BerryBak BVS2300 2-channel TPA3255 power amp module

[eric00886]

I suppose, that's kind of what DSP is.
There is a substantial thread on the TPA3255 version here for those who have not seen it.

The supplied ACPWorkbench PC software allows for stereo parametric EQ, plus adjustments for dither noise, noise cancelling etc. Owned two and was able to set them up as active crossovers for a stereo pair of speakers. They are limited as the DSP processing ( ADAU1701) is not sophisticated.

Like most Wondom stuff, they were not not designed for optimal measurements. I could hear hiss 6" from the speaker. But still they sounded great to me in normal use. Not distinguishable from other TPA3255 amps. Except of course that with REW or other Room EQ software you can better match your speakers to the room. For $50 (plus the cost of a power supply) it gives a hobbyist who does not want to solder a lot of tweaking fun.

On the other hand, my ears are old and I cannot hear the difference between High End DACs fed through a comparator box

Ah, I had no idea this thread existed—awesome! Heading over to check it out and learn a few things!

 

[pma]

This should be on the board:

BTW, Bluetooth connection with my iPhone works as it should, so it is another nice option. I may make some measurements via Bluetooth later.

 

[pma]

So, below are measurements via Bluetooth wireless. 5W/4R7 THD/THD+N (SINAD) and also white noise (44.1kHz data) response.

iPhone XR has been used as a data source.

 

[DonH56]

Significantly worse than the previous direct measurements, yes?

 

[pma]

Significantly worse than the previous direct measurements, yes?

Yes, significantly worse, and it has been expected taking into account Bluetooth compression and the BP1048 chip in the signal path. It seems to be acceptable for everyday background listening but not for a concentrated or critical listening. This is my subjective impression so far.

 

[pma]

More measurements in Bluetooth mode :

1. 5W/4R7/1kHz


2. 89W/4R7/1kHz

There is almost no difference between distortion and SNR at 5W and 89W.

3. Julian Dunn jitter test

Very poor result. More on jitter methodology below:

https://www.collinsaudio.com/Prosound_Workshop/Julian%20Dunn_Jitter.pdf

4. DIM test (sine 15kHz + square 3.18kHz)

Better than I have expected, except for the noise floor modulation.

More on the system noise and amp noise limits

5. System noise, amp off, complete measuring system with input dividers, 0dBFS = 43.368Vrms

S/N of the system is 122dB.

6. Same as (5), but amp is on

 

[ElJaimito]

For almost the same price I would get this Tpa3255 PFFB module full balanced

New Dual TPA3255 Module + PFFB / Fully differential / OP amps DIP8 @ Great price

Hi amigos, As usual, I'm sharing my little discoveries from the web. I made a good find, so I'm sharing it. I found a TPA3255 dual chip module with PFFB and DIP8 op amps at a pretty good price! SPEC : Dual Chip TPA3255 (PBTL) Version A : Fully Balanced Version B : SE inputs Power : 150W / 8R...

www.audiosciencereview.com

See attached file for measurements

Page has gone.

 

[pma]

Significantly worse than the previous direct measurements, yes?

One more direct comparison:

BT is V5.0 with the chip used. I suspect that what we see is an AAC compression used in the iPhone (data source).

 

[pma]

Talking about lossy compression codecs, the paper linked below seems interesting to me:

https://arxiv.org/pdf/2511.11527v1

Maybe @j_j would kindly add some comment?

/I think I should note that I got very different results when I have sent files through aac codec from iPhone than are the results in the paper/

 

[j_j]

Talking about lossy compression codecs, the paper linked below seems interesting to me:

https://arxiv.org/pdf/2511.11527v1

Maybe @j_j would kindly add some comment?

/I think I should note that I got very different results when I have sent files through aac codec from iPhone than are the results in the paper/

I've no use for anything done with PEAQ of any variety. Listening tests themselves are the only real guide. What's more, the various "level xx" settings mean nothing in particular to me, and the elimination of HF at very low levels is not necessarily audible unless you crank the gain to 11. The statements about "richness" and such in the paper need some evidence to back up the claims. I didn't read the whole thing, after a bunch of rather revealing comments that made it pretty obvious that perhaps, just perhaps, people were listening with their eyes, I remembered I'm retired and I don't have to deal with that stuff any longer.

PEAQ makes it easy to tune a codec to the PEAQ perceptual model, which is not a listening test, or anything of the sort.

 

[pma]

I didn't read the whole thing, after a bunch of rather revealing comments that made it pretty obvious that perhaps, just perhaps, people were listening with their eyes, I remembered I'm retired and I don't have to deal with that stuff any longer.

One of the reasons of my question was that even the graphs they published (both spectra and spectrograms) are in controversy to my measurement. Figure 2 of the paper shows:

I compared the original file spectrum (another music sample) to the spectrum of the file recorded via Bluetooth and Apple aac codec and got very different result. The spectrum is cumulative with peak hold over 1 minute sample length:



and this is the spectrogram:



Of course I have the data recorded and tried an ABX test - with negative result.

Data are attached below, sample music is shorted to 15 s (so the cumulative spectrum would look different that what I posted - if someone tries to analyze).

 

[j_j]

I compared the original file spectrum (another music sample) to the spectrum of the file recorded via Bluetooth and Apple aac codec and got very different result. The spectrum is cumulative with peak hold over 1 minute sample length:

View attachment 548529

and this is the spectrogram:

View attachment 548532

Of course I have the data recorded and tried an ABX test - with negative result.

Data are attached below, sample music is shorted to 15 s.

Where was this paper published? What's the actual published reference?

There are any number of questions, and I'm a bit surprised that this would get into a technical journal as it stands, but then again, I've been surprised before.

 

[pma]

Where was this paper published? What's the actual published reference?

The only info I have comes from this web link:

Evaluation of Audio Compression Codecs

arxiv.org

... and it looks like a Thesis:

Evaluating Audio-Compression Techniques

Perceptual quality of audio is the aural accuracy and fidelity perceived by the listener. It is how humans respond to the accuracy, intelligibility, or fidelity of aural media. High-fidelity audio represents high accuracy and minimal distortion and is the standard against which compression...

research.ualr.edu

Duong, Thien Tuong, "Evaluating Audio-Compression Techniques" (2024). Theses and Dissertations. 1222

 

[john61ct]

One of the reasons of my question was that even the graphs they published (both spectra and spectrograms) are in controversy to my measurement.

Do you mean, in conflict with?

 

[pma]

Do you mean, in conflict with?

I meant “different”, based on looking at the images posted in the paper linked in post #51. But I admit that I have been confused by the lower quality of graphics in that paper.

 

[pma]

Bluetooth stepped sine measurements

I have persuaded my REW setup and MacBook Pro to work with BerryBak with MacBook as a sound source via Bluetooth. So I have been able to make stepped sine measurements. Below are plots for THD+N vs. power and THD vs. frequency. The result is better than I have expected. Some patience is needed due to Bluetooth intrinsic drifts.

 

[pma]

I was curious how the BVS2300 would work with Topping DX5 if the Topping is in the Bluetooth mode and the signal is sent wireless from my MacBook. Below is the result, and it is better (as expected) than if we use the on-board BVS2300 Bluetooth module. We get noise better of some 10 dB and also we can reach the full power of the amp module.

And still, even through DX5 in Bluetooth mode, we can see the difference in passband frequency response with different DAC filters.

 

[pma]

THD vs. output power compared to NC252MP :

 

[daniboun]

Hi team,

There is a new version from Wondom with a dual TPA3255 with PFFB + DSP : WONDOM AA-JA33288 JAB6
This module is design with a modular approach )


SPEC :

 

[daniboun]

More measurements in Bluetooth mode :

1. 5W/4R7/1kHz
View attachment 547885

2. 89W/4R7/1kHz
View attachment 547886
There is almost no difference between distortion and SNR at 5W and 89W.

3. Julian Dunn jitter test
View attachment 547887
Very poor result. More on jitter methodology below:

https://www.collinsaudio.com/Prosound_Workshop/Julian%20Dunn_Jitter.pdf

4. DIM test (sine 15kHz + square 3.18kHz)
View attachment 547888
Better than I have expected, except for the noise floor modulation.

More on the system noise and amp noise limits

5. System noise, amp off, complete measuring system with input dividers, 0dBFS = 43.368Vrms
View attachment 547890
S/N of the system is 122dB.

6. Same as (5), but amp is on
View attachment 547893

Given that this module only supports SBC and AAC codecs, we couldn't expect great results here... with LDAC the situation would have been different, see Amir's tests on this point...