Monday, October 8, 2018

LM3886 Based Stereo Power Amplifier

Another transfer from my web site to this blog... This is a project I did back in the early 2000s.  I used it for several years, but in the last few years it has been sitting in a box in my basement.  Maybe I'll pull it out soon...

LM3886 Based Stereo Amplifier
Several years ago, National Semiconductor came out with some very high performance, easy to use audio power amplifier ICs.  I was in need of an extra amplifier so I could biamp some of my home-built electrostatic loudspeakers so I tried the LM3886 chip.

This part was chosen because of the ease of use, power output, turn-on and off thump suppression, low distortion, and built-in protection against shorts and thermal runaway.  There isn't much more to ask of a power amp than that.  When driving electrostatic speakers, you can't have too much protection!

Download a data sheet and app notes for the chip here.

There are some people who claim to have "golden ears" and feel that no application note-supplied schematic diagram is ever good enough so they insist on "tweaking" to make "improvements".  The problem is that most of them are not engineers and have no idea what the potential consequences of their "improvements" may be.  For example, for a few years when these chips first became popular with the audiophile crowd, it was all the rage to use minimal power supply filter caps for "best sound".  We're talking about 500 uF on each rail of the supply for each amplifier chip used.  This is woefully inadequate and leads to distortion at fairly low volume levels as the power supply sags under the load.  The problem was that some of the golden ears saw the IC's great power supply rejection ratio spec and figured that it meant the chip could tolerate 10V of ripple in the power supply.  The pendulum has swung the other way and now many audiophiles are putting proper amounts of energy storage into the power supply.

Another common tweak was the elimination of the "Zobel" network at the output of the amplifier because it "sounded bad".  After a few people burned up some expensive speakers they came to realize what the network is for (to prevent high frequency oscillation) and that dead speakers sound much worse than Zobel networks, so these days most people are using the Zobel network.  I have used my amp to drive electrostatic speakers, a notoriously difficult load due to the capacitive reactance, and never had ANY stability problems whatsoever.

My amp was built using a schematic diagram pulled right off the NS data sheet, with the power supply added.  Maybe the tweaking mentality about manufacturer supplied schematics comes from not understanding proper construction practices.  One thing that is key to getting good performance from an amplifier is to use proper grounding.  Grounding isn't normally shown on a schematic diagram because it depends on the mechanical layout of the amplifier circuit and chassis.
 Construction practices such as keeping high impedance circuits (such as amplifier inputs) away from components and wires carrying large currents (power transformer, power supply and amplifier output wires) and using a "star" ground contribute more to high performance than $10 resistors and gold plated connectors.

Usually, the housing is the most expensive part of the amp. It has to dissipate heat from the amp ICs and power transformer, it has to look acceptable, and it has to be large enough to hold everything.  My amp's housing came from a scrap yard in Dallas where I was able to buy it for $1 per pound- about $15.  The box originally housed a cell phone diversity antenna amplifier.

This is the box I bought at a scrap yard in Dallas.  It is a diversity amplifier and switch from a cell phone network.

This is the stuff that was inside the box.

The side walls are 1/4" thick aluminum, welded at the corners, making this a super sturdy box that can dissipate a lot of heat.  The top plate is 1/2" thick and is covered with  fins to dissipate even more heat.  The bottom plate, not shown, is 1/8" thick.

There were a bunch of connectors going through holes in the sides of the box.  I was able to use some of them, but not others, so they were filled using Bondo auto body filler.  First I taped some scraps of PCB material to the inside of the box then smeared the Bondo into the hole from the outside.  After the Bondo hardened, I sanded the outside surface flush with the rest of the aluminum.  It has worked out quite well and I recommend this technique to anyone who is recycling surplus enclosures.

Unneeded holes filled with Bondo auto body filler.

I added a few more holes where I needed them, enlarged a couple of the existing holes, then applied the finish..

One of my other hobbies is restoring antique radios.  I have one very special radio in my collection- a 1927 Neutrowound Super Six.  This radio must have looked like it came from outer space back in 1927.  At that time most radios were plain, rectangular wood boxes.  This one was is a steel box with a very unusual look provided by the nickel plated tube covers that protrude through the top, and the exotic painted finish- cracked blue over orange!

Neutrowound Super 6.  Neutrowound didn't last long after they made this model- they clung to battery power while the rest of the country was stringing AC power lines everywhere.

I like the finish on that radio a lot so I decided to try to duplicate it on my amplifier.  The radio was spray painted but I didn't have access to spray equipment so I had to use brushes to apply the finish.  I selected a beautiful metallic blue (not quite the same as the radio) and a bright orange for the base coat.  The result wasn't a very good match for the radio, but it came out sort of interesting anyway.
The finishing steps were primer, orange acrylic base coat, acrylic crackle medium, metallic blue acrylic, and finally an acrylic clear coat for protection.


Orange base coat, then a layer of clear crackle media.

Blue coat added after crackle media, and finished with clear coat.

Next came construction of the amplifier. The first channel was hay-wired because so few parts were required, but  I didn't like the look of it much, so the second channel was built on perf board.  The plan was to eventually go to printed circuit boards (more on that later).  All grounds were brought to a single "star" point on the chassis to minimize the possibility of ground loops and associated hum/noise.

This is the literal configuration of a "star" ground.

320VA power transformer, more than adequate for a 30W per channel amplifier.

The filter caps in the power supply total about 100,000 mF, more than enough, but at such a low voltage, capacitance is cheap ($6 total for the four caps), so why not?
The volume control is mounted close to the input jacks on the rear panel to keep the signal wires short and far away from the power transformer.  The shaft extension is a piece of 1/4" diameter brass tubing.  Alignment of the volume control shaft and the front panel hole were not perfect so I made a flexible coupling from a piece of urethane air tubing and two tie-wraps.  It works well- if you turn the knob past the stop in the pot, it slips a little, limiting the stress on the pot.

Eventually I got around to ordering some printed circuit boards and rebuilt the two amplifier channels.  I prefer PCBs because circuits that are hay-wired together may develop short circuits if  wires have to be moved around to perform repairs.

Vital statistics:
Weight: 18 lbs, 2 oz.
Size: 13" x 10" x 5"
Power transformer: 2x 18V @ 8.9A
Caps: 4x 27,000 mF @ 35V
Supply Voltage (no signal) +/- 26.5V
Volume control: ALPS dual 20K

Yes, of course, it sounds great!


  1. I like that you paid attention where it's needed, grounding and plenty of filter capacitance. Then, the amplifier is not limited by anything outside it's design. Well done!

    1. Thanks! I've been out of the audio thing for several years now, but yes, good grounding and clean/adequate power are always a good place to start.


Leave comments or a questions here and I'll try to post a response as soon as I can.