problem with paralleled LM3886s, help?

[mnturner]

hey guys,
i've just finished a paralleled LM3886 based power amp and i'm having a weird problem.
The two LM3886 amplifiers are on separate boards and have their inputs connected by a wire between the two boards.
One board has the zobel network and damped series inductor, the output of the other board again is connected to this one by a wire.
When i turn it on, even if the volume pot is set to 0, a weird low pitched noise, which isn't particularly loud, comes out of the speaker and the mains fuse blows after a second or two.
If I disconnect the wire connecting the outputs of the two boards however, both of them work fine. this problem only occurrs when the outputs are actually connected.
I've read that if the two circuits outputs aren't well matched, they might try to 'drive each other'? i have used 1% resistors in all places that have an effect on the output signal.
I've attatched the schems of both boards. Where it says 'to board 1' etc are the points at which the two boards are connected.

Anyone have any idea what the problem might be/ how i can fix it?

edit: forgot to mention, the series inductor is 12 turns of enamelled copper wire around the 10 Ohm resistor on the output of the first board. and the reason I have used 5 parallel resistors at the output of each is that the LM3886 documentation recommends .1 Ohm resistors but I couldn't get 1% accuracy .1 Ohm resistors with high enough power rating, so I used 5 x 1 Ohm resistors to get close. 

[mnturner]

just got a 'scope on it and, with a square wave input, as i increase the volume pot there is high frequency ringing on the square wave which gets worse as i increase volume, then at about 2/3 volume it breaks out into high freq. oscillation. this is happening with both boards in exactly the same way when they are not connected to one another.
does this sound like a ground loop or a problem with my feedback loop? i can't figure it out.

[phatt]

I maybe missing something here but at a glance,,, this is the worst possible way to blow up 2 amps.
I think you need to do some research.
You can *Bridge* two outputs and send out of phase to the inputs but then the speaker terminals float from ground.

Sorry to tired to think now.
Phil.

[mnturner]

  I think the schems I posted are unclear...I'll redraw and post it soon.
I'm not trying to bridge them; bridging would give me twice the voltage swing and twice the current draw from each amp as they see half the load impedance each. I'm paralleling them so they see twice the load impedance giving half the current draw from each amp.
The circuit as a whole is pretty much taken from the LM3886 application notes, with some alterations of my own which I've probably screwed up causing my problems.

 

[phatt]

pin 1,5, 3 look funny so I wait for a better schem.

Meantime,,,Is this what you are trying to build?
http://www.shine7.com/audio/pa100.htm
Phil.

[teemuk]

I think that's it. The vital parts in the project are not only the current sharing resistors in output and input but also the fact that most of the resistors are matched at least 1% and in many cases as close as 0.1%.


If I remember correctly, the National's application notes for a parallel LM3886 design also encourage the use of DC servo.

[J M Fahey]

Sorry to bring bad news, but parallel output power amps spell B-A-D E-N-G-I-N-E-E-R-I-N-G to me.
I hate the concept. The (in) famous Marshall Mode Four, probably the most unreliable guitar amp ever, just uses 4 x TDA7293 amps, paralleled and bridged.
Anyway, it *may* work, sort of, but not on separate boards.
please: what power into what load (ohms) with what supply voltage are you using?
Yea, I know, it's an "official" application note, yet .....  :loco What were they thinking?

[mnturner]

Hey,
here's the new schem; this is how the amp is looking right now after some tinkering.

Phil; yeah, I saw that guy's site and pretty much based my circuit off that with a couple of small changes, eg. the inclusion of the volume pot. The pinouts are correct, I just had to mess around with the symbol in EagleCAD to get the wires to match up with the pins.

Teemu; after it initially didn't work, I hand matched all resistors in the signal path (input, feedback network) to about 0.5%. I think the app. note only really specifies the use of DC servos in the 200W bridged/parallel circuit, it seems to suggest you can get away without them, as the successful builds on DIYaudio seem to have.

JM Fahey; I was a little skeptical about it when I first thought about attempting to build it, but after I saw the guy's site that Phil posted, as well as a bunch of mostly successful builds of this type on DIYaudio I decided the project must at least be plausible? Not that I'm disagreeing with you; you clearly know more about this that I do, but why exactly would it be impossible on two boards? I understand that connecting them by wires introduces another potential source of interference, but i tried to keep the wire connecting the inputs as short as possible (about 3-4cm, the boards are mounted one above the other), and i can't imagine that connecting the outputs by a wire could be a problem interference-wise.

The power supply is an unregulated 24-0-24 t/f with 9,400uF at each rail, which is putting out about +/-38V.
I have tried with 4Ohm and 8Ohm loads as well as unloaded and the same problem occurrs.

Most testing has been done unloaded, as I can't imagine that this problem can be good for speakers.
At this stage, both boards work perfectly with the outputs unconnected, but whenever i connect them, they oscillate rail to rail at high frequency and as a result heat up very quickly.
I have connected the positive and ground terminals of my scope probe to each output (positive end of probe to output of one board, ground lead of probe to output of other board) with the boards unconnected, in an attempt to get an idea of how different the ouput signals were, and there was no visible difference *except* when the volume pot (25k, log) was between about 2 and 4 0'clock, when there was a high frequency signal of about 150-200mV difference between the two outputs. When the volume was turned up more that this, this signal dissappeared and the difference once again became negligibly close to 0mV.
I tried to tested this because I was under the impression that the biggest problem with the parallel setup is that differences in the outputs cause the two amps to drive each other, although I don't know if that would actually cause the oscillation I'm seeing at the output when the boards are connected.

I'm pretty confused. I'm open to the possibility that the fact that the two amps are on two different boards is causing problems, as JM Fahey suggests, but I can't think of any other possibilities as individually the boards work fine and there appears to be so little difference between the outputs (except for that weirdness I just described).

Anyway, thanks for the replies. Anyone have any other ideas with this new info or should I just scrap the whole thing?

[phatt]

You know,,,
with respect,,, a one wheel car might have a better chance

JM Fahey and Teemu are 2 Seriously talented chaps with years of experience under the hood. So I think if it was a great idea they would have already persued your concept.
I do understand that when you start out playin with stuff you get all sorts of crazy ideas that at the time look like they may just work,,, hum if I do this,,, and IF I can find some rare .1% poofoo valves.
Then you end up filling your shed with square wheels.
If it makes you feel better,,, I've got a whole shed of (good idea at the time) projects that never worked.
In close on 30Years of messing around with 100's of circuits I've only designed/built a couple of circuits that I'm seriously proud of.

If you can't let it go then may the force be with you 8|
Phil.

[teemuk]

There's no reason why a parallel circuit couldn't work but it is extremely sensitive to matching of the amplifier blocks, proper noding, layout etc.

A schematic only tells us half of the story.

[rowdy_riemer]

In close on 30Years of messing around with 100's of circuits I've only designed/built a couple of circuits that I'm seriously proud of.

That reminds me of something a professional photographer told me. He said he only liked about 2 or 3 pictures out of every hundred he took.

[mnturner]

haha ok, so i'll probably end up looking at this from that perspective and calling it one of the hundred pieces of junk i need to get through before i end up with something i'm proud of.
i have to admit the main thing thats keeping me going with this is stubbornness :lmao:; having read about other people building these successfully i'm frustrated by not being able to do it myself, also i'm extremely curious as to what the problem actually is.
I think I'll redesign the PCB and make sure there are no grounding issues, with both chips on one PCB.
If that fails I'll give it up and take one out, that way i've still got a single lm3886 gainclone. (even if it is only 50W into 8Ohms  )

[teemuk]

Alternatively, you have yourself a stereo setup. ;-)

[armstrom]

Alternatively, you have yourself a stereo setup. ;-)

That's the way I plan to go with my dual LM3886 setup. Of course there's nothing wrong with a 100W+ single speaker amp, but the reality is that most amps in that power range have a pair of 10 or 12" speakers. There's nothing to stop you from having a mono preamp split into both channels. Then you can still claim the 100-120W power output and include a stereo effects loop for next to no effort. Just makes for a much more flexible build overall. Keep in mind the Roland JC-120 was also "just" a 2x60W stereo amp with mono inputs, some built in stereo effects and an external stereo effects loop.
-Matt

[J M Fahey]

In a nutshell, the problem is this:
modern SS amps are practically perfect voltage sources, with "almost" zero internal impedance.
They achieve that by having massive internal gain and equally massive feedback, which for DC is almost always 100%.
That's why the output DC voltage , which could easily slam against a power rail, gets reduced to less than 50 milliVolts.
They do that by supplying to the load as much current as needed (tens of amperes if necessary) to bring output voltage to where the input differential amp says it should.
Now imagine two amplifiers with joined outputs: you have two "brains", absolutely ignorant of each other, driving the "muscle" (output transistors) to get a certain output voltage, on which they do not agree. I do not care about the load, they will try to force the other equally powerful opponent's output to submit to their will, killing each other in the process. Or at least oscillating madly, overheating, the works.
Neanderthal solution nº 1): put a 4 ohm 50W resistor in series with each output, the node feeds the speaker. It works, sort of, although you gain no power and waste a lot.
A somewhat-less-Neanderthal solution nº2) : you match all resistors to 1% or better, match chips from a large batch or at least use those from the same box, same manufacturing code, draw a special board where all traces have the same resistance and inductance, mount them to the same large, thick heatsink to minimize thermal differences, lower the 4 ohm resistors to 0.1 ohm, relax and think "what a great Engineer I am " . Well, not bad for a Neolithic village Engineer in around 4.500 B.C.
Yes, it may work, but, why worry?
Increasing chip count carries the complexity up where it pays to use a boosted-chip or even discrete design, where you can design for as much current as you want because, remember, parallelling chips you only get more current capability, nothing else.
As an example, in the 70's I sold a lot of 400W RMS into 1 ohm bass amplifiers, using 8 x selected RCA or Motorola 2N3055H as output devices. They drove 8x12" LEEA speakers (the local ALTEC licensed factory), consider I used the equivalent of 8 Altec 12" woofers in big boxes,  they *easily* beat Ampeg SVT's in Stadiums.
3) The real beauty of using 2 chips lies in using them bridged, because they supply *a lot* of power into a single 8 ohm speaker.
4) Of course, Teemu's and Armstrom's idea of using them "in stereo", each one driving its own speaker, is excellent. Even if you are not interested in stereo per se, you can send the signal from a single preamp to both of them.