2025: Does somebody needs Hypex etc...?

[TurtlePaul]

All the TPA3255 amps with PFFB that I've see have too much noise at high frequencies on multitone. I really don't like this behavior. It's too much for 2025. Something is wrong with this chip's high frequencies. The Purifi are much cleaner in this test up to 20 kHz.

It is -100 dB until 16 khz. I get that you may not like the shape of the picture. But I am convinced I cannot hear that.

 

[Yevhen]

All the TPA3255 amps with PFFB that I've see have too much noise at high frequencies on multitone. I really don't like this behavior. It's too much for 2025. Something is wrong with this chip's high frequencies. The Purifi are much cleaner in this test up to 20 kHz.

I guess it's not a noise, but IMD. The low frequency subcarriers are mixed up with the high frequency ones

 

[TurtlePaul]

Maybe that 3e Audio has a global feedback, but it's clearly not working well at high frequencies. That 15kHz curve stays at -65dB

It is working. Harmonic distortion at 15 khz makes tones at 30 and 45 khz. You cannot hear them. If they increase the frequency of the feedback (don't bend back the poles) then the amp will start to become unstable with certain speakers.

 

[Yevhen]

It is -100 dB until 16 khz. I get that you may not like the shape of the picture. But I am convinced I cannot hear that.

You'll hear it quite clear on a speaker with very low 3rd/5th harmonic distortion in the tweeter. Something like 0.1% for H3, H5 and H7

 

[Yevhen]

Harmonic distortion at 15 khz makes

IMD at 15kHz mixes sound of you cymbals with the sound of your bass and voice

 

[Mike123]

It is -100 dB until 16 khz. I get that you may not like the shape of the picture. But I am convinced I cannot hear that.

The point is not whether I will hear this noise specifically. The point is that this test shows problems with the amplifier under conditions of a real music signal. Where and how they can manifest themselves - I do not know. But I do not want to test it on myself. At the beginning of this thread I wrote that Purifi is obviously transparent and has no problems with any signal. TPA3255 potentially has problems.

 

[Yevhen]

If they increase the frequency of the feedback (don't bend back the poles) then the amp will start to become unstable with certain speakers.

Fully agree. I tried to build a few amps with a very deep feedback and oscillation was the biggest challenge . That's why I admire the designs from Bruno as he managed to keep the amp very stable while having the extremely high open loop gain

 

[TurtlePaul]

OK, so I came into this thinking that Hypex existing because people with expensive systems were willing to pay a little more for overkill. However, now I see that quite a few people are convinced that they would be able to hear the difference between a 100 dB SINAD and 110 dB SINAD amp.

 

[Mike123]

However, now I see that quite a few people are convinced that they would be able to hear the difference between a 100 dB SINAD and 110 dB SINAD amp.

This is not true. I tried to explain. AmirM does not only test the sinad at 1 kHz. He does many tests. Some are more complicated than sinad. And some tests show unstable operation of the amplifier, which can potentially cause transparency problems in certain parts of the phonogram. Yevhen gave an example of THD from frequency on the last page. There, too, everything is not so good. At 10 kHz, there is not -110 and not -100 dB THD. But -72 dB. Is it very bad? I don't know. Is it very audible? I don't know. But if there are much better alternatives that do not have such problems (and do not have any problems at all) and that do not cost as much as a Boeing wing - then why buy a far from cheap and rather mediocre amplifier?

 

[TurtlePaul]

This is not true. I tried to explain. AmirM does not only test the sinad at 1 kHz. He does many tests. Some are more complicated than sinad. And some tests show unstable operation of the amplifier, which can potentially cause transparency problems in certain parts of the phonogram. Yevhen gave an example of THD from frequency on the last page. There, too, everything is not so good. At 10 kHz, there is not -110 and not -100 dB THD. But -72 dB. Is it very bad? I don’t know.

The test is run at 45 khz bandwidth. It is not a test of audibility. Maybe Amir should explain this in the reviews or run the 5 kHz test at 19 khz or 20 khz bandwidth.

 

[dominikz]

All the TPA3255 amps with PFFB that I've see have too much noise at high frequencies on multitone. I really don't like this behavior. It's too much for 2025. Something is wrong with this chip's high frequencies. The Purifi are much cleaner in this test up to 20 kHz.

IMHO the typical multitone measurement with flat spectrum ("white" spectrum) is almost like a torture test, and not a signal resembling any realistic sounds.

No sane music or movie would push a high output level at both 100Hz and at 10kHz.

As such, this typical measurement may overstate issues which may be of little practical significance.

On the other hand, we can do the same kind of measurement with a multitone signal with falling level vs frequency (i.e. "pink" spectrum), which is IMHO much more representative of real music.

Here's how one TPA3255-based amp measures with both kinds of signal here:

Notice how there's no N+D increase with frequency when using the "pink" spectrum signal.

So far I saw no evidence that this increase in distortion at high frequencies (and levels) causes any audible issues. Sure, from engineering perspective I'd prefer it wasn’t there, but I for one am not going to lose sleep over it.

 

[Yevhen]

The test is run at 45 khz bandwidth. It is not a test of audibility. Maybe Amir should explain this in the reviews or run the 5 kHz test at 19 khz or 20 khz bandwidth.

See those THD tests only as the representation of the nonlinearity of the amplifier. You'll never hear harmonics until they are like 10% or so. However, based on my experience, the multitone test more accurately represents the quality of the amplifier.

 

[Yevhen]

Is it very audible? I don't know.

I lived for 2 years with NAD D 2030 V1. I think it was based on the Hypex's UcD102 OEM, and should be close to TI chips in terms of IMD performance. The sound was punchy and clean in mid and bass regions, but a bit muddy >5kHz. I never took care of it with Monitor Audio Bronze speakers but after getting a better speaker (Totem with Wavecor tweeter and a ribbon-based DIY afterwards), it became quite obvious.

UcD102 OEM (sounds muddy to me):

Purifi 1ET400A (sounds transparent to me)

Intermodulation products are 35-40dB lower.

 

[Mike123]

but I for one am not going to lose sleep over it.

Of course, I also don't suffer from the simple fact that a $2 multimedia chip has been elevated to the rank of high-quality audio . The thread asked why purifi if there is TPA? So I'm answering this question.

 

[Mike123]

The test is run at 45 khz bandwidth. It is not a test of audibility. Maybe Amir should explain this in the reviews or run the 5 kHz test at 19 khz or 20 khz bandwidth.

TI is a serious company. It conducted these tests itself. This picture is from the datasheet for the TPA3255. It also contains measurements of the THD from a frequency with an operating band of up to 80 kHz and standard ones up to 20 kHz. Here you can see that the TND grows with frequency. It grows strongly for a device that claims to be high-quality. As a computer amplifier, this is normal. As a basis for a high-quality system, in my opinion, it is not normal. But I do not forbid anyone who likes this amplifier to buy it .

 

[valerianf]

The TPA3255 is priced $3.34 for 2k units at Digikey.
Will it do the same job as a board full of components developed during several years by very high end EE?
For sure not.
I may have a similar IC in my car audio system, it is fine, but for I will not buy any product using it for my home theater system.
For the surround channels I am using a Yamaha WCA 50 using ICE modules.
The SQ is satisfying.

 

[RexrothPigeon]

Will it do the same job as a board full of components developed during several years by very high end EE?
For sure not.

Why not? (genuine question)

 

[valerianf]

"Why not? (genuine question)"
Santa Claus doesn't exist.

TI is making some audio circuits.
Long time ago they bought Burr Brown that was a top notch company for analog ICs.
I do not believe that TI is top notch in audio products.
My AVR (Onkyo TX RZ30) has a TI DAC.
I am aware that it is not at the level of the ESS or older AKM DAC.
However, It is sufficient for my usage.
Same remark about the class D amp products.

 

[RexrothPigeon]

I do not believe

audiofaithreview.com seems to be available for purchase, fyi

 

[valerianf]

The THD+N (Sinad) jump in the high frequency is so obvious that it is a pass.