testing L-Mosfets

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've recently finished my sterio opti-mos amplifier project, was sounding fantastic for a week.

Yesterday, I just turned on the amp and had very loud beating sound from the speakers. I immediately turned off the amp and it also smelled like something went :hot: .

The 10Ohm 1/2W resistor between the input ground and the connection leading to the star ground was destroyed and now reads 410ohm and 630ohm respectively.

I originally thought the smell came from the burned resistors but I followed my nose and came to the mosfet devices :( .

It seems that one of the N channel device (2SK1058) was blown in the ordeal. Whenever I mounted the device on to the heatsink, the multimeter showed that the 'tab' (source) was in contact even though there's an insulator.

I took off the device and now the other mosfets no longer show a connection to the heatsink but the problem is this mosfet.

I did a quick test with the multimeter by using the diode test to turn on the device (red lead to gate, black lead to source), and then turn off by shorting the gate/source lead and the readings were perfectly fine.
Also, before I took off the device, I even turned the amplifier back on and everything sounded fine again. It smells pretty bad but none of the fuses went and the music was sounding great again.

I'm not quite sure what's happend but, I will need to fix the heatsink and source conducting issue since it's the same on the other channel. (Multimeter even shows that my speaker out connection is touching the chassis)

Is there any further testing I could do to see if the mosfet's really dead? Also any other details to check on? I would really hate to get new mosfets since I've spent too much time and money on this.
Typically you can measure with a multi-meter set to ohms. The following is true for a MOSFET that is out of the circuit.

Measuring from gate to any other electrode should produce infinite resistance.

Measuring from source to drain can produce a low resistance (a few ohms) or a high resistance (a few megohms) depending on the presence of any residual charge on the gate. (Remember that the gate is a pretty good capacitor with extremely low leakage.)

First discharge the gate by touching all three leads at the same time with your fingers. Now if you measure source to drain with the meter, you should get several megohms.

Next charge the gate by measuring gate to source. Be sure to note the polarity of the meter for this test. Now, being careful not to touch the gate with anything whatsoever, measure source to drain again. You will either get a few megohms or a few ohms depending on the polarity of your meter and the polarity of the device.

If you now charge the gate with the opposite polarity (by measuring gate to source with the opposite polarity), you should get the other reading (a few ohms or a few megohms) when measuring from source to drain.

Good luck.
great, the mosfet's still alive! thanks for the advice.

Now it seems I get a very faint crackle/buzz (audible only when I put my ears to the speaker) from the channel I did the mosfet test on. Could it be that the mosfets are functional but damaged?

Found another problem with my soft start circuit, the resistors are blown. :smash:
peak power through the soft start resistors is enormous. 5Apk is about 1250wpk in the resistor.
What did you use?
I have seen 3 by 5w 15r (series) suggested and 3 by 5w 150r (//). I think the thicker wire in the 15r series connected will have more thermal inertia to survive that big peak.
You could couple up the fet as you would test for Vgs. A low voltage DC power supply some resistors and a multimeter is all you need (look up Pass & others for the circuit). You could also check the rise in current as you adjust the gate voltage to get an estimate of the mho & compare the good ones, looking for a mismatch.
Thanks andrew, I haven't thought about that test. I'm sure I did something like that even at school.

The resistors I was using were four 180ohm 5W in parallel. It seems like they might have been dead from the get-go. They were always cold and now they show an infinite resistance on the meter.
I replaced the resistors with 4, 220ohm 5W resistors and now the soft start circuitry is actually doing something instead of just closing a relay.

It also stopped the humming of my transformer for some reason (not complaining!).

Now my amplifier seems to have a DC fault, both channels around 23.1V. It's odd though, if I put my voltmeter from the Line ground to +Vcc, the DC fault is gone (capacitor charging through Vcc?) then back when I take the voltmeter out.

Right now I took out the 10ohm burned resistors and just left the input ground floating. I don't have an input source so this shouldn't matter.
Yeah, it seems the floating input ground was what was causing the DC fault.

I really do think the mosfets were somehow damaged though. The readings from the good one and suspected one are just too different (>10%). They smell bad too.

Also the channel with the faulty mosfets seem to have a -5mV DC offset whereas the 'good' channel is around +1.8mV offset.

I suppose I'll have to get new mosfets (2SK1058), and see if that solves the crackling problem.
Hi Whalefat,

Haven't seen a reply from you as whether you have solved your problem or not so I'll just throw in my experience with some version 2 boards.

I built 15 200 watt Opti Mos channels for my home theater system. All worked fine on first test with one exception which had a high DC offset voltage on the output and would only drive my 8 ohm test load at about 1/3 power, it eventually blew an output device during efforts to find the fault.

The fault turned out to be a bridge on the printed circuit board that was a manufacturing defect. I'm sorry but I don't remember the exact location but it was betwen two of either the input or the Vas stage tansistor pads.

Randy apologized profusely to me for this, it turned out he'd had a defective batch of boards and although he thought he'd culled them all from his stock this one crept in and I missed it during my mass production efforts.

Also for what it's worth I've found that the L-Mosfets can test fine with a multimeter but still be defective at operating voltages. I found this out when I had another self induced failure. I had one of my cases opened up like a butterfly to do some changes to my power supply circuits. I drove it into oscillation by letting the output and input cables get too close together whilst driving a dummy load. The Zobel network inductor rapidly rose to a high enough temperature to melt the solder even though I hit the power switch as soon as I noticed something. Testing the output devices with a multimeter said they were OK but they wouldn't perform when in circuit, replacing them solved the problem. An expensive basic lesson.

I'm now about to launch into upgrading to Randy's new Totem pole design and increasing the number of channels to 18.

Probably overkill but it keeps me out of trouble. By the way if you are ever on the other side of the Rockies in Calgary give me a call, I'm in the phone book.

Mike Singleton
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.