Sorry, but the above made me think that you'd put on only two of those 10-ohm resistors on the blown side. You know what, don't change to 10 ohms, just leave it the way it was, and use the amplifier normally. And if these ICs blow again during usage, just add the emitter follower then.
I just touch to see if things are abnormally hot, not sure of what the IR measures.
Last edited:
Ok, Jumper installed, amp idling fine, temperatures are the same on thermal cam.
Fets read 35c and feel like that to touch ... IC reads the same as above (60-65c) and I can just hold my finger on for about 30seconds so it is hot.
Centre of the IC on the side without fets at the moment reads 45c and feels like it.
---------------------------------------------------------------------
Yes the 1st fet failed when I had installed only 2 of the 10ohm resistors (it's well detailed), so I thought it was the imbalance of the gate resistors that caused the fault?
----------------------------------------------------------------------
I have a feeling that the 2nd fet that failed did so because it was also maybe partially damaged along with the 1st fet but still read ok on the low voltage tester (this was the 1st test with 10ohms both sides) The reason I think this is because when I switched to 240v I noticed the green channel active LED flashing for the couple of seconds before the 2nd fet failed .....
This time with a different fet I noticed the LED stays solid green and the amp idles fine although it has the same high? temperature in the centre of the IC.
It could of course be because there was 10ohm gate resistors installed instead of 6.8ohm but from what I have read and from what you've said 10ohm is not an unusual value to use but I have no previous experience of Class D so that's why I am here asking advice.
??? So you think now it is too risky to try 10ohm gate resistors?
I'm gonna scope the IC's later anyway with my battery scope, it's not greatly accurate but it will show square waves etc. ok and it's isolated (needs charged for a while though) ...
P.S. ... The amp is drawing 41watts at idle is that unusual?
Fets read 35c and feel like that to touch ... IC reads the same as above (60-65c) and I can just hold my finger on for about 30seconds so it is hot.
Centre of the IC on the side without fets at the moment reads 45c and feels like it.
---------------------------------------------------------------------
Yes the 1st fet failed when I had installed only 2 of the 10ohm resistors (it's well detailed), so I thought it was the imbalance of the gate resistors that caused the fault?
----------------------------------------------------------------------
I have a feeling that the 2nd fet that failed did so because it was also maybe partially damaged along with the 1st fet but still read ok on the low voltage tester (this was the 1st test with 10ohms both sides) The reason I think this is because when I switched to 240v I noticed the green channel active LED flashing for the couple of seconds before the 2nd fet failed .....
This time with a different fet I noticed the LED stays solid green and the amp idles fine although it has the same high? temperature in the centre of the IC.
It could of course be because there was 10ohm gate resistors installed instead of 6.8ohm but from what I have read and from what you've said 10ohm is not an unusual value to use but I have no previous experience of Class D so that's why I am here asking advice.
??? So you think now it is too risky to try 10ohm gate resistors?
I'm gonna scope the IC's later anyway with my battery scope, it's not greatly accurate but it will show square waves etc. ok and it's isolated (needs charged for a while though) ...
P.S. ... The amp is drawing 41watts at idle is that unusual?
Last edited:
Here are the pics:
1) HO on the side with fets fitted:
2) HO on the jumpered side without fets:
3) LO on the side with fets fitted:
4) LO on the jumpered side without fets:
Scope set to AC coupling to get a decent size waveform ... (10x compensated probe).
Amp idling, no load, gain knobs at minimum.
What do make of that?
1) HO on the side with fets fitted:
2) HO on the jumpered side without fets:
3) LO on the side with fets fitted:
4) LO on the jumpered side without fets:
Scope set to AC coupling to get a decent size waveform ... (10x compensated probe).
Amp idling, no load, gain knobs at minimum.
What do make of that?
I think your blown side IRS2092 was not damaged, only the MOSFETs were. Note that I'm assuming the high-side to be w.r.t VS and the low-side to be w.r.t -80V.
The gate charging seems to be really slow even with 6.8 ohms , looks like those MOSFETs need an emitter follower buffer stage after the IC. I'm quite sure the IR2110 (SMPS) HO/LO waveforms look much better than these.
Also, how many MHz is that scope ? There's absolutely no ringing to be seen !!!
The gate charging seems to be really slow even with 6.8 ohms , looks like those MOSFETs need an emitter follower buffer stage after the IC. I'm quite sure the IR2110 (SMPS) HO/LO waveforms look much better than these.
Also, how many MHz is that scope ? There's absolutely no ringing to be seen !!!
Last edited:
My take is:
MOSFETs and IC (w/o MOSFETs) = OK.
IC driving MOSFETs = hotter than usual.
There are mainly two reasons for heating of the ICs:
1) The current (source/sink) possibly exceeds the rating (as we've already seen).
2) The drive voltage used is 15V, when 12V should suffice in most applications. The switching losses are proportional to the square of this voltage, due to which the losses at 15V would be at 156% of those at 12V. (15*15 / 12*12 = 1.56.)
However, since the original design was for 15V, it may be better to leave it like that and add an emitter follower buffer instead, to solve both 1) and 2) rather neatly. But, since you had the amplifier working without all that, you could just leave it right there and see how things change in the long run, after which, any further failures would be clear indicators of the necessity for the EF buffer.
10 ohms is not a bad value at all, but there's an extra internal resistance within the MOSFET itself (often labelled Rg int.) which, for your MOSFET, has not been given by the manufacturer. This value could be upto 2-3 ohms and therefore, it maybe best to leave it at 6.8 ohms, especially because the amplifier was completely functional at that value.
Something less than 20W would be more of a proper idling power for this kind of amplifier.
Yes WRT the correct locations .. and yes agree only mosfets were damaged ... I've noticed now that one of the channel panel leds flashes if there is a fault with the mosfets. I've noticed twice now that if these lights flash at any time then even if the fets have tested correctly with DMM & transistor tester that a fault will occur when at proper running voltages. Might be some kind of undocumented fault indication.
Yes they do .... but that is a better scope too though.
Ummm it's not great, it's an old Velleman HPS140 .... 10Mhz, it's best used for just seeing if a wave is present, but the PK-PK voltage reading of the square wave is quite accurate (tested on my sig gen).
I see, yes that's the way it is ... I'm hoping to get some more info on typical running temp, I'll maybe write to IR and ask.
Well the circuit boards in this are dated 2008 and the guy I got it from is a regular performer (every weekend + some week nights) and has had it from new, so it seems to have lasted a while with fairly frequent use. I got a chance to test the speakers attached to this (after the original fault occurred) with proper full range frequency sweep impedance tests and one was still working but had a partially shorted voice coil so the impedance was below 2ohms. I'm pretty sure this is what caused the original fault.
Last edited:
Woops missed this bottom paragraph.
Yes I've seen Rg int. on other fet spec. but I've still got a feeling it may have been a dodgy fet when both sides were 10ohm as the led was flashing, just before the fet failed (I've seen that twice now) ... hmmm mind you both time either one or both sides had 10 ohm resistors.
Yes I'd have been much happier all round with much lower idling power but there's nothing else obviously wrong.
Can you tell me which reference design the emitter follower buffer is used in so I can have a look to see what would be involved?
I think it was Iraudamp9, the high-power one, 1.7kW, but one could easily use the commonly available BD139/140 in place of the itsy bitsy SOT-23 transistors. But before you add the buffer, it may be a good idea to check the rise/fall times and gate voltage wave-forms of the healthy side on a good scope. I strongly suggest leaving it the way it was.
And, one more thing is that the gate driver itself would have a finite output impedance of few ohms (1-2ohms), that is once again, not specified by the manufacturer.
And, one more thing is that the gate driver itself would have a finite output impedance of few ohms (1-2ohms), that is once again, not specified by the manufacturer.
Last edited:
Ok thanks I'll have a look, yeah don't worry I'm not in a rush to mod unless I have to, just interested in learning.
I've written to Infineon about the 2092 temp. ... I'll report back with the reply ....
That's very beautiful with fast rise/fall times, no problems seen.
So your wave-forms indicate the need for faster gate charging which was why the 10ohm value didn't work. You actually needed to reduce (not increase) the gate resistor. And since that can't be done without the emitter follower buffer, I think this is where you stop.
However if you would like to experiment, you may proceed with adding the buffers.
So your wave-forms indicate the need for faster gate charging which was why the 10ohm value didn't work. You actually needed to reduce (not increase) the gate resistor. And since that can't be done without the emitter follower buffer, I think this is where you stop.
However if you would like to experiment, you may proceed with adding the buffers.
Last edited:
Yes that is the total draw measured at the mains socket.
I'd forgotten all about that .... they have requested further info:
Can you please provide the following inputs:
1. What is the ambient operating temperature?
2. Is the captured data a steady-state value?
3. What is the calculated power dissipation in the IRS2092?
4. Are you using the SOIC16N or DIP16 package?
I'm not sure about how to properly calculate number 3. in this instance .... can you help?