Questions on building High-Powered subwoofer Amp.

[EWorkshop1708]

I just built a new Subwoofer amplifier. Posted the schematic.

My design goal is 300W+ depending on the transformer I get will determine if I run 4 ohm or 2 ohm. For 4 ohm I plan on getting a Trans that can sustain +/- 60V rails. Amp has 5 x 80V 4700 uf caps per rail for the output, and 2 seperate 4700, 1 per rail for the VAS stage PCB. Currently for testing, I'm using a ~200W 35-0-35 trans giving me 52V rails at no load until I get a bigger transformer.


Questions about turning this amp into a finished project:

Any recommendations about changing some of the resistors? Do I need to lower my 4.7ohm resistors on my drivers to 3.3 or 2.2 ohm, or are they fine?

Also wondering about a preamp. The gain isn't set real high on the amp, and I'm pondering either raising the gain, or adding a preamp to get full power with line level signal. Which is better? I AM concerned with noise, and right now the amp makes 0 hum or AC noise, and with only the slightest amount of hiss at full volume, I'm hoping to keep it this way.

Also the enclosure, for looks I'm thinking of going with clear, or slightly blue Plexiglass. My heatsinks will be fan cooled and inside the enclosure, also with appropriate cooling. Will I have to worry about the amp's shielding not being in a metal enclosure?

Thanks for any suggestions!
EW

EDIT: The 100ohm resistor in the offset adj is now a 470. Oops, forgot to update the pic.

[EWorkshop1708]

Transistors in the amp:

PCB Diff stage, and sink trans in CCS - 150V 50mA 2SC2632 and the sense trans in the CCS says F422 all of these came from CRT monitor.

Drivers are 350V 4A MJE15034/35 on the main heatsinks and also the VAS is MJE15035 on a 3" tall heatsink.

Outputs are MJL4281/4302 350V 15A and there's 5 pairs, 10 transistors in total. Each set of NPN's and PNP's have their own seperate heatsink, NO pads under the transistors, just white grease and screwed directly to heatsink, but the drivers are mounted to them with silicone pads + nylon washers.

[EWorkshop1708]

Pictures of my amp.

DIY in progress.

[EWorkshop1708]

4700 uf capacitors

[EWorkshop1708]

PCB

The toroids on the PCB filter the incoming +/- rails

[EWorkshop1708]

Closeup of outputs on heatsink. The driver is also on heatsink, but hidden under the 1nf cap.

[EWorkshop1708]

[darkfenriz]

hmmm.. I doubt if your double heatsink with no washers is a good idea. I not big expert on this, but you lose one great feature of a heatsink: a shielding, instead of this you get an antenna. Also wires between output NPNs and PNPs become significantly long for currents this high, their inductance can aslo be an issue for feedback. Just suggestions.
About amp project: this beast seems OK. although I personally would go for lower current VAS and predrivers. If I'm not wrong, you run more than 15mA VAS. Could you explain why do you parallel two transistors in Vbe multiplier? Does positive temperature coefficient provide better temp. compensation when Vbe multiplier transistors are paralleled?
regards

[EWorkshop1708]

Originally posted by darkfenriz
hmmm.. I doubt if your double heatsink with no washers is a good idea. I not big expert on this, but you lose one great feature of a heatsink: a shielding, instead of this you get an antenna. Also wires between output NPNs and PNPs become significantly long for currents this high, their inductance can aslo be an issue for feedback. Just suggestions.

Thanks for the comments and you bring up some very good points.

I did the parallel VBE Bias transistors because of the dual heatsinks. Not sure if it's any better or not, just something I tried out. I haven't had any problems with temp stability at all. Bias sets easily and doesn't change very much, even when I'd make the amp hot on purpose by cranking the bias to 350 mA

I was thinking of using a bunch of pads under the transistors, but I figured since the collectors are connected among the 5 pairs, I could just screw them to the heatsink with grease and get better thermal contact. I haven't had any HF or RF problems, and actually the amp is pretty noiseless, but I may try adding a zobel to each heatsink as extra protection. I really don't want an antenna

Being the physical layout of things, I can't go with shorter wire on the outputs, but would going to a thicker wire, or parallel wires prevent any inductance? I'm currently using 18 AWG.

Also, yes, the VAS Transistor does run a fairly high current 15-16 mA and it has a large heatsink, and it along with the two 2.2K resistors and 200V 470uf bootstrap cap take up a lot of the PCB. I was thinking of going predrivers, but I haven't built any circuits yet with 3 stage outputs other than with breadboards, and didn't want to risk instability, so I figured a strong VAS would make up for it. I may try predrivers though.....

[darkfenriz]

I did the parallel VBE Bias transistors because of the dual heatsinks.

Oh, yes stupid me! This should have been clear. This probably works as a worst case compensation, I mean if one heatsink is hotter, than it comenasates it correctly and the cooler is over-compensated. Clever.

Thicker wire will probably have lower resistance, but for higher frequency, when its inductance becomes significant, it will be not much better, because HF signal propagates mainly on surface of a wire. I'd choose a speaker wire (not silver of course ).

About VAS, it's OK to run VAS current that high (15-16mA), but your VAS transistor is pretty high power, so it may have low current gain (hfe,beta). Than I would run higher tail current in LTP. You have something like 3mA tail current, so 1.5mA per leg! That's low. If VAS bjt has hfe=30, than 1/3 of leg current is sucked to base of Vas bjt. That may couse non-linear behaviour. Give more tail current, like 8mA, possibly high within dissipation rate of LTP transistors.
Cheers