AMB Laboratories DIY Audio

[adamboon]

Hi,

Apologies if this has been covered, I have tried to look around the board, but am struggling to find what I'm after.

We have just purchased an AMEK Langley Recall Audio Console (Desk) for Studio Use. I have built a σ22 to power some Mic Pre's in the past with great success. Now the Langley is Hungry! it would like around 20A per rail and I need +/- 18V and a +5 (also a +48 for Phantom, but this only needs to be a couple of amps). So I'm thinking of adapting the σ22 for my +/- 18V and a σ11 for my +5 rails. I have lined up a few 500VA Transformers (1 each for + and - 18V, and one more for +5V) and was considering some beefy 20A Diodes and was originally going to use 4 Mosfet's per rail rather than the standard 2, but have found some higher capacity MOSFET's also. I would really appreciate any advise anyone has. Even if it's to not use this design for this purpose (although a hint on the best option would be appreciated.)

Thanks for your time.

Adam

[amb]

A stock σ11 or σ22 with other default parts and onboard heatsinks can support up to about 1A continuous (with short duration peaks much higher than that). Adding more paralleled MOSFETs and using higher current rectifiers would increase the current capacity, but 20A is a bit far-fetched even if you use larger MOSFETs and huge heatsinks. You can work the numbers (see guide on the σ11 or σ22 website, under "board & heatsinks").

Paralleling more MOSFETs will increase the total effective gate capacitance, especially if you use larger MOSFETs which have even higher gate capacitances. The circuit's compensation capacitors will need to change accordingly in order for it to remain stable into various loads, but it will be at the detriment of bandwidth.

[adamboon]

The only other reg design I know is current bypassed LM regs... Is this a better way to go? Any other ideas?

[amb]

Which "current bypassed LM regs" are you referring to?

20A is a lot of current, and difficult to implement with linear regulators due to high heat dissipation.

[mikeg88]

What about building more than one σ22 (possibly sharing the same trafo) and use them in parallel, so that collectively they are able to accommodate the 20A current draw?

[amb]

What about building more than one σ22 (possibly sharing the same trafo) and use them in parallel, so that collectively they are able to accommodate the 20A current draw?

The output impedance of the σ11/σ22 is practically zero ohms, which will look like a dead short to the other board if their outputs are connected in parallel. You shouldn't connect two regulators together like that, just as you wouldn't do so for two power amps.

However, taking that concept (much) further, if the circuit being powered is a collection of clearly-defined blocks, each of which can be separately powered, then one could use many σ11/σ22 boards, each one serving a block.

[testla]

I also need a high current version of the a22, and like the other poster I thought of just doubling the original output devices. I noted your comments about capacitance and was wondering what the practical effect it would have when driving a Class A power amplifier. I need 10A at 34V continuously. Since I will be mounting the larger rectifier and large can caps separately and bottom mounting the board to a large heatsink I thought that mounted this way, the original board might handle it, but would really like more of a safety margin. What is your opinion of a single board's max continuous amps on an unlimited heatsink? If I do need to double the ORIGINAL devices, which capacitor would I need to change, and to what value? Many Thanks

[amb]

Do the math and you'll soon find out that just by having a very large heatsink won't help much beyond a certain level, because of the MOSFETs' TO-220 casing and the insulating pad's thermal resistances. 10A continuous is a lot of current, do some calculations to see what kind of heat dissipation will be required of each MOSFET...