AIYIMA A20 Stereo 2.1 Amplifier Review

[obsessive9119]

I understand, thank you for sharing this information. Also, what kind of operational amplifiers (OP-AMPs) are you using? Are they stock or something else?

I’m firmly in the “don’t waste your time” camp wrt opamp rolling. See https://www.audiosciencereview.com/...parkos-ss3602-opamp-rolling-in-fosi-p4.61946/

 

[RafaRafita]

I understand, but 2 x Sparkos exceeds the value of the entire amplifier? Does it actually work?

 

[staticV3]

I understand, but 2 x Sparkos exceeds the value of the entire amplifier? Does it actually work?

Sparkos doesn't do anything, but sunk cost fallacy and confirmation bias will make you hear an improvement regardless.

It's the only reason why the brand still exists. There is no objective merit.

 

[RafaRafita]

Intelligent answer. Thank you very much. I wouldn't have bought it anyway.

 

[Head_Unit]

It is amazing how far we have come in this category of amplifier.

Yessssss...but none of these things have a remote control? Not all of these amps are dirt cheap, and I wouldn't think everyone wants to use these on a desktop.

 

[georgeT]

nice, my fosi V3 stereo should arrive next week, fml. though it was only $90 shipped with 48v power supply included.

 

[caught gesture]

Any experts at reading measurement graphs care to comment on this from Aiyima. Does it show a 12dB/octave 2nd order filter? There seems to be an unresolved conflict of opinion as to what HPF this amplifier has.

 

[hipoagu]

View attachment 493901
Any experts at reading measurement graphs care to comment on this from Aiyima. Does it show a 12dB/octave 2nd order filter? There seems to be an unresolved conflict of opinion as to what HPF this amplifier has.

That is 12db/octave. Basically you measure how response varies as you duplicate or half the frequency. In this picture the response of the filter at 100hz is -13dB and at 50hz(half of 100hz) it is -25dB. Hence the slope of this filter is 12dB/octave.

 

[caught gesture]

That is 12db/octave. Basically you measure how response varies as you duplicate or half the frequency. In this picture the response of the filter at 100hz is -13dB and at 50hz(half of 100hz) it is -25dB. Hence the slope of this filter is 12dB/octave.

That was my reading of it as well. It is just that other people are insisting that the HPF is 6dB/octave based on their own measurements or interpretation of Amir’s measurements.

 

[NTK]

That was my reading of it as well. It is just that other people are insisting that the HPF is 6dB/octave based on their own measurements or interpretation of Amir’s measurements.

This is Amir's measurement. The - 3dB point is ~80 Hz. At 1 octave higher (~160 Hz), the attenuation is about -1 dB.

If you look at the graph from Aiyima, the -3 dB point is ~200 Hz. At 1 octave higher (~400 Hz), the attenuation is close to 0 dB. It is not the same filter as the one Amir measured (after considering the change in the filter cutoff frequencies).

 

[caught gesture]

This is Amir's measurement. The - 3dB point is ~80 Hz. At 1 octave higher (~160 Hz), the attenuation is about -1 dB.

If you look at the graph from Aiyima, the -3 dB point is ~200 Hz. At 1 octave higher (~400 Hz), the attenuation is close to 0 dB. It is not the same filter as the one Amir measured (after considering the change in the filter cutoff frequencies).

On the Y axis Amir measures RMS Level, on the Aiyima measurement it is Relative Level. Does this make a difference? I’m having a hard time finding out what the actual filter is in my amplifier with this conflicting information as to whether it is a 1st or 2nd order filter.

 

[NTK]

On the Y axis Amir measures RMS Level, on the Aiyima measurement it is Relative Level. Does this make a difference? I’m having a hard time finding out what the actual filter is in my amplifier with this conflicting information as to whether it is a 1st or 2nd order filter.

No. dB is a relative measure. For example, a drop of 3 dB means the signal power is reduced by half, whether it is a drop from 0 to -3 dB, or a drop from 100 to 97 dB.

 

[Mark185]

I performed my test measurement of the A20 acoustically using a Umik1 microphone and Room EQ Wizard software. This means I was measuring both the roll off of the speaker and the electrical roll off of the A20.

Aiyima does not indicate exactly what the A20 amplifier is driving (or loaded with) in their test graph. If the A20 is driving a closed enclosure speaker, that by itself will have a 12 dB/octave roll off at low frequencies. When I performed my measurement, it was with the amplifier driving small sealed speakers which had 12 dB/octave roll offs due to physics. I arrived at my 6 dB/octave measurement by comparing test data with and without the high pass filter switch engaged on the front panel.

I have found Aiyima to be a wonderful company to deal with as a customer with other products I purchased so this discrepancy is baffling.

 

[caught gesture]

@AIYIMA can you please explain your test protocol in reference to your HPF claim so that we, your customers, are absolutely clear as to what the parameters are of the product we have purchased.

 

[Mark185]

I posted new set of measurements on the other thread

Aiyima’s new balanced amplifier- A20

I also want to refer to the HPF function here... it's a big advantage for 2.1 desktop setups (mainly made using small monitors with low bass performance coupled with a medium size subwoofer). HPF could make the big difference in this scenario. Such feature is not something to be overlooked...

www.audiosciencereview.com

I think I should have been posting there from the beginning. I am not trying to compete with Amir. I am just posting my user experience.

 

[Mark185]

@AIYIMA I owe you an apology about claiming the high pass filter is 6 dB/octave. I am an electrical engineer and I should have known better than to publish a negative finding without thoroughly reviewing my test procedure, test data, and theory of filter design. It has been 40 years since I studied filter design and I forgot that the rolloff rate is not 12 dB/octave initially but that it converges to a final rate of 12 dB/octave about 2 octaves below the cutoff frequency. I also forgot about the effect of the 'Q' factor.

My mistake was that I measured rolloff rate close to the cutoff frequency I set on my amplifier. Setting was 80 Hz and I measured from 80 down to 40 Hz. If you look at the plot below, you can see that the initial response varies considerably depending on 'Q' factor.

 

[Ropeburn]

For future reference: cutoff frequencies of filters are defined to be the -3dB point. Below (HP) or above (LP) that is where the actual slope is.

So for example if you look at a -3dB point of a highpass that is set to 100 Hz, and then it's close to 0dB at 200 Hz (one octave higher), you can't conclude anything from that - because you're looking into the wrong direction.

You have to look at the slope below the -3dB point. It's easily visible how gently rounded the filter starts attenuating towards that point, and only reaches proper slope steepness below that.

This is perfectly normal analog filter behaviour. If you're unfamiliar with it, probably because you're used to modern digital filters showing different behaviour like steeper slopes, edgier "knees", logically more perfect behaviour and such, it's easy to fall into a mental trap and think the classic analog one is somehow behaving wrongly. It isn't.

 

[Mark185]

For future reference: cutoff frequencies of filters are defined to be the -3dB point. Below (HP) or above (LP) that is where the actual slope is.

So for example if you look at a -3dB point of a highpass that is set to 100 Hz, and then it's close to 0dB at 200 Hz (one octave higher), you can't conclude anything from that - because you're looking into the wrong direction.

You have to look at the slope below the -3dB point. It's easily visible how gently rounded the filter starts attenuating towards that point, and only reaches proper slope steepness below that.

This is perfectly normal analog filter behaviour. If you're unfamiliar with it, probably because you're used to modern digital filters showing different behaviour like steeper slopes, edgier "knees", logically more perfect behaviour and such, it's easy to fall into a mental trap and think the classic analog one is somehow behaving wrongly. It isn't.

That is exactly my mistake. I have been living in the digital world for a long time and forgot what I was taught when I was starting out.

 

[Ropeburn]

That is exactly my mistake. I have been living in the digital world for a long time and forgot what I was taught when I was starting out.

I made a loosely related mistake: when the slope thing came up, I took the -6dB claim at face value and didn't check anything, then proceeded to condemn the filter as useless and more or less Aiyima as lying.

Bad move, I should've known better and check first.

Meta-self improvement: don't just look at a mistake, but especially at how you arrived at it. Then try changing the way of thinking that led you there. It'll help avoiding a lot more future mistakes in general.

 

[NTK]

I made a loosely related mistake: when the slope thing came up, I took the -6dB claim at face value and didn't check anything, then proceeded to condemn the filter as useless and more or less Aiyima as lying.

Bad move, I should've known better and check first.

Meta-self improvement: don't just look at a mistake, but especially at how you arrived at it. Then try changing the way of thinking that led you there. It'll help avoiding a lot more future mistakes in general.

Regardless of how you and @Mark185 feel, I stand by my assessment that the HPF filter in the unit Amir tested was first order. The cutoff frequency was set to 80 Hz.

The HPF response was textbook first order response. Yes, you can adjust the Q of a second order analog filter to get close to this response, but the cutoff frequency (as defined by the filter poles) will not be 80 Hz. Only a first order filter or a second order Butterworth ( Q = √(1/2) ) will give -3 dB attenuation at the cutoff frequency, and I am possitive that this response is not second order Butterworth.