One of the PFFB (Post-Filter Feedback Technology)-based Class D audio amplifiers showcased in a recent writeup of mine was Fosi Audio’s V3 Mono, which will get sole billing today:

It interestingly (at least to me) originally launched as a Kickstarter project in April 2024:

As the name implies, it’s a monoblock unit, intended to drive only a single speaker, with both single-channel XLR balanced and RCA unbalanced input options.

I own four functional (for now, at least) devices, plus the nonfunctional one whose insides we’ll be seeing today. Why four? It’s not because I plan on driving both front left and right main speakers and a center speaker and subwoofer, or for that matter, the two main transducers plus two surrounds. Instead, it’s for spares, notably ones obtained pre-higher tariffs, and specifically to do with that dead fifth amp.

Design evolution, manufacturing, and reliability issues

Before I go all Debbie Downer on you, I’ll begin with the good news. The V3 Mono is highly reviewer-rated (see, for example, the write-up from my long-time tech compatriot Amir Majidimehr) and has also garnered no shortage of enthusiastic feedback from owners like Tim Bray, who had heard about it from Archimago (here’s part 2). Alas, amidst all that positive press are also a notable number of complaints from folks whose units let the magic smoke escape, sometimes on just the first use, or whose amplifiers had more modest but still annoying issues.

Mis-wired connections

I’ll start with the most innocuous quirk and end with the worst. Initial units were mis-wired from the PCB to the speaker banana plugs (due, I actually suspect, to a fundamental PCB trace layout issue) in such a way that they ended up with inverted-polarity outputs, i.e., signals being 180° out of phase from how they should be.

This wasn’t particularly a problem if all the units in your setup exhibited the issue, because at least then the phase was consistently inverted. However, if one (or some, depending on your setup complexity) of them were in phase and other(s) were out of phase, the inconsistency resulted in a collapsed stereo image and overall decreased volume due to destructive interference between the in- and out-of-phase speakers.

The same goes if you mixed-and-combined out-of-phase V3 Monos with in-phase other devices, whether from other manufacturers or even from Fosi Audio itself. The fix is pretty easy; connect the red speaker wire to the black speaker terminal of the affected V3 Mono instead, and vice versa, to externally reinvert the phase back to how it should be. But from my experience with these units, it’s not possible to discern if a particular device is wired correctly without disassembling it; this guy’s sticker-based methodology, for example, didn’t pan out for me:

As commenter @TheirryG01210 wrote in response to the above video, “A better way to figure out if phase is correct is to check that cables are cross-connected (left solder pads cable goes to the right banana socket and vice versa).”

That’s spot-on advice. Here, for example, is one of my functional units, which has the wires un-crossed, therefore operating in an inverted-output fashion. That said, this approach looks like how it should be wired, right? Therefore, my conjecture that this actually is inherently a PCB layout issue, with wire-swapping the cheaper, easier workaround to the alternative costlier and otherwise more complicated board “turn”.

My photo also matches one of the two in this Audio Science Review discussion thread post:

The other picture in that post shows the wires crossed; it’s not clear to me whether this is something that the owner did post-purchase with a soldering iron or if Fosi Audio revamped units still in its inventory, after discovering the problem and prior to shipping them out:

Conceptually, it matches the from-factory crossed wiring of my other three functional devices, along with today’s teardown victim, although the wire colors are also swapped with my units:

But color doesn’t matter. A crossed-wires configuration is what’s key to a correct-phase output.

The next, more recently introduced issue involves gain-setting inconsistency. Look at the most recent version of the “stock” image for the product on Amazon’s website, for example:

And you’ll see that the two gain-switch options supported for the RCA input (the switch doesn’t affect the XLR input) are 19 dB and 25 dB. That said, the gain options shown in the online user manual are instead 25 dB and 31 dB, which match the original units, including all of mine:

Here’s the key excerpt from an email by Fosi Audio quoted in a relevant Audio Science Review post (bolded emphasis is mine):

We would like to confirm whether your V3 mono gain is the old version or the new version. Since V3mono does not have a volume adjustment knob. It has already obtained a large power output when it is turned on, so we have reduced the gain of 31db to 25db, and 25db to 19db in the new version, which can effectively ensure the stable output of V3mono, safe use and extend the service life.

Loud “pop” sound

Which leads to my last, and the most concerning, issue. After a seemingly random duration of operation, but sometimes just the first use, judging from comments I’ve seen on Audio Science Review, Amazon, Fosi’s online store, and elsewhere, the amplifier emits a loud “pop” and the sound disappears, never to return.

The front panel light still glows, and you can still hear the “click” when the amp initially turns on or transitions out of standby in response to sensing an active input source (or when you transition from one input to another, for that matter), but as for the output…nothing but the sound(s) of silence. This very issue happened with one of the devices I purchased brand new, fortunately, within the return-for-full-refund period.

Several of the other V3 Monos I acquired “open box” off eBay also arrived already DOA. In one particularly mind (and amp)-blowing case, I bought a single-box two-device set. When I opened it up, one of the amps had a piece of blue tape stuck to the top with the word “good” scribbled on it. Yep, the other one was not “good”. 

What the eBay seller explained to me in the process of issuing a ship-back-for-full-refund is that when large retailers get a return, they sometimes just turn around and resell it discounted to eBay sellers like her, apparently without remembering to test it first (or, more cynically, maybe just not caring about its current condition).

A blown-output case study

Today’s victim (1,000+ words in) was another eBay-DOA example. In this case, the seller didn’t ask me to return it prior to issuing a refund, and it therefore became a teardown candidate, hopefully enabling me to discern just where the Achilles’ Heel in this design is.

To Fosi Audio’s credit, by the way, the pace of complaints for this particular issue seems to have slowed down dramatically of late. When I first looked at the customer feedback on Amazon, etc., earlier this year, comments were overwhelmingly negative. Now, revisiting various feedback forums, I see the mix has notably shifted in the positive-percentage direction. That said, my cynical side wonders if Fosi and Amazon might just now be nuking negative posts, but hope springs eternal…

I’ll start with some overview shots of our patient, one of which you’ve already seen, as usual, accompanied by a 0.75″ (19.1 mm) diameter U.S. penny for size comparison purposes (the V3 Mono, not including its bulbous suite of external power supply options, has dimensions of 105 x 35 x 142 mm and weighs 480 grams).

Remove the two side screws from the back panel:

And the front panel slides right out:

The “Aesthetic and Practical Dust-Proof Filter Screens” (I’m quoting from Fosi Audio’s website, though I concur that they both look cool and act cooling) also then slide right out if you wish:

Removing two more screws on the bottom:

Now allows for the extraction of the internal assembly (here again, you saw one photo already):

PCB extraction and examination

The front and back halves of the “Sturdy and Durable All-Aluminum Alloy Chassis” are identical (and an aside: pretty snazzy shots, eh?):

Returning to the PCB topside (with still-attached back panel), let’s take a closer look:

One thing I didn’t notice at first is that none of the ICs are PCB-silkscreened as to their type (R for resistor, for example, C for capacitor, L for inductor, U for IC, etc…), far from their specific-device identifying number (R1, C3, L5, U2…). Along the left side, top to bottom, are:

• The three-position switch for on, auto, and off operating modes
• The power status LED
• The two-position XLR-vs-RCA input selector switch, and
• A nifty two-contact spring-loaded switch that’s depressed when the front panel is in place. I suspect, but didn’t test for myself, that it prevents amplifier operation whenever the front panel is removed.

Note, too, the four screw heads in between the two multi-position switches, along with the ribbon cable. Look closely and you’ll realize that the first three items mentioned are actually located on a separate mini-PCB, connected to the main one mechanically via the screws (which, as you’ll see shortly, are actually bolts) and electrically via the ribbon cable.

And in fact, the silkscreen marking on the mini-PCB says (among other things) “SW PCB” (SW meaning switch, I assume) while the main PCB silkscreen in the lower left corner says…drumroll…”MAIN PCB”.

Why Fosi Audio went this multi-PCB route is frankly a mystery to me. Until I noticed the labeled silkscreen markings (admittedly just now, as I was writing this section) I’d thought that perhaps the main board was common to multiple amplifier product proliferations, with the front panel switches, etc. differentiating between them. But given that both boards’ silkscreens also say “Fosi Audio V3 MONO” on them, I can now toss that theory out the window. Readers’ ideas are welcome in the comments!

In the middle of the photo are two 8-pin DIP socketed chips, op-amps in fact, Texas Instruments NE5532P dual low-noise operational amplifiers to be precise.

They’re socketed because, as Fosi Audio promotes on the product page and akin to the two Douk Audio amplifiers I showcased in my prior coverage, too, they’re intended to be user-swappable, analogous to the “tube rolling” done by vacuum tube-based audio equipment enthusiasts.

Numerous (Elna) electrolytic and surface-mount capacitors (along with other SMD passives) dot the landscape, which is dominated by two massive Nichicon 63V/2200μF electrolytic filtering capacitors (explicitly identified as such, along with the Elna ones, by visual and text shout-outs on the V3 Mono product page, believe it or not). And one other, smaller Texas Instruments 8-lead IC (soldered SOP this time) on the bottom toward the right bears mentioning. It’s marked as follows:

N5532
TI41M
A9GG

Its first-line mark similarity to the previously mentioned NE5532P is notable, albeit potentially also coincidental. That said, Google Image search results also imply that it’s indeed another dual low-noise op amp. And it’s not the last of them we’ll see. Speaking of which, let’s next look at the other half of the PCB topside:

There it was at the bottom; another socketed TI NE5532P! Straddling it on either side are Omron G6K-2P-Y relays. At the top are even more relays, this time with functional symbol marks on top to eliminate any identity confusion: another white-color one, this time a Zhejiang HKE HRS3FTH-S-DC24V-A, and below it a dark grey HCP2-S-DC24V-A from the same supplier.

Remember when I mentioned earlier that after one V3 Mono stopped outputting amplified audio, I could still hear relay clicks when I toggled its power and input-select switches? Voila, the click-sound sources.

Those coupling capacitors are another curious component call-out on the V3 Mono product page; they’re apparently sourced from German supplier WIMA. The latter two, on either side of the aforementioned PCB solder pads that end up at the speaker’s banana plug connectors, are grey here but yellow colored at Fosi Audio’s website, so…

To the left of the red coupling caps is a grey metal box with two slits on top and copper-color contents visible through them; hold that thought. And last but not least, along the right edge of the PCB are (top to bottom) the power-input connector, two hefty resistors, the XLR input, and the RC input. The two-wire harness in the lower corner goes to the aforementioned gain switch.

Insufficient thermal protection?

Now for the other side:

That IC at far left was quite a challenge to identify. To the right of an “AB” company logo is the following three-line mark:

TNJB0089A
UMG992
2349

Google searches on the text, either line-by-line or its entirety, were fruitless (at least to me). However, I found a photo of a chip with a matching first-line mark here. About the only thing on that page that I could read was the words “AB137A SOP16”, but that was the clue I needed.

The AB137A, is (more accurately was) from the company Shenzhen Bluetrum Technology, which Internet Archive snapshots suggest changed its name to Shenzhen Zhongke Lanxun Technology at the beginning of this year. The bluetrum.com/product/ab137a.html product page no longer seems to exist, nor does the link from there to the datasheet at bluetrum.com/upload/file/202411/1732257601186423.pdf. But again, thanks to the Internet Archive (the last valid snapshot of the product page that seems to exist there is from last November) I’ve been able to discern the following:

• CPU and Flexible IO
  High-performance 32-bit RISC-V processor Core with DSP instructions
  RISC-V typical speed: 125 MHz
  Program memory: internal 2 Mbit flash
  Internal 60 KB RAM for data and program
  Flexible GPIO pins with programmable pull-up and pull-down resistors
  Support GPIO wakeup or interrupt
• Audio Interface
  High-performance stereo DAC with 95 dB SNR
  High-performance mono ADC with 90 dB SNR
  Support flexible audio EQ adjust
  MIC amplifier input
  Support Sample rate 8, 11.025, 12, 16, 22.05, 32, 44.1, and 48 kHz
  Four-channel Stereo Analog MUX
• Package
  SOP16
• Temperature
  Operating temperature: -40℃ to +85℃
  Storage temperature: -65℃ to +150℃

So, there you have it (at least I think)!

The other half of this side of the PCB is less exciting, unless you’re into blobs of solder (along with, let’s not forget, another glimpse of those hefty resistors), that is:

But it’s what’s in the middle of this side of the PCB, therefore common to both of those PCB pictures, that had me particularly intrigued; you too, I suspect. Remove the two screws whose heads are on the PCB’s other side:

Lift off the plate:

Clean the thermal paste off the top of the IC, and what comes into view is what you’ve probably already suspected: Texas Instruments’ TPA3255, the design’s Class D amplification nexus:

At this point in the write-up, I’m going to offer my conjecture on what happened with this device. The inside of the metal plate, acting as a heatsink, paste-mates with the TPA3255:

while the outside, also thermal paste-augmented, is intended to further transfer the heat to the bottom of the aluminum case via the two screws I removed prior to pulling the PCB out of it:

Key to my theory are the words and phrases “bottom” and “thermal paste”. First off, it’s a bit odd to me that the TPA3255, the design’s obvious primary heat-generation source, is on the bottom of the PCB, given that (duh) heat rises. The tendency would then be for it to “cook” not only itself but also circuitry above it, on the other side of the PCB, although the metal plate-as-heatsink should at least somewhat mitigate this issue or at least spread it out.

This leads to my other observation: there’s scant thermal paste on either side of the plate for heat-transfer purposes, off the IC and ultimately to the outside world, and what exists is pockmarked. I’m therefore guessing that the TPA3255 thermally destroyed itself, and with that, the music died.

Wrapping up

Before I forget, let’s detach that mini-PCB I mentioned earlier. Here are the backside nuts:

And the front-side bolt heads:

Disconnect the ribbon cable:

And you already know what comes next:

Not too exciting, but I’ve gotta be thorough, right?

At this point, it occurred to me that I hadn’t yet taken any main-PCB side shots. Front:

Left side:

The back:

The right side:

And after removing the two screws surrounding the XLR input:

I was able to lift the back panel away, exposing to view even more PCB circuitry:

In closing, remember that “grey box with two slits on top and copper-color contents visible through them” that I mentioned earlier? Had I looked closely enough at the V3 Mono product page before proceeding, I would have already realized what it was (although, in my slight defense, the photo is mis-captioned there):

Then again, I also could have identified it via the photo I included in my previous write-up:

Instead, I proceeded to use my flat-head screwdriver to rip it off the PCB in the process of attempting to more conservatively detach just its “lid”:

As I already suspected from the “copper-color contents visible through the two slits on top”, it’s a dual wirewound inductor:

from Sumida, offering “superior signal purity and noise reduction, elevating the amplifier’s sound performance,” per Fosi Audio’s website.

Crossing through 3,000 words, I’ll wrap up at this point and turn the keyboard over to you for your thoughts in the comments!

—Brian Dipert is the Principal at Sierra Media and a former technical editor at EDN Magazine, where he still regularly contributes as a freelancer.

 Related Content

• Class D: Audio amplifier ascendancy
• Audio amplifiers: How much power (and at what tradeoffs) is really required?
• Class D audio power amplifiers: Adding punch to your sound design
• How Class D audio amplifiers work

The post The Fosi Audio V3 Mono: A compelling power amp with a tendency to blow appeared first on EDN.