I got 4 x Sanwu TPA3118's biamping my B&W P5 speakers. These speakers are rated 200W at 8ohms.
At normal listening and a tad above these 3118's really sound nice. But they do not like playing louder. And Im also convinced that I can hear the distortion..
Would I gain anything trying to find 4 x TPA3116 boards instead? These are rated 40W higher each board. Problem is Im lead to believe these Sanwu boards sound the nicest in the budget end of class D.
I have also been looking at the IRS2092 but Ive read that they are noisy? I already think the 3118 are noisy even though I lowered gain to 20dB. Also I cannot find any comparison of sound between the TPA's and IRS2092. Probably not a jump people take because of big difference in power supply? So If anyone have experience with both I would like to hear.
At normal listening and a tad above these 3118's really sound nice. But they do not like playing louder. And Im also convinced that I can hear the distortion..
Would I gain anything trying to find 4 x TPA3116 boards instead? These are rated 40W higher each board. Problem is Im lead to believe these Sanwu boards sound the nicest in the budget end of class D.
I have also been looking at the IRS2092 but Ive read that they are noisy? I already think the 3118 are noisy even though I lowered gain to 20dB. Also I cannot find any comparison of sound between the TPA's and IRS2092. Probably not a jump people take because of big difference in power supply? So If anyone have experience with both I would like to hear.
I've tried both TPA3116 and IRS2092 - no contest, in an optimized circuit the IRS is miles ahead. Primarily because you can give its signal stages (the OTA front-end) a very clean low impedance supply because the power pins are brought out of the chip. To my ears there seems to be something lacking in the TPA3116's bass, no matter the power supply details.
Another important matter is the IRS2092 reference design has low impedance input (3k) so needs a good (read classA buffer, not an opamp) driver for it.
Another important matter is the IRS2092 reference design has low impedance input (3k) so needs a good (read classA buffer, not an opamp) driver for it.
I'm also interested to hear opinions about this.
@abraxalito - can you recommend an IRS2092 board which you think has optimized implementation. Thanks.
@abraxalito - can you recommend an IRS2092 board which you think has optimized implementation. Thanks.
None that I have seen has low enough supply impedance on those analog rails. But its no big deal to add the caps necessary on yourself. If you're interested in optimizing IRS2092 there's a great long thread with some useful insights towards the end here : $175 Class D amp--120 wpc
I used to own several of the IRS2092-based Classdaudio amps and they did sound terrific. I prefer the TPA overall (at least a well-designed one like the gmarsh wieners), but my memory of the IRS comparison was more about taste. The IRS yes had solid bass and a more rounded tubey presentation. But depending on the speakers you have to have a nice beefy power supply - I had 60k uf and even 90k at one point. Bigger the transformer the better too IMO. Then an enclosure... All that starts to add up.
Damn good amps though.
Damn good amps though.
From a single amp the B&W's are probably well above 8 ohm for outputfilter resonancefrequency. Sanwu has filter designed for 4 ohm load.
I used to own several of the IRS2092-based Classdaudio amps and they did sound terrific. I prefer the TPA overall (at least a well-designed one like the gmarsh wieners), but my memory of the IRS comparison was more about taste. The IRS yes had solid bass and a more rounded tubey presentation. But depending on the speakers you have to have a nice beefy power supply - I had 60k uf and even 90k at one point. Bigger the transformer the better too IMO. Then an enclosure... All that starts to add up.
Damn good amps though.
The thing got my attention the most when you said IRS2092 has more tubey sound? Can you open little more? I used to own Topping T-amp. Pretty good (tubey? don't know) but to my ears it was little dark (tubey again?). Then I got the TPA3116 and could not believe my ears because I paid $25 for it and it was VERY good. Then I got the TDA7498 even cheaper but did not like it as much as the TPA3116. The bass was less but the overall was not as musical as the TPA3116. I liked it better than the T amp though.
I am very interested in the IRS2092 now since you mentioned and had also the 3116.
Would you recommend a cheaper IRS2092 option instead ClassD Audio?
Thanks for all the feedback. Seems like a hot topic. I read a couple of people now that prefer the TPA's so I decided to take one last stab at the 3116. Should collectively have 160watts more to offer than the 3118's. Bought 4 red boards in BTL with balanced input. My DAC has balanced out so thats nice. https://m.aliexpress.com/item/32699597173.html
Have been looking into the IRS2092. There are the infamous $20 board. Still need a BIG transformer and rectifiers. Or a dual rail switchmode. It will end up being a costly upgrade because of the big difference in power supply from the TPA's. Also saw some IRS2092 boards with integrated rectifier circuit. These seem interesting and easy to modify.
Have been looking into the IRS2092. There are the infamous $20 board. Still need a BIG transformer and rectifiers. Or a dual rail switchmode. It will end up being a costly upgrade because of the big difference in power supply from the TPA's. Also saw some IRS2092 boards with integrated rectifier circuit. These seem interesting and easy to modify.
The thing got my attention the most when you said IRS2092 has more tubey sound? Can you open little more? I used to own Topping T-amp. Pretty good (tubey? don't know) but to my ears it was little dark (tubey again?). Then I got the TPA3116 and could not believe my ears because I paid $25 for it and it was VERY good. Then I got the TDA7498 even cheaper but did not like it as much as the TPA3116. The bass was less but the overall was not as musical as the TPA3116. I liked it better than the T amp though.
I am very interested in the IRS2092 now since you mentioned and had also the 3116.
Would you recommend a cheaper IRS2092 option instead ClassD Audio?
I only owned the Classdaudio amps, I don't know about the comparative quality of the ebay/Sure versions so can't speak for them. Don't want to get too verbose, but I guess I would describe the CDA amps as kind of 'warm', 'cohesive' presentation. The audiocircle link abraxalito posted above will take you to a lot more detailed impressions (including my own, some years ago).
One note of caution though is that the IRS amps were IMO *very* sensitive to input impedance as well as output matching to speakers. There were definitely a few folks who found their IRS amps shrill or lean in their setup, totally the opposite from most reports. Particularly people who had speakers that needed a lot of power. At the time for instance I had a Salas DCB1 with 1k output impedance and there was definitely a warm rounded tube presentation, then I reduced to 220 or so and it changed to wider, leaner, but still pleasant. I played around alot with values and the CDA amp was very sensitive. The more expensive SDS amps have some special kind of buffer that helps somewhat. I don't think the ebay/Sure versions do.
The TPA is also sensitive, but just when it comes to having the right output filter for the speakers.
That's my .02.
FWIW I replaced a Class D Audio SDS-470 amp with a pair of DUG's tpa3116 mono boards. This is for my living room, which isn't for critical listening. But I didn't hear any obvious downgrade.
There was nothing wrong with the SDS-470, except that I now realize it was grossly overpowered for my needs... IIRC it was capable of something like 600 WPC at 4R which is ridiculous (for me anyway). So even idle it put off more heat that I wanted (still class D though, so way better in heat/power usage compared to class AB). The tpa3116 chips are capable of about 1/10 the power, and I still have plenty of headroom. In other words, my priority in this particular application is mostly heat/power.
I have been thinking about IR2092 again, particularly ChocoHolic's SystemD LiteAmp. This design includes post-filter feedback and also an input circuit to help out with the chip's input sensitivity. Too many projects going on now, though; it will have to stay in my "wishlist" for the time being.
Also, tpa3116 vs 3118 is really a non-issue. The circuit is the same, the only difference is the location of the thermal pad (on top for 3116, bottom for 3118). Given the same implementation of 3116 or 3118, I would expect them to sound and measure identically.
In general I like modest listening levels, so I've yet to encounter a situation where the tpa311x didn't cut it. But if I needed more power, I'd definitely look at IRS2092, SystemD LiteAmp in particular.
There was nothing wrong with the SDS-470, except that I now realize it was grossly overpowered for my needs... IIRC it was capable of something like 600 WPC at 4R which is ridiculous (for me anyway). So even idle it put off more heat that I wanted (still class D though, so way better in heat/power usage compared to class AB). The tpa3116 chips are capable of about 1/10 the power, and I still have plenty of headroom. In other words, my priority in this particular application is mostly heat/power.
I have been thinking about IR2092 again, particularly ChocoHolic's SystemD LiteAmp. This design includes post-filter feedback and also an input circuit to help out with the chip's input sensitivity. Too many projects going on now, though; it will have to stay in my "wishlist" for the time being.
Also, tpa3116 vs 3118 is really a non-issue. The circuit is the same, the only difference is the location of the thermal pad (on top for 3116, bottom for 3118). Given the same implementation of 3116 or 3118, I would expect them to sound and measure identically.
In general I like modest listening levels, so I've yet to encounter a situation where the tpa311x didn't cut it. But if I needed more power, I'd definitely look at IRS2092, SystemD LiteAmp in particular.
Maybe off topic... I used to use TPA3116/18 for years and loved it. But if you have speakers that need more than say 25w or 30w, they just don't cut it and the bass is always not going to be as good as the IRS2092. But if going with a big dual PSU needed for the 2092, why not just go with a real class AB discrete MOSFET amp? After 13 class AB amps built (and a few class A thrown in), I am convinced that the sound is superior to class D. Here is what I am using now - a CFH9 - clean, dynamic, able to drive square square waves, probably good for 150w/ch - very nice sounding. If you can solder, the parts are about same in cost as one of these $20 IRS2092, except for the big heatsink.
http://www.diyaudio.com/forums/solid-state/294834-cfh7-amp-46.html
Easy to make and maybe $10 in parts:
http://www.diyaudio.com/forums/solid-state/294834-cfh7-amp-46.html

Easy to make and maybe $10 in parts:

Last edited:
Main reason I'm playing around with class D is the effeciency. Or else I would just have kept my heavy AF, frying hot, class AB amp 🙂 I'm really not a bass head. Maybe IRS2092 might be too much? I just got a BMC-2 DAC. Sounds awesome, but I sometimes feel the 3118 needs a loudness button. Now I tried ordering those cheap red 3116 mono boards with balanced input. They look nice. Will lower gain and change output filtering on them and have a listen. Might also try another DAC. My Aune had less detail but sounded somewhat more fun.
I just got this DAC - sounds great and has native high speed rates with USB. XMOS U8 and Burr Brown PCM5102A.
http://www.aliexpress.com/item/32409065830/32409065830.html
http://www.aliexpress.com/item/32409065830/32409065830.html
Last week I made a small modification on my red TPA3116 board, I remove the capacitor near TPA3116 IC and replace with the bigger one. IMHO the lower frequency now sound better. It change the mid sound also. For those who interested :
http://www.diyaudio.com/forums/class-d/237086-tpa3116d2-amp-post4884272.html#post4884272
http://www.diyaudio.com/forums/class-d/237086-tpa3116d2-amp-post4884272.html#post4884272
- Status
- Not open for further replies.
- Home
- Amplifiers
- Class D
- TPA3118 vs TPA3116 vs IRS2092