LM3886 Stereo Power Amplifier Kit

I recently posted a review of a volume control, and today I'm continuing the series of reviews about all kinds of audio equipment. This time, I'm looking at a power amplifier based on the fairly popular LM3886 chips.


To start with, I should say that I'm not an audiophile. For the most part, what matters to me is the music I'm listening to, rather than the equipment I use to play it. But that doesn't mean I don't care what I listen to music on; it's just a secondary concern for me.
You also won't find phrases like “airy mids,” “transparent highs,” “stubborn bass,” or “muddy soundstage” in this review. To me, an amplifier is just an amplifier, and it has one job: to amplify the audio signal with as little added distortion as possible.

In part, I stopped paying attention to equipment because the pursuit of perfection never ends. There's always something a little more expensive, but better and more powerful, and it goes on forever. I don't want to become one of those people who, instead of listening to music, start breaking in speaker stands, listening to USB cables, buying stands for those cables, and so on.

Also, this review features lots of components with brand-name markings: Dale, Nichicon, Wima, Omron. But I can't guarantee that they're genuine, or that they're counterfeit. In this case, they're just pretty little parts to me :)


There were four things that sold me on this amplifier:
1. The price
2. The fairly positive reviews of the LM3886 chips.
3. The fact that it's a kit, not a finished product.
4. Its good looks.

How could I pass up a beauty like this?


The only technical specifications listed were:
Supply voltage: acceptable range from 2 × 15 V to 2 × 28 V (refers to the transformer's output voltage, for example, 25-0-25)
Load impedance: 4–8 Ω
Output power: max. 68 W + 68 W
Board size: 155 mm × 80 mm



The kit comes without a heatsink, transformer, or input filter capacitors, but the product page has links to:
1. A 200 W transformer with 2 × 25 V + 2 × 15 V windings
2. A heatsink
3. Filter capacitors
4. A board with a potentiometer
5. A potentiometer without a board

Apparently, you can even order an assembled board for about $4 extra.

In the end, I received a fairly well-packed parcel that was supposed to contain everything needed.


The kit includes:
1. A printed circuit board
2. A set of components for assembly
3. An assembly diagram


I was pleased with the assembly diagram. It not only shows where to install the components, but also includes a key at the bottom translating the resistor codes into the familiar values shown on the diagram.


The printed circuit board looks like a solid 5 out of 5: it has a solder mask, silkscreen, and plated traces. The only thing missing, perhaps, is the component values printed on the board, but there's a detailed assembly diagram for that.



Judging by the markings, this board revision was designed in May of this year, though in this case that could just as well be the production date for the batch.


The board is also marked with the kit name: 3886AMP DC Servo.


The components are split into two groups: the main batch, which includes capacitors, chips, terminals, and relays, and a separate bag containing resistors and a couple of diodes. Most of the main components were simply tossed together, but nothing was damaged in transit.


Once sorted, everything looks a little easier to make sense of.


Honestly, I was surprised that the components weren't just undamaged—their leads hadn't even been bent during shipping.


Let's move on to assembly. Since there's a lot of routine stuff involved, I'll tuck it under a spoiler.


And so I've reached the final stage of assembling this board: the power amplifiers, LM3886.


I can't vouch for the authenticity of these components, so I'll just include photos of them. The amplifiers have fully insulated packages. I seem to remember seeing a version with a non-insulated package, and I looked for information on the seller's page but couldn't find any. Maybe I was confusing it with another model.


The seller does say you can choose between two lead-forming options: the standard one for vertical mounting and an alternative one for horizontal mounting. The first option is sent by default; to get the second, you need to mention it in the order comments.


A few key LM3886 specs. According to the listed specifications, maximum output is 68 W into a 4 Ω load with a ±28 V supply, or 50 W into 8 Ω with a ±35 V supply. In my case, I used inexpensive speakers rated at around 20 W with an impedance of about 3.5 Ω.


The LM3886 needs a heatsink—not just because it needs one in operation, but because the correct installation procedure is to screw the ICs to the heatsink first, and only then solder them to the board.
I used a heatsink you've seen in some of my other reviews. It's a little undersized for this amplifier, by about a factor of two, but since I wasn't planning to run the amp at high power, it's more than adequate for this experiment.

First, mark and drill holes in the heatsink for mounting the ICs.


Next, screw the ICs to the heatsink, but don't apply thermal paste yet. At this stage, the main thing is to position them as they'll sit in the finished assembly. Once they're in place, solder the ICs. I usually tack one lead with solder, check that everything is straight, and then solder the remaining leads.


Unscrew the heatsink, apply thermal paste, and reinstall it, tightening the mounting screws properly this time. The kit included screws in two sizes. The idea was to use the longer ones with small hex heads, but the heatsink fins got in the way while I was tapping the holes, so I ended up using the shorter screws with large Phillips heads instead. No washers are needed in this case.



And that's the main assembly done. After this, you can wash off any flux residue and check that all the connections are soldered.


I didn't use any extra flux; the flux in the solder was more than enough. The board was a pleasure to solder, and all the components were relatively new and soldered without any trouble. I usually populate the board first, bending the leads as I go, but this time I left them straight and trimmed them almost flush with the board, which makes the assembly look a little neater. Once the assembly is completely finished, I'll probably coat the board with Plastik-70 lacquer to protect it from possible oxidation.


But I still hadn't settled the question of the power-supply filter capacitors. As I understand it, this circuit needs capacitors in the 10,000–15,000 µF range, rated for 35 V, but I don't have any like that at home, and the ones I ordered are still on their way.
In the end, I dug up two pairs of capacitors at home: 4,700 µF ones made locally and imported 10,000 µF ones. Both types are rated for 25 V, but the local ones are older.


Out of curiosity, I measured their specs first and, as expected, went with the imported ones. While doing so, though, I noticed something interesting: the local capacitors have less variation in their readings than the imported ones.


Since I have no plans to use these capacitors with this amplifier long-term, and don't want to shorten their leads, I simply bent them over after soldering them in.


That's it. The only thing missing, really, is the power transformer. It's also a shame the heatsink isn't black anodized—it would look much better.


Before moving on to the tests, checks, and so on, I need to digress. I'm very biased against the sound of IC-based amplifiers, and I probably wouldn't have ordered a kit like this if it weren't for the positive reviews online. In fact, my experience with IC-based amplifiers ended a long time ago—probably 20 years ago or more. The last non-discrete amplifier I built is shown in this photo. It's only half assembled here, and I never got around to using it.
By the way, not long ago I showed a subwoofer amplifier in the comments. Imagine my surprise when I accidentally found another one, almost completely assembled. I’d even forgotten I’d made two of them :)


This is how the amplifier from the review was supposed to look once installed in an enclosure. I’ve actually seen this amplifier for sale fully assembled, and I think it looks great. It’s just a shame that shipping heavy items like equipment enclosures from Tao is so expensive :(


It’s usually recommended to power up a device for the first time using a transformer with an incandescent bulb connected in series with the primary winding. This helps avoid some of the problems that can arise from a short circuit on the board. With a discrete amplifier, the first power-up is sometimes done without the output transistors installed.
In my case, things were a little simpler. I used a very low-power transformer with a secondary voltage of 2 × 24 V and a power rating of no more than 20 W. In the event of a short circuit, its output voltage would simply sag because of the high winding resistance.
After I applied power, the LED lit up, the relay switched on three seconds later, and that was it—the amplifier was ready to use.


By the way, about the connections. There’s another terminal block next to the input one: it’s the output for the power indicator LED, connected through a 4.3 kΩ resistor to the 24 V regulator output. Also, only two of the four mounting posts are connected to the board’s ground. One is shown here, and the other is near the output terminal block. Only the one shown in the photo is marked as ground, though. I also thought it was better to connect to the enclosure at a single point rather than two, so be careful.


Next, I connected the volume control I’d shown earlier, a pair of speakers (only one is connected in the photo so far), and a source—in this case, a small Transcend player. Obviously, the whole setup looks a bit... ridiculous, but it is what it is. Later I used a computer as the source, but that didn’t really change anything.


You can see in the photo above that only one output is connected. I only switched it on out of curiosity, but I was a little surprised by how well it worked. It did get quite noticeably hot, though. The second thermal image shows that the left-hand chip gets much hotter; that was the channel under load.


After listening to music like that for a while, I loaded the second output too, and everything kept working—quite well, in fact, considering the low-power transformer, basic speakers, and “source.” After about 15 minutes at a fairly high volume with both channels under load, the transformer was the hottest component, and it started giving off a rather unpleasant smell. The chips, on the other hand, had warmed up to about the same temperature.

I understand that the transformer doesn’t necessarily need to be rated for 20 W to deliver 20 W of music power, but I was still a little surprised by how powerful and good-quality the sound was. By my estimate, the total power delivered to the load was at least 30 W—but that was with a music signal. With a sine wave, of course, things would be different.
By the way, the manufacturer recommends a transformer rated at around 150 W for this amplifier, but links to a 200 W model. In my opinion, there’s simply no point in getting a transformer rated above 150 W.


To be honest, I initially connected the player just to check whether the amplifier worked at all. In the end, I didn’t turn the music off for about an hour and a half—I really got carried away :)

Naturally, this little transformer wasn’t suitable for testing, so I connected a TPP270 transformer. The primary was wired for 240 V, and the secondary for the highest possible voltage. The transformer is rated at 57 W, but our transformers are pretty good quality, so you can safely load them beyond that.
With no load, I measured 26 V across the filter capacitors, so they were operating slightly above their rated voltage.


I listened to music for much longer this time—several hours at a fairly high volume. The chip packages were around 62 °C, and the transformer was slightly warm.


All right, let’s move on. I should at least test it with some instruments.

First, two tests: measuring the DC and AC voltage at the output. The first test checks that the zero-adjustment circuit is working correctly. With the second, I tried to get some idea of the noise level at the output with the input shorted. Obviously, the second test doesn’t tell us much, but it may still be useful.
1, 2. The multimeter showed zero DC at the output.
3, 4. About 0.3 mV AC. I should mention that the amplifier does produce a little noise—not hum at 100 Hz, but actual noise. You can only hear it from about 10–15 cm away from the tweeter. The 100 Hz hum is completely inaudible.


After that, I checked how the amplifier behaved with a sine wave at the input and different load resistances. Somehow I didn't have 4 and 8 Ω resistors at home, so I ran the tests with high-power 10 Ω resistors. This gave me three test load resistances: 3.3, 5 and 10 Ω. By the way, my speakers have a resistance of about 3.5 Ω, so I also tested at 3.3 Ω, the closest value to a real-world load.

1. 3.3 Ω load, output swing just before clipping: output voltage 10.94 V, input voltage (DC) 19.55 V, calculated output power: 36 W
2. 3.3 Ω load, clipping begins: output voltage 11.46 V, input voltage (DC) 19.35 V, calculated output power: 39 W. The lower half-cycle clips first; as the input voltage is increased further, both half-cycles clip.
3. 5 Ω load, output swing just before clipping: output voltage 12.505 V, input voltage (DC) 21.1 V, calculated output power: 31 W
4. 10 Ω load, output swing just before clipping: output voltage 14.08 V, input voltage (DC) 23.5 V, calculated output power: 20 W



In principle, the stated and measured power are roughly the same. The problem with getting full power was the low supply voltage.


This graph is easier to read, especially since it includes a test with a 6 Ω load, close to my 5 Ω test load.
According to this graph, with a 21 V supply and a 6 Ω load, the output power should be about 26 W. I got 31 W with 5 Ω.


The sine wave looked good, but the square wave was a real disappointment: every test showed an overshoot on the signal's leading edge. I was quite surprised, so I even turned on the oscilloscope's second channel to see what was happening at the amplifier input. The yellow trace is the input.
1. 3.3 Ω load, output voltage: 9.2 V
2. 5 Ω load, output voltage: 9.8 V
3. 10 Ω load, output voltage: 10.4 V
4. No load.

I tried to set the output swing as close to the maximum as possible. Increasing the input signal further produced a normal-looking square wave; raising the voltage even more triggered the amplifier's protection. Both channels behaved identically.


I hadn't been testing for very long, but the test channel's chip had already heated up to about 100 degrees.


After all that, I decided to experiment with the RMAA software. I've known about it almost since it first appeared, but I hadn't worked with amplifiers for many years, so this was my first time actually using it. The tests were run using my computer's built-in sound card, since I haven't managed to get the computer with the Creative Live 5.1 up and running yet. I'm not even sure it would give better results.

Overall, the test results are rather strange, and in this case I'm more inclined to blame my inexperience testing this kind of device with RMAA than the amplifier itself.


But one thing I definitely agree with is the boost in the low- and high-frequency ranges. While listening to music, I had already noticed one drawback: there's a slight dip in the midrange. I thought it might be a quirk of the player I was using as the signal source. But the frequency response test showed roughly the same thing, which I don't like, since in my view an amplifier shouldn't affect this parameter at all.


I'm not yet confident in the test results, so I plan to keep experimenting and repeat the tests in a future review of another amplifier. Maybe I'm doing something wrong, but I learn quickly. As a small confirmation, here are the RMAA results for the volume control from my previous review.


The tests were run with the same computer and the same cables.


Video version of the review


As usual, here's a brief summary of this amplifier.
On the plus side, the kit is exceptionally well put together. I had no problems with assembly, finding the components, their quantity, or soldering. And on top of that, the amplifier worked after assembly. There isn't a single trimmer on the board, so even a complete beginner can build it. All you need to get it working is a transformer, two capacitors, and a heatsink. The heatsink should have about twice the surface area of the one I used.

Despite the odd test results, it sounds pretty good. I listen to all kinds of music, and these tracks were part of the test:
Jean Michel Jarre - Teo And Tea
Schiller - Deep Dance
Radiorama
Ken Laszlo
Shui Mu Nian Hua - Fly With Me
Cafe del Mar
Himekami
Keiko Matsui
Dire Straits
Pink Floyd
Andrea - Macho Man
Didier Marouani 1987 Space Opera
And plenty more.

Based on all the tests, I can say it sounds perfectly decent. I listen to music at home all the time, so I checked something that matters a lot to me: listening to this amplifier doesn't get tiring.

That said, I wasn't impressed by either its response to a square wave or the RMAA results. The latter may well be down to a mistake on my part, but I'm confident in the oscilloscope traces. I should also mention that the frequency response isn't flat, and I could hear that when listening to music. At first, I thought it was just the player sounding bad.

I'm not done experimenting, and I think there'll be more reviews of devices for scaring the neighbors playing music. Still, I don't think this review turned out all that well.
That said, I find the kit itself interesting, especially given its low price and how easy it is to assemble. Even so, I'd put this amplifier in the mid-range class. Perhaps my bias against the sound of "chip amps" is influencing my opinion here.

The kit costs $21.31 including shipping to the forwarding agent; shipping from there depends on various factors. The kit weighs 166 grams, according to the order page at the forwarding agent.