Blown MOSFET -- what did I do wrong?

[klausman]

Note: I am a beginner trying to learn, so advice is preferred over admonition, thank you. Also, yes I've watched the "how to not blow up your osc" video. As far as I can tell, I made none of the mistakes mentioned in the video

So I was working on a 90s Sony amp (TA-FB930R). The amp was working mostly correctly (beyond a scratchy volume pot), so I hooked up my laptop’s external USB soundcard (no output signal just yet) and my oscilloscope like in the attached image.

A few notes:

• the amp is double-insulated (two nested boxes on the back panel), and only has a Europlug power lead, so it does not have any mains earth connection.
• the laptop was not plugged into its power supply, so it did not have a mains earth connection, either
• according to the schematics of the amp (and tested with a DMM), the signal shielding of the input I used (AUX) is directly connected to speaker negative
• The oscilloscope (Rigol DS1054Z) does of course have mains earth and that is connected to the probe terminals/probe crocs

When I switched on the amp, one of the power MOSFETs blew up instantly (and violently, including flames).

That in turn makes me believe that “somehow” I shorted its output to ground — but I don’t understand how this happened. AIUI, any input source could have opted to short signal shield to earth, no? And that would mean that the GND path through the probes/osc would not be the reason this happened?

Now I am about 95% sure I didn’t do this, but it’s the only explanation I have:

MAYBE I made a mistake connecting the probes to the speaker terminal and put speaker pos on the osc GND, effectively shorting speaker pos and neg (i.e. via the connected GNDs on the osc, I connected speaker pos to signal shield, and thus speaker neg).

So in order not to repeat exploding MOSFETs, I was wondering if my measuring setup was alright, or if I missed something there. I also wonder if having no load at all (no speakers) on the terminals is a terrible idea. My gut tells me no, but I've clearly been wrong in the past....

Thanks in advance!

[ahsrabrifat]

Before clipping the oscilloscope probe ground, use a multimeter to check for voltage differences between oscilloscope ground and the amp’s circuit ground.

[klausman]

Before clipping the oscilloscope probe ground, use a multimeter to check for voltage differences between oscilloscope ground and the amp’s circuit ground.

Yeah, I will definitely do that next time (or any time I am not 100% sure about the thing I am clipping the probe to). Fortunately, all I blew was a $25 MOSFET (though finding a matching replacement 2S1529-Y will be interesting). It would have been so much worse had I ruined my new oscilloscope!

[TheoB]

Some amplifiers drive both the plus and minus. This is standard for 12V car amplifiers. Make sure all grounds are floating except one (here your scope). And don't connect the ground leads of the probes to different grounds. It's even better to leave one ground floating if you are not sure the ground points are firmly connected. Current will flow through the ground shield of your probes otherwise.

[BillyO]

You basically shorted your "-" output to the amplifier ground.  There is no guarantee that the "-" output is at ground level.  Scope probe shields can pass quite a lot of current, as you found out.

Is there no schematic available for that amp?

[klausman]

Some amplifiers drive both the plus and minus. This is standard for 12V car amplifiers. Make sure all grounds are floating except one (here your scope). And don't connect the ground leads of the probes to different grounds. It's even better to leave one ground floating if you are not sure the ground points are firmly connected. Current will flow through the ground shield of your probes otherwise.

I had suspected that, since some high-power bridged amplifiers are the same, but this amp is not running in bridge mode. I checked the schematic (I fortunately have the service manual), and the signal ground is indeed connected to the Speaker Negative (I also tested it with a DMM (amp powered off of course)). As for the general advice to "only have one croc biting something": yes, I learned that lesson, even if it wasn't the problem here. At a minimum, it reduces the chances of making a mistake with a ground conn.

You basically shorted your "-" output to the amplifier ground.  There is no guarantee that the "-" output is at ground level.  Scope probe shields can pass quite a lot of current, as you found out.

Is there no schematic available for that amp?

Yes, see above. I'd share the PDF of the service manual here, but I don't know if that is a big No-No in this forum (copyright and all that). If you have access to Hifi Engine's library, the service manual for the Sony TA-FB930R is the relevant one. Might be available elsewhere, too, of course.

[tooki]

You basically shorted your "-" output to the amplifier ground.  There is no guarantee that the "-" output is at ground level.  Scope probe shields can pass quite a lot of current, as you found out.

Is there no schematic available for that amp?

It is, and in this case the speaker negative output is indeed at ground. Looking at the service manual (which I spent more time studying than necessary, since by pure chance I own the same amp) the input grounds do eventually end up at the central star ground of the system. (Amusingly, the star ground in these amps is an actual component, a stamped solid copper thingie soldered into the board. See the crop of a photo I found online of the successor model’s board. The board also has big fat solid wire jumpers for all the high-current connections.)

So to me the real question is “what happened?” The power supply is indeed floating, the grounds end up in the same place, so what actually occurred?

[tooki]

Yes, see above. I'd share the PDF of the service manual here, but I don't know if that is a big No-No in this forum (copyright and all that). If you have access to Hifi Engine's library, the service manual for the Sony TA-FB930R is the relevant one. Might be available elsewhere, too, of course.

Sharing service manuals here is common. Moreover, the one in question is also available on the Internet Archive with no login needed, as well as on Elektrotanya (a fantastic resource for which no login is required for a limited number of downloads, but I recommend creating a free login since you then get unlimited, no-delay downloads).

Before clipping the oscilloscope probe ground, use a multimeter to check for voltage differences between oscilloscope ground and the amp’s circuit ground.

Yeah, I will definitely do that next time (or any time I am not 100% sure about the thing I am clipping the probe to). Fortunately, all I blew was a $25 MOSFET (though finding a matching replacement 2S1529-Y will be interesting). It would have been so much worse had I ruined my new oscilloscope!

FYI, in the service manual, did you see the section about adjusting the bias after transistor replacement? (Page 4). In that section, they refer to using a “slide regulator”/“slidack”, which is basically a gibberish translation — what they mean is a variac. Should you not have a variac, but are near Zurich, I have a spare one at work.

[klausman]

FYI, in the service manual, did you see the section about adjusting the bias after transistor replacement? (Page 4). In that section, they refer to using a “slide regulator”/“slidack”, which is basically a gibberish translation — what they mean is a variac. Should you not have a variac, but are near Zurich, I have a spare one at work.
[/quote]

I indeed am in Zurich (Wiedikon, to be precise). Thanks for the offer!

I saw the bias adjustment instructions, and reading them:

1. Rotate fully the bias adjusting semi-fixed resistors (RV500, RV600) to the MIN position (counterclockwise).
2. Connect a digital voltmeter to the CN500 and CN600.
  (table of pinouts here)
3. Set the slide regulator to 0V, and press the ⏼ button (ON) on the set.
4. Raise gradually the slide regulator voltage up to the rated voltage.
5. At this time, confirm that the SP relay is turned on (it will click).
6. Adjust the RV500 and RV600 so that a reading of digital voltmeter is 10 ± 1 mV.
7. Return the slidack to 0 V, and press the ⏼ button (OFF) on the
set

To me that sounds like "set bias trimmer to 0, slow-start the amp (but do go to 230V before doing anything else), do bias adjustment, then slow-stop the amp".

I am not sure the Variac is really needed unless there is suspicion of a fault somewhere? Or am I missing something? I would have done the adjustment like described, just without a Variac.

[klausman]

So to me the real question is “what happened?” The power supply is indeed floating, the grounds end up in the same place, so what actually occurred?

I have two hypotheses:

• Maybe I did not set up the extra scope probes on the outputs correctly and transposed probe and probe ground of one, effectively shorting the right channel + to ground via the oscilloscope. Unfortunately, I didn't have the presence of mind to take a picture of the setup just after the fault occurred. In my defense, my mind was a bit frazzled after the pop+smoke+fire.
• Perhaps I assembled the boards/conenctors in the wrong way. One connector set can be transposed and this would result in the input board connector carrying the signals and grounds from the inputs directly into the main board inputs (the pinout  even is the same!), i.e. intstead of connecting CN102-B-CN300 (input board to tone board) and the connector from the volume pot PCB to the main board (CN303->CN501) maybe I ran the connection from the input board to the main board (CN102->CN501).

I can't rule out either scenario to 100%, but I think either (or worse, doing both) could spell death for a MOSFET. Rest assured, once the repairs have been done (and I have quadruply checked my work), I will be very, very careful with testing.

[SteveThackery]

Just a general question: wouldn't you expect the output stages to have some sort of protection against short circuit?  Pops and flames seems a pretty undesirable way of responding to an accidental short on the speaker output.

[tooki]

FYI, in the service manual, did you see the section about adjusting the bias after transistor replacement? (Page 4). In that section, they refer to using a “slide regulator”/“slidack”, which is basically a gibberish translation — what they mean is a variac. Should you not have a variac, but are near Zurich, I have a spare one at work.

I indeed am in Zurich (Wiedikon, to be precise). Thanks for the offer!

No kidding! I am near Bhf. Wiedikon myself! (Though I work up at Irchel. I actually have a variac at home, too…)

I saw the bias adjustment instructions, and reading them:

1. Rotate fully the bias adjusting semi-fixed resistors (RV500, RV600) to the MIN position (counterclockwise).
2. Connect a digital voltmeter to the CN500 and CN600.
  (table of pinouts here)
3. Set the slide regulator to 0V, and press the ⏼ button (ON) on the set.
4. Raise gradually the slide regulator voltage up to the rated voltage.
5. At this time, confirm that the SP relay is turned on (it will click).
6. Adjust the RV500 and RV600 so that a reading of digital voltmeter is 10 ± 1 mV.
7. Return the slidack to 0 V, and press the ⏼ button (OFF) on the
set

To me that sounds like "set bias trimmer to 0, slow-start the amp (but do go to 230V before doing anything else), do bias adjustment, then slow-stop the amp".

That’s how I understood it, too.

I am not sure the Variac is really needed unless there is suspicion of a fault somewhere? Or am I missing something? I would have done the adjustment like described, just without a Variac.

I’m not really sure, either. Perhaps there is some failure mechanism we don’t know of that this procedure protects against. Sony service manuals do not normally specify using a variac, so there must be some reason they say to use it here.

[klausman]

I indeed am in Zurich (Wiedikon, to be precise). Thanks for the offer!

No kidding! I am near Bhf. Wiedikon myself! (Though I work up at Irchel. I actually have a variac at home, too…)

I live just around the corner, then (Weststrasse). Once I got all the spares back, I'll DM you about the variac, if that's ok?

[tooki]

So to me the real question is “what happened?” The power supply is indeed floating, the grounds end up in the same place, so what actually occurred?

I have two hypotheses:

• Maybe I did not set up the extra scope probes on the outputs correctly and transposed probe and probe ground of one, effectively shorting the right channel + to ground via the oscilloscope. Unfortunately, I didn't have the presence of mind to take a picture of the setup just after the fault occurred. In my defense, my mind was a bit frazzled after the pop+smoke+fire.
• Perhaps I assembled the boards/conenctors in the wrong way. One connector set can be transposed and this would result in the input board connector carrying the signals and grounds from the inputs directly into the main board inputs (the pinout  even is the same!), i.e. intstead of connecting CN102-B-CN300 (input board to tone board) and the connector from the volume pot PCB to the main board (CN303->CN501) maybe I ran the connection from the input board to the main board (CN102->CN501).

I can't rule out either scenario to 100%, but I think either (or worse, doing both) could spell death for a MOSFET. Rest assured, once the repairs have been done (and I have quadruply checked my work), I will be very, very careful with testing.

Scenario 2 definitely wouldn’t cause the issue, since the tone and volume controls are completely passive. So connecting the input board directly to the main board would simply feed the input signal to the amp at full volume. Maybe not ideal, but exceedingly unlikely to cause damage.

Even in scenario 1 I wouldn’t expect catastrophic damage. With no input signal, the output signal should be 0V, too, so no current would flow. The outputs are current-monitored, so the protection circuit should kick in if you did apply an input signal with the output shorted, I would think.

[tooki]

I indeed am in Zurich (Wiedikon, to be precise). Thanks for the offer!

No kidding! I am near Bhf. Wiedikon myself! (Though I work up at Irchel. I actually have a variac at home, too…)

I live just around the corner, then (Weststrasse). Once I got all the spares back, I'll DM you about the variac, if that's ok?

Sure!

[klausman]

Just a general question: wouldn't you expect the output stages to have some sort of protection against short circuit?  Pops and flames seems a pretty undesirable way of responding to an accidental short on the speaker output.

I honestly don't know. The amp[ does have some protection circuitry on the SP-TM board, but I don't know what specifically it guards against. One part of my planned repair is to figure out what could have shorted the MOSFET in question, or if "somehow" more than its design voltage made it through from the PS.

I'll see if I can test everything but the main board separately. Unfortunately, this amp does not have a "pre out"/"main in" type of loopback on the back panel. Otherwise I'd test those two parts of it separately.

[JoanBS]

It could have been caused by a faulty ground connection.
Have you measured the voltage between the ground and the AC power terminals?

[klausman]

It could have been caused by a faulty ground connection.
Have you measured the voltage between the ground and the AC power terminals?

The amp itself has no ground (Europlug), so no ground connection to fail there. Haven't yet measured the voltage between signal ground or chassis and the AC points, since the amp is currently sitting disassembled on the workbench while I wait for parts.

[JoanBS]

According to the diagram the amplifier is grounded via the probes and the oscilloscope.
The correct thing to do would be to ground the amplifier chassis and connect the oscilloscope via an isolation transformer (not ground).

[tooki]

According to the diagram the amplifier is grounded via the probes and the oscilloscope.
The correct thing to do would be to ground the amplifier chassis and connect the oscilloscope via an isolation transformer (not ground).

Why should it matter, since the amp’s power supply is galvanically isolated already through its transformer? Since everything else in the entire setup was floating, the scope’s “real” ground shouldn’t make any difference.

[klausman]

For those still following along, I've posted an update here: https://www.eevblog.com/forum/beginners/hifi-amp-output-mosfet-matching-(or-lack-thereof)/msg5837897/#msg5837897