TA202X vs TDA8920BTH

[Ronj]

Hello,
I am a new member here, and very new to the world of Class D amps too. I have a few TA2020 and TA2024 based amplifiers with me, mostly the ones I assembled from finished PCBs I bought. I also have a Philips MCD908/12, which uses TDA8920chips and a tubed buffer amp stage.
Strangely I feel that the soundstage of my MCD908 is not as wide and deep as the Tripath amps. Is it just my perception or any scientific fact behind it? From the specs, I can see MCD908 is pretty poor. As the spec says the amp section is only good for 100HZ – 20Khz and 10%THD at the rated power. But what about at the normal listening volumes? Or say at 6-7ws? Is the TDA8920 chip is the culprit for this mediocre performance or other electronics such as the 12AX7 Tube buffer stage?
I am interested to know other than in power, how TDA8920chips compare against TA2020 or say TA2022 (which has almost the same amount of RMS out when compared to TDA8920)?
I wanted to assemble a standalone TDA8920BTH based amp may be by buying from one those guys from China who sell the PCBs. Has anyone here compared these chips? And what the experts have to say about this?
Thanks
Ron
P.S here are the datasheets for these chips

http://www.nxp.com/documents/data_sheet/TDA8920B.pdf
http://g.dubuc.free.fr/pdf/TA2022.pdf
http://www.e-ele.net/DataSheet/TA2020.pdf

[Ronj]

It seems like I have asked something really sacrilegious; looking at the number of replies!

[sangram]

I have the 8920BTH boards from Connex, and I have to say I'm ashamed because I have not yet tried them out in normal mode, one board failed while running and neither worked without serious whine in BTL mode.

Maybe I'll spin the surviving board in stereo mode and give you some fodder, but don't expect a direct comparison. I've never heard the 2022, and probably will never do so.

[sangram]

Right. Comparison is made with a PA100 based on LM3886, speakers used are Dynaudio BM6 studio monitors. PA100 amp is built P2P, with Holco resistor in feedback path but no exotic components anywhere.

The 8920BTH sounds muddy, cloudy and with no soundstage at all. The sound is, to put it mildly, lumpy. These may work for non-critical applications in small subwoofers or multimedia systems for PCs etc, but not for any kind of critical audio use. They were a waste of money for me - $80 for two modules was $80 too much. I'm heartbroken

[Ronj]

Thanks for that info pal....

What was the +- rail voltage you were using? And what's the value of rail caps on these boards?

By the way Philips make MCD908 and another MCD series DVD component system with these chips... And they sell it for SGD1000 here with 3way bookshelf speakers...

I have seen their(Connex) boards... seems like they use Micrometals cores?

Thanks
Ron

[sangram]

The Connex boards take a transformer - I used a 36VCT transformer, should be around 24V rails. The rail caps are 10,000uf each, up from the stock 6800uF.

I would also direct some of the issues towards the fact that it hasn't run in at all, though I believe an amp reveals most of its tonality in the first few hours of running. This one was pathetic, I hope the sound improves with some running in but I wouldn't be sure.

I think Cristi winds the inductors himself - not too familiar with cores. I got the TDA boards because I thought they would be a nice summer system - low heat and reaosnable sound. They are low heat, but the sound is terrible. Just as well the other board blew while testing.

Just to be clear - I don't think Cristi had anything to do with the sound itself. It's a pretty well-built board, there are proper MKP caps on the input, and the rest of the components seem pretty high quality. I think the limitations can be attributed to the chip's performance limits.

[Ronj]

Thanks a lot.... That's some useful info...
I wonder what other members have to say about this... I have seen some posts here saying TDA8920B does sound good.
Ron

[Cristi]

for a successfull amplifier is not enough just the way how are made the components, and modules, it also count the way how the final setup is implemented, the interconnection between them.

Originally Posted by sangram

one board failed while running and neither worked without serious whine in BTL mode.

Originally Posted by sangram

I would also direct some of the issues towards the fact that it hasn't run in at all

well, if i remember clearly (because if not, i can read one of the 44 emails which we exchanged trying to help you solve the problem) both boards were working at the beggining. then on one board the IC suddenly died. i offered to send you a replacement ic, but you told me that the board was completely damaged because of too much heat during desoldering. then i sent you a board instead, which you told me after a while that you haven't receive-it......

Originally Posted by sangram

I think Cristi winds the inductors himself - not too familiar with cores.

well, again, there are few types of inductors which can be used. i just chose one of the best. the alternatives are general purpose inductors, with saturation current of 3-4A, high DC and AC resistance, interwinding capacitance radiating a looooot of EMI everywhere and with huge core losses above 150KHz, like most of the other boards with TDA892x are using. (a closer look at some photos can see that) moreover, the Micrometals cores cost about 3-4 times more than the other inductors. if one prefer those instead of the Micrometals ones, no problem, i can use them instead.
the T68-2 core is made from a material which have low losses up to few MHz.

[steph_tsf]

MAX9744 : Modulo D - ELEKTOR.com | Electronics: Microcontrollers Embedded Audio Digital Analogue Test Measurement
TA3020 : ClariTy 2x300W - ELEKTOR.com | Electronics: Microcontrollers Embedded Audio Digital Analogue Test Measurement

[sangram]

Cristi - it wasn't an issue with you or your work, that was pretty clear. I did appreciate your help even then and still do - so there's nothing negative here that is directed at you

It was just a comment on my experience, and you'll notice I did say the module failed while running so it was operational when I got it. The remaining one does work in stereo mode, unfortunately I did not manage to appreciate the performance quite so much. And I think it is the chip itself to be blamed rather than your implementation.

I have absolutely no clue which cores perform well and which don't, so the Micrometals comment was above my head. Thanks for the clarity - I am sure you use the good stuff!

btw, I still haven't received the board you sent me, I'm sure it got appropriated by someone in transit. I am still waiting for an opportune moment to place my next order and then we can work out how to get the second board up and running.

@RonJ: Cristi offers multiple chips, and my boards are indeed the 8920B version. I got them because my speakers are proper 8 ohm impedance with no awkward dips, so the 8920B should have been able to cope with them just fine at the voltages I was running them at, even in bridged mode. It was a simple desoldering job but I botched up the faulty board - Cristi's support was excellent in this regard. He offered to send me one more board after the replacement went missing. I asked him to put that on hold knowing our postal system, hoping to club the recovery board with a subsequent Class AB amp.