Little Lebowski Urban Achiever (LLUA) Mark I Power Amplifier

s y n o p s i s
	Build a cheap amp: battery powered and portable
	Use a cheap power amplifier chip
	Choose the right cheap version
	Drives your pair of cheap 8Ω speakers
	Excel at being cheap

What happens when you take an underprivileged chip and try to raise it up by its bootstraps? Success? Well that depends on your definition of success....

For our purposes, we want to build a portable stereo amplifier that accepts an Ipod line level or headphone jack signal, and can power a pair of small 8Ω speakers.

In this project, we'll build a stereo power amplifier based on the lowly LM/JRC386 power amplifier chip, a device most commonly found in devices where sound reproduction fidelity is the least important design consideration. The 386 is found in a myriad of crappy sounding boxes, like clock radios, 10 dollar portable amplifiers and other devices of dubious audio performance.

But it is cheap, sounds relatively decent if tuned right, and can be powered by anything from a 9 volt battery up to an 15 volt DC wallwart.

Design Parameters

In designing this amplifier, we want:

	Cheap	cheap parts, cheap components, and a cheap power supply.
	Power supply flexibility	from a 9 volt battery up to 18vDC from a wall wart.
	Portability	the ability to disconnect from our house or office and wander the woods with a stereo amplifier of small size, just like Ann Coulter's heart.
	R&D budget not to exceed 12 dollars	cheap parts, cheap components, cheap wine, and a cheap power supply
	Power Output	claim that we can output 1 watt, but that's into a 32Ω load, so in reality we'll generate about half a watt. Downhill. With a backwind.

It just so turns out that the 386 easily meets all those goals.

But Which 386?

There are many variations on the 386 chip from several manufacturers. Here's an overview:

Part #	Manufacturer	Supply Min	Supply Max	Output Power
LM386N-1	National Semiconductor	4	12	325mW @ 8Ω, 6v supply
LM386N-3	National Semiconductor	4	12	700mW @ 8Ω, 6v supply
LM386N-4	National Semiconductor	5	18	1,000mW @32Ω, 16v supply
NJM/JRC386-D	New Japan Radio Company Ltd	4	12	500mW @16Ω, 9v supply

You'll get the most power out of the LM386N-4, and the LM parts are usually the most easily obtainable. However, I have found the stability, reliability and performance of the NJM/JRC part to be better than the National Semiconductor parts. Your mileage, of course, will vary. You can install a DIP-8 socket on your protoboard and try different versions of the chip. See which one you like the best. If you are going to power the circuit at 15 volts, the LM386N-4 is an awesome part.

The Schematic

Here is the LLUA Mark I schematic. It is divided into power supply, left channel, and right channel so you can bask in the design mastery I have exhibited by slightly modifying a bog-standard datasheet example and tacking on some power supply filtering. The 386 is distinctive in that is be used in everything from a guitar distortion pedals, to mass-produced clock radios to DIY projects like the venerable Headbanger headphone amplifier. Many designs, like mine, are slightly-modified derivatives from the National Semiconductor datasheet.

For my version, I made a few small changes.

The 386 has a very narrow band of clean headroom. It will distort quickly, so I wanted to be sure to tweak the input signal and match it to the highest pre-clip gain factor of the chip.
Since the LLUA designed with the output levels of an IPod in mind, the R3/R5 pair at the input takes care of taming the input signal a bit. It works great for line level, and for headphone output, you may have to lower the volume on the Ipod just a bit.
I left off the usual gain components you will find between pins 1 and 8. Anything other than the stock gain of 26dB (pins 1 and 8 disconnected) and it will start to distort. So there isn't much point in adding components to increase the gain.
The power supply section has a diode (D1) in place for reverse polarity protection. That's because I get tired of burning out chips by hooking up the wrong polarity AC/DC adaptor :)
Finally, I've found that a 100nf cap for pin 7 bypass works well across all the power sources I used.

(Click on the schematic for a larger version)

Let's Build One!

Now that we have the general idea of what it is, let's build one. First off, the parts list.

Part #	Description
Semiconductors
U1, U2	JRC/NJM386D or LM386N-4 power amplifier
D1	1N4002 diode
LED1	LED power indicator
Resistors, all 1/4watt 5-10% tolerance, metal film
R1	1KΩ (Limits current to the LED, adjust to taste)
R2	47Ω
R3, R5	22KΩ
R4, R6	100Ω
Potentiometers
VR1	B10kΩ Dual-Gang, Log or LinearTaper
Film Capacitors, 50-100 volt
C2, C5, C9	47 nanofarads
C4, C8	100 nanofarads
Polarized Electrolytic Capacitors, 35-50 volt
C1	100uf, although you can use 220, 470 or higher)
C3, C7	10uf
C6, C10	220uf
Hardware
SW1	SPST switch
J1	2.1mm power jack (or whatever works for your power connection)
J2	1/8" stereo jack
J3, J4	1/4" mono output jacks, although you can use speaker terminals or whatever is convenient for you to cheaply connect cheap speaker wires.
Enclosure	Whatever you like, preferably metal, and definitely cheap

I used a protoboard from Wright Hobbies to get everything mounted.

Enclosure and Finishing Up

I used a Hammond 1591DTCL clear polycarbonate enclosure. I drilled all the holes, used stand-offs to mount the board, and some euro-style terminal strips for the speaker connectors. Turned out rather nice.

(Click images to zoom)

Use and Listening Tests

To properly test my masterwork, I first departed the beavis basement and headed upstairs to the kitchen for a nice little glass of $7/btl Chianti. For input, I used the wonderfully convenient Sendstation dohickey connected to my ipod. The speaker connectors were then attached to a pair of my Stealth Cheap Speaker Buy of the Year™. Speaker wire was some nasty looking old Radio Shack grey stuff, but then again, I could never discern an actionable problem with cheap speaker wire.

The first test was using a 9v battery as a power source. It sounded passable. Mids and trebles were there, but the bass was a bit flaccid and otherwise droopy. Jumping up to a 12v 500mA source improved that. But not by a huge amount.

Overall conclusions? It was fun to go back over all the LM386 experiments I'd done in the past. And to work out some basic changes that would make the same chip behave nicely in the hi-fi world. Bottom line is that this design will sound slightly better than if you hooked up your clock radio hooked up to better speakers.