Part I: Fosi Audio V3 Mono Amp; Class D + PFFB, TI TPA3255 - Retail box with single 48V/10A power supply and filter. [With initial measurements.]

Over the last few months, I've been hearing a lot about this Fosi Audio Mono V3 (~US$140) amplifier which is typically sold as an individual unit with 48V/5A power supply, or as dual amps with 48V/10A supply, DC filter, and splitter cable (currently ~US$280).

I have already seen a number of reviews and YouTube videos typically with the pre-release product supplied by the company; honestly guys, many of the reviews seem superficial with essentially "Trust me, bro!" opinions rather than presentation of facts. For this review, let me grab the 2 amps + single 48V/10A kit from the Amazon retail pipeline to have a look at the final product. This review is not sponsored by Fosi Audio. With measurements, I believe this would be as bias-free as a review can be.

As you might already be aware, this inexpensive amplifier represents an evolution of their TI TPA3255 Class D "chip amp" product line. The company incorporated community feedback and developed this as a monoblock design (through Kickstarter funding, as usual due diligence needed when participating in crowd-funding) with no volume control so as to optimize resolution, accepts true balanced (XLR/TRS) input, and also incorporates Post-Filter Feedback (PFFB) to lower output impedance (increase damping factor) for load independence.

The company publishes a power rating of 240W into 4Ω, 1% THD; 120W into 8Ω. That should be more than enough for most reasonably efficient speakers of normal load, in normal smallish listening rooms.

Let's take a good look...

I. In The Box

Once I opened the outer generic cardboard shipping box, we see the actual black retail box above. Simple, classy with contents packaged well among the foam protection:

With all the pieces out, we see the large 48V/10A power supply, the 2 monoblocks, some wires including the splitter cable (upper right), and there's a smaller metal DC filter box (lower right):

That DC filter box (you can get this separately from places like AliExpress) I believe was included after some early feedback that the single power supply unit could pick up noise. Presumably the idea is that there's some kind of noise coming from the power supply DC output causing the complaint?

Small blue LED on the other side when plugged in. Weight = 1.1kg.

That's the beefy 48V/10A power supply above. In principle the splitter will feed the equivalent of 5A for each monoblock as needed. The idea here presumably is that you can save some space and use up fewer power outlets with this single box rather than 2 power supplies.

And there's the amplifier. It's pretty cute, all metal, matte finish which doesn't pick up fingerprints too easily, and I think looks quite elegant with the slightly tapered front top and bottom edges given the price. Dimensions are about 14.5cm x 10.5cm x 3.5cm, weighing just over 500g each.

There are a couple of switches up front - ON/Auto/Off for power control. The 'Auto' setting puts the device to sleep after 10 minutes without a signal - front amber power LED will go dim. The RCA/XLR switch obviously for which input to use.

Out back, we have RCA input with selectable +25/+31dB gain levels. This is convenient if you're feeding a low-level signal from a pre-amp or DAC with volume control. With the rated 240W into 4Ω 1%, we're looking at maximal 31Vrms so if using the +25dB gain setting, the source should be able to provide up to 1.75Vrms. With the +31dB gain setting, at least up to 0.9Vrms. 

Whenever we increase gain, we potentially increase noise. Thus for highest resolution, logically one should keep it at the +25dB setting if using RCA. Most modern source devices (DAC, preamp, streamers...) should have no problem providing at least 1.75Vrms output.

Note that the gain switch applies only to the RCA input. We'll measure the balanced input gain later.

I like the convenient balanced combo input which accepts both XLR and 1/4" balanced TRS connectors. The multi-way speaker jacks are of typical quality for this class of inexpensive amplifiers. They work fine even if not the most robust and accepted banana plugs firmly. On the right side of the connectors is the barrel connector to the power supply; up to 48V in.

A look at the other side of the box; it's symmetrical with orange round side ports for airflow. The device does get hot with prolonged high-power usage when running test tones and such. In normal use with music, it gets warm to touch but I didn't find it uncomfortable.

Let's open this up and have a look at the heat-sink situation: 

The enclosure is easy to open up. Just grab a hex screw driver and remove 2 screws from the back which will release the front panel. Then remove 2 shorter screws from below that hold the mainboard down to the casing and it'll slide out.

Fosi advertises the use of higher quality components - Nichicon, WIMA, and ELNA caps, Sumida inductor.

Fosi Audio V3 Mono product material.

The enclosure was intended to be opened up by the end-user in that there's mention made about the ability to do your own op-amp "rolling". Mid left of my picture of the mainboard, we see the two NE5532 opamps in a row which can be switched to alternates like Sparkos SS36XX, NJR MUSES02, OPA2604, Burson variants, etc. Based on Fosi's FAQ, they don't have to be the same type of opamp and if you want to switch only one for most impact, they recommend the more middle/inner one. 

[FYI, my previous explorations into using MUSES02 with a DAC suggested minimal effect, so personally I would not be tempted to fool around with this. In principle, if an op-amp switch made a significant difference, we should be able to show an objective difference in the DACs. The NE5532 is already a fine opamp for hi-fi audio.]

Notice a third op-amp bottom right near the RCA input. This is for conversion of unbalanced RCA to a balanced signal to feed the amp. I don't think there's much discussion about switching out this part. We'll see next time when I measure distortion the effect of using this RCA path instead of the balanced input.

Flipping the exposed mainboard to have a peek at the underside, we can see that there's a metal plate between the TI TPA chip and case. Notice the thermal compound that aids with heat transfer. Feel free to grab some non-conducting thermal compound like the Arctic MX-4 and expand that coverage area for even better heat spreading.

Let's remove the 2 screws holding the metal plate for a look at the TI TPA3255 chip:

There she is! Notice how small it is compared to everything else. I rubbed off the thermal paste to get that picture of the chip, obviously will reapply some fresh paste when reassembling.

A note about reassembly. The front power LED might lose alignment with the faceplate hole where the light shines through; you might need to gently move the LED around and turn on the unit a few times to see if the LED is shining through with good alignment before screwing that front plate back into place.

The picture above gives you an idea then what this all looks like when connected:

Dual mono amplifiers ← Splitter cable ← Filter box ← Power supply

Cable length is about 3' from power supply to wall and 4.5' between power supply to the filter. Shorter 20" splitter cable between the filter box and 2 amps so it's likely that filter will be sitting on your equipment rack or table just behind the 2 amps.

Basic power measurements for the above set-up show 18W both channels when turned on and idle. In "Auto" mode when asleep, this drops to <10W. Class D devices are obviously highly efficient.

Let's talk about the "noise" that the DC filter is supposed to remove. I think this is something to do with either humming or perhaps ground noise that is transmitted across, heard when one of the amplifiers is turned off.

Alas, I noticed this noise with my retail unit even with the DC filter installed. For my set-up using the RCA pre-outs from the Yamaha A/V receiver connected by HDMI to the MiniPC as multichannel Roon endpoint, I can still hear some odd noises coming out of the contralateral speaker when I turn one of them off.

It doesn't happen with the balanced input. It's picking up ground noise through the RCA jack. While an annoyance, so long as you keep both amps on (or auto) as intended, it should not be an issue.

For reference, here's how the noise sounds like:

Archimago · Fosi Audio V3 Mono - Noise Using RCA In, 48V - 10A PS, DC Filter

Thankfully, it's not too loud and a minor annoyance since I always have the 2 amplifiers on and there's no problem when used with balanced input (let's just say the balanced input is much preferred as we'll talk about next time). I suspect the susceptibility to and amount of noise can be idiosyncratic for each system, so YMMV whether you experience this issue. If you're likely going to be annoyed by this and must use RCA input, the only reasonable solution is to buy the individual amps with separate 48V/5A power supplies rather than the single shared 48V/10A version.

Let's move on and get some initial measurements of these amplifiers.

Fosi Audio V3 Mono: Testbench torture time! 😨

II. Volume Gain, Oscilloscope and Switching Noise

With the amplifiers on the testbench, let's check a few core parameters and have a peek at the Class D switching noise.

First, I measured the volume gain and got the following:

RCA (+25dB setting): +25.5dB
RCA (+31dB setting): +31.0dB
XLR/TRS input: +21.5dB

The RCA +25 and +31dB gain settings are pretty much as expected. With a +21.5dB gain on the XLR, 2.6Vrms balanced input would adequately push the amp to full rated power (should be no problem with modern balanced DAC and preamps typically hitting 4V).

As a Class D switching amp, we should see some low-level very high-frequency noise on the digital oscilloscope:

Yup, high frequency stuff can be easily seen. Nice level balance between the two amps with only 0.01V difference on the 1Vrms 1kHz sine. Nothing surprising on the 2V 5kHz square wave.

While I usually use my Rigol oscilloscope to grab a high-frequency FFT, just as easily I can turn on the FFT function with the digital scope and get a reading on that switching noise. As a reference, I'll play a 50kHz tone:

The switching noise is sitting at just over 600kHz. This is typical of TI TPA devices as can be seen also with the Topping PA5 Mk II+. With no signal playing, that switching noise is around 100mVrms; nothing to worry about and normally filtered out through the inductance in your speakers.

III. Output Impedance and Damping Factor

Implementing Post-Filter Feedback (PFFB) means there's external feedback applied to the amplifier chip output. In principle, this will correct errors in distortion (again, which we'll measure in Part II), but more importantly, we should see a reduction in output impedance across the audible frequencies compared to a TI TPA design without this feature.

For more information, have a look at TI's Application Report on this dated 2017. Obviously the use of the PFFB is not a new idea. The Topping PA5 Mk II+ amplifier, discussed last year I believe implements its own PFFB circuit in order to improve the output impedance of their design.

As you can see, the AIYIMA has increasing impedance particularly >10kHz although most of the spectrum sits around 0.1Ω.

There you go. We can see the effect with the PFFB compared to the earlier-generation AIYIMA A08 Pro measured about a year ago. This low output impedance translates to excellent damping factor into a 4Ω load of >100x average across the audible spectrum:

Note the small increase in output impedance (and reduction in damping) above 10kHz, no big deal.

IV. Frequency Response

With low output impedance, this therefore implies that we should see excellent load invariance including flattening out of the frequency response connected to complex speaker loads:

Very nice, Fosi Audio. I'm seeing -0.25dB down at 20Hz and at most -0.9dB up at 20kHz using a typical hi-res DAC like the SMSL DO100 (dual ES9038Q2M) with its flat frequency response. Notice I've included the frequency response of the AIYIMA A08 Pro which does not incorporate PFFB to the TI TPA3255 design as a comparison when connected to the Sony speaker. Notice with these speakers, it's not a massive change in frequency response. We're talking less than 0.5dB difference at 20Hz and around 1dB at 20kHz. Might not even be noticeable even though objectively easy to measure so don't freak out if you already have a good TPA3255 amp without PFFB (I see that Erin tried an ABX with the Fosi ZA3 and was unable to differentiate, don't know which speakers).

For reference, here's the impedance, phase and EPDR of the Sony speaker:

Sony speaker impedance (~6Ω nominal), phase
and EPDR (down to ~3Ω).

Let's have a closer look at the frequency response and phase characteristics of the Fosi V3 Mono connected to the Sony SS-H1600 bookshelf:

Good flat frequency and phase response. Note however that the phase graph sits at -180° which means inverted polarity (this is with balanced input). Looking around, I see that indeed these early retail units shipped before August 2nd, 2024 have inverted polarity which has since been corrected supposedly.

V. Summary, Part I

I think this post is long enough, let's stop here for now. 🙂

I trust this has been a good initial overview of the hardware including a peek inside the box. As you can see, this is the retail Fosi Audio V3 Mono amp x 2 with 48V/10A power supply kit being tested. You can also purchase the single mono amplifier with 48V/5A version.

While the name is similar, this is not the older Fosi Audio V3 stereo nor the ZA3 stereo/mono versions so just be mindful of these variants. All these amplifiers are based on the TI TPA3255 Class D chip.

There are certainly many things to like in this dual mono box design. I like the adjustable RCA gain and especially given the price, I think the build quality is very good. While this package includes a DC power supply filter, notice that noise was still seeping through when one of the amplifiers is turned off while connected with unbalanced RCA input. Perhaps there were various types of noises; for example, maybe the DC filter was able to block switching noise from the power supply itself?

Despite that anomaly, I didn't notice a problem in daily use when both devices are connected and put into "Auto" shutdown mode after 10 minutes. The obvious solution is to opt for two mono units with separate 48V/5A power supplies if this is a concern for you when using RCA in.

Something that irks me with some amps is the speaker "pop" when turning the amp on/off. I do notice a turn-on "click" with the Fosi V3 Mono but it's minor and there's no sound when turning off. The relay "click" from the amp itself is often louder than what's coming from my speakers.

In these initial measurements, we can easily see the benefit of the Post-Filter Feedback (PFFB) with low output impedance and hence good load invariance. Compare the results here with earlier generation TI TPA3255 amps like the AOSHIDA A7, Fosi Audio TB10D, and AIYIMA A08.

Since I was early in line for one of the first 2-channel retail boxes shipped before August 2nd, as you can see, the amps show an inverted polarity (said to be fixed after August 2). This should not be a problem when both amps are the same polarity (audibility discussed a few years back) but be mindful of this possible issue if you're mixing and matching with other amps.

[Hmmm, did any of the many reviews of the pre-release units pick up this polarity inversion over the months, was this only an issue with the early batch of retail boxes!?]

Already, I've spent a number of evenings listening to music through these amps and can say it sounds very good with balanced input. Next time, in Part II, let's explore the output power, resolution/distortion characteristics, and speak more about subjective listening impressions.

--------------------

I see that Pacific Audio Fest 2024 is happening September 6-8. This show is the North Western sister of Capital Audio Fest (coming November 8-10 in Rockville, Maryland) and Southwest Audio Fest (next March 21-23, 2025 in Dallas).

This is their 3rd year and I noticed the dates have been pushed later each time. This time of the year could be tight for me given the start of the university academic year at work, plus kids back to school. Hopefully it's alright for others. If I can swing it, will see if I can pop by this year for a listen and check out what's new in the world of "high end".

As we wind-down August (already!?), I hope you've been enjoying some decent weather and nice music. Cheers, audiophiles! 😃