BerryBak BVS2300 2-channel TPA3255 power amp module

[pma]

This is a detailed description and measurement of BerryBak BVS2300 amplifier module, purchased from Soundimports at € 49,95 (VAT included) + € 10,95 shipment cost, total € 60,90.

The package contains these items:



i.e. the amplifier module + complete necessary wire set.

Built quality is good and the PCB board is well described with connector pin names.
The amplifier works either with analog line input or Bluetooth wireless. There is a switch on-board to change the input source.



After doing the preliminary check of module function



I have put it into the metal case with two MeanWell RS-100-24 power supplies connected in series (this is what I have had in stock).

https://www.czech-meanwell.cz/fotky10575/fotov/RS-100-spec.pdf

As the amp module is supposed to be supplied from DC PSU of 24V – 48V range, so I have set the PSU modules to give 47Vdc total.

Specifications per module datasheet

The datasheet of the amplifier module specifies this:

Let's what I am going to measure.

Measurements - general

All measurements are done with 4.7ohm/200W and 8ohm/200W load resistors, because I do not have a 4ohm/200W resistor at the moment. Distortion is measured with 96kHz sampling frequency and 20Hz – 20kHz bandwidth (unless otherwise noted). Topping DX5 and D10s are used as generators and E1DA Cosmos (with appropriate input dividers) as measuring ADC, with REW software. BREVE FPS 250 isolation transformer is used when needed. Distortion vs. power is measured with one channel driven only. Distortion vs. power stepped sine measurement lasts 4 minutes and is continuous without cooling intervals and it covers 40dB amplitude range approx.

1. Input impedance

About 10kohm

2. Frequency response and output impedance

Measured at 4.5V output voltage into 4.7 ohm and 8 ohm load

The response depends on load impedance quite strongly above 5kHz. Output impedance is inductive (20uH).

Output impedance measurement:

3. SINAD at 5W/4.7ohm

SINAD is 74.9dB. THD+N is dominated by noise and spurious frequencies, which cannot be avoided by wiring or grounding scheme. THD only (without N) is low at about 0.003%.

4. Output noise voltage

722.5uV (20Hz-20kHz), 439.8uV A-weighted. Numbers are higher than declared by the manufacturer.




5. THD and THD+N vs. power at 1kHz

Power for THD+N 1kHz/4.7ohm/0.01% is 143W, 180W for 0.1% and 201W for 1%.
Power for THD+N 1kHz/8ohm/0.01% is 89W, 112W for 0.1% and 127W for 1%.

Output power is reasonably high and the module did not go into protection mode during testing, even if the power sweeps were repeated many times. Maximum heatsink temperature was up to 60°C.

6. THD+N vs. frequency at 16V output voltage

We can see ugly rise of distortion below 200Hz. This is because the manufacturer has decided to use very non-linear high K ceramic capacitors in the signal path, for cost and size reasons. The plot shown is typical for these parts. Distortion rise above 2kHz is typical for TPA3255 non-PFFB designs.

Note: left channel has had lower noise than the right channel. Lch load was 4R7, Rch load was 8R. That's why we can see lower THD+N in the midband with 4R7.

7. CCIF IMD 19+20kHz at 25W

This is not bad for the TPA3255 amplifier and the distortion keeps almost same up to maximum power.

8. RMAA test report

measured close to maximum unclipped power into 4R7 may be found below as a pdf file.

Conclusion

The amplifier did not meet manufacturer's specifications regarding output noise, input impedance and THD+N at 5W. However, it has reasonably high power and quite good thermal management (better than AIYIMA A07 and much better than Topping B100 that I tested as well). It seems it could be quite reliable.
Taking into account its very low price, it could be an option for a DIYer who seeks for cheap amplifier with enough power. It is not the amp for those who run for SOTA paper parameters, in such case try something like Sylph Audio, but at 3x higher price.
The manual that can be downloaded from supplier's web is instructive and describes all the board connectors that have to be used when building a complete amplifier. Also a plus point is that all the necessary cables assembled with connectors are a part of delivery.

Edit: Bluetooth measurements are added in this post.

 

[DonH56]

Nice work! A few questions:

1. Any concerns about the big electrolytic cap wedged between the coils and next to the heat sinks (will it get hot)? Is it another decoupling cap or coupling cap?
2. Are the nonharmonic spurs in the distortion and noise plots related to the power supplies, amp module, or both? Or something else?
3. Why the steps in the high-frequency distortion? (Is this normal for this amp? Function of test system attenuators? I have not been following.)
4. IMD seems high relative to the distortion curves, but I probably lost track of where all the measurements are on the curves.
5. Thoughts on replacing the ceramic caps with film to see if the low-frequency distortion improves? I noticed the noise floor also rises at low frequency.

I agree with you, looks like a good low-cost amp for DIY projects!

 

[Redacted]

Would also make an excellent amp for a FRFR guitar rig (I'm not being sarcastic).

 

[pma]

Nice work! A few questions:

1. Any concerns about the big electrolytic cap wedged between the coils and next to the heat sinks (will it get hot)? Is it another decoupling cap or coupling cap?
2. Are the nonharmonic spurs in the distortion and noise plots related to the power supplies, amp module, or both? Or something else?
3. Why the steps in the high-frequency distortion? (Is this normal for this amp? Function of test system attenuators? I have not been following.)
4. IMD seems high relative to the distortion curves, but I probably lost track of where all the measurements are on the curves.
5. Thoughts on replacing the ceramic caps with film to see if the low-frequency distortion improves? I noticed the noise floor also rises at low frequency.

ad 1) I do not know, almost every TPA3255 module manufacturer places the caps this way. They are 105°C capacitors and I have measured maximum of 60°C on the heatsink surface. They are additional rail decoupling caps.

ad 2) I do not know. Changing wires, grounding and input shorted only without signal source makes no difference. The same PSU with AIYIMA A07 results in clean noise floor.

ad 3) It is a THD+N measurement with 20kHz BW. Thus you can see only H3 of 6.67kHz basic frequency and H2 of 10kHz and thus the steps. Everything above 10kHz is only noise. Please see the same behaviour in 3255 datasheet.

ad 4) I do not understand what you mean.
Ah, I see. Should be as in the plot below:


ad 5) Yes, I am sure. But there is no space on the board, and the budget module is not worth such effort, to me.

 

[pma]

Would also make an excellent amp for a FRFR guitar rig (I'm not being sarcastic).

Yes. And for sub as well. The LF distortion (0.3% at 20Hz) would be in any case inaudible and masked by speaker. Compare to subwoofer amps tested recently by @amirm .

 

[amirm]

Nice work. Promoted to home page.

 

[pma]

Nice work. Promoted to home page.

Thank you!

--------
BTW, Input impedance measurement added to post #1.
Rigorous measurement of output impedance is difficult with this amp, as there is a lot of HF rubbish at the output that interferes with the extremely low signal. Adding a filter is not possible as it changes the output impedance. There are two 10uH output coils per channel but no usual 1uF MKT foil capacitors on the module.

Edit: I found the way how to measure output impedance even in such conditions. Measurement added to post #1.

 

[MCH]

Thanks a lot for the review. Doesn't it have DSP? I thought that would be the advantage of these boards. Otherwise, for 60 eur or even less you can get a finished A07 max or a douk A5 finished amp.

 

[Power Pop 23]

Nice work. Promoted to home page.

I don't yet see this review promoted to the home page.

 

[GGroch]

... Doesn't it have DSP? I thought that would be the advantage of these boards. Otherwise, for 60 eur you can get a finished A07 max or a douk A5 finished amp. Often even less.

Berrybak/Sure/Wondom make several TPA3255 models based on this board. In the US, I see the BDM-8 with Bluetooth and DSP/PC interface for $39 on Amazon. Also the BDM-8a, with no Bluetooth or DSP for $21 (after 40% coupon on page). There may well be other differences.

The Amazon offerings have a fan which I understand seldom fires up at all except under high load and may not be necessary in most applications.

Wondom is about to ship a model JAB6 dual TPA3255 PFFB board with docking/replaceable DSP and Bluetooth modules that looks interesting.

 

[pma]

Doesn't it have DSP?

Yes, as per the manual datasheet. I have not tried it yet, and I would need to build a USB to J302 cable. Which should be easy.

This cable is not included in the delivery.

 

[amirm]

I don't yet see this review promoted to the home page.

Ah, I did it on the phone and accidently also checked to not show up there. Duh! It is fixed now.

 

[DonH56]

ad 1) I do not know, almost every TPA3255 module manufacturer places the caps this way. They are 105°C capacitors and I have measured maximum of 60°C on the heatsink surface. They are additional rail decoupling caps.

ad 2) I do not know. Changing wires, grounding and input shorted only without signal source makes no difference. The same PSU with AIYIMA A07 results in clean noise floor.

ad 3) It is a THD+N measurement with 20kHz BW. Thus you can see only H3 of 6.67kHz basic frequency and H2 of 10kHz and thus the steps. Everything above 10kHz is only noise. Please see the same behaviour in 3255 datasheet.
View attachment 546928

ad 4) I do not understand what you mean.
Ah, I see. Should be as in the plot below:
View attachment 546930

ad 5) Yes, I am sure. But there is no space on the board, and the budget module is not worth such effort, to me.

Thanks Pavel!

I was unclear on the IMD plot comment, sorry. Yes, the frequency response went too high, but looking at it IMD amplitude also looks high relative to THD, maybe due to the rolloff. Ideally IMD would be about 9.54 dB above THD but the IMD looks like it would be more than that. Not worth the effort to discuss/test further.

I was surprised the coupling caps had such a profound impact on the distortion at low frequency but it is not something I have considered much. Ceramic coupling caps are common for high-speed data streams but those have much higher frequency content (10's ~ 1000's of MHz). I always used film caps for audio coupling. I believe you, you've much more experience with them at audio (and probably above), just caught me by surprise. Makes sense given their hysteresis and voltage nonlinearity after I did a little digging.

 

[Hiten]

Newbie post.
Dont know why proper PFFB is not implemented ? Just requires few resistors and capacitors, and some neasurements. Do designers lack implementation knowledge ?
If more power is not required even chip amps like LM3886 can outperform these i guess.

 

[john61ct]

Newbie post.
Dont know why proper PFFB is not implemented ? Just requires few resistors and capacitors, and some neasurements.

I think less than five in a hundred such amps even try

and maybe one or two out of those five do a good job at it.

So either A. it's a lot harder than you say B. most "designers" are not good at their job or C. their employer does not care

maybe because most of their target market does not care

I suspect it is not very important unless trying to get high SPL?

 

[sam_adams]

We can see ugly rise of distortion below 200Hz. This is because the manufacturer has decided to use very non-linear high K ceramic capacitors in the signal path, for cost and size reasons. The plot shown is typical for these parts.

Thanks, @pma, for calling this out. Viewers at home should make note of this for future reference.

See attachments for reference material.

 

[pma]

Thanks, @pma, for calling this out. Viewers at home should make note of this for future reference.

See attachments for reference material.

Yes, that's it. Thank you for mentioning these documents to public.

 

[pma]

I was surprised the coupling caps had such a profound impact on the distortion at low frequency but it is not something I have considered much.

Hello Don and thanks for your comments. Some time ago I posted here LF distortion from various types of capacitors, including high K ceramics. And also a review of PRE-TC10, which is a nice product, unfortunately degraded by use of ceramics in the EQ circuit, as also described in the publications posted by @sam_adams .

 

[daniboun]

Many thanks for the measurements
Well done )
Another non PFFB TPA3255...

 

[pma]

Newbie post.
Dont know why proper PFFB is not implemented ? Just requires few resistors and capacitors, and some neasurements. Do designers lack implementation knowledge ?
If more power is not required even chip amps like LM3886 can outperform these i guess.

This question is to be addressed to the designer of the module.

My thoughts:

• older implementations of TPA3255 (and this is probably one of them, though the manual is dated year 2023) did not use PFFB and there is still a plenty of modules without PFFB on the market,
• design cost reason, just take the TI EVM board design (published July 2016), and you are done, without further R&D costs,
• PFFB design with TPA3255 is not without possible stability issues (depending on complex load impedance) and again you need time to tune it.

All and all, successful PFFB design (if you do not do a mere unscrupulous copy of another existing design) reflects in higher price (see here) and probably is not requested for a mass budget product.