TPA3255 4-channel module review - $12.50 per channel but is it any good?

[dukanvadet]

I was thinking of getting 2 3e boards but this seems just as good for my needs (and cheaper). If i swapped the caps for some 63v 105c it should be no problem using a 48v SMPS i assume? I allready have the power supply so it would be nice to be able to use it. There will be fan cooling either way.

 

[reVintage]

We see a lot of 2-channel TPA3255 based modules, but 4 channel ones are rarer. I bought this one off AliExpress for $50 including shipping - that's $12.50 per channel!

Build quality is OK but not great. The power and speaker I/O is all screw terminals, where better modules use spades. The capacitors are rated at 50V which is a bit tight - but then again the vendor does suggest a 36V supply. We'll see later that this is sensible as raising the power supply voltage to 42V doesn't do much to help performance. The heatsink is fine. There are no pins to expose the TPA3255 control and status signals, which is a bit annoying if trying to build a more sophisticated amp with soft-start, or clipping detection.

Does it do what a TPA3255 should do in terms of power and distortion? Yes it does . It's wonderful to be able to get this level of performance so cheaply. Full details below...

The test rig is a Topping D10 Balanced, a 150W/4.7ohm dummy load, a DIY AES17-25 filter (https://github.com/mattclarkdotnet/aes17_filter_25), and a Cosmos ADC. Power was from my bench DC supply - I have some cheap AliExpress supplies to test later.
View attachment 322645

The TPA3255 Single Ended THD+N should be this (from the TI data sheet), so about 0.01% THD+N at 1W, and 60-70W available before distortion takes off:
Starting with frequency response it's as expected, just 1dB down at 20KHz

At 36V power performance is as expected, topping out at around 70W, with 0.01% THD+N at 1W
EDIT: my bench supply was the limiting factor, it actually does more like 85W at 0.1% THD:

Increasing the power supply to 42V gives pretty much the same result: EDIT - this is a limitation of my bench supply

The 1KHz FFT is very clean:
View attachment 322653

Hi,
Great with someone who has a technical approach here. Did you do any FFT´s at 1W too? Will go SE for my midbass 60-350Hz with just below 100dB sensitivity.
Cheers
Lars

 

[mcdn]

Hi,
Great with someone who has a technical approach here. Did you do any FFT´s at 1W too? Will go SE for my midbass 60-350Hz with just below 100dB sensitivity.
Cheers
Lars

Hi! No I don't think I did. This module is actually BTL with 2 TPA3255 chips to give 4 channels, as corrected in the original post. The 1W performance is entirely noise dominated in the range you are interested in, at about -70dB: https://www.ti.com/document-viewer/TPA3255/datasheet#specifications/SLASE403745

With speakers that sensitive you really need to look for amps with low noise, not high power.

 

[reVintage]

Thanks for pointing out the noise problem! Your FFT indicates the noise floor being in the ballpark of -120dB rel. 5W 4ohm(2,5W rel. 8ohm) and s/n 89,6dB. 1W 8ohm is 4dB down from 5W 4ohm. This makes -89.6-4=-85.4 at 1kHz and adjusted to 60-350 ca 6dB more, ie -80dB.
I use DHTtube/hybrid amps and discrete built up FET L-R filters for the 350-20kHz horn system. My idea is to do a temporary low budget system for lower registers using one stereo Fosi V3 per channel driving the subs BTL with DSP and rebuilding one channel for SE driving the midbass. The output capacitor will be a part of a cascaded 12dB L-R filter with the first part of the cascade before the power amp. The output capacitor will also protect the vintage Altec 414-8A drivers from DC. The Hornresp sims of the midbass system indicates close to 100dB but I haven´t measured. The subs are low sensitivity that really needs the BTL. Will anyway try it, to see if I can bare the noise.
The actual reason I asked was from pure curiosity as I don´t favor dominating odd harmonics, which in the midbass wouldn´t be a problem at all .

 

[mcdn]

Your FFT indicates the noise floor being in the ballpark of -120dB rel. 5W 4ohm(2,5W rel. 8ohm) and s/n 89,6dB. 1W 8ohm is 4dB down from 5W 4ohm. This makes -89.6-4=-85.4 at 1kHz and adjusted to 60-350 ca 6dB more, ie -80dB.

Maybe kind of? There's a reason we use THD+N or SINAD rather than eyeballing a noise floor. Noise floors are artificially lowered though FFT sampling. Use the datasheet as your guide. For subs the TPA3255 is a great solution, but for the high sensitivity parts be careful. Sounds like you know what you are doing though, good luck!

 

[reVintage]

This is the circuit I mentioned: