Building a symmetrical psu B1 buffer

[regiregi22]

Originally Posted by wushuliu

Are there any recommended heatsinks for the IRFPs that would be good for hot-rodding?

The bigger ones you could fit/afford. I wouldn't go bigger than a CPU heatsink for each mosfet.

Regards,
Regi

[Tea-Bag]

I used an L-channel Aluminum piece from Home Depot, and Thermal epoxied two CPU heat sinks to each side. Only get's a little warm. But I might go further down with current at some point.

[wushuliu]

Originally Posted by Tea-Bag

I used an L-channel Aluminum piece from Home Depot, and Thermal epoxied two CPU heat sinks to each side. Only get's a little warm. But I might go further down with current at some point.

Hrm, so i take it something like this wouldn't be good enough...

[AndrewT]

it states that the sink is suitable for To220, not To247

[wushuliu]

Originally Posted by AndrewT

it states that the sink is suitable for To220, not To247

Ah, ok, thanks andrew.

[Salas]

Bolt 'em Mosfets to the metal floor. Cheap & easy. Is your box's floor a bit thick with enough area?

[wushuliu]

Originally Posted by Salas

Bolt 'em Mosfets to the metal floor. Cheap & easy. Is your box's floor a bit thick with enough area?

Not at the moment, it's in an ABS, but I've ordered a Par-metal anodized enclosure, so that should do the trick!

[Salas]

Yes it will. Will be shielded better in a grounded metal case too.

[wushuliu]

I have some 1800uf caps lying around, would this be too low a value to substitute the 4700ufs?

[oenboek]

Finished my B1 this weekend. Today I added heatsinks, raised the current and changed the caps parallel to the 5 LEDs to MKP18371 100nF (which makes more sense then a huge cap in my opinion).

Some remarks on the build:

• All of a batch of 36 LEDs were from 1,691 to 1,718V (AVAGO HLMP-K101). By lowering current resistor to 12,59Ohm, I have currents -111 and +102mA. Output voltages are -8,68V and +8,97V at the moment. With lower current (34Ohm resistor) I had -9,25 and +9.58V.
• IDSS are less then 0,01mA different for each pair, easy to sort out of 40 pcs. Output offset voltages are around 1mV now. At higher voltage they were about 5 times smaller. I definitely need to raise voltage again. I guess I will add another LED or a regular diode to the chain.

I did some optimisations besides the previous ones:

• I bypassed the delay relais as I don't want this contact in the loop. I use the relais to ground the signal. This is a very easy change on the PCB. Just connect the 2 unused pins of the relais-switches to the ground, and link the middle and output pins of the switches together (by channel off-course).
• I saw that temperature has a big influence on offset voltage. I reduced this by linking the 2 fets of each pair together. That lowers a lot the thermal drift. At the moment I have just a metal clamp by pair, but I will put some aluminium to improve. I don't like the DC drifting, even if it is slow. It can cause woofer movements on a DC-coupled poweramp.

Next steps:

• I will avoid the long PCB-track to the left input. I will connect the inputs next to the 220R-resistors and cut the PCB-traces. Yes, I agree, it's of no importance but I like complete symmetry in audio.
• I'll put the DCB1 in my preamp instead of tubes (now a DIY CleoV-amp from Triode Dick). I stopped the tube-amp some weeks ago and work passive now. The quality of the amp went down after some years. I don't want to start changing tubes regularly.

Next is put a LDR-pot instead of the 50k plastic I have from my tube. The LDRs will be driven by an Arduino. I have some issues setting the very low LED-currents, but by putting an exponential amp between the DAC and the LED it starts to get better. But this design isn't finished yet (at all). Don't tell me the "George" design is enough. It is enough, I know, but I want to integrate some other stuff, and I wanted to play with an Arduino and DACs and ADCs around. It's a nice combination of audio, electronics and software.

My power-amp is a Hypex. First impressions with the DCB1 this WE were a smoother sound and basses were a bit tighter, although not louder. I keep you informed after shortening wires and listening some more.

In any case the DCB1 is a very nice design. Without coupling caps is really great. The PCB is nice and easy. I'll build a 2nd one for fun, and to use as a universal buffer. I'll make some pictures, and show statistics of leds and fets later.