place is so much easier than trying to troubleshoot, especially as some components cover the pins of others.

2) Despite my purist tendencies, I should have socketed the TA3020.

3) I need more desoldering practice! But "Good judgment comes from experience. Experience comes from bad judgment".

I'll sit back an enjoy my Truepath for now...