AIYIMA A07 TPA3255 Amplifier Measurements and Review - LM4562 (and OPA2134) option

[TheWalkman]

There have been some reservations about my no-recommending of the 48V PSU, like this:



So here is my reasoning:
1) the amplifier triggered off itself after 1 minute of 100W/4ohm single channel test, as already mentioned in this thread. This is probably not so important, but
2) I have just made a temperature measurement on the heatsink, with the K-thermocouple. The box was partially closed. I have set the power to 50W/4ohm maximum when the signal reaches 0dBFS. Then I used Amy Winehouse / Our Day Will Come track, which has DR=7 and the track was played in a loop. Only one channel of the amp was driven. The average power was much less than 50W, due to music programme with DR=7. And again, only one channel. After opening, there is a smell of melting plastic.
View attachment 106465
After 12 minutes, the heatsink temperature was 60°C. An this amplifier is specified as 2x300W/10% THD .....
So this is my reasoning and recommendation, run the amp with 32V power supply.

PMA,

Interesting observations.

Though I was running my A07 with a quality power brick, my amp wasn’t running hot (actually, after a loud listening session with the top off, the heatsink was just warm to the touch.) I was powering it with with a Mean Well 48V, 220W power brick (GMS 220B48).

however, based on your comments, I switched the 48V supply with another Mean Well brick I had in my collection, this time a 36 V, 160 W (GSTA136). Though this supply obviously has less peak wattage, I’m still getting great sound from the amp.

it seems any penalty of using the smaller supply appears minimal. Perhaps it will increase the longevity of the A07.

FWIW, I am running the A04 with the stock 32V brick.

I also the stock op amps for 1656’s in the A04 and LM4562s in the A07.

I have both amps hooked to a pair of Unifi UF5 floor standers via an MCM auto speaker switch.

Though I haven’t done any direct A-B comparison. I’m very pleased with both of these amps though I think the A04 is my favorite. It seems to be clearer, better dynamic range and, seemingly, low end response.

Though there is an incredible mod thread about A04 mods over on DIYAudio, I’ve decided to put the cases back on and leave well enough alone. These little amps simply sound phenomenal for the price, despite some apparent graphs to the contrary.

That said, for someone just getting into hi-fi, especially on a limited budget, I would wholeheartedly recommend either amp as great starting points. For $75 with a power brick, you just can’t go wrong. Add some decent bookshelf speakers, a DAC (dongle) and a cellphone loaded with music, $500 will buy an incredible listening system the quality of which I could have only of dreamed of when I was a college kid just getting into this hobby.

 

[pma]

Still waiting for the SMPS delivery, there seems to be some shipment delay.

In the meantime, I made some measurements with another power supply, 25Vdc unregulated, 270VA, so there is enough current available if requested.

The power was measured, with both channels driven and loaded by 4ohm each. 5 minute test at continuous power. The power measured was 2 x 50W/4ohm at THD < 1%, with the 25V/270VA power supply. The case temperature remained mild, little bit warm, during the test. I expect to get the same or very similar result with the SMPS, maybe even better, because SMPS output voltage may be set up to 26.8V by the trimpot.

L and R channels were nicely matched re distortion, below is the plot.

The next plot shows distortion vs. output voltage (/power) at 100Hz, 1kHz and 6kHz

 

[pma]

This is the "new" method of measuring IMD, using two tones, 2kHz and 17kHz, when amplitude of the 17kHz is 100x smaller (-40dB) compared to 2kHz tone amplitude. This was chosen not to torment the poor class D by high pitch tone of high amplitude, that would never occur in a natural music signal, as complained by several members. OK, but the small high pitch tone (or overtone) is always superimposed on the signal with lower frequency, so the method is perfectly valid and there may hardly be more complaints about too big high frequency tone. It is barely visible on the oscilloscope screen only as a slight change in the shape of the lower frequency tone. However, our class D amplifier does not like it. The measurement was performed with one channel driven, 37W/4ohm, 27V SMPS power supply and 48kHz measurement bandwidth. And this is the result, a nice forest of harmonics with 1kHz spacing, though there is no 1kHz input signal. Nice intermodulations, indeed, with a very small 17kHz tone. I believe @restorer-john might like it.

 

[pma]

So this would probably be my final contribution to A07 measurements. I have made a thorough THD and IMD vs. amplitude measurements of my AIYIMA A07 TPA3255 amplifier, with LM4562 opamps (reduces noise and distortion slightly compared to original NE5532), powered from Mean Well RS-100-24 SMPS set to 27Vdc voltage. One channel driven, 4ohm load. A set of frequencies 100Hz-1kHz-3kHz-6kHz-12kHz. IMD 250Hz+8kHz and 13+14kHz. Measurement bandwidth 48kHz. I am trying to be as accurate as possible.

IMO the behaviour measured (I am speaking about non-linear behaviour from low power 250mW upwards) can be extrapolated to all amplifiers with TPA325X circuit and without PFFB. Probably to most class D amplifiers with the LC filter outside feedback loop. I am not saying that to class D generally. At least we know that Hypex NC400 (and similar Hypex products) and Purifi modules behave much much better, without measured issues.

But there is a plenty of class D amplifiers with similar design as this AIYIMA A07 that will behave very similarly. And this non-linear behaviour, together with load-dependent frequency response, rise a potential of audible sound differences on music signals.

 

[Tangband]

So this would probably be my final contribution to A07 measurements. I have made a thorough THD and IMD vs. amplitude measurements of my AIYIMA A07 TPA3255 amplifier, with LM4562 opamps (reduces noise and distortion slightly compared to original NE5532), powered from Mean Well RS-100-24 SMPS set to 27Vdc voltage. One channel driven, 4ohm load. A set of frequencies 100Hz-1kHz-3kHz-6kHz-12kHz. IMD 250Hz+8kHz and 13+14kHz. Measurement bandwidth 48kHz. I am trying to be as accurate as possible.

IMO the behaviour measured (I am speaking about non-linear behaviour from low power 250mW upwards) can be extrapolated to all amplifiers with TPA325X circuit and without PFFB. Probably to most class D amplifiers with the LC filter outside feedback loop. I am not saying that to class D generally. At least we know that Hypex NC400 (and similar Hypex products) and Purifi modules behave much much better, without measured issues.

But there is a plenty of class D amplifiers with similar design as this AIYIMA A07 that will behave very similarly. And this non-linear behaviour, together with load-dependent frequency response, rise a potential of audible sound differences on music signals.

View attachment 107522

Thanks for all the good work.

 

[pma]

This is how the BTL mode (bridge) of the TPA3255 works. There are 2 outputs (for each channel), OUT+ and OUT-, and the load is connected between them. These outputs are at 1/2 PSU voltage, with no input signal, referred to signal ground. They are out of phase, i.e. OUT- is inverted with respect to OUT+. This enables to get twice output voltage swing compared to each OUT vs. ground. Let's measure OUT+, OUT- and (OUT+) - (OUT-), which is the voltage across the load. Let's take a square wave input. Blue channel is OUT+ vs. ground, Yellow channel is OUT- vs. ground and Red trace is the signal that the load would see. We can see that imperfections and ringing on both OUT+ and OUT- are reduced by subtraction. This is fine to certain degree.

 

[howard416]

pma,

In a multi-channel setup, would you have any reservations at all using this type of amp (i.e. no PFFB) under ~1500 Hz?

 

[howard416]

This is the "new" method of measuring IMD, using two tones, 2kHz and 17kHz, when amplitude of the 17kHz is 100x smaller (-40dB) compared to 2kHz tone amplitude. This was chosen not to torment the poor class D by high pitch tone of high amplitude, that would never occur in a natural music signal, as complained by several members. OK, but the small high pitch tone (or overtone) is always superimposed on the signal with lower frequency, so the method is perfectly valid and there may hardly be more complaints about too big high frequency tone. It is barely visible on the oscilloscope screen only as a slight change in the shape of the lower frequency tone. However, our class D amplifier does not like it. The measurement was performed with one channel driven, 37W/4ohm, 27V SMPS power supply and 48kHz measurement bandwidth. And this is the result, a nice forest of harmonics with 1kHz spacing, though there is no 1kHz input signal. Nice intermodulations, indeed, with a very small 17kHz tone. I believe @restorer-john might like it.

I am pretty sure that this is worse than the performance of many tweeters...

 

[pma]

Some more measurements disclosing the HF nonlinearity of all TPA3255 based amplifiers. Measurement bandwidth 48kHz. PSU 27Vdc SMPS.

 

[pma]

As already mentioned here and in another threads this amplifiers, as any class D with output LC filter outside FB loop, is optimized for a defined value of pure resistive load and does not like complex load frequency dependent impedance. This can be shown on a measurement of THD/frequency distortion plots where distortion into resistive load is compared to distortion into complex speaker dummy load (I have shown the circuit before). We can see much higher distortion into the dummy load above 100Hz even though the dummy load impedance magnitude is never below 5.5 ohm. Distortion rise below 100Hz is a result of a nonlinear ferrite core coil used in the dummy load.
In other words, this topology likes resistive loads but does not like most of real life speakers.

Measurement bandwidth is 40kHz

Distortion in %

Distortion in dB

Dummy load impedance. Brown = magnitude, green dashed = phase.

To compare, the same load, distortion measurement with a low distortion class AB amplifier - what a difference!

 

[Ralferator]

As already mentioned here and in another threads this amplifiers, as any class D with output LC filter outside FB loop, is optimized for a defined value of pure resistive load and does not like complex load frequency dependent impedance. This can be shown on a measurement of THD/frequency distortion plots where distortion into resistive load is compared to distortion into complex speaker dummy load (I have shown the circuit before). We can see much higher distortion into the dummy load above 100Hz even though the dummy load impedance magnitude is never below 5.5 ohm. Distortion rise below 100Hz is a result of a nonlinear ferrite core coil used in the dummy load.
In other words, this topology likes resistive loads but does not like most of real life speakers.

Measurement bandwidth is 40kHz

Distortion in %
View attachment 122884

Distortion in dB
View attachment 122885

Dummy load impedance. Brown = magnitude, green dashed = phase.
View attachment 122887

To compare, the same load, distortion measurement with a low distortion class AB amplifier - what a difference!
View attachment 122897

But can you hear it?

 

[Solveit]

What"low distortion class AB amplifier " is it?

 

[memphiskat]

As already mentioned here and in another threads this amplifiers, as any class D with output LC filter outside FB loop, is optimized for a defined value of pure resistive load and does not like complex load frequency dependent impedance. This can be shown on a measurement of THD/frequency distortion plots where distortion into resistive load is compared to distortion into complex speaker dummy load (I have shown the circuit before). We can see much higher distortion into the dummy load above 100Hz even though the dummy load impedance magnitude is never below 5.5 ohm. Distortion rise below 100Hz is a result of a nonlinear ferrite core coil used in the dummy load.
In other words, this topology likes resistive loads but does not like most of real life speakers.

Measurement bandwidth is 40kHz

Distortion in %
View attachment 122884

Distortion in dB
View attachment 122885

Dummy load impedance. Brown = magnitude, green dashed = phase.
View attachment 122887

To compare, the same load, distortion measurement with a low distortion class AB amplifier - what a difference!
View attachment 122897

PMA or anyone - Could you list a TPA32xx design that uses a PFFB and so doesn't have this non linearity behavior?

 

[Bruce Morgen]

To compare, the same load, distortion measurement with a low distortion class AB amplifier - what a difference!

To the eye, perhaps -- to the human ear listening to real world loudspeakers, not so much.

Then we get to consider the cost, size, and quiescent power consumption of that "low distortion class AB amplifier." Of course the more advanced and costly Class D designs -- e.g. Putrifi's Eigentakt -- also eliminate that issue entirely, apparently by anchoring the source side of the feedback loop after the output filter circuit.

 

[howard416]

Would a Zobel network eliminate the increased distortion from a non-resistive load powered by a Class D amp?

 

[Synergy4]

Self oscillating class D amplifiers with a large amount of global feedback (LC filter inside feedback loop) are less dependent on stable 4-8 ohm impedance & can tolerate erratic speaker impedance over frequency, due to the higher damping factor.

Hypex amplifers: https://www.diyclassd.com/diy-amplifier-modules/

Amplifier kits with Self oscillating class D amplifiers....

VTV Purifi: https://vtvamplifier.com/vtv-purifi/

VTV Hypex: https://vtvamplifier.com/vtv-hypex/

VTV Pascal: https://vtvamplifier.com/28456-2/

 

[Bruce Morgen]

Self oscillating class D amplifiers with a large amount of global feedback (LC filter inside feedback loop) are less dependent on stable 4-8 ohm impedance & can tolerate erratic speaker impedance over frequency, due to the higher damping factor.

Hypex amplifers: https://www.diyclassd.com/diy-amplifier-modules/

Amplifier kits with Self oscillating class D amplifiers....

VTV Purifi: https://vtvamplifier.com/vtv-purifi/

VTV Hypex: https://vtvamplifier.com/vtv-hypex/

VTV Pascal: https://vtvamplifier.com/28456-2/

Just a tad pricier than a sub-$100 A07+SMPS.

 

[DualTriode]

Hello,

You talk about this amplifier with a NE5532 operational amplifier and then with a LM4562 operational amplifier.

Is there a socket on the PCB that allows for rolling the operational amplifier(s)?

Thanks DT

 

[Bruce Morgen]

Hello,

You talk about this amplifier with a NE5532 operational amplifier and then with a LM4562 operational amplifier.

Is there a socket on the PCB that allows for rolling the operational amplifier(s)?

Thanks DT

Yes, the A07's opamps are socketed. That said, the supposed advantages of op amp rolling have been widely debated for years and, for the most part, persuasively debunked in terms of audibility -- especially in a relatively undemanding application like the front end of a sub-$100USD Class D power amp. I think it's very telling that TI manufactures both the LM4562 and the NE5532 -- and the latter was chosen for their own TPA3255AVM evaluation board.

 

[DualTriode]

Thanks for the news about the sockets.

Hear tell the LM4562 has less noise and better PSRR. More important is the quality of the power supply.

I will post here test results.

Thanks
DT