Part II: Fosi Audio V3 Mono Amp; Class D + PFFB, TI TPA3255 - Retail box with single 48V/10A power supply and filter. [Power, Distortion, and Subjective Impressions.] And on rejoicing over "Solved Problems" in audio.

Gotta love those Nordost-like flat speaker cables from AliExpress. The 90° locking banana connectors are a bit on the large side however. 😉

As we saw last week in Part I, I started reviewing the Fosi Audio V3 Mono amplifier, early August 2024 retail release, in the form of the stereo with 48V/10A power supply kit. We've already had an extensive look externally and inside the device, seen the benefit of PFFB for load invariance as well as discussed a couple of the issues I ran into (inverted polarity and the ground noise with 2 amps connected to one battery).

This time in Part II, let's proceed deeper into the measurements to explore the amount of power this little amplifier can provide, some details on distortion characteristics, and as usual, ending off with subjective impressions and summary of the product.

I. Power

With a single 48V/10A power supply for 2 channels, in theory we could achieve 480W total. Starting with a 4Ω load, let's see how much power the amplifier can produce with both channels going:

 

As usual, I set my "high-fidelity" threshold at 0.1% THD+N or -60dB (a reasonable level based on conservative blind test data a few years back). As you can see, this amplifier is able to push out about 200 "hi-fi watts" per channel; 400W total; not bad at all!

Let's plot the output level and look at the linearity performance:

Notice the clipping into -3dBFS input level on the right from the DAC. Linearity looks excellent even down to -120dBFS with less than 0.5dB deviation. This is a near-DAC-level performance.

We can measure the same into 8Ω:

Excellent. Almost exactly half of the number of watts across the doubled resistance load with around 100W into 8Ω, less than 0.1% THD+N.

Since I'm testing the dual-amp kit with single large power supply, I wondered what would happen if I used that 48V/10A brick to power one mono amplifier only without the cable splitter or included DC filter. Would we achieve any more power or lower distortion as compared to both channels driven?

With the single amp connected to the 48V/10A power supply (no DC filter or splitter cable), we see marginal improvement in the output power to 206W into 4Ω using the 0.1% THD+N threshold; an increase of 7W. And we can overlay the THD+N vs. Level graphs to show the slight improvement in resolution at higher power running mono vs. 2-channels driven:

I tested with and without the DC filter; was not able to measure a difference.

Since I don't have the Fosi 48V/5A power supply, I don't know if this would be the same as the 48V/10A driving a single amplifier. Regardless, the improvement in power is really just a handful of watts around the 200W mark, resolution improvement is mostly below -90dB THD+N (around 0.003%) and above 5.5Vrms (>7.5W into 4Ω). IMO this is not audibly significant.

Obviously, I would not loose sleep over this slight difference and agree with Fosi that a single 48V/10A power supply feeding two amplifiers is totally fine. (I suppose if you want to be extreme, you could buy another 48V/10A power supply and go dual mono with overspec'ed power supplies!)

II. Single-Tone Distortion

Moving along, let's look at the FFT with single-tone tests.

Here's the 1kHz THD+N for 2 channels driven, single 48V/10A power supply, 1W and 5W into 4Ω and 8Ω using balanced input (I used an XLR to 1/4" TRS cable I had handy from the SMSL D100 DAC to Fosi amplifier):

Average of better than -90dB at 1W and -98dB THD+N at 5W into 4Ω, both channels driven - clean low-level harmonics, very respectable results regardless of amplifier price.

Compared to the balanced input, here is the 2Vrms into 4Ω (1W), 1kHz tone using RCA inputs with +25dB and +31dB gain settings:

Notice the deterioration in THD+N/SINAD with unbalanced RCA input compared to balanced TRS. As expected, when pushing the gain higher from +25 to +31dB, we introduce more noise and distortion, hence higher THD+N. Specifically, we can also see that the unbalanced input picked up a bit of 60Hz powerline hum; for some reason worse in the left amp.

Clearly, for this amplifier, stick with balanced input (XLR or TRS) for better resolution; we're seeing a worsening of 10dB on the THD+N/SINAD using the RCA input. Having said this, it's important not to freak out because depending on your system, room, ears, this might still not be audible.

Here's a look at right and left channels 1kHz THD+N with a higher 8Ω load, 6.3Vrms or 5W:

About THD+N -96dB average with balanced input. Very good. This would handily beat many amplifiers out there of much higher cost.

 

Performance at 1kHz, while a reasonable proxy for an amp's resolution is just at one frequency. Let's have a look at the THD(+N) vs. Frequency graph - here's 5W into 4Ω:

No problem here as distortion stays significantly less than -60dB/0.1% across audible frequencies with typical low-pass filter at 20kHz. Notice though that as frequency increases above 2kHz, the odd harmonics (3rd and 5th) start creeping up.

Let's now increase the measurement bandwidth to 96kHz (192kHz samplerate), and take off that 20kHz low-pass filter so we can include ultrasonic harmonics. Let's have a look again at that THD+N vs. Frequency graph at different power levels from 1W to 100W into 4Ω:

Amplifier fed with balanced input.

Interesting eh? With the filtering in REW turned off (so that we can measure ultrasonic content like 15kHz tone, second harmonic at 30kHz), total noise level is increased, thus the higher THD+N result (which is why at 1W THD+N is only around -76dB instead of closer to -90dB). In particular, most interestingly, notice the frequency-dependent increase in distortion as power goes up.

Should we be worried? No, I don't think so. While this is interesting as an objective measure which future generations of chip amps I hope can improve on, it's IMO one of those objective findings that isn't audible yet I've seen some folks seem to be concerned. We might want to discuss this another time...

III. Multi-Tone Distortion

Moving along, let's have a look at dual-tone intermodulation distortions. Here's my typical panel with various IMD test signals (discussed here):

Quite a bit of data which we can get lost in, but no need to worry since this all looks pretty good; as you can see, all IMD results measure much less than 0.1%/-60dB. Feel free to compare this with the Topping PA5 Mk II+ and also the Hypex NC252MP for relative performance.

While the CCIF/ITU-R tone is very commonly measured and shown, it's primarily a marker of high-frequency distortions given the loud 1:1 19 & 20kHz tones and not realistic when playing music. Sure, it would be great to have perfect results in future generations of these Class D amps, however, the SMPTE and Linkwitz are more representative of distortions that occur within sensitive frequencies of human hearing. 

As a marker of how well the amplifier handles transients, let's have a peek at the TIM ("Transient InterModulation") test signal (96kHz bandwidth):

For reference: These squarish transients are what the TIM waveform
is asking the amp to reproduce cleanly without added distortion (24/192 signal).
More demanding that what the amp will be asked to reproduce with music.

Excellent. All distortion products below -110dB even at the higher 25W into 4Ω. That 10V TIM test is asking the amplifier to swing ~20V peak-to-peak cleanly in about 10.5µs.

Next, here's the 1/10 Decade Multitone 32 at 2V into 4Ω:

Notice excellent results with the balanced input with distortion level and noise around -100dB from tonal peaks in the audible range. Notice I've turned on the peak levels along with average to make sure there are no unusual spikes of noise across the 32-sample capture (about 2 minutes continuous complex signal). We've already seen above in Section II that RCA input isn't performing as well. Here again, the RCA input set at +25dB gain is significantly worse than with balanced XLR/TRS in.

Finally, time for the Triple-Tone Total Distortion and Noise (TD+N, 48/960/5472Hz) test which is what I've been using over the years as a single-number measurement for resolution comparisons among amplifiers. It's a more complex test signal (than a single 1kHz sine) that will show us both harmonic and intermodulation distortion patterns in the FFT. The frequencies chosen are in line with actual music (bass, mid, "presence" tones), and measuring at 2Vrms output into 4Ω, it's within the "first watt" level where most of the nuances we hear are to be found with loudspeakers of normal sensitivity (say ~85-95dB/2.8V/m) in typical small listening rooms.

In principle, I think an amp that's able to perform well on this test (something like -90dB would be excellent) would likely demonstrate very clean rendering of nuances and good low-level transparency.

Results above show very good resolution with TD+N right at -90dB when using XLR/TRS balanced input. 

Again, clearly the balanced in performance is superior to the RCA single-ended/unbalanced. Higher noise when using RCA +31dB compared to +25dB gain while the distortion pattern remains very similar. As expected, 60Hz powerline hum with harmonics picked up using unbalanced input.

IV. Subjective

I had a listen to the amplifiers on my system connected either as stereo amps directly fed by the balanced Sabaj A20d 2022 DAC (based on ESS ES9038Pro) output or as part of my multichannel system fed by the RCA pre-out from my Yamaha receiver. As usual, main speakers are the Paradigm Signature Reference S8v.3; Paradigm SUB1 subwoofer active with multichannel. I listened with and without either Acourate (stereo) or Audiolense (multichannel) DSP filter active.

[For those who have used DSP room correction, it's clear that this makes significantly more difference than changing hi-fi amps; don't forget to think in parallel "domains"!]

Orchestral Manoeuvres In The Dark's (OMD) Sugar Tax (1991, DR13) while not one of their canonical works (compared to Architecture & Morality or Organisation), is a fun synthpop guilty pleasure of mine from the early '90s. "Pandora's Box (It's a Long, Long Way)" (inspired by the story of Louise Brooks) is a favorite on this album. This song and the album in general sounds great with crosstalk cancellation (XTC) to give a wide soundstage through the Fosi V3 Monos. Excellent detail separation throughout these complex multilayered high tempo tracks.

Another track I like on this album is "Call My Name" with its very danceable '80s vibe contrasting with the melancholic lyrics. "Apollo XI" consists of a synthpop beat overlaid with samples from the first moon landing is cool as well.

As usual, another nice sweeping orchestral soundtrack from Ramin Djawadi, this time for House Of The Dragon, Season 2 (2024, DR10). While not as popular as Game Of Thrones, I've been enjoying this prequel spin-off. Let's hope the story arc develops well and doesn't conclude in the same way like GoT's disastrous final episode and 8th Season in general! I guess everyone's still waiting for George R. R. Martin to get the final books done.

Although I didn't think the Season 2 soundtrack was as good as Season 1 ("The Crown of Jaehaerys" still my favorite), there are some nice themes here like the final track "All Must Choose" with its lovely piano intro (nice rendering of attack and natural decay though the Fosi), the track builds in complexity with increasingly intense low-end later on. There's a fragile yet at times jarring violin part on "Remembering Those Who Came Before" that stood out among the tracks.

Check out "There Will Be No Mercy" for a propulsive track with excellent dynamics and some deep bass. Subwoofer recommended!

Virginia To Vegas' greatest hits (2024, DR5) is pure modern compressed pop confection. Derik John Baker hails from Toronto and we've got some very well done Top 40 tracks here like "We Are Stars", "Lights Out", "Just Friends", "Selfish" and the current one played here on Canadian radio "TIME!". Pretty catchy stuff with a Millennial vibe in some of the lyrics about relationships and challenges of that generation. Regardless, I think as humans we share more commonalities of experiences through the phases of life than differences among the generations so material like the nostalgic memories of our youth and anxieties of getting older on a track like "TIME!" certainly resonates.

The music itself is of the modern pop variety and to be honest, after awhile, all music with complex synths, a ton of plug-in effects, and Autotune on the vocals with high dynamic compression can end up feeling generic. The Fosi Audio amps were not challenged by content like this and rendered the (limited) dynamic range, complex layers with ease. Notice modern pop like this uses complex phasic effects which inherently expands the soundstage on speakers without needing further XTC DSP unlike many of the older albums like the OMD above.

Time for a modern rock/blues recording - Little Feat's Sam's Place (2024, DR7). I see that Stereophile had a review article on this album last month and gave the sonics 5-stars. I'm not sure I would be so generous but that's a subjective call since I would have liked the sound to have been a bit more punchy and dynamic but at least tonally I appreciate the richness of this recording.

Anyhow, if you enjoy traditional rock/blues, I would certainly recommend this album. The contrast and interplay between lead singer Sam Clayton's bass vocal range and guest Bonnie Raitt on "Long Distance Call" is nice along with great guitar work and harmonica. Love the energy and fun vibes on "Can't Be Satisfied" (by Muddy Waters) as well, very nice electric guitar part. Check out the studio video to that track.

The Fosi V3 Mono sounds well balanced and "rich". Some people claim that Class D can sound excessively "thin". Well, maybe years ago this could have been the case but not with modern Class D amplifiers capable of hi-fi performance (low noise, full frequency response, low distortion, excellent transient response); they're just neutral-sounding these days.

Lastly, let's talk about Post Malone's new one - F-1 Trillion [Long Bed] (2024, DR12 multichannel/Atmos). The "Long Bed" just refers to the extended edition with an extra 9(!) tracks. I integrated the Fosi V3 Monos to my 5.2.2 system (taking into account the polarity inversion with this early retail model, flip the red/black terminals to speaker). Interesting country crossover album for a guy typically known for his hip-hop, pop, R&B with many country genre collaborators here. Regardless, he still sounds like his usual Autotuned synthetic vibrato vocals with some country twang. 😏

A number of these I think will be popular, "Pour Me A Drink" (with Blake Shelton), "Guy For That" (with Luke Combs) and "I Had Some Help" (with Morgan Wallen) seem to be targeted for the radio. We've even got some Dolly Parton here ("Have the Heart"). "M-E-X-I-C-O" (with Billy Strings) and "Losers" (with Jelly Roll") are fun and catchy as well.

As a dad with a teenage daughter, I found "Yours" very touching.

We've got some pretty enjoyable immersive multichannel mixes here though not the best or most impressive I've come across. The Fosi V3 Monos I think are good multichannel system amps given the flexibility of just adding one here and there with auto-power-on for energy savings. It would certainly make a very fine, clean-sounding center channel amp for example.

Overall, I'd say my subjective impressions "check out" given the results we're seeing objectively. These simply sound like hi-fi amplifiers with clean output, produces excellent dynamics when called for, provides great transients, and tight bass within the power envelope. As usual, make sure to pair them up with good speakers (IMO, important hardware component).

Subjectively, I also thought the balanced input sounded superior compared to the RCA-in even before I did the measurements but that's not a blind-tested impression.

V. Summary

So, here's my summary score card for the Fosi Audio V3 Mono amplifier - specifically the 2-amps, single 48V/10A power supply, retail package:

These are quite powerful little mono amps that are able to push up to around 200W into a 4Ω load and around 100W into 8Ω with less than -60dB THD+N (0.1%). The post-filter feedback (PFFB) circuitry has done a great job maintaining load-invariance as reflected in the frequency response and damping factor of >100x into 4Ω.

So let me summarize with a few general "take home" points then:

1. For best performance, use balanced input. It's convenient having the RCA +25dB and +31dB gain input options though. However, as an audiophile looking for the highest fidelity, you're simply not getting the best of this amplifier if you use the RCA in.

2. If you have one of the early units (before early August 2024), be mindful that the device might have inverted phase/polarity. This is significant if you mix-and-match with newer Fosi V3 Mono amps and in multichannel systems mixing amps.

3. I like the 10 minute auto-sleep/turn-on function when signal present. Nice little power-saving feature without using 12V trigger wiring although there is a reassuring immediateness when hooked up with a wired trigger.

4. The sound quality is excellent with this unit. Even with 2 amps powered through one 48V/10A power supply, the measured power and distortions did not deviate significantly from what I saw when pairing a single amp with that 48V/10A power supply (with or without DC filter).

5. Unless you really need to save space with one large power supply or reduce the number of plugs used for the stereo channels, I suspect going with dual mono amp + dedicated 48V/5A power supply would be best. While I don't think the potentially lower distortion from having a dedicated power supply is significant, the issue with the ground noise (discussed in Part I) when one of the two amps is turned off when fed RCA input, even with the supplied DC filter, will be avoided. Another potential benefit is that you will have more freedom to place the amps further apart right next to your speakers. Using balanced cables to feed the amp and very short speaker cables would be excellent audiophile practice to minimize distortions and cable effects (not that I think there's usually any problem); back in the day, this is what I used to do with the Emotiva XPA-1L monoblocks and short lengths of Canare 4S11 cables.

6. I guess there's something psychologically cool about monoblocks which tells others "I'm an audiophile". Go for it if this turns you on. 😉 But seriously guys, these days good stereo amplifiers already can achieve plenty of power and crosstalk of less than -100dB; not that we even need that much.

7. While I don't consider this blog primarily a "review site", I'm of the opinion that if you don't need the full 200W-4Ω / 100W-8Ω power of the Fosi V3 Mono amp, you might want to consider the Topping PA5 Mk II+ (as reviewed previously). Although it provides 1/2 the power at 100W-4Ω/50W-8Ω, and it costs about the same, resolution is objectively higher, I think has better build quality, RCA input is less noisy, has a convenient volume control, doesn't have the soft "click" when turning on, and for me has been trouble-free since last year with regular use. The 12V trigger on the Topping works well.

[I bought the Topping amp just as I did this Fosi V3 Mono through standard retail channels so I'm not biased by any "freebies" from the companies. Just my impressions as an independent user.]

Although I have no direct experience, the Topping PA7+ amplifier is a dual-chip TI TPA3255-based device with company specs of 245W x 2 into 4Ω, 1% distortion. This is equivalent in power output to the Fosi V3 Mono, also with excellent measured performance. More expensive though. Consider this model as another option to consider if you're in the market for these kinds of amps.

Despite being tortured on the testbench for a few evenings to the point of over-temperature and over-current protection being triggered on a few occasions (just turn off and on to reset), the Fosi V3 Mono amps survived without spewing out any of its magic smoke. I think these should be reasonably robust in intended daily music reproduction. As I showed in Part I, if you get a chance to open up the box, double check on the thermal compound and you might want to put a bit more at the interface between the heat spread plate over the amp chip and the casing which acts as heatsink.

It's great witnessing the evolution of these low-priced (~US$300 for 2-channels) TI TPA chip amplifiers over the last few years! While perhaps there are further optimizations to be done, we're probably at the limits of maximal fidelity for the TI TPA3255 chips. Already, the power and resolution from something so inexpensive and small is very impressive. Needless to say, an amplifier like the Fosi V3 Mono's performance is at a level that would challenge and exceed many Class A/B designs at multiples of the price.

Despite the few issues discussed, great job Fosi Audio for an enjoyable product. Given the price and performance, in many ways, this product can be considered a kind of benchmark for comparisons with other affordable, excellent-sounding Class D amplifiers.

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

Evolution of the sports car - primate powered to autonomous? 🙂

On "Solved Problems" in High-Fidelity Audio Reproduction...

To end, I thought I'd address a comment that Matthew Poes made in his YouTube review of the Fosi V3 Mono near the end (24:05). I came across this while researching what others were saying about the product (emphasis mine):

"The one thing that makes me a little bit sad is that I can't think of a good speaker to match this with. I mean, there's lots of cheap speakers that aren't bad, but I can't think of any speaker that's like $200 a pair or $400 a pair that provides state of the art performance for $400 and the compromise is build quality or something like that, it just doesn't exist..."

Well, this is why I suggested 5 years ago within my pie-graph of system component costs, that the power amplifier should be a mere "8%" of the total - that percentage is getting smaller: 😁

Seriously, audiophiles, it's very much OK to spend thousands on DACs and amplifiers if you want! But the point is, you don't have to in order to achieve equivalent "high-fidelity" sound quality. Buying luxurious products is totally within one's rights, but make sure to be realistic in recognizing that "high fidelity", transparent sound is already achievable on a "beer budget". This is the natural path of technological progress and it's awesome!

We see this all the time in tech. Back in 2016, the MSRP for the nVidia GTX1080 GPU was US$600 if you wanted some awesome 3D gameplay. These days a mid-level RTX4060 is US$300; half the price with typically more than 25% faster performance. So too, as high-fidelity technologies improve, we will see equivalent if not better performance at lower cost. At least with computer components we can experience truly improved speeds like frame-rates or image quality (like ray-tracing); obviously not so much in audio sound quality regardless of incessant claims month after month about a sensational "best sounding" new product from reviewers and shills to drive sales.

I rejoice in this rather than feel "a little bit sad" that there are no speakers around the same price range that would match the Fosi V3 Mono's level of performance! There's nothing wrong with pairing $300 amps like these that can produce objectively near-state-of-the-art load-invariant performance with $50,000 near-state-of-the-art-speakers is there? Of course not, just as there's absolutely nothing wrong with using generic 12AWG copper $100 speaker cables with those $50,000 speakers either because it doesn't make any meaningful difference if we were to honestly do blind volume-controlled listening comparisons! This is exactly what we can foresee happening with power amps and DACs because high-fidelity performance is much easier to achieve in the electronic domain than performing electromechanical transduction producing full-range audible soundwaves or achieving rooms with ideal acoustic characteristics.

DACs and power amplifiers are more than likely "solved problems" that objectively keep getting better fidelity-wise, price-wise, efficiency-wise, beyond the needs of human hearing within the limitations of our usual rooms (again when tested honestly with blinded comparisons of course!). Some folks in the "high end" or even "ultra high end" audio Industry who make money on all kinds of luxury stuff will of course not want to believe this. Imagine how this would look if amplifiers like these were used in an audio show (like the Pacific Audio Fest next week 🤔) paired with 5-figure speakers?! This kind of talk might even be considered "dangerous" because it undermines a kind of psychology, or culture of price to sound-quality correlation that has been perpetuated in the Industry for decades as to what owning an "audiophile" system means. I think it's emotionally more healthy to just be truthful even if it might deviate from certain prevailing opinions.

Regardless, it's up to each of us to figure this out for ourselves...

Hope you're all enjoying the music, audiophiles!