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

[pma]

AIYIMA A07 TPA3255 Amplifier Measurements LM4562 option

I received my A07 yesterday, on January 6, 2021, shipped from Italy. Purchased at Amazon.de and as stated on their website, it arrived without a power supply.

I am testing the A07 with 2 different power supplies. One of them is a linear regulated PSU with 30Vdc voltage and current 1.5A max. This is not originally intended for audio, but for instrumentation. It has low noise of 143uV (BW 24kHz) and very very low hum components. However, due to low output current the usable power is limited. The second is an unregulated linear 50Vdc power supply which is able to give 10A current. It has a 10 000 uF filter capacitor bank.

When measuring A07, one has to bear in mind that it is a BTL (bridged) amplifier, thus both speaker output terminals are “Live”. There is always a DC voltage of ½ PSU voltage at each terminal. The speaker sees the voltage difference, so no problem, but SE input soundcard could not be used for measurements, as it would short one of the output terminals to ground. Please also note that in case of grounded 48Vdc power supply, each of the A07 output terminals would be at hard 24Vdc vs. ground. This will also happen with class II power supply but some of the audio chain components in class I, thus grounded.

To measure such amplifier, floating or balanced input instrument must be used.

Another point to consider is class D output spectrum. A07 operates with carrier frequency of 600kHz and 183 mVrms level, as per following measurement.

The output spectrum shows 600kHz carrier + harmonics to 10MHz

Such spectrum would make the measuring audio system crazy, so during all measurements in audio band a differential R-C-R lowpass filter is used with -3dB corner at 40kHz. Wideband measurements (with the digital scope) were made without the filter.

Wideband amplitude frequency response

was measured into 4ohm and 10ohm loads, from 100Hz to 780kHz, to show the interaction of the LC filter with the load resistance and also to show carrier frequency and filter roll-off. As a signal source, wideband noise generator was used.

This is the frequency response into 4ohm. Note the 50kHz LC filter cut-off visible in individual output branches.

This is the frequency response into 10ohm. Note the 50kHz LC filter cut-off and also gain peaking at 45kHz because of lower damping of the output LC by 10ohm load.


Audio band measurements

The amplifier was tested with supplied NE5532 opamps and also with LME4562 opamps. I will always note which one was used. I will also note if the 30V regulated linear power supply or unregulated 50V linear power supply was used.

Input impedance

Input impedance = 23 kohm at 1kHz

Noise measurements – all with 30V regulated linear power supply

Noise with NE5532 was -73.4 dBV linear 24kHz BW (with the 40kHz lowpass filter as in all further measurements). This would make S/N 86.4 dB re 5W/4ohm and 101.2 dB re maximum power, which I estimated at 150W/4ohm (later).

With LM4562, the noise was -74.9dBV (24kHz), which would make S/N 87.9 dB re 5W/4ohm and 102.7 dB re 150W/4ohm.

Please note slightly lower noise and also better rejection of mains multiples.


THD and THD+N (SINAD) at 1kHz/5W/4ohm measured with LM4562

With the 30V regulated PSU, we get very good suppression of mains components and the SINAD of 84.3 dB. As we can see from noise measurement, this is mainly affected by the residual noise.

With the 50V unregulated PSU, we can see rise of mains components, SINAD worsening to 83.2dB and, more importantly, the appearance of high order harmonics, 7th and more. Both is not appreciated.

THD and THD+N at 1kHz/20W/4ohm

With the regulated 30V supply, we can see nice absence of the mains components. LM4562.

With the unregulated 50V supply, please note the rise of mains components and high order harmonics. Both is not appreciated. NE5532 (the opamp makes not much difference here).


THD and THD+N at 1kHz/100W/4ohm, unregulated 50V power supply

The amplifier switched off itself after about 1 minute of this test, due to overheating. However it has survived.


THD vs. amplitude at 1kHz and 6kHz / 4ohm, unregulated 50V power supply

Please note the strange nonlinear behavior at 6kHz. This is completely in conformance with TI SLAA788A document, Fig.87


THD vs. frequency at 5W/4ohm, regulated 30V power supply

CCIF IMD 19+20kHz, 4ohm, NE5532, 30V regulated PSU

Conclusions

This is a small, tiny, cheap amplifier. Concerning its ultra low cost, it does not behave badly. At lower frequencies it measures well. Better SINAD parameters would be obtained if the intrinsic noise was lower. Care should be taken because both output terminals are always at ½ PSU voltage, with respect to the amplifier signal ground and metal case. The results obtained highly depend on the power supply used. My recommendation would be a linear, regulated power supply. Maybe also a SMPS with very clean output. Remember that output spectrum of higher harmonics depended strongly on power supply cleanliness. Another issue is frequency response peaking with high load impedance. This may easily happen with some speakers when impedance at tweeter band frequencies rises.


I will be continuing in this review later, together with listening impressions.

---------------------------------------
Measurements are continuing ...

So we have a problem. Below is the frequency response into JBL Control1 Pro, measured at amp terminals. There is a big interaction of the output filter with the speaker impedance and this may be audible. I think it is the peaking between 3kHz and 10kHz that might be audible.

Continued on January 9, 2021
I have just measured the frequency response into my speaker dummy load. The impedance plot of this load is shown below

The result of AIYIMA A07 is very disappointing

The response peaking is too big and shows resonant behaviour of the output LC filter with this load.
So I can say that even the sonic result with AIYIMA A07 will strongly depend on the speaker connected. This is certainly not a good news. A speaker with impedance close to resistive in midrange and high frequency area is recommended.

 

[Matias]

Geat test! If I may suggest I had some difficulty comparing graph to graph as there are several variables at play here:
- opamp choice
- power supply choice
- power output

Could you maybe group the graphs when changing only 1 of those variables? Makes it easier to read what has changed.
Thanks!

 

[pma]

Geat test! If I may suggest I had some difficulty comparing graph to graph as there are several variables at play here:
- opamp choice
- power supply choice
- power output

Could you maybe group the graphs when changing only 1 of those variables? Makes it easier to read what has changed.
Thanks!

I was thinking about the optimal method, but as I do not have the AP or a system with similar noise and resolution specs, I need to change input divider (balanced stepper attenuator) ratio according to level measured and simultaneously measure the amp output level. If I did not that, the N of the THD+N would rise too high and would mask the amp noise component. And I have no automated system. It took me about six hours to make these measurements and optimize the system. I believe it is a bit difficult for orientation, but at the same time I believe that you who know what is it about and can read the graphs will collect the useful information. Especially because I am concentrated to a bit different issues than Amir. I think this may be a good complement to his measurements and both together give a good view. Please note that our measurements give very similar results where they overlap.

 

[pma]

The output terminals are so closely spaced (14 mm!) that I have to make an adapter to 29 mm standard spacing if I want to connect my speakers!
My latching speaker banana plugs have just 14mm diameter and they make a short if connected to A07 .

Another linear PSU, 49V

So yes, it produces sound!!

 

[Eetu]

Thanks for the measurements! I have the A04 with the LM4562 upgrade ordered so glad to see it's not making things worse

 

[pma]

Listening impressions

I connected the amp to the JBL Control 1 Pro speaker placed above my workbench - and I did not like the sound. I am quite sure that it was for the reason that this speaker is poor and the sound is awful with anything.

Then I moved the amp to my listening room, connected to my regular speakers
http://www.troelsgravesen.dk/CNO-T25_Pavel.htm
played some jazz and rock music - and everything was fine. The sound was punchy, well controlled, with good imaging.

Then I put Beethoven's No.2 Symphony with Wiener Philharmoniker directed by Simon Rattle to my CD player and made a "critical listening". The sound was fine, little bit on a dry side, but no bigger complaints.

In general, the amp sounds fine. I would not take it as my main amp, but as a second amp in the office, workshop or bedroom it would be fine. I have some concerns about heat and reliability, the case gets hot even at moderate sound level after some time.

 

[pma]

Listening impressions

I connected the amp to the JBL Control 1 Pro speaker placed above my workbench - and I did not like the sound. I am quite sure that it was for the reason that this speaker is poor and the sound is awful with anything.

Unfortunately, there might have been a technical reason. I have just measured a frequency response of the A07 when loaded with JBL1 Pro and it is not very nice. There is an interaction of the amp output LC filter with the speaker impedance resulting in FR modulation, and this may be IMO audible. I mean the peaking between 3kHz and 10kHz. The review was just updated with this measurement.

--------
As a comparison, frequency response measured with my BUF2 class AB headphone amplifier into the same speaker. This headamp has very low output impedance so it is able to drive speakers, of course at low power.

 

[Ralferator]

AIYIMA A07 TPA3255 Amplifier Measurements LM4562 option

I received my A07 yesterday, on January 6, 2021, shipped from Italy. Purchased at Amazon.de and as stated on their website, it arrived without a power supply.

I am testing the A07 with 2 different power supplies. One of them is a linear regulated PSU with 30Vdc voltage and current 1.5A max. This is not originally intended for audio, but for instrumentation. It has low noise of 143uV (BW 24kHz) and very very low hum components. However, due to low output current the usable power is limited. The second is an unregulated linear 50Vdc power supply which is able to give 10A current. It has a 10 000 uF filter capacitor bank.

When measuring A07, one has to bear in mind that it is a BTL (bridged) amplifier, thus both speaker output terminals are “Live”. There is always a DC voltage of ½ PSU voltage at each terminal. The speaker sees the voltage difference, so no problem, but SE input soundcard could not be used for measurements, as it would short one of the output terminals to ground. Please also note that in case of grounded 48Vdc power supply, each of the A07 output terminals would be at hard 24Vdc vs. ground. This will also happen with class II power supply but some of the audio chain components in class I, thus grounded.

To measure such amplifier, floating or balanced input instrument must be used.

Another point to consider is class D output spectrum. A07 operates with carrier frequency of 600kHz and 183 mVrms level, as per following measurement.
View attachment 104309


The output spectrum shows 600kHz carrier + harmonics to 10MHz
View attachment 104310


Such spectrum would make the measuring audio system crazy, so during all measurements in audio band a differential R-C-R lowpass filter is used with -3dB corner at 40kHz. Wideband measurements (with the digital scope) were made without the filter.

Wideband amplitude frequency response

was measured into 4ohm and 10ohm loads, from 100Hz to 780kHz, to show the interaction of the LC filter with the load resistance and also to show carrier frequency and filter roll-off. As a signal source, wideband noise generator was used.
View attachment 104311

This is the frequency response into 4ohm. Note the 50kHz LC filter cut-off visible in individual output branches.


View attachment 104312

This is the frequency response into 10ohm. Note the 50kHz LC filter cut-off and also gain peaking at 45kHz because of lower damping of the output LC by 10ohm load.


Audio band measurements

The amplifier was tested with supplied NE5532 opamps and also with LME4562 opamps. I will always note which one was used. I will also note if the 30V regulated linear power supply or unregulated 50V linear power supply was used.

Noise measurements – all with 30V regulated linear power supply

Noise with NE5532 was -73.4 dBV linear 24kHz BW (with the 40kHz lowpass filter as in all further measurements). This would make S/N 86.4 dB re 5W/4ohm and 101.2 dB re maximum power, which I estimated at 150W/4ohm (later).
View attachment 104315


With LM4562, the noise was -74.9dBV (24kHz), which would make S/N 87.9 dB re 5W/4ohm and 102.7 dB re 150W/4ohm.
View attachment 104316

Please note slightly lower noise and also better rejection of mains multiples.


THD and THD+N (SINAD) at 1kHz/5W/4ohm measured with LM4562
View attachment 104317

With the 30V regulated PSU, we get very good suppression of mains components and the SINAD of 84.3 dB. As we can see from noise measurement, this is mainly affected by the residual noise.

View attachment 104318

With the 50V unregulated PSU, we can see rise of mains components, SINAD worsening to 83.2dB and, more importantly, the appearance of high order harmonics, 7th and more. Both is not appreciated.

THD and THD+N at 1kHz/20W/4ohm
View attachment 104319

With the regulated 30V supply, we can see nice absence of the mains components. LM4562.


View attachment 104320

With the unregulated 50V supply, please note the rise of mains components and high order harmonics. Both is not appreciated. NE5532 (the opamp makes not much difference here).


THD and THD+N at 1kHz/100W/4ohm, unregulated 50V power supply
View attachment 104321

The amplifier switched off itself after about 1 minute of this test, due to overheating. However it has survived.


THD vs. amplitude at 1kHz and 6kHz / 4ohm, unregulated 50V power supply
View attachment 104323

Please note the strange nonlinear behavior at 6kHz. This is completely in conformance with TI SLAA788A document, Fig.87


THD vs. frequency at 5W/4ohm, regulated 30V power supply
View attachment 104324


CCIF IMD 19+20kHz, 4ohm, NE5532, 30V regulated PSU
View attachment 104325

Conclusions

This is a small, tiny, cheap amplifier. Concerning its ultra low cost, it does not behave badly. At lower frequencies it measures well. Better SINAD parameters would be obtained if the intrinsic noise was lower. Care should be taken because both output terminals are always at ½ PSU voltage, with respect to the amplifier signal ground and metal case. The results obtained highly depend on the power supply used. My recommendation would be a linear, regulated power supply. Maybe also a SMPS with very clean output. Remember that output spectrum of higher harmonics depended strongly on power supply cleanliness. Another issue is frequency response peaking with high load impedance. This may easily happen with some speakers when impedance at tweeter band frequencies rises.


I will be continuing in this review later, together with listening impressions.

---------------------------------------
Measurements are continuing ...

So we have a problem. Below is the frequency response into JBL Control1 Pro, measured at amp terminals. There is a big interaction of the output filter with the speaker impedance and this may be audible. I think it is the peaking between 3kHz and 10kHz that might be audible.

View attachment 104380

Do you think the better op amp is audible?

 

[pma]

Do you think the better op amp is audible?

No.
But I think that power supply choice is crucial and that the speaker used should have as flat impedance response as possible, especially above some 3kHz. The later IMO would be audible.

 

[Bruce Morgen]

Those little JBLs are gawdawful in several respects -- and I say that as a long-time JBL admirer and the owner of four pairs of more or less "vintage" models -- so I'm not surprised that they have a problematic impedance curve.

 

[whazzup]

No.
But I think that power supply choice is crucial and that the speaker used should have as flat impedance response as possible, especially above some 3kHz. The later IMO would be audible.

Thanks for the measurements!

Not sure if this is the same Control Pro model you have, but looking at JBL's published impedance: http://files.tse.si/jbl-ds-control1.pdf

The speaker's impedance seems to drop to 2+ ohms at lowest points. But how does this low impedance react with the LC filter that's outside of audible FR range, to impact the FR within audible range?

 

[pma]

Thanks for the measurements!

Not sure if this is the same Control Pro model you have, but looking at JBL's published impedance: http://files.tse.si/jbl-ds-control1.pdf

The speaker's impedance seems to drop to 2+ ohms at lowest points. But how does this low impedance react with the LC filter that's outside of audible FR range, to impact the FR within audible range?

I am attaching the plot from my JBL specs

Now the plot that I have measured

It is +/- the same. The problem is not in the bass area, but above 2kHz. This is the part that interacts with the output LC. Please read the phase plot and see that the impedance is capacitive from 3kHz to 8kHz. This capacitance is added to output LC filter capacitance and changes the amplifier FR.


Now the plot of my CNO-T25

Please note that it is flat above 3kHz and the phase angle is close to 0°, so it is close to resistive load above 3kHz.


And this completely fits to listening impressions. With CNO-T25, the sound is fine, just a little bit "dry".

With JBL Control1 Pro, the sound is harsh in higher mids frequency band.

 

[amirm]

There is an interaction of the amp output LC filter with the speaker impedance resulting in FR modulation, and this may be IMO audible.

0.5 dB at 5 kHz is audible to you??? Likely your left and right speakers are more different than each other this.

 

[pma]

0.5 dB at 5 kHz is audible to you??? Likely your left and right speakers are more different than each other this.

Yes, the peaking followed by the decay is audible. You can simulate it in your EQ and make a scientific DBT . BTW I have been listening through only one speaker in mono, just according to your and Harman recommendations .

 

[JohnYang1997]

I would say if the speaker has a bit excess energy in that region. The amplifier can push it to bothering. I used to fine tune eq on 0.2db to 0.5db basis a couple of years ago with my er4.

 

[pma]

I would say if the speaker has a bit excess energy in that region. The amplifier can push it to bothering. I used to fine tune eq on 0.2db to 0.5db basis a couple of years ago with my er4.

John, have you made some comparisons between the LC out-of-FB and inside FB (PFFB), re impedance interaction. I think you have designed some class D amps, haven't you.

 

[TNT]

LM4562 and regulated PSU with 30Vdc?

//

Listening impressions


Then I put Beethoven's No.2 Symphony with Wiener Philharmoniker directed by Simon Rattle to my CD player and made a "critical listening". The sound was fine, little bit on a dry side, but no bigger complaints.

In general, the amp sounds fine. I would not take it as my main amp, but as a second amp in the office, workshop or bedroom it would be fine. I have some concerns about heat and reliability, the case gets hot even at moderate sound level after some time.

 

[pma]

LM4562 and regulated PSU with 30Vdc?

//

Yes and then with another unregulated PSU, 49V but less current capability than the 50V/10A one, however with better output capacitor bank filter. The last one is able to give 80W/4ohm max. with the A07. The high order harmonics visible with the 50V/10A PSU have disappeared with the 49V PSU (I call it No.2). This is interesting and probably a result of better capacitor filter bank and bypassing ESR/ESL. Even earlier with other amps I found this high order harmonics modulation due to poor PSR of the amp. But the listening impressions with both PSUs, regulated and 49V No.2, has remained the same.

Distortion plot with the unregulated 49V PSU No.2, 56W/4ohm.

 

[pma]

Someone has asked what brand of capacitors is used in the output LC filter. So, this one:

 

[JohnYang1997]

John, have you made some comparisons between the LC out-of-FB and inside FB (PFFB), re impedance interaction. I think you have designed some class D amps, haven't you.

Only early stage. I'm working on topological design than testing fr variation with these loads. (I'll be using PFFB type design anyway) All I can easily do now is just regular 4ohm or 8ohm. There are also many other stuff having higher priority.