how to get as much clean power as possible from 386 chip

[snoof]

I've heard that the NTE458 has more gain than a J201, but I don't know if that translates to buffer useage.

[R.G.]

A power amp is best thought of as a power SUPPLY that has some added stuff to let some of the power out under specially controlled conditions. This is true of ALL power amps, in ALL circumstances. You just can't get more power out than your power supply can supply. The power amp parts will limit this further.

Once you specify a load resistor - like an 8 ohm speaker - you have locked in what power can possibly be obtained from your amp's power supply. You don't particularly need to know anything about the amp to compute what the maximum possible power is. The real power with a real amp will always be less.

Example:
Let's say you have a 9V battery that can supply 10,000A of current. How much power can that supply put into an 8 ohm speaker if we don't use power conversion tricks to up-vert the voltage?

If we had a perfect amp, it would produce a peak of 9V and a valley of 0V across the speaker. If the signal is a sine wave, the way the EIA says to measure power, then the speaker is getting a sine wave of 9V peak to peak. That's 4.5V peak, which is 4.5V/1.414 = 3.182Vrms. The power in the speaker from that is V*V/R = 1.27W. That's all there is. Any real amp cannot produce an output of exactly 9V and 0V, but only something less, like maybe 8.5V and 0.5V, or 8V and 1V. The power from that is correspondingly smaller, by the same calculation. The amp is voltage limited in what it can supply.

OK, let's say that the speaker is 2 ohms. What's the power?
Same calculation, 3.182^2/2 = 5.06W. This is the same trick that Thomas Vox amplifiers played. All of their bigger amps used the same power supply voltage, +/-31V. This is sufficient to make 30W into 8 ohms. The baby amps (Buckingham and Viscount) use an 8 ohm load and produce 30W. The Royal Guardsman uses a 4 ohm load and produces 60W. The Super Beatle uses four 8 ohm speakers in parallel for 2 ohms and produces 120W from the same power supply voltages.

The power can be current limited as well.
Let's say we want to do our 5W amp from a 9V battery. Will that work?
No amp is perfectly efficient. A push-pull amp is about 78% efficient at best when it's driven to full power. That means it wastes 22W for each 78W of audio it produces. So our 5.06W output needs 6.48W of DC power. At 9V, that's 9/6.48 = 1.38A.  A 9V battery stores something like 0.16 to 0.48 Ampere-hours of charge, so the battery (if perfect; it's not, so the reality is worse) would last 0.115 to 0.347 hours. That's 6.9 to 20 minutes. It's going to eat a lot of batteries. 8-)

If you want more power, you need to supply more power. Some quick choices are the LM386 at up to about a watt, the LM1875 at up to 20W, and the LM3886 at up to 50W. National Semi actually makes a new power amp driver chip that will drive discrete power transistors at up to +/-100V. If you can feed it the power from an appropriate supply, you could theoretically get something like 300-500W out of that setup. But the power amp driver and transistors will be a trivial part of the expense. The power transformer, rectifier diodes, filter caps, and *wiring* will cost you about $0.50 per watt if you buy new. Not to mention being deadly dangerous to mess with. The OUTPUT of the amplifier would be an electrocution danger.

[Mark Hammer]

To reply to a few well-argued points....

1) Larger speakers DO move more air, and hence are more "efficient"....when they start moving.  However, moving more air, and moving it in a linear fashion are two different things.  My bias is that if you want maximum clean volume and ideal bandwidth, then the goal is a balance between smaller speaker size and overall efficiency. Birt's use of a pair of smallish-medium speakers strikes me as ideal and exactly what I'm advocating.  A 3" speaker will likely deliver all treble in the guitar signal but, certainly lack in the bass department and the overall air-movement department.  Conversely, attempting to bludgeon a 12" cone with a half-watt is likely to yield a fairly bass-heavy signal with a muffled top-end.  That may be great if the goal is creamy distortion all the time, but not so great if the goal is clarity and capacity to do things like reproduce Strat cluck or Tele twang.  When it comes to 386-based amps, something in the 4"-8" range is the ideal.  You'd be surprised how much bass you can get out of a 6" in the right cab, while still getting plenty of bite.

2) How loud is loud enough, and how much wattage is really required for loud enough?  Owners of single-ended 6V6 or EL84 amps can tell you with great conviction that you would be very pleasantly surprised by how loud a measely 4-6W can be is the speaker and accompanying cab is efficient.  Not necessarily enough to compete with a drummer and bassist, but more than loud enough to break leases or have people yell at you to turn it down.  Two watts can be pretty loud.

3) Although the various 8-pin and 14-pin power-amp chips can often handle supplies of more than 9v, and loads as low as 2-4 ohms, the datasheets often indicate that the chip does not fare well with both lower speaker loads and higher supply voltages.  In other words, do not plan around running a 386 unsinked with a 3.2R speaker and 12V supply......at least not for very long.  Of course, supply the right amount of heat-sinking/siphoning and perhaps the chip would be able to pass the amount of current required for that wattage into that load.  All who suggest there is a diminishing return, given the awkwardness of sinking an 8-pin dip versus a TO-220 style chip (e.g., LM1875) are correct....but that doesn't mean it can't be done.  The constraint is coming from what is commercially available not from what is possible.  For example, there are mountains of heat sinks for old 486DX CPU chips sitting around, whether in surplus places or on your old clunker that you keep meaning to excavate from the basement.  These often have closely-spaced fins on them that are 3/4" high.  Why not take a hacksaw to one and tranform it into a bunch of smallish footprint vertical heatsinks?  You could probably fit 4 fins or so over an 8-pin chip.  Splat a little thermal compound between heatsink and chip top and secure it to the top of the chip with a piece of solid-core wire running over it and soldered to the board, and the chip could likely take a little more heavy-duty service.

[birt]

Your speaker theory is well explained. In the ideal speaker you would want to move as much air as possible and a cone without any weight. A big speaker has a heavy cone and will be slow, so the higher tones get less accurate or dissapear. A small speaker has a light cone but has problems with the bass tones.

the closest you can get to the ideal speaker is to use a lot of small speakers together so you move a lot of air with tight bass and enough treble. something like the 5" speaker bass cabs http://www.philjonesbass.com/PRODUCTS/SPEAKERS/8T&24B.htm

[nokaster]

i plan on running my ruby with bassman mod, with a 386-3 chip on a 9V power supply through an 4 ohm speakerload. (2x12")
will i need a heatsink, or will the chip be able to handle the 4 ohms without problems?


thx!

[Mark Hammer]

You'll certainly hear it through a 2 x 12 cab, though that's a whole lot of cone to move, and at this point I have no idea of the sensitivity or rating of the speakers in question.

As for the lifespan of the chip, I would imagine that a 386-3 would be more or less comfortable in the setting you describe, as long as you aren't pushing it too hard at the front end.  remember that the current it attempts to pass (i.e., wattage produced) is a function not only of the power supply and load, but the instruction that the input signal and gain setting directs the chip to produce.

Looking at it from the other end, you could probably plug your $2 bargain store headphone to the output terminals of a 500W Crown power amp, even though they are really intended for use with chips that put out 400mw in a cheap radio.  The headphones would last forever under those circumstances, as long as the input signal does not direct the amp to crank out all the current it is capable of.  So, the converse is true as well.  Keep the signal level modest enough, and it should all work out fine.  But again, is 1/2W enough to push all that cone?  Maybe, and maybe not as much as you'd like to think.

[Timebutt]

I think the chip will be just fine connected to the 4 ohm cab, my 386 (bassman mod as well) is running at 12V and I usually run it with the volume and gain completely open: chip doesn't get warm, just a bit after a while. As for volume: it depends a lot on the sensitivity of the cabinet but I think it puts out a fair amount of volume for such a small circuit. Certainly enough for a practice amp.

I really like the overdrive my Ruby produces, but then again: I mic the little cabinet and send the signal through a medium PA system, sounds really huge running it through that setup

[Nasse]

I think cutting bass frequencies before amp helps (was this said yet)

[PerroGrande]

Another great thread! 

Another resource to check out when it comes to amps, power, efficiency, and the like is Elliot Sound (ESP).

http://sound.westhost.com/

Things that are certainly on my Christmas list for next year -- that 9v 10,000A battery that RG has... I don't want it to run an amp... just to run *all* my pedals... forever...  (and maybe for some welding).