No not for a living, I am retired now (health issues), from being former musician/sound engineer/guitar tech ... the electronics I studied when I was young was put aside completely until a couple of years ago when I had time to pick it up again and because everybody I know is involved in pro-audio there is regularly somebody looking for a repair (I can't get about now but still do what guitar repair I can manage too) ... as previously mentioned though Class D has been late to take over here and I believe we are about to see a lot more failed gear as the 3 & 5 years warranties run out. None of the few local music shop repairers will touch Class D because it's so hard to get schematics & they are not set up for soldering to the large heatsink/ground planes.
I spoke to a guy recently who contacted a distributer for the brand "Alto" & was told that there are are no schematics sent out to anybody (even previous official service people for them) for their Class D anymore. The lucky thing I've noticed though is most Class D seem to follow the data sheet typical use diagram very closely .... 🙂
I spoke to a guy recently who contacted a distributer for the brand "Alto" & was told that there are are no schematics sent out to anybody (even previous official service people for them) for their Class D anymore. The lucky thing I've noticed though is most Class D seem to follow the data sheet typical use diagram very closely .... 🙂
While waiting on parts arriving, I was having a read at what I could find in a search for "irs2092 irfb4227 gate resistors" on this forum.
I found a few people using 6.8r resistors & one using 4.7r with irs2092 & irfb4227 ... e.g,
"My first attempt at a pcb was a disaster with the 2092 kept resetting on medium volumes. I had to set OC setting to as high as possible to get a decent volume.
I lowered the gate resistor to 4R7 and that helped quite a bit as I was using irfb4227 mosfets which have a large gate charge. A gate driver IC helped a lot especially when I went on to use two pairs of mosfets".
So the good point to using higher value gate resistor in my case is cooler IC ... are there any bad points that may occur?
Also I am testing with a dummy load of 8ohms both sides, could changes I make that are fine with 8ohms cause problems with lower ohm loads?
I found a few people using 6.8r resistors & one using 4.7r with irs2092 & irfb4227 ... e.g,
"My first attempt at a pcb was a disaster with the 2092 kept resetting on medium volumes. I had to set OC setting to as high as possible to get a decent volume.
I lowered the gate resistor to 4R7 and that helped quite a bit as I was using irfb4227 mosfets which have a large gate charge. A gate driver IC helped a lot especially when I went on to use two pairs of mosfets".
So the good point to using higher value gate resistor in my case is cooler IC ... are there any bad points that may occur?
Also I am testing with a dummy load of 8ohms both sides, could changes I make that are fine with 8ohms cause problems with lower ohm loads?
The slower charging of the gates would cause a slight increase in THD and switching losses, but is unlikely to be any noticeable. The typical application circuit from the datasheet uses 12V / 10 ohms = 1.2A. Note that the thermometer has measured a case (not junction) temperature close to 60*C which, clearly, should be hot to touch.
The OC is over-current protection blanking time, not related to gate-drive. The conduction losses for lower loads would be higher but I guess the amplifier should have adequate heat-sinking, as it seems to be rated at 800+800 watts.
The OC is over-current protection blanking time, not related to gate-drive. The conduction losses for lower loads would be higher but I guess the amplifier should have adequate heat-sinking, as it seems to be rated at 800+800 watts.
Not my day today ... I couldn't wait so I got 4x 10ohm through hole resistors, shortened the leads as much as possible, tinned them, fitted them, turned back on and instant burning smell ... I immediately turned off .... then noticed 2x of the 4x 10ohm resistors still on the bench, (I'd only fitted 2x of them).
So that was one pair of fets now with 10ohm gate resistors and the other pair still with 6.8ohm when I switched on.
I'll be back once I've investigated the damage ... 🙁
So that was one pair of fets now with 10ohm gate resistors and the other pair still with 6.8ohm when I switched on.
I'll be back once I've investigated the damage ... 🙁
So 1x of the irfb4227 mosfets has bitten the dust .... no way to tell if the irs2092's survived.
Secondary voltages are still there so SMPS looks to be ok.
I have no spares of either so now I will have to try and find somewhere with parts, that wasn't easy the first time round ... 🙁
Secondary voltages are still there so SMPS looks to be ok.
I have no spares of either so now I will have to try and find somewhere with parts, that wasn't easy the first time round ... 🙁
Blown MOSFETs are usually straight shorts across the DC bus. Open-circuit failures are quite rare.
To test the modulator, you feed both the input and output of the working channel into those of the blown one (only after removing the MOSFETs) so as to complete the feedback loop. You may then look at the gate pulses from the suspected part to verify if it's functional. You may also check for similarity with the working channel, as both have the same input signal fed into them.
There was a high-power reference design, iraudamp9, that used low valued gate resistors. However, that design also used extra emitter-follower buffers instead of driving the gates directly from the IRS2092 (Fig. 24 of the below).
https://www.infineon.com/dgdl/iraudamp9.pdf?fileId=5546d462533600a40153569aee022bff
To test the modulator, you feed both the input and output of the working channel into those of the blown one (only after removing the MOSFETs) so as to complete the feedback loop. You may then look at the gate pulses from the suspected part to verify if it's functional. You may also check for similarity with the working channel, as both have the same input signal fed into them.
There was a high-power reference design, iraudamp9, that used low valued gate resistors. However, that design also used extra emitter-follower buffers instead of driving the gates directly from the IRS2092 (Fig. 24 of the below).
https://www.infineon.com/dgdl/iraudamp9.pdf?fileId=5546d462533600a40153569aee022bff
Last edited:
Yes the blown fet is shorted S - D.
Can you explain that in more detail please? Also can you confirm where to connect the scope ground?
Yes I'd come across that yesterday.
Can you explain that in more detail please? Also can you confirm where to connect the scope ground?
Yes I'd come across that yesterday.
Last edited:
The scope ground goes to the -80V DC bus. Since this is not earth potential, you'd need to float the scope to measure like this.
And, as the high-side pulse (HO) is referred to VS, its amplitude would then be (Vs + 15V). That is, for a +/- 80V supply, VS swings 160V, HO swings (VS+ 15) = 160+15 = 175V, whereas LO would do only the usual 15V.
The IC is not necessarily damaged. I haven't been able to blow one such HVIC from International Rectifier yet. These devices are quite robust.
And, as the high-side pulse (HO) is referred to VS, its amplitude would then be (Vs + 15V). That is, for a +/- 80V supply, VS swings 160V, HO swings (VS+ 15) = 160+15 = 175V, whereas LO would do only the usual 15V.
The IC is not necessarily damaged. I haven't been able to blow one such HVIC from International Rectifier yet. These devices are quite robust.
Last edited:
Yes thanks for confirming that & I can float the scope.
It was more the connections between the 2x IC's that I wasn't totally sure about ... can you tell me from what exact points you mean?
The idea is to feed both ICs with the same signal. However, since one of the power bridges is blown, the IC corresponding to that channel cannot function (since it is a closed-loop modulator). We, thus need to provide an artificial feedback from the VS of the working channel to fool the questionable IC into thinking that it is indeed driving a half-bridge, and then check its gate signals to see if they're similar to those of the working IC.
Alternatively, you could try measuring across the HO and LO of the IC to see if it's shorted out from inside, and if OK, then also compare any readings to those of the healthy IC, for confirmation, all without any power being applied to the amplifier.
Alternatively, you could try measuring across the HO and LO of the IC to see if it's shorted out from inside, and if OK, then also compare any readings to those of the healthy IC, for confirmation, all without any power being applied to the amplifier.
Page 19 of another reference design, iraudamp5, shows a thermal image of the design at 15W power, according to which, almost all components on the board, including the gate-driver, are at temperatures close to 60*C. Now, if that's deemed as normal by the manufacturer, then maybe you could just leave things the way they were (Rg = 6.8 ohms), especially if you think that that was what blew the MOSFET.
That's interesting ... yes I also found this page where the IPR 1600 is torn down and tested, if you look at the camera temperature colour scale the thermal images taken show very similar temperatures on the IRS2092's to what I was seeing & the guy there gives the amp the thumbs up (no mention of excessive temperature). See here: https://www.abeltronics.co.uk/amptesting.php?z=peavey_IPR-1600
I think the damage was done by the accidental difference in gate resistors value (nothing else was changed).
I've got a couple of other things to get out of the way but I'll be back soon with some more results.
Just for confirmation, yes I thought it would be VS ... so connecting good VS and IN to respective VS and IN on blown side?
Yes, it's VS that needs to be copied to the blown side, and the 60*C on the thermals may also not be very true. Have you tried touching the ICs while running the amp? I've never seen any of these IR chips get that hot while operating. 60*C should be sufficient to melt some candle wax !!
Ok ... that didn't go too well ... 😱
There were no fets installed on the "bad" side .... so fets that tested good were installed on the "good" side, jumpered good VS over to bad VS (jumpered from D on Q104 to D on Q204), fired up on the lightbulb limiter to check for shorts, there were none.... switched to full 240v and Pooof! .... one of the "good" fets failed.
That was with 10ohm gate resistors on both sides.
So at the moment I'm left with 2x fets that test ok and the SMPS is still functioning properly.
There were no fets installed on the "bad" side .... so fets that tested good were installed on the "good" side, jumpered good VS over to bad VS (jumpered from D on Q104 to D on Q204), fired up on the lightbulb limiter to check for shorts, there were none.... switched to full 240v and Pooof! .... one of the "good" fets failed.
That was with 10ohm gate resistors on both sides.
So at the moment I'm left with 2x fets that test ok and the SMPS is still functioning properly.
Last edited:
I've posted another specific thread just to see what temperature other's have also seen the 2092's at, as I'm still seeing some saying they get hot ...
Last edited:
Actually, what I meant was the same old 6.8 ohms on the good side. See, now you don't know if it's the 10ohm or the IC that blows the MOSFETs. Had it been 6.8 ohms on the good side, you could've easily found that out !! You need to make only one change at a time.
As far as I understand, heat from the neighbouring components (only if really hot) could increase the ambient temperature around the gate driver. Then there can be IR crosstalk between objects photographed by the thermometer, which causes some of the radiation from an object to be influenced by its neighbours.
If "hot" is the way the IC is supposed to be then I think it's better to include the emitter follower (small/no heatsink required) shown in one of the reference designs, for better overall reliability.
Last edited:
Well LOL it would've helped if you had specifically said that, I am not psychic 😵 ... I am well aware to make one change at a time ... to me changing both sets of gate resistors was one change and it was you who told me to do it. I also can't see how raising the gate resistance would be detrimental. Please be more clear in future as to exactly what you mean, don't take it for granted that I understand something you haven't written.
I do not know if it is supposed to be hot or not, you have said both and seem to have settled on that they should be cooler and I am following what you tell me.
So I will try again with the last 2 fets & change the resistors back to 6.8ohms as I now know clearly that is what you meant ... 🙂
I have reset my camera and checked the emissivity ... I have set it to 0.95 as that is what I have seen recommended for IC cases.
I have the last two good fets installed on the "good" side, all 6.8ohm resistors re-installed (no VS jumper yet) and the amp is idling.
Here are shots with only the crosshair temperature showing the centre of the "good" IC, the cooler part of the IC and the pcb beside the IC:
Contact thermometer says 37.5c ... but the IC is directly in front of the fan and the long probe is being cooled by the air from the fan.
The IC is HOT to touch ... I can just keep my finger on it.
Ok I'll be back soon after installing the jumper again ....
I have the last two good fets installed on the "good" side, all 6.8ohm resistors re-installed (no VS jumper yet) and the amp is idling.
Here are shots with only the crosshair temperature showing the centre of the "good" IC, the cooler part of the IC and the pcb beside the IC:
Contact thermometer says 37.5c ... but the IC is directly in front of the fan and the long probe is being cooled by the air from the fan.
The IC is HOT to touch ... I can just keep my finger on it.
Ok I'll be back soon after installing the jumper again ....