• 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!

Fosi Audio V3 Amplifier Review

Rate this amplifier:

  • 1. Poor (headless panther)

    Votes: 10 2.0%
  • 2. Not terrible (postman panther)

    Votes: 44 8.8%
  • 3. Fine (happy panther)

    Votes: 235 47.1%
  • 4. Great (golfing panther)

    Votes: 210 42.1%

  • Total voters
    499
Hello :)

Concerning OPAMPs, I am not going to address here SUBJECTIVE considerations that are often linked to a kind of 'myth', 'fashion', various opinions concerning listening or even commercial orders...

I will try to remain OBJECTIVE (my goal is not to convince but to inform) first by observing the circuit used and then by relying on the manufacturers' data concerning the OPAMPS.

The circuit used has capacitors at the inputs of the OPAMPS for two reasons:

- a bipolar OPAMP creates OFFSET if these input impedances are not identical then we use these 'ARTEFACTS' to remedy it, even if it does not correct the OFFSET totally...

- the manufacturer, thanks to this 'TECHNICAL PIROUETTE', will be able to offer a device at a much lower cost price to gain more margin (a NE5532 DIP8 costs about 0.50 euros at MOUSER)

That being said, another important parameter is the value of this input resistor (actually it's impedance).

Here is a comparative table from TEXAS INSTRUMENTS to observe the behavior of BIPOLAR OPAMP and FET according to the input resistance. It will be noted that after a certain value, it is obvious that a FET OPAMP will be preferable to a BIPOLAR OPAMP.

Comparison OPAMP BIPOLAR-FET.jpg


The additional advantage of an OPAMP FET is that it will not create OFFSET if the input resistors are not identical:

we can then do without capacitors, i.e. a 'simplified' circuit without having the various 'effects' of an additional component on the signal.

It can be added that a chemical type capacitor will not work properly if it does not have a minimum potential difference across its terminals, which is the case here while an 'MKP' type capacitor (e.g. WIMA MKS2) is perfectly designed for this use.

Here is another comparison table that shows the 'THD+N' according to the value of the input resistor ->

OPAMP Comparison.jpg


The reading being done, I leave you 'alone' to judge but I will simply indicate that the OPA1642 (double version of the OP1641) is modern and with a very reasonable cost while the OPA827 is of older technology (although excellent) and only exists in a single version, which implies using two of them on a 'special' adapter whose overall cost will be much higher.

So finally, here is the modified diagram as it should be after observing my remarks ->

DIAGRAM FOSI AUDIO V3 STEREO + MOD V2 (15 -12 -2024).jpg


I simply added 220pF capacitors 'close' to the OPAMPs (the slots are empty on the PCB) which allow for an improved FFT such as the one I proposed for the O-NOORUS D1 amp because the circuitry is comparable almost in every way.


Additional addition:

The OPA1656 is also a very good choice that remains just as affordable while in the following table the OPAMPs OPA2828 (double opa828) and OPA2156 are starting to enter a much higher 'price range'.

OPAMP.jpg


Have a nice day ;)
 
Last edited:
Hello :)

Concerning OPAMPs, I am not going to address here SUBJECTIVE considerations that are often linked to a kind of 'myth', 'fashion', various opinions concerning listening or even commercial orders...

I will try to remain OBJECTIVE (my goal is not to convince but to inform) first by observing the circuit used and then by relying on the manufacturers' data concerning the OPAMPS.

The circuit used has capacitors at the inputs of the OPAMPS for two reasons:

- a bipolar OPAMP creates OFFSET if these input impedances are not identical then we use these 'ARTEFACTS' to remedy it, even if it does not correct the OFFSET totally...

- the manufacturer, thanks to this 'TECHNICAL PIROUETTE', will be able to offer a device at a much lower cost price to gain more margin (a NE5532 DIP8 costs about 0.50 euros at MOUSER)

That being said, another important parameter is the value of this input resistor (actually it's impedance).

Here is a comparative table from TEXAS INSTRUMENTS to observe the behavior of BIPOLAR OPAMP and FET according to the input resistance. It will be noted that after a certain value, it is obvious that a FET OPAMP will be preferable to a BIPOLAR OPAMP.

View attachment 414096

The additional advantage of an OPAMP FET is that it will not create OFFSET if the input resistors are not identical:

we can then do without capacitors, i.e. a 'simplified' circuit without having the various 'effects' of an additional component on the signal.

It can be added that a chemical type capacitor will not work properly if it does not have a minimum potential difference across its terminals, which is the case here while an 'MKP' type capacitor (e.g. WIMA MKS2) is perfectly designed for this use.

Here is another comparison table that shows the 'THD+N' according to the value of the input resistor ->

View attachment 414099

The reading being done, I leave you 'alone' to judge but I will simply indicate that the OPA1642 (double version of the OP1641) is modern and with a very reasonable cost while the OPA827 is of older technology (although excellent) and only exists in a single version, which implies using two of them on a 'special' adapter whose overall cost will be much higher.

So finally, here is the modified diagram as it should be after observing my remarks ->

View attachment 414100

I simply added 220pF capacitors 'close' to the OPAMPs (the slots are empty on the PCB) which allow for an improved FFT such as the one I proposed for the O-NOORUS D1 amp because the circuitry is comparable almost in every way.


Additional addition:

The OPA1656 is also a very good choice that remains just as affordable while in the following table the OPAMPs OPA2828 (double opa828) and OPA2156 are starting to enter a much higher 'price range'.

View attachment 414115

Have a nice day ;)
This is excellent work. The additional 220pF COG capacitors will improve opamp stability but introduces an imbalance at 3255 inputs (0.3dB & 15 degrees @ 20KHz). I'm unsure of the impact of this imbalance on 3255 operation. The imbalance may be corrected by replacing C1/C4 with 1K/2200pF/10uF networks.
Dare I ask how your modifications 'sound'?
 
Hello.

Are you talking about imbalance?...

The circuit of the FOSI AUDIO V3 STEREO is identical to the O-NOORUS D1 that I also own and with the same modifications:

how does it sound ?

Better than some other Class-D amplifiers ->

Spectrum comparison - CLASS D AMPLIFIERS.jpg



Now, if this is indeed the case:

I know few people who are able to hear a difference of 0.3dB from 20KHz and even less a phase 'deviation' of 15°...

... except for a person with sonar in an old submarine or a bat :D

That said, you are right since it is customary to talk about a frequency reproduction range between 20 and 20000Hz in a (most) linear way (possible) when referring to a so-called 'Hi-Fi' device.

So as you can see on the graph, this is what was sought.

(Even the TOPPING PA5 does less well...)
 
Last edited:
Hello.

Are you talking about imbalance?...

The circuit of the FOSI AUDIO V3 STEREO is identical to the O-NOORUS D1 that I also own and with the same modifications:

how does it sound ?

Better than some other Class-D amplifiers ->

View attachment 414242


Now, if this is indeed the case:

I know few people who are able to hear a difference of 0.3dB from 20KHz and even less a phase 'deviation' of 15°...

... except for a person with sonar in an old submarine or a bat :D

That said, you are right since it is customary to talk about a frequency reproduction range between 20 and 20000Hz in a (most) linear way (possible) when referring to a so-called 'Hi-Fi' device.

So as you can see on the graph, this is what was sought.

(Even the TOPPING PA5 does less well...)
The imbalance is between the 3255 A & B inputs. Presently the A path is direct, the B path has an opamp lowpass filter. The lowpass filter causes the imbalance.
 
Hello.

Here is the 'MOD' represented (just LEFT Channel), where is the 'problem' according to you ?

FOSI AUDIO V3 STEREO (LEFT CHANNEL).jpg
 
Hello.

Here is the 'MOD' represented (just LEFT Channel), where is the 'problem' according to you ?

View attachment 414355
3255 input B has imbalance (-0.3dB, 15 degrees @ 20KHz) with respect to input A due to C46. The imbalance may be compensated by changing C1 to a 1K/2200pF/10uF network. The opamp (1/2) stability should be verified in SPICE with the network attached before making the change. I'm unsure if the network will fit in the space available. If there is space I can give details of network connections.
 
Last edited:
Hello.

Have you carried out a simulation to advance this ?

Can you publish it please, it's interesting ;)

However, if C46 creates this 'imbalance', wouldn't it be easier to remove it (C46) rather than create an additional network to 'compensate' for it ?
 
Hello.

Have you carried out a simulation to advance this ?

Can you publish it please, it's interesting ;)

However, if C46 creates this 'imbalance', wouldn't it be easier to remove it (C46) rather than create an additional network to 'compensate' for it ?
Why was C46 added?
 
Hello Trevor.

Thank you for intervening to raise this issue and I just realized my mistake due to too much eagerness on my part to share the 'MOD' :(

So here is my error corrected and I simply ask you to 'validate' that it restores the balance for each entry ;)

Only LEFT channel ->

FOSI AUDIO V3 STEREO (ERRATUM).jpg


Both channels ->

DIAGRAM FOSI AUDIO V3 STEREO + MOD (CORRECTED 17 -12 -2024).jpg


Once again I apologize for this (basic) mistake and thank Trevor again for his important intervention :cool:
 
Hello Trevor.

Thank you for intervening to raise this issue and I just realized my mistake due to too much eagerness on my part to share the 'MOD' :(

So here is my error corrected and I simply ask you to 'validate' that it restores the balance for each entry ;)

Only LEFT channel ->

View attachment 414607

Both channels ->

View attachment 414608

Once again I apologize for this (basic) mistake and thank Trevor again for his important intervention :cool:
C39 & C41 are useful, in that they limit the input noise bandwidth (even more so than R21/C64 & R25/C65) without any negative effects. I would keep C39 & C41.
 
Hi.

The R21/C64 & R25/C65 sets are used to avoid 'receiving' HF emissions in some countries that can transmit in 'radio frequency' quite close to these frequencies, for other countries they are indeed totally useless.

That said, you didn't answer me about the correct balance for each input of the TPA3255 chip after my circuit correction (replacement of R9, R10, R15 and R16 from 10K to 15K).

DIAGRAM FOSI AUDIO V3 STEREO + MOD V3 - FINAL (17 -12 -2024).jpg
 
Hi.

The R21/C64 & R25/C65 sets are used to avoid 'receiving' HF emissions in some countries that can transmit in 'radio frequency' quite close to these frequencies, for other countries they are indeed totally useless.

That said, you didn't answer me about the correct balance for each input of the TPA3255 chip after my circuit correction (replacement of R9, R10, R15 and R16 from 10K to 15K).

View attachment 414692
The previous schematic eliminated C20 & C46, that was correct. Your new resistor values shown in orange are correct.
 
THX Trevor.

So according to your opinion, this last scheme allows for balance for all channels of the TPA3255, do we agree ?
 
Hello Trevor :)

Thank you for your answer ;)

Can you show us the comparative simulation of signals with and without these capacitors (C20 & C46) on curves (levels and phase) ?

Thank you in advance.

Kind regards.
 
I've kind of lost the thread of what you guys are trying to achieve with these mods, versus just using Equalizer APO or similar to slightly reduce treble in the 10khz+ area?
 
Hi Joe smith.

Do you think a similar APO Equalizer costs less ?
 
? The Equalizer APO and Peace software is free, I stream from PC to DAC to preamp to amp - so that works fine for streaming uses.

I guess I can see the need for PFFB type-tweaking if one's playing LPs or CDs and one has very sensitive/younger hearing (at 66, my hearing tops out at about 11,500). But for those type of sources, with Fosi's new P4 preamp that I am testing, there are nice tone controls that can be put -in-circuit to slightly drop treble if needed...I think that unit will be out in a month or two. Or, other preamps, of course. Or an in-line EQ box like the one from Schiit.

I personally do like the original Aiyima A07 amp, the A07 MAX and the Fosi V3 "as is" - with my hearing level, no compelling need to upgrade for the PFFB amps that are coming out now - it does seem though, given their more accurate FR, that these designs will replace this older group of amps fairly quickly for new customers...

I applaud you guys' initiative to modify the last crop of amps, but given the pace of change in this small D amp market, seems like the PFFB change pretty much solves the FR issues...now the next thing, I suppose, is improvement re all of the various user concerns over the power brick matching and reliability and improvement of real-use power ratings. I can see one of these folks, or O-Noorus, going further on the reliability/safety/accurate power ratings side of things...
 
Indeed, you are right about O-NOORUS that I know very well...
... Speaking of which, what do you think of their products and their 'attitude' towards their customers?
 
Last edited:

Hey quick question could anyone tell me what the input impedance of this fosi v3 amp is?

It's a glaring oversight that the input impedance is not specified in the manual or in Fosi's FAQ on their website.
The one clue we might have is that in the photo of the amp with its case open the volume control pot is 50k ohms (labeled "A50K", ganged, presumably audio taper).
Usually the resistance from signal input to signal ground sets the input impedance, but I don't know if that's the case with this design.

FWIW, the Fosi V3 Mono amps have input impedance of 25k ohms for the unbalanced (RCA) input and a very low 5k ohms for the balanced (XLR) input.
Perhaps the V3 stereo amp's input Z is similar, but Fosi isn't saying...
 
Back
Top Bottom