Reducing output voltage

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello!

I'm building an amp based on the LM3886, and now (I think) I've got some problems with the voltage supply. The supply gives out about ± 44 V wich is a bit too much for the chip (isn't it?). The amp works pretty well for a few minutes, and then the chip blows.
Is there any simple way to reduce the voltage to say ± 35-40 V, or do you know any othe chip that can handle that much voltage?
I'm using a 230V/2x30V 260 VA toroid in the power supply. I already thought of reducing a few turns from the secondary windings but found it impossible, because the toroid lies in a plastic "cup" wich is filled with some sort of white glue.
The easiest (and the most expencive) way would probably be to buy a new toroid, but that's the last option.
Any suggestions?
 
Technically, +-44V is too much. The chip is spec'd for up to +-42 V, so you aren't too far out of range. If, however, you are using the chips in a bridged amp and/or you are trying to drive a load of less than 8 ohms, then you definitely will run into problems with that voltage. 8 ohm would be the absolute minimum for a single chip, and you can forget about bridged altogether.

The chip can survive +-44 V, but the key is reducing the power dissipation so that it will remain happy. To do this, you need to design the amp so that less current flows (for a given output voltage). Less current translates directly into less power dissipation in the chip. There are two ways to accomplish this: us a higher impedance load, or use more than one chip in parallel.

If you use two (or three or four) chips in parallel, they will each provide a fraction of the current output, and therefore each chip heats up a lot less.

If you are using a 8 ohm load, I would recommend two chips in parallel. Using just one might be OK, but only if it has a very large heatsink and is in the normal (T) not isolated (TF) package. If you are using a four ohm load, consider Three chips or maybe even four. You can drive an almost arbitrarily low impedance by just paralleling more and more chips (and as long as your power supply is up to it). Refer to National's application note AN1192 for schematics of parallel amps. (it is not as simple as connecting the pins of multiple chips in parallel - that could result in disaster).

You didn't mention how big your heatsink is. You are using a heatsink aren't you? It is not optional.
 
dhaen said:
If there is a centre hole in the toroid, you can add some turns, and connect in series - out of phase. It will need 2 sets of new windings.


jackinnj said:
if you put in more "secondary" turns you increase the ratio of secondary to primary, ergo hoc the output voltage goes up.
Although I wasn't clear enough about which winding, I did mention the phasing ;)
 
Whow! You guys reply fast!

I did some further research of my own, and found this regulator circuit with just one zener-diode, a resistor and a Darlington. Just finished building it, and so far it seems to work fine (haven't tested it under load yet). At idle it gives +36.7 V and - 35.5 V with about 0.02 V ripple. Is that good enough for the amp?

The amp I'm building is based on a single chip and yes, I AM using a heatsink:). It will be driving a small home theatre subwoofer with Focal 27 V driver (2x6 ohm voicecoils).
 
The schematic is from Elektor electronics May 1995. Actually I bought the amp several years ago as a kit from a local electronics store. I used it to drive my computer sub for a while, but now it has been unused for a long time. I really don't know what type of amp it is, but the schematic looks a bit more complicated than any schematics I've seen here so far.
I'm too a bit suspicious if it will be able to drive the sub, but if it won't I'll just build another one and drive each voicecoil with its own amp. The sub won't be driven at any insane SPL as I live in a flat and my neighbours would probably disapprove...
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.