Topping A90 Headphone Amplifier Review

[JustAnandaDourEyedDude]

I recall reading somewhere that with balanced (differential?) amplification, it is possible to output say L+ and L- that are phase-inverted relative to each other, and are referenced to amp ground, thus doubling the voltage swing. Though, not having much background in electrical or electronics engr., I cannot say that I really understand what this means.

I should stop posting about things I know little about. My Likes count is trailing my Posts count by quite a bit. At this rate, I am never going to catch up with AmirM.

 

[Tatteredmidnight]

I should stop posting about things I know little about. My Likes count is trailing my Posts count by quite a bit. At this rate, I am never going to catch up with AmirM.

There you go buddy

 

[JohnYang1997]

@JohnYang1997 Congrats on this stellar design! When I was on the Apos website, I was blown away by one of the pictures listing specs. It was showing maximum output voltage swing of 49Vpp in balanced mode. This seemed ground breaking for an integrated solution. Now that I see the PCB showing the power supplies, I don't get it. The TPA6120A can only swing +/-12.5Vp (assuming symmetry which it slightly isn't) on +/-15V rails. Each chip looks to have its outputs in parallel through 3.3 ohm resistors (not sure how that drops down to 0.1 ohm on the output but must be due to the feedback loop). Each chip is a phase of the output (L+,L-,R+,R-) so that is 25Vpp on the balanced output at max gain.

I'm not sure where the 49Vpp comes from. Are you suggesting I buy two and parallel the outputs? Sounds like a fun challenge but not what was meant here. I'm just curious what I'm not seeing to come up with that number.

Thanks!


Michael

It's been explained in previous posts.
There are firstly 4 tpa6120 chips, 8 channels of tpa6120s.
Also that's why it's called NFCA. If i only put tpa6120 in parallel it won't have this performance. Not even close.

 

[JohnYang1997]

Yes, but it's not 49Vpp so that's what I was curious about.

Bruh.... balanced out.....

 

[MalinYamato]

I recall reading somewhere that with balanced (differential?) amplification, it is possible to output say L+ and L- that are phase-inverted relative to each other, and are referenced to amp ground, thus doubling the voltage swing. Though, not having much background in electrical or electronics engr., I cannot say that I really understand what this means.

isnt the point in differential amplification not to refer to a fixed ground when arriving at the final output voltage level?

 

[JohnYang1997]

which wedding do you recommend?
SMSL M400 MQA or Topping D90 MQA?

The much higher output impedance about 1.2kohm of M400 won't work correctly with A90.

 

[ReaderZ]

The much higher output impedance about 1kohm of M400 won't work correctly with A90.

DX7 Pro RCA out to A90 should work well?

 

[JohnYang1997]

DX7 Pro RCA out to A90 should work well?

Use XLR in real world. It's much less hassle.

 

[Paolosnz]

The much higher output impedance about 2kohm of M400 won't work correctly with A90.

it does not work well? or from problems? I understand that D90 is the chosen one

 

[ferongr]

sorry, I meant T1.2

I'm one of those weird people that chose the T1.2 over other options at that price range and can drive them to comfortable listening levels with a smartphone. They don't need a lot of power.

Edit: Note that with headphones I like listening at conservative listening level.s

 

[JohnYang1997]

it does not work well? or from problems? I understand that D90 is the chosen one

The M400 output omits buffer after passive LPF. Hence large and reactive output impedance. It's not a poor design per se but not compatible in this case.

 

[MalinYamato]

The much higher output impedance about 2kohm of M400 won't work correctly with A90.

is it really so? isn't there an establish IEEE standard regarding impedance in interconnects?

 

[ReaderZ]

Use XLR in real world. It's much less hassle.

My speaker is taking up the XLR... Headphone is for fun and speaker is my main rig, and it only take XLR...

 

[HammerSandwich]

Nice! So how do you think it would perform using the balanced out's to powered monitors?

Better than the speakers would.

 

[MalinYamato]

I'm one of those weird people that chose the T1.2 over other options at that price range and can drive them to comfortable listening levels with a smartphone. They don't need a lot of power.

must be quite sensitive cans, yet I am sure those would sing better with an external amp.

 

[MWeston]

It's been explained in previous posts.
There are firstly 4 tpa6120 chips, 8 channels of tpa6120s.
Also that's why it's called NFCA. If i only put tpa6120 in parallel it won't have this performance. Not even close.

Okay my mistake then. The picture looks like their outputs are directly summed through resistors. I know they are still inside the feedback path but I thought they were just summed for more current output. I hope this is fun discussing the design.

 

[MalinYamato]

Okay my mistake then. The picture looks like their outputs are directly summed through resistors. I know they are still inside the feedback path but I thought they were just summed for more current output. I hope this is fun discussing the design.
View attachment 65454

impressive knowledge! I have a better idea than getting a A90 for myself, Ill let you solder one for me instead.

 

[JohnYang1997]

Okay my mistake then. The picture looks like their outputs are directly summed through resistors. I know they are still inside the feedback path but I thought they were just summed for more current output. I hope this is fun discussing the design.
View attachment 65454

Yes they are in parallel. But that's not just it. The whole nested feedback thing and composite amplifier thing.....

 

[thefsb]

Multiple measurements can be made from a single output. Here, the input to the amp is varied and the output waveform captured. One measurement extracted from this is the power. The other is distortion+noise (THD+N).

The input value is not shown because it is of less interest here. The sweep is from 10 millivolts to 4 volts from what I recall.

OK. That confirms my understanding. X is not the independent variable of which Y is a function.

 

[JohnYang1997]

is it really so? isn't there an establish IEEE standard regarding impedance in interconnects?

Majority of DACs have 100ohm for rca out and 200 for BAL out. This is fine. M400 omitted the output buffer after they choosing the passive LPF path. The impedance is about 600ohm per strand.