• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

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

OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,728
Location
Prague
Below please find AIYIMA A07 output noise measurement.

A07_noise.png
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,728
Location
Prague
Below is the measurement of THD vs. power, with @90kHz measuring bandwidth and at frequencies 1kHz, 5kHz, 10kHz and 15kHz. The power supply used was Meanwell SMPS RS-100-24 set to 27Vdc output voltage.

A07_2R5_@90kHz_1k-15k.png


It is the same measurement that I made with Hypex UcD180HG in another thread:
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,728
Location
Prague
AIYIMA A07, step response into 6.8 ohm load. With 8 ohm, the overshoot would be even higher. The overshoot indicates to worsening stability with higher ohmic load. This is why we make square and step response tests, as @restorer-john knows, but many others do not.
If the load is removed, the amplifier stops working. It starts to work again when the load is re-connected. It might have been interesting to make the same test with Topping PA5.

A07_step_6R8.png
 

friarminor

Member
Joined
Sep 16, 2021
Messages
36
Likes
17
Doesn't have a clue on what the measures practically mean but amazed at the diligence of using metrics to an otherwise realm of 'personal auditory preference'. I am on the market for a new amp (checked CA CX series, Rega and Musical Fidelity) and never felt more paralyzed yet at same time taking comfort that I'm not spending in vain.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,728
Location
Prague
AIYIMA A07 frequency response into 6.8 ohm, 10 ohm and no load ("open" circuit). Please note peaking with higher load impedance (alternatively seen as overshoot in my previous post). Please note the +27dB peak with no load. These are resistive loads only. If there is a complex impedance, the peak may travel down from 50kHz to some 15 kHz, as was already shown earlier.
The peak with no load indicates big instability and the amp switches off itself at higher output level.

A07_FR_6R8_10R_infinity.png
 

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,755
Likes
16,215
AIYIMA A07 frequency response into 6.8 ohm, 10 ohm and no load ("open" circuit).
Could you please confirm if with a high ohmic value load (like for example used as a preamplifier) it behaves the same as with "no load"?
 
Last edited:
  • Like
Reactions: pma

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,585
Likes
38,284
Location
Gold Coast, Queensland, Australia
That "no load" response would destroy piezo tweeters...

PS, Pavel what output level (1W/5W/full power) is the FR plot above taken at?
 
  • Like
Reactions: pma

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,585
Likes
38,284
Location
Gold Coast, Queensland, Australia
Could you please confirm if with a very high load (like for example used as a preamplifier) it behaves the same as with "no load" which actually is "infinite" load?

Nobody would use the A07 as a preamplifier, surely.
 
  • Like
Reactions: pma

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,755
Likes
16,215
Nobody would use the A07 as a preamplifier, surely.
Yes, that is very unlikely, still would like to see it though for example for the case it is used to drive a pair of passive loudspeakers and active sub(s) in parallel through their high input and due to some reasons the passive loudspeakers get disconnected.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,728
Location
Prague
Could you please confirm if with a high ohmic value load (like for example used as a preamplifier) it behaves the same as with "no load"?

This "no load" is in fact 20 kohm, input impedance of my measuring balanced divider.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,728
Location
Prague
That "no load" response would destroy piezo tweeters...

PS, Pavel what output level (1W/5W/full power) is the FR plot above taken at?

It was about 20W for 6.8 ohm and 10 ohm. With "no load", I had to reduce the output voltage level (let's not speak about power in such case ...) of some 20 dB, for the understandable reason - the amplifier was shutting down during the resonant peak. This reduced level is also the reason why "no load" frequency response is erratically noisy above the resonance.
----------------------------

P.S.: I hope everybody understands that this behaviour is not specific of AIYIMA A07 only. It may be generalized to all class D amplifiers with output LC filter outside feedback and to PFFB designs according to TI SLAA788A application report are not free from this issue.

1639122273423.png


1639122396365.png


Contrary to this, Hypex UcD and NCore have frequency response almost independent on load impedance.

It is a pity that more in-deep tests are not performed in ASR main reviews.
 
Last edited:

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,585
Likes
38,284
Location
Gold Coast, Queensland, Australia
It is a pity that more in-deep tests are not performed in ASR main reviews.

It would also appear these particular PFFB Class Ds would be unable to be switched on a comparator at medium/high levels as used in A/B tests due to the momentary shutdown when the break before make relays operate.

I would really like to see typical usage scenarios and fault conditions investigated more. Amps need to be overdriven massively, have DC injected, have short circuited speaker leads and pulled out cables.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,728
Location
Prague
Contrary to this, Hypex UcD and NCore have frequency response almost independent on load impedance.

I have made the same measurement as in post #106, now with UcD180HG amplifier. Output level was 12V, load 6.8 ohm, 10 ohm and 20 kohm ("no load"). The plot covers same 53 dB range in Y-axis, to make a fair comparison. All the 3 responses overlap, UcD180 frequency response is independent of load impedance. That's a big difference compared to TPA32xx and all similar chips, regardless if in PFFB or no PFFB mode.

UcD180HG_FR_6R8_10R_20k.png
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
This (passive) PFFB is just a buzzword. At high frequencies the feedback factor is very small and thus the open-load peak is damped only a little bit, if any.
Best way to mitigate is use the highest possible LC-resonance and damp it down (convert to heat) with a substantial snubber -- seldom seem in the commercial amps which usually stick with the datasheet recommendation.
 

ogonex

Member
Joined
Sep 24, 2021
Messages
69
Likes
27
Ta (pasywna) PFFB to tylko modne hasło. Przy wysokich częstotliwościach współczynnik sprzężenia zwrotnego jest bardzo mały, a zatem szczyt przy otwartym obciążeniu jest tłumiony tylko nieznacznie, jeśli w ogóle.
Najlepszym sposobem na złagodzenie jest użycie najwyższego możliwego rezonansu LC i wytłumienie go (przekształcenie w ciepło) ze znacznym tłumikiem – rzadko pojawiają się we wzmacniaczach komercyjnych, które zwykle trzymają się zaleceń podanych w arkuszu danych.
especially measurements on a clean resistor :)
 

julbo

Member
Joined
Jan 10, 2022
Messages
33
Likes
33
Location
Mediterranean
I learned tons of stuff with your thread @pma, many thanks really for your effort and patience :D
Ordered a Picoscope as my Hantek handheld is barely 6~7bit ENOB with my latest measurements and does not have the wealth of app support.
Learned that ARTA tools are a real gem.
Crap, should have read this before ordering an AIYIMA A07 :facepalm: At least I'll remeasure and confirm your findings, although I don't have access to sensitive RT spectrum analyzer.
Maybe I'll manage to eq the hell out of it or compensate in the tweeter xover, in the hope that I might salvage something out of this
Do we expect the hailed Toppings PA5 to behave better given a presumed active PFFB ?
 
  • Like
Reactions: pma
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,728
Location
Prague
Do we expect the hailed Toppings PA5 to behave better given a presumed active PFFB ?
Thank you :). Yes we do expect for Topping PA5 to behave better (however not perfect) for the reason that a similar issue is pushed higher in the frequency axis. This is re Zout.

Talking about class D generally, even the best performers Purifi and Hypex are not free of high frequency nonlinearity issue (yes >20kHz distortion components, to be clear).


Talking about AIYIMA A07, it will end up in a dustbin soon. Though I use it only occasionally mostly for tests, and it has not operated more than a couple of hours summed, the right channel started to emit high frequency hiss noise time after time, without any special reason, even if nothing is connected to the input. What would one expect at $90, it has done its job on my test bench. Not recommended, by myself.
 
Last edited:
Top Bottom