Sorry for being vague I'll try to be clearer from now on, I understand all the basic principles, what I'm looking for are more specific details.
I understood that the amplifier was to be tested on it's own with "a transformer supply".
My question above was about:
What type/specification of transformer supply you use, e.g. single bipolar output supply?, multiple supplies? ... something different to what I imagine?
How/where are the supply or supplies connected to the amplifiers different voltage rails?
Do you test parts of the amp in a sequence?
From your last comment, are you still using the variac to bring up power to the transformer supply, .... or using current limiting on the transformer supplies outputs, etc.?
What general potential problems are you looking for?
Any other details that are worth being aware of?
It's best practice "details" of how you guys do this that are of most value to me, I have a good idea of the principles of how this is done ... I just wonder about how exactly Class D designers/veterans go about it.
All your help/advice is very well appreciated!
1) Similar specification as the SMPS provides, if dual then dual, if single then single. Just a step not to involve the SMPS. Lower voltage etc. is OK for testing purposes.
2) After the SMPS secondary rectifier, with SMPS removed / unpowered.
3) Outputs result only if both modulator and power stage are OK and negative feedback loops prevent you from further breaking the amplifier into smaller parts. You may however test channel by channel if you want.
4) That's your choice. A class-D amplifier is more or less like any other. But SMPS is different from transformer that's all.
5) Dead shorts, abnormal heating of any part or any say some kind of protection (if present) getting unnecessarily activated.
6) The SMPS might need a minimum load to function properly, in case the amplifiers are removed from operation.
here is an infineon writeup about LLC resonant converters
https://www.infineon.com/dgdl/Appli...n.pdf?fileId=db3a30433a047ba0013a4a60e3be64a1
https://www.infineon.com/dgdl/Appli...n.pdf?fileId=db3a30433a047ba0013a4a60e3be64a1
Thank you both for the info.
After the secondary side rectifiers there are 3x outputs, +-30v, +- 80v and the Bias circuit
So I have a variac & two linear supplies here: 1x variable 0-30v single output with current limiting, 1 x fixed 15v bipolar output, without current limiting.
Is it possible for me to test the amp with what I have or what else would be required?
After the secondary side rectifiers there are 3x outputs, +-30v, +- 80v and the Bias circuit
So I have a variac & two linear supplies here: 1x variable 0-30v single output with current limiting, 1 x fixed 15v bipolar output, without current limiting.
Is it possible for me to test the amp with what I have or what else would be required?
Lower than usual power? how would that be done properly?
On the Mosfet front RS here has some stock in, not a lot to choose from but here is the best of what is available:
On Semi NTP095N65S3HF Rds(on)= 95mohms, Qg= 66nC
Toshiba TK31E60X Rds(on)= 88mohms, Qg= 65nC
Just keep the volume low that's all. Both MOSFETs seem to be OK.
I insisted on low Rds on because its value increases by almost 100% between 25*C (datasheet value) and 125*C (real-life) junction temp.
Yes, I used what you said previously to search out replacements. The TK31E60 were the best I could find ... there was nothing better available without lead times into next year. In the data sheet the typical Rds on is slightly better too at 73mohms.
My initial plan was to repair the SMPS then install everything apart from the output Mosfets, then check everything up to the driver IC (IRS2092) outputs before fitting the output fets.
I've been reading about shoot-through but not sure of the best way to check for and avoid conditions that will lead to this. Have you any tips for checking everything is ok before finally fitting output fets? (I've already eliminated all the obvious shorted components etc.). What should I be looking for or avoiding on the scope etc.?
Last edited:
As far as I understand, the IRS2092 is a closed-loop modulator that wouldn't work without the output bridge and therefore your initial plan may not succeed as expected.
You do not need to worry about shoot-through unless you have 1) played around with the dead-time control on the modulator chip and / or 2) used gate resistors and / or MOSFETs that are vastly different from the original ones.
Abnormal heating of the MOSFETs, well, mostly. However, if you want to see the high/low side switching waveforms simulataneously, then you would once again need a differential scope, as was the case with the SMPS half-bridge.
You do not need to worry about shoot-through unless you have 1) played around with the dead-time control on the modulator chip and / or 2) used gate resistors and / or MOSFETs that are vastly different from the original ones.
Abnormal heating of the MOSFETs, well, mostly. However, if you want to see the high/low side switching waveforms simulataneously, then you would once again need a differential scope, as was the case with the SMPS half-bridge.
Ok will look into that.
Good stuff, still all original, just wanted to confirm.
Yeah won't be doing that again without proper diff probes. Ok I have a thermal imager so temperature will be watched closely..
Can you give me a ballpark temperature range that you would expect to see with properly functioning fets? ... what would be abnormal?
Another thing I've been wondering about is whether a load is required on the outputs while testing or if that is a bad idea for any reason?
Eventually got back to finishing this, the amp seems to run pretty cool ... here's a couple of thermal images showing the power supply mosfets and the output mosfets (I ended up having to hand form the D2PAK shapes from TO-220's as that was all I could get) but judging by the results it looks as though they worked out) ... these images are after about 40mins of constant bass heavy music (amp controls fully clockwise and peaks hitting the limiters):
Power Supply (these are the fets you helped me choose instead of IGBT's)
The highest temp on the screen is 38.9c & the tab of the top mosfet is only 28.2c (these temps are pretty accurate too as I've tested the thermal imager)
Amplifier (Channel B), Channel A is actually slightly cooler as it is nearer the fan.
The hottest component is the IRS2092 on the right at 56.9c & the fets (IRFB4227) are much cooler, in fact one leg of the top fet is only 31.6c (the coolest point on the screen).
Thanks so much to you all for the info & advice, it's certainly given me a lot more confidence with Class D & it's well appreciated!
Tell me though ... what is the best way for me to monitor the speaker outputs on this amp with my scope?
Power Supply (these are the fets you helped me choose instead of IGBT's)
The highest temp on the screen is 38.9c & the tab of the top mosfet is only 28.2c (these temps are pretty accurate too as I've tested the thermal imager)
Amplifier (Channel B), Channel A is actually slightly cooler as it is nearer the fan.
The hottest component is the IRS2092 on the right at 56.9c & the fets (IRFB4227) are much cooler, in fact one leg of the top fet is only 31.6c (the coolest point on the screen).
Thanks so much to you all for the info & advice, it's certainly given me a lot more confidence with Class D & it's well appreciated!
Tell me though ... what is the best way for me to monitor the speaker outputs on this amp with my scope?
Congratulations on being able to repair a professional amplifier successfully !! This is something many struggle with, eventually ending up scrapping the whole unit.
Any amplifier that's supposed to provide 2x800W clearly needs to draw over 5A from a 240V supply and therefore it's best to be prepared for worst-case power, especially in a professional application. That change was so necessary.
By the way, what was the approximate ambient temperature at that time ? Nothing to worry, just a casual question.
IRS2092 is hot because of the low value of the gate resistors (6.8 ohms). Note that this driver is rated only for 1A, and the peak current is currently 15V / 6.8 = 2.2A. A slightly higher value (say 10-15 ohms) would reduce the die temperatures to normal levels.
However, the IR2110 in the SMPS section is rated for 2A, and would therefore do better I think. If not, adjust the gate resistors to about 7.5 to 10 ohms to keep within the recommended limits of operation.
If the speaker return/ground is earthed (that is if it connects to the utility earth), then you could just check the outputs normally as the SMPS has isolation (by virtue of its own transformer).
If it's not earthed, you could float the oscilloscope by running on battery or UPS (with utility earth disconnected) and then check the outputs. Since the outputs are single-ended and have a common return, both channels of the oscilloscope may be used simultaneously as well.
Last edited:
Thanks newvirus ... but I have actually repaired piles of pro amps & other equipment back in the day and other classes of amp recently (just finished an EV CP2200 class H yesterday) ... it's only Class D that is new to me, I studied Music Technology and Electronics in the early 90's but then took a long spell completely away from it as a pro musician and sound engineer and have only come back to electronics more and more in the last couple of years, I completely missed the introduction of Class D and it's only recently that we have started seeing a lot more of them here (most pro-audio amps locally here are still AB, G or H), I'm still super rusty on some stuff and I've only repaired 2 other Class D with really simple faults. I decided with this one rather than just directly replacing the blown components that I would try and learn as much as possible about Class D and that's why I'm here rather than one of the dedicated repair based forums. Your ongoing help has been great, I know sometimes my questions might seem stupid but really I'm just looking for as much detail as possible in the answers and just like to double check even simple stuff, that's just the way I am.
Yeah I know what your getting at ... I'm in Scotland here so the temperature will have been between about 17-21c ... it's been warmer the last few days so probably nearer 20 or 21c. I never thought, I should have noted that at the time. I'll make sure and do that from now on. The thermal image temps are very accurate though compared against all the other various contact thermometers I have here.
Gate resistor adjustment is what I've been trying to read up on the last couple of days, your confirmation of what I've been reading is well appreciated.
Ok got you, I will adjust the gate resistors, I have very few SMD resistors here, is there any reason why I shouldn't use through hole resistors for testing?(I've got the required values in TH).
Yes thanks for confirming that, I'll get back to it a.s.a.p. and report back with some more images, I have an isolation transformer here if required.
Well I meant the patience that you've shown, which is something I don't have, for some reason. I'm sure that your previous experiences have helped you, as Class D is not rocket science, but just a slightly different SMPS.
21*C is nice, especially since datasheets are written for 25*C. You may use 0.25W TH / SMD resistors for gate drive. If SMD, try to have 1206/1210 or bigger, avoiding 0805 and smaller.
They are 1206 on the pcb, so I've decided to order an E12 series of 1206 as I'm getting more SMD stuff all the time & this seems like a common size I see.
LOL patience, I have a terrible problem with that too, just really busy with many things so there is always something else to do to keep me occupied.
Yes I pretty much understand Class D now enough to repair but to make improvements in reliability is even better ... in many ways it actually seems much simpler than other Classes ... I was just expecting it to be much harder from what I'd previously heard/read.
The amp fets and IRS2092 are just direct replacements for the original parts so what you see in the thermal images above is the way it was running from the original design, it's interesting to see that it can be easily improved.
I'll be back with more info when the resistors arrive next week ...
LOL patience, I have a terrible problem with that too, just really busy with many things so there is always something else to do to keep me occupied.
Yes I pretty much understand Class D now enough to repair but to make improvements in reliability is even better ... in many ways it actually seems much simpler than other Classes ... I was just expecting it to be much harder from what I'd previously heard/read.
The amp fets and IRS2092 are just direct replacements for the original parts so what you see in the thermal images above is the way it was running from the original design, it's interesting to see that it can be easily improved.
I'll be back with more info when the resistors arrive next week ...
It's always nice to be able to use your improved understanding for making repairs, but in case you do this for a living, making these amplifiers too reliable (bulletproof) might just result in you running out of jobs !!!
All the big people like Crown, Peavey, QSC, Yamaha etc. know of repairs being a nice way to keep making money, even after a product has been sold.However, if you're someone who owns / rents out equipment to others, then it's always great to have something that's unbreakable / long living.