LM3886 amplifier Gainclone

LM3886 amplifier Gainclone

A few years ago, National Semiconductor released some very high-performance, easy-to-use audio power amplifier circuits, the LM3886. I needed an additional amplifier to biamp my homemade electrostatic speaker, so I experimented with the LM3886 chip. The LM3886 amplifier was chosen for its ease of use, performance, low distortion, and built-in short-circuit and thermal instability protection. For a power amplifier, few things are more important than these. When driving an electrostatic speaker, you can never have too much protection.

Some people with "golden ears" feel that no applied notebook circuitry is good enough, and they claim that "optimization" and "improvement" are necessary. The problem is that most of them are not engineers and are unaware of the potential consequences of their "improvements." For example, a few years ago, when these chips were popular among audiophiles, using minimal power supply filter capacitors in pursuit of "best sound quality" was all the rage. We're talking about using only 500 uF on each power rail of every LM3886 amplifier chip used. This is clearly insufficient, causing the power supply to drop under load at lower volumes, resulting in distortion. The problem is that some "golden ears" saw a large power supply rejection ratio (PSRR) IC specification and assumed that this meant the chip could tolerate a 10V power supply ripple. Pendulum has swung in the other direction, and now many music lovers use ample power stored in their power supplies.

LM3886 Gainclone amplifier chip amplifier circuit diagram

Another common change is the removal of the "Zobel" network from the LM3886 amplifier output, because it "sounds bad." After several people burned out some expensive speakers, they realized the network was there to prevent oscillations, and speakers without the Zobel network sounded worse, so most people now use the Zobel network. I use my amplifier to drive electrostatic speakers, which is particularly difficult due to the capacitive reactance, but I've never had any stability issues.

My amplifier uses a circuit pattern from the NS datasheet with an additional power supply. Perhaps the mindset is that manufacturers don't understand actual build practices when refining their models. For the LM3886, proper grounding is crucial for good amplifier performance. Grounding is often not shown in diagrams because it depends on the mechanical design of the LM3886 amplifier chassis. Build practices such as keeping high-impedance circuitry (like the amplifier input) away from high-current components (transformers, power supplies, and the LM3886 amplifier output path), and using star grounding, higher-performance $10 resistors, and gold-plated connectors all help. Click here to see a diagram of star grounding.

Typically, the chassis is the most expensive part of an amplifier. It must dissipate heat from the amplifier and power ICs, have an acceptable appearance, and be large enough to house all the components. My amplifier chassis came from a junkyard in Dallas; I bought it for $1 per pound, or about $15. The box was originally a mobile phone antenna diversity unit, used to house the amplifier.

The sidewalls are 1/4-inch thick aluminum, welded at the corners to make it a heavy-duty box that can dissipate a lot of heat. The top plate is 1/2-inch thick and features heat sinks to dissipate even more heat. The bottom plate (not shown) is 1/8-inch thick.

The enclosure had many port holes on the side. I couldn't use them, but I filled the other holes with Bondo putty. First, I put some device bits containing the PCB into the enclosure, then coated the holes from the outside with Bondo. After the Bondo cured, I sanded the outer surface flush with the rest of the aluminum. It worked very well, and I recommend this technique to anyone recycling surplus assets.

I drilled some holes where needed, enlarged some existing holes, and then applied the goal.

Another hobby of mine is restoring old radios. One particularly special radio in my collection is a 1927 Super Six Neutrowound. This radio must have looked quite spacious back then. At that time, most radios were simple rectangular wooden boxes. This one, however, is a steel box with a very unusual appearance, fitted with nickel tubing coverings protruding from the top, and painted with an exotic blue crackle lacquer and orange!

I really liked the finish on the radio, so I decided to try replicating it on my LM3886 amplifier. The radio was painted, but I don't have spray paint equipment, so I had to use a brush. I chose a nice blue metallic paint (not quite the same as the radio) and a bright orange as the base coat. The result wasn't great for the radio, but it was kind of fun anyway.

The coating process begins with an orange acrylic primer, followed by an acrylic crackle medium, acrylic, metallic blue, and finally a layer of clear acrylic protective varnish.

Next came the construction of the LM3886 amplifier. The first channel was hand-wired because there were few components required, but I didn't like its appearance, so the second channel was built on a perf board. The plan eventually shifted to a circuit board (later). All reasons were to place a single "star" point on the chassis to minimize the possibility of ground loops and associated hum/noise.

The transformer is a device that could be 320VA, which is a bit of an exaggeration, since the amplifier is about 30 watts per channel at a specified supply voltage.

The filter capacitors plugged into the socket total about 100,000 uF, which is more than enough. They are low voltage, large capacity, and cheap (only $6 for four options), so why not?

The volume control is located near the rear inlet to keep the signal short away from the transformer. The tree expansion is a 1/4-inch diameter brass tube. The alignment of the volume tree and the front hole wasn't perfect, so I made a flexible coupling using a piece of polyurethane tubing and two cable ties. It works well; it slips slightly when you turn the knob past the potentiometer stop by emphasizing the potentiometer.

I finally found the time to fabricate the printed circuit board and build two LM3886 amplifier channels. I prefer PCBs because hand-wired circuits can easily short-circuit when moved for repairs.