My experiences with chinese generic TPA3251 and TPA3255 amplifiers

[Deleted member 9685]

I bought them with two 32V 5A adapters from aliexpress.com because two is one, and one is none. A chinese 32V 5A adapter costs about 17~18 USD. The total cost was about 120 USD.
Either amplifier works perfectly with a 32V 5A adapter.

• Chinese generic TPA3251 amplifier // ~39 USD
  • It has a cheap potentiometer that exhibits channel imbalance below 8'o clock.
    • You just need to remember to keep the volume above 8'o clock. If you need quieter volume levels, decrease DAC volume or software volume.
  • Other than channel imbalance at low volume levels, it doesn't have much white noise.
  • White noise is not audible beyond 7cm with environmental noise.
  • Because the main capacitor is a 35V one, it may not last very long.
• Chinese generic TPA3255 amplifier // ~45 USD
  • It has a better potentiometer that exhibits channel imbalance at very low volume levels
  • It has substantially more white noise in the right side, but it becomes virtually inaudible at about 15cm away with ambient noise.
  • When I disassembled the amplifier, I accidentally severed the connection between bluetooth antenna and bluetooth chipset which resides on the right side of the amp.
    • Perhaps, that's why there's more white noise in the right side? The generic TPA3251 amp has no bluetooth functionality.
  • Although it has more white noise in the right side, because it has less channel imbalance, it is more useable than my TPA3251 amp.
  • White noise from either speaker is still inaudible at my listening position(~60cm away).
  • It has a 50V capacitor, so it may last longer than the generic TPA3251 amplifier.

Both sound okay with onboard soundcard unless the volume goes beyond 1'o clock where my computer's power supply noise starts becoming audible.
If you are low on money, both amplifiers are useable and serviceable until you gather enough money to upgrade.
For an upgrade, I'm thinking about allo.com's upcoming TPA3255 amplifier which is rumored to be released in the beginning of 2020.

 

[renaudrenaud]

Why do not use directly the EVM from Texas with the 3255? Is it a bad board relative to Chinese clones?

 

[Deleted member 9685]

Why do not use directly the EVM from Texas with the 3255? Is it a bad board relative to Chinese clones?

Because a board is not a finished product. I want a finished product.

 

[renaudrenaud]

Because a board is not a finished product. I want a finished product.

I understand perfectly. Anyway, if I am a bit perfid I can argue most of this kind of boards are to be finished by the buyer...

 

[bravomail]

You could have tried cheap IcePower products from Ali. They r worlds better than TPA3xxx .

 

[bigx5murf]

Regarding the white noise you hear. I know many of the Chinese TPA3116 amps come set with a gain of 36db, which is too high. If you're able to de-solder, and solder in tiny SMD resistors, you can often lower the gain to around 20-22db, which will eliminate the white noise completely, and usually gets rid of channel imbalance as well. I've done this with both the single and dual chip versions of the breeze audio tap3116.

According to this, generic TPA42255 gain is set at 28db
https://quantasylum.com/blogs/news/quick-look-a-no-name-tpa3255-design

 

[Deleted member 9685]

I know many of the Chinese TPA3116 amps come set with a gain of 36db, which is too high. If you're able to de-solder, and solder in tiny SMD resistors, you can often lower the gain to around 20-22db, which will eliminate the white noise completely, and usually gets rid of channel imbalance as well. I've done this with both the single and dual chip versions of the breeze audio tap3116.

Or, you can buy a pair of 6~12dB RCA attenuators.

 

[Lorenzo74]

You could have tried cheap IcePower products from Ali. They r worlds better than TPA3xxx .

do you have data that support this statement? are you referring to datasheet comparison only? many thanks

 

[ZetZet]

do you have data that support this statement? are you referring to datasheet comparison only? many thanks

amirm has tested one of them, idk about others, but it was pretty good. https://www.audiosciencereview.com/...power-125asx2-class-d-amplifier-review.10442/

 

[dwkdnvr]

amirm has tested one of them, idk about others, but it was pretty good. https://www.audiosciencereview.com/...power-125asx2-class-d-amplifier-review.10442/

Based on the AP results that 3E Audio posted on their site, their 3255 board performs right in line with the IcePower - not sure there is a lot of gap between them. Granted, those are manufacturer provided plots, but they seem credible based on the datasheets.

 

[Lorenzo74]

amirm has tested one of them, idk about others, but it was pretty good. https://www.audiosciencereview.com/...power-125asx2-class-d-amplifier-review.10442/

thank you. I'm pretty ok with ICE, it is a consolidated technology. I was more interested in objective measurements of TPA3255 implementations, TI EVM and Chinese ones.
best
L.

 

[somebodyelse]

https://quantasylum.com/blogs/news/ti-tpa3255-evm
https://quantasylum.com/blogs/news/quick-look-a-no-name-tpa3255-design

In summary the TI board hits datasheet specs. The no-name one doesn't, and uses a regulator that isn't rated for the supply voltage they usually claim when they're selling them. 3E have measurements matching the TI datasheet ones, but with AP plots that aren't lifted from the TI datasheet. I'd love to see them measured independently, but I don't think it's happened yet.

 

[Lorenzo74]

https://quantasylum.com/blogs/news/ti-tpa3255-evm
https://quantasylum.com/blogs/news/quick-look-a-no-name-tpa3255-design

In summary the TI board hits datasheet specs. The no-name one doesn't, and uses a regulator that isn't rated for the supply voltage they usually claim when they're selling them. 3E have measurements matching the TI datasheet ones, but with AP plots that aren't lifted from the TI datasheet. I'd love to see them measured independently, but I don't think it's happened yet.

thank you, I actually remembered that report made more than 2 years ago by quantasylum.
all in all it returns THD+N of 0.034 @192W (not bad for 50€)
my speaker couldn't even dream such low distortion level and my ear will not even reach 18kHz so -1.4dB there will be ok.
Which is according to you the best (wherever is made) DIY Amp Board below 100€ in the range of 250W?
many thanks
L.

 

[somebodyelse]

250W into what load, and at what distortion? The more traditional audio community are still quoting power into 8R and sensible levels of distortion, while most of the class D chip amps are quoted into 4R at 10% distortion, so be sure you're making a fair comparison. The high frequency response is also sensitive to the load with a lot of the class D chip amps, and they usually seem to be optimised for 4R. TI have an application note on applying post-filter feedback (PFFB) which reduces this sensitivity, and reduces distortion in general somewhat. This isn't implemented on their EVM but they do have a worked example for the component values to add it yourself. I don't know whether or not 3E have PFFB on their boards.

The 3E boards look good to me, but there may be something better - I haven't been looking too closely at what's available. There are some high power boards based on an open design from International Rectifier that I don't remember the name of - they measure pretty well but don't have much in the way of output protection.

 

[Lorenzo74]

250W into what load, and at what distortion? The more traditional audio community are still quoting power into 8R and sensible levels of distortion, while most of the class D chip amps are quoted into 4R at 10% distortion, so be sure you're making a fair comparison. The high frequency response is also sensitive to the load with a lot of the class D chip amps, and they usually seem to be optimised for 4R. TI have an application note on applying post-filter feedback (PFFB) which reduces this sensitivity, and reduces distortion in general somewhat. This isn't implemented on their EVM but they do have a worked example for the component values to add it yourself. I don't know whether or not 3E have PFFB on their boards.

The 3E boards look good to me, but there may be something better - I haven't been looking too closely at what's available. There are some high power boards based on an open design from International Rectifier that I don't remember the name of - they measure pretty well but don't have much in the way of output protection.

many thanks.
building PFFB is not good for one time only (DIY) .
traditional audio community is far from me, it's a long time now that legacy 8Ohm has been (by most) traded for 4Ohm
apple to apple 250W, 1% distortion 4ohm is fine to me, lower power returns inaudible distortion.
I also found many class D to be more resilient vs impedance variation (Hypex Purity) vs equivalent class AB.
IRS2092 LJM 20d seems very good but need +- high voltage power supply, too expensive for the game
what about this:

https://www.ebay.it/itm/Hifi-TPA325...a=1&pg=2045573&_trksid=p2045573.c100033.m2042

bigger audio capacitors and air core inductance... worth the game vs 3E (i know how we can judge it from picture...)

your opinion is of value
my Best
L.

 

[chips666]

https://www.audiosciencereview.com/forum/index.php?threads/which-tpa3255-amp-do-you-recommend.9148/
Hello
The TPA3255EVM and 3e audio are good boards
Both do not have PFFB
Enjoy...

 

[somebodyelse]

I knew I'd posted something similar a long time ago - it'll be in that thread somewhere.

many thanks.
building PFFB is not good for one time only (DIY) .
traditional audio community is far from me, it's a long time now that legacy 8Ohm has been (by most) traded for 4Ohm
apple to apple 250W, 1% distortion 4ohm is fine to me, lower power returns inaudible distortion.
I also found many class D to be more resilient vs impedance variation (Hypex Purity) vs equivalent class AB.

Indeed - but they're somewhat better and more expensive. They also have feedback after the filter.

IRS2092 LJM 20d seems very good but need +- high voltage power supply, too expensive for the game
what about this:

https://www.ebay.it/itm/Hifi-TPA3255-Class-D-Power-amplifier-board-BTL-High-Power-amp-150W-150W-L21-23/193076135301?_trkparms=aid=111001&algo=REC.SEED&ao=1&asc=20140423084956&meid=a8877d69341249568ec4d1d8b03dbe09&pid=100033&rk=6&rkt=8&sd=132832662854&itm=193076135301&pmt=1&noa=1&pg=2045573&_trksid=p2045573.c100033.m2042

bigger audio capacitors and air core inductance... worth the game vs 3E (i know how we can judge it from picture...)

your opinion is of value
my Best
L.

Bigger caps but further from the chip which is also important for performance - without measurement it's hard to know if performance will be better or worse because of this. The filtering arrangement is unusual - not quite sure what's going on there. Since you brought up Hypex as an example earlier you might like to have a look at their white papers, particularly the 'UCD and EMI' one, then look at this filter arrangement again.

 

[snip3r77]

is there such a thing as quality implementation?

 

[somebodyelse]

Define 'quality' - at the upper end for this chip family you've got something like XRK971's assembled board with quality components, good layout and PFFB, which, according to the measurements in the diyudio thread, exceeds the performance from TI's PFFB application note. Bare boards and BOM from Mouser are also available, but still more expensive than the TI EVM. XRK971 says in that thread that the 3E implementation is pretty good, and it's a fraction of the price. Below that you've an assortment of cheaper boards that cut various corners to get the price down, at the cost of performance in the very few instances they've been measured, and sometimes badly implemented muting at power on and off. That's not to say they're necessarily bad value at the lower price point, but to some extent you seem to be getting what you pay for.

 

[snip3r77]

Define 'quality' - at the upper end for this chip family you've got something like XRK971's assembled board with quality components, good layout and PFFB, which, according to the measurements in the diyudio thread, exceeds the performance from TI's PFFB application note. Bare boards and BOM from Mouser are also available, but still more expensive than the TI EVM. XRK971 says in that thread that the 3E implementation is pretty good, and it's a fraction of the price. Below that you've an assortment of cheaper boards that cut various corners to get the price down, at the cost of performance in the very few instances they've been measured, and sometimes badly implemented muting at power on and off. That's not to say they're necessarily bad value at the lower price point, but to some extent you seem to be getting what you pay for.

is there a 3E implementation that is finished?