Want to build a SS PA

[lowell]

JM,
So is it or is it not ok to run 2 LM3886 circuits in parallel into 4 or 8ohms? Right now I'm building an amp with 1 chip. However I'd like to build another one with twice the power and use 2 LM3886.

Secondly the datasheet says maximum differential input is V+ or V-. So then my supplies being +/-30v, my input voltage should be no more than 30v peak to peak?

[J M Fahey]

I don't like running zero internal impedance generators in parallel, meaning I don't run SS poweramps in parallel, period.
It's perfectly safe to run them *bridged* though.
The beauty of chipamps is their simplicity, once you double or triple it you are better off with a simple discrete amp.
Besides, all parallelling accomplishes is allowing it to run 2 speakers instead of one. Big deal!!
In my mind the *practical* options are:
1) Single 3886, with adequate PSU voltages to run, say, one good (Jensen/Eminence/Celestion) in 4 or 8 ohms to over 50W RMS: good loud sound to play your typical pub. Equivalent to a Bandit/Fender Performer , etc. 8 ohm speakers being most common.
2) Two 3886, each driving its own 12" as above. *Loud*, good for larger venues, a theater, playing in the open (not Woodstock but some local open air event) Equivalent to a Twin/CrateG212, etc.
3) Two 3886, voltages for 50/4, bridged, they supply 100/8, into a *beefy* 12" (one of those 150W Eminences, an EV, etc.) *Loud* as in 2) but into a very compact package (it will be heavy, though)
You can also build it into a head and drive an 8 ohm 4x12" (or a lighter/smaller 4x10")
4) A practical option : build a 1x12" combo as in 1), but with two 3886: one powers the internal 12"; the other goes to a jack to drive an extension 12".
You have a compact light 50W amp for general purpose use, which turns into a 100W 2x12" one for larger venues.
5) Forget about parallelling, even more parallelling/bridging, Marshall MF350 has proven that it does *not* work.
Good luck.
EDIT: I forgot. If your amp runs from +/-30V and your preamp from the same or less (+/- 15V) there's no way you can have more than +/-30V peaks, so relax.
The *only* way you could have a problem with that is if you drive your chip straight from, say, a 12AX7 fed 250/300V (a Twin preamp or similar). And even then, a couple back to back 15V Zeners come to the rescue.

[R.G.]

National Semiconductor has an application note that shows circuits for paralleling, bridge, and bridge+parallel connections for the LM3886. As JM notes, paralleling low impedance sources is tricky. National does it by using precision resistors and small-value resistors in series with the "parallel" outputs to equalize currents.

I very much agree with JM's #2. Drive one 12" from each LM3886. One LM3886 is a good match to the cheap toroids and one 12. In fact, my favorite scheme is to make on speaker box per 12", and put the LM3886 into the box. With a good preamp setup you can then take as many speaker boxes as you like with you to suit the loudness to the venue, and not have nearly the problem of carrying speaker cabs.

I like soft limiting ahead of solid state power amps. You can make a soft clipper from some diodes and a pair of small MOSFETs that will do no clipping up to a volt or two, then soft clip above that. Set the input sensitivity of the power amp by changing the feedback resistors so that the clipper activates before the power amp does, and you'll never have the power amp clipping audible.

[lowell]

I like soft limiting ahead of solid state power amps. You can make a soft clipper from some diodes and a pair of small MOSFETs that will do no clipping up to a volt or two, then soft clip above that. Set the input sensitivity of the power amp by changing the feedback resistors so that the clipper activates before the power amp does, and you'll never have the power amp clipping audible.

RG could you draw a schematic of this soft clipper circuit or direct me to one?

[lowell]

Okay so 1 LM3886. What kind of "bleeder" resistors could I use on these 10,000uf caps? I'm thinking something less than your typical 220k is suitable. Also, for a mains fuse. Fyi I'm using a thermistor (same one as in Fender Deville) 60watts\120v=500ma. So a 1a fast blo mains fuse?

[R.G.]

RG could you draw a schematic of this soft clipper circuit or direct me to one?

I'd have to draw it up and upload it. May take a day or two.

What kind of "bleeder" resistors could I use on these 10,000uf caps?

I'd decide how much power I could waste in the bleeders and calculate resistance from that.

If you can afford to waste 1/2W, you will need 1W resistors. The voltage on the caps will still be about 32Vdc each. Power is voltage squared over R, so R = voltage squared over power.

R = (32*32)/0.5 = 2048 ohms. That says that a 2K resistor across a 32V source will disssipate 1/2 W.

When power goes down, if the bleeder is the only load (it's not, but say it is) then the voltage declines to 63% of its previous value every time constant, which is R * C.

So for 10,000uF, R = 2K, time constant is 2000*0.01F = 20 seconds. So the voltage is
32V*0.63 = 20V at 20 seconds, 12.7V at 40s, and 8V at one minute. Five time constants is generally assumed to be steady state, and that's 3.2V at one minute forty seconds.

In point of fact, the amp will continue to suck the filter caps dry unless it's very much damaged.

I'm thinking something less than your typical 220k is suitable.

I would do the same thing for a tube amp; decide how much power to waste, then pick the resistor for that power. 220K and 50uF is a time constant of 11 seconds, so the run down to five time constants is twice as fast, 55 seconds to 49.6V from 500V charged.

Also, for a mains fuse. Fyi I'm using a thermistor (same one as in Fender Deville) 60watts\120v=500ma. So a 1a fast blo mains fuse?

Never use a fast fuse for a mains fuse. The inrush surge will pop them until they're a nuisance. Start with a 1A slow blow.

[lowell]

"Never use a fast fuse for a mains fuse. The inrush surge will pop them until they're a nuisance. Start with a 1A slow blow. "

Even if I'm using a thermistor?

[R.G.]

"Never use a fast fuse for a mains fuse. The inrush surge will pop them until they're a nuisance. Start with a 1A slow blow. "
Even if I'm using a thermistor?

Yep. A fast blow fuse doesn't help for what primary/mains fuses do. They do not protect anything in the amp except accidentally. They are there to stop fires from starting. The thermistor would cut down on inrush, maybe even enough. But my personal preference is to use slow blow.

I guess the right way to say that is that a thermistor and a fast blow *might* work fine. A slow blow will work if sized properly.

[lowell]

Ok sounds good thanks RG!