All,
I'm planning to build a home theater amplifier with 8channels x 100W into 8ohms Class D.
I've been looking for a reference design that I can tweak, combine and build. I spent a good part of the last 5 years playing with Class D at UHF/L-Band. But I don't know much about class D at Audio.
I've looked at the TDA895x series, and the IRS2092 drivers. I think I've settled on the IRS2092 based on flexibility. I have a couple questions:
1) Are there any other good Class D kinda monolithic reference design ( a al Self/Cordell) suggestions that I missed?
2) Any problem with feeding 8 channels (Crosstalk?) with one power supply other than supplying the needed VA?
3) On the IRS2092 - do I have to supply the 12V Vcc referenced to -Vss....Not sure how to do that...
Thanks,
Scot
I'm planning to build a home theater amplifier with 8channels x 100W into 8ohms Class D.
I've been looking for a reference design that I can tweak, combine and build. I spent a good part of the last 5 years playing with Class D at UHF/L-Band. But I don't know much about class D at Audio.
I've looked at the TDA895x series, and the IRS2092 drivers. I think I've settled on the IRS2092 based on flexibility. I have a couple questions:
1) Are there any other good Class D kinda monolithic reference design ( a al Self/Cordell) suggestions that I missed?
2) Any problem with feeding 8 channels (Crosstalk?) with one power supply other than supplying the needed VA?
3) On the IRS2092 - do I have to supply the 12V Vcc referenced to -Vss....Not sure how to do that...
Thanks,
Scot
Well does anyone know the answer to the third question? Seems like several people have worked on the IRS2092. What do you do for the 12V supply?
Thanks for any help...
Scott
Thanks for any help...
Scott
Last edited:
I'm looking for a schematic for IRS2092 as well. Do you really need 100W on all 8 channels? You can use LM3886 for the low power parts of the system, and a more powerful amp for the sub, depending on what setup you have.
And about the crosstalk thing, I don't think it's an issue if you use large enough filter caps.
And about the crosstalk thing, I don't think it's an issue if you use large enough filter caps.
After going through the irs2092 design cycle a few times I would recommend buying in ready made modules.
The 2092 is fussy about pcb layout. Correct decoupling is vital.
I also had a weird problem on power down with high value power supply capacitors.
On power down I would get a siren type noise through the speaker for a few seconds then there would be a huge thump. IR told me I needed a reset circuit on the 2092 to stop this. I had to design a PIC micro based circuit to monitor VCC and shutdown the 2092 when the VCC sagged.
The output mosfets need careful decoupling and snubbers.
I also had to put some work in with output mosfets as I wanted high power.
If you are only using 100 watts then the irf4019's work OK without gate buffers.
You also need to get the dead time right, not just on the 2092 but also the gate resistors.
The 2092 is also a bit fussy about no load. I found it got in one of two states. It would either sit at +17VDC or it would oscillate OK.
If it was going to be tested with no speaker then the output filter capaictor needs to be high voltage as the output filter rings with no load on the output.
The 2092 is fussy about pcb layout. Correct decoupling is vital.
I also had a weird problem on power down with high value power supply capacitors.
On power down I would get a siren type noise through the speaker for a few seconds then there would be a huge thump. IR told me I needed a reset circuit on the 2092 to stop this. I had to design a PIC micro based circuit to monitor VCC and shutdown the 2092 when the VCC sagged.
The output mosfets need careful decoupling and snubbers.
I also had to put some work in with output mosfets as I wanted high power.
If you are only using 100 watts then the irf4019's work OK without gate buffers.
You also need to get the dead time right, not just on the 2092 but also the gate resistors.
The 2092 is also a bit fussy about no load. I found it got in one of two states. It would either sit at +17VDC or it would oscillate OK.
If it was going to be tested with no speaker then the output filter capaictor needs to be high voltage as the output filter rings with no load on the output.
Ready made modules aren't better either. At least the cheap ones. And the manufacturers don't provide the schematic for troubleshooting.
Sent from my Nexus 4 using Tapatalk
Sent from my Nexus 4 using Tapatalk
That's an interesting idea. Would reduce the transformer requirements.
Certainly many home theater receivers operate under assumptions that only a few channels are pushed hard at the same time -- I wonder if that's really true -- or if the amps are just distorting a lot.
nigelwrite7557 said:
Yeah, I'm a little nervous about this. I've done Class D at 500 MHz but at lower power levels. It can be a real pain. I know that grounding is critical.
The power down circuit sounds like a pain to implement - I'm not a fan of writing microcode.
Many of the kits I've seen have the power supply on board - which would make it more difficult to integrate I think.
FOr 2092, I'm considering adopting the SystemDLite (sp?) that Chocoholic has put together. But I would combine the gainboard and amp on the same PCB.
Thanks for the thougths.
Scott
If they don't work at least with ebay you can send them back.
Ebay isn't as bad as it used to be, bad sellers get jumped on pretty quickly these days. Also check their feedback before buying.
Or spend a little more money on a better pcb.
At least buying ready made they shouldn't have any of the problems I came across with the 2092.
MAybe surf on over to Carvin and look at their prosound amps that use those chips. Else, classdaudio.
I don't mean to be rude, but your PCB layouts are below average and the ready made ones on Ebay are slightly above average. its better to take a course in PCB layout and gain some experience in high speed layout with simulation modelling and off you go! but that's hard work only few on this forum will undergo Chocoholic is one of them.
If your system is made of multiple small speakers and a sub, then each small speaker can be driven by a chip amp. LM3886 has enough headroom for such a task. And you can add an active filter before the amp.
Then the only issue is the sub, which needs more power, but I guess you can use multiple chip amps either in bridge or parallel mode. Never tried this one, but I will.
- Status
- Not open for further replies.