Class T Tripath

I have just recently discovered Tripath and the class T technology. I am thinking about building a big amp from one of their devices.

How is the sound quality? If you put a scope on the output do you see alot of HF noise? Does it emmit RFI/EMI. Does it pick up alot of RFI/EMI?

Does anyone out there have scope pictures of the output signal reponse with sine and square waves?


-Eli Hughes
 

leroy

Member
2001-10-31 10:52 pm
Digital amplifiers do have excellent measurement values. I know Sharp introduced a SACD-player and digital amplifier about a year ago (I assume it to be class T) and they perform very well.
RFI/EMI is rejected by low pass filtering of the output signal and carefully shielding the amplifier/power supply part.
 
I have built a custom Class-T amplifier which used the TA0102A driver which has now been discontinued. I have double sided circuit boards with silkscreen, soldermask, and plated through holes which I may or may not be able to sell. The sound is very good and is better than most amplifiers in consumer products. There are past postings where we discussed this in more detail. there is a little tiny bit of high frequency noise which may ride on the audio wave, however, due to the processing on the chip aboulutely none of it is in the audio band. It is all out-of-band noise. When the amplifier "clips" the internal DSP has a value of constants which "overflow" and causes a high frequency ripple on the tops and bottoms of the sinewaves. The great thing is that, unlike its linear counterparts, you cannot really hear the clipping occur. That is why the chips have an "OVERLOADB" output pin on them so that you can control a visual reference (LED driver circuitry) to indicate the listener of this occurance. Overall the Tripath products (when designed and implemented correctly offer more than exceptional sound quality and unbelievably high efficiency). The particular unit which I have built when tweaked correctly can deliver 190W x 4 into 4 ohms and the tiny heaksinks are no more than room temperature to body temperature at full output power. Also the great thing is that the losses are nearly constant over the entire load range meaning that the higher the output power the losses stay pretty much constant and this means a higher efficiency and a constant temperature. I could talk about the "T" forever if you want. Anybody can email me with questions....

This is a link which I started a few months back on the topic and actually had some of the Tripath Applications Engineers get in on the discussion. Here is the link. For others do a search on our forum page for "Tripath" or "Class-T".

http://www.diyaudio.com/forums/showthread.php?s=&threadid=3906&highlight=Tripath
 
The Class T amps are basically third-order delta-sigma modulators followed by a switching power stage.
In this respect they are similar to the Sharp 1-bit amplifier.
The main differences are the following: Sharp uses a 7th order d-s modulator with the same sample rate as DSD uses (twice that much for the newer generation amplifiers) which is not very easy to design in a stable fashion but gives a high S/N ratio for a given maximum switching frequency (0.5 times the sampling frequency).
Tripath on the other hand uses a very high sampling rate combined with a third order only noise-shaping loop and holds the maximum switching frequency down by a pulse qualification circuit that follows the modulator. In this respect it is quite clever.

Although I could imagine better working principles, the Tripath modules have some advantages. For their price they include quite a lot of functionality and you can get complete evaluation boards and suggestions for board layout etc.
Just remember that any switching amplifier is an RF circuit used to process audio. If you don't have sufficient experience in RF and EMI design you wouldn't get something like that running properly.

Enclosed is the result of the simulation of a third order s-d amplifier model. It's maximum output voltage is just 1 Volt but that doesn't matter for seeing how the priciple works. Shown are the switching waveform at the output of the switching stage and the voltage acros the load. The input frequency is 10 kHz and the amplitude 10% below clipping. I chose such a high input frequency for better visibility.
You can see how the switching frequency decreases with increasing amplitude, that's why there is some ripple on the sinusoid's maximum amplitude (simply because the output filter's efficiency is worse at lower switching frequencies). You can also see how the signal across the load is lagging behind the switching pattern simply because of the output filter's phaseshift.
There is also an FFT of the aforementioned signals enclosed, showing the working principle of the noise shaping. K3 of this circuit is 0.3% approx but would be better for lower input frequencies. One property of noise-shaping delta-sigma modulators is the increased signal quality for lower input frequencies. I think the constant THD figures statted for the class-T amplifiers over the entire audio range is mainly due to the additional filter that is used during THD measurements.

I was able to audition one of the Sharp amplifiers and it was indeed sounding good, but at the same price conventional amplifiers sounds as good IMHO. There is an amp manufactured by Belcanto (Evo) using Tripath's technology, that received good reviews but I wasn't able to hear it so far. But for an almost empty box it is quite expensive at approx 4000 Euros. Also an amplifier manufactured by british ESLAB (also using Tripath technology) got very good reviews apart from the S/N ratio which was 17 dB worse than the manufacturer's statement. It's price is in the same range as the Belcanto

Regards

Charles
 

Attachments

  • sd_amp.zip
    19.5 KB · Views: 155
When I reread my last posting I found the following sentence to be a little misleading:

If you don't have sufficient experience in RF and EMI design you wouldn't get something like that running properly.

I didn't want to discourage anybody from building any type of switching amp of course.
What I wanted to say is that if you don't have any experience in RF and EMC design then go for a worked out example (a Tripath evaluation board or their PCB design for example) it would save you a lot of headache.

Regards

Charles
 
"Class-T" is a registered trade mark / brand name belonging to Tripath Technologies Inc. in Santa Clara, CA. It is genericly a digital amplifier which uses a non-constant switching frequency by use of DSP techniques. There are other companies who have similar technology. The old 38-pin quads were about $21US apiece for the TA0102A and a few bucks more for the the higher models. Each module is two channels. The new and improved TA3020 chip is a 48pin DIP with inproved performance and functionallity.

BeanZ