Version 2 of power supply

[profdc9]

I am working on my version 2 of a linear power supply.  The first one worked fine, but I think I can do better.  The idea of it is to make it so that when you tear down and recycle the bits of old linear power supplies (the transformers and capacitors for example) you can reuse them with this new power supply PCB.  This design is a constant voltage (up to 30 volts) and/or constant current (perhaps up to 20 amps, depending on dissipated power).   This design uses very common op-amps (LM358) and up to four pass transistors in parallel (TO-220 TIP41C or 3055-type).

The new design has a few improvements, including a connection for analog or digital meter voltage and current outputs on the PCB, and provides 5 volts low current (< 50 mA) output for digital panel meters.  Also, a connection for an external analog control of the voltage and current if you want to use an Arduino to control the power supply. 

[johnf]

Well done
you might not have had to complicate the design so much with the negative rail by using opamps that go to the negative rail like the old CA3130 and its later clones they go a few 100 mv below the negative rail
your choice of tip41 is curious as they have a limited SOAR area for a 30 volt supply each tip41 is limited to 2 amps so 4 of them gives a max of 8 amps
even dropping to 20volts the soar is only 3 amps per device so 12 amps max1

a good rule of thumb i have always used for smoothing caps is 4700uf per amp to keep ripple current in the caps and ripple volts to the circuit in check --your caps C1, C10, C11 footprints seem a little small

otherwise a good effort!!

[Hydron]

I have not looked through it fully, but there are a couple of areas that I'd think about:
- Diodes in parallel - they don't look like they will be heatsunk together, so equal current sharing may be an issue
- Tap switching relays may be useful to limit dissipation; these are common in the linear supplies that you're looking to upgrade (though I do not know what downsides they have, as I haven't designed one myself).

[profdc9]

I am aware of the current sharing issue with diodes.  If I could find a TO-220 or some similar case power diode (especially cheap) it would be better.   But I was thinking that I could use some heatsink compound and use the copper layer on top of the board to equalize the temperature, or the same fan used to blow on the transistor heatsinks could also help cool the diodes.  Otherwise I could use one of the "brick" full-wave rectifiers.  I'm just trying to keep the costs down, anticipating that most of the components used to assemble this in practice are cheap fleabay/Shenzhen generic components.

For the capacitors, I was thinking that the large capacitor bank would be connected off-board, and a smaller capacitor on the PCB which a smaller inductance for surge currents.  So I just put locations for three capacitors, thinking that one or two might be used to place capacitors off-board.

The TIP41C goes up to 6A, but mostly the four are there so that if you drop a high voltage to a low one with high current, there are many transistors to share the heat load with.  Also, a TIP3055 can go up to 10 A, but I think the heat load will be a problem before the current is maxxed out.

I haven't put relays in this design but I will consider it for a future design.  However, I am trying to make the simplest, more adaptable design I can.  It's not elegant, given that a lot of power could be dissipated in the transistors, so if a high voltage is used, a large heatsink and fan is likely required.

Thanks for any feedback you can provide,

Dan

I have not looked through it fully, but there are a couple of areas that I'd think about:
- Diodes in parallel - they don't look like they will be heatsunk together, so equal current sharing may be an issue
- Tap switching relays may be useful to limit dissipation; these are common in the linear supplies that you're looking to upgrade (though I do not know what downsides they have, as I haven't designed one myself).

[johnf]

Tip41 SOAR only gives 6 amps up to 10 volts across it after that it quickly falls off as I have pointed out
this SOAR assumes perfect contact with heatsink and the heatsink being unlimited
I note that the 6amps you quote is the absolute maximum for the device
If you want a reliable supply I would not go past 3 amps per device for up to 10 volts across device dropping to 2amps at 20 volts across and 1 amp @30 volts

When you look carefully at the humble 2N3055 it is a 115w die but the thermal resistance of die to case limits dissipation to about 32 watts on an unlimited heatsink
parallel diodes are a no no they will not share, the first to heat up will hog all the current, and your diode will tend back to beach sand followed rapidly by the others
Oh and yes diodes treated like that tend to go shortcircuit which will toast your opamps

[T3sl4co1l]

Seems like a lot of diodes for not a lot of capacitors?

Tim

[profdc9]

Ok, I changed it so there is space for four bigger diodes (P600 type) and bigger capacitors, with bigger holes to solder pads for external capacitors.

I also changed the footprints of the transistors to allow for TO-220 or TO-247 type pass transistors which has a higher power dissipation package.

I thought TO-220 was good up to 50 watts, but I suppose even with a good fan it will still get too hot, perhaps even to do 30 watts per device.  If so, then I had better go with the TIP3055 (TO-247 package).   I don't want to use the 2N3055 as the TO-3 package is expensive and rare, and is relatively hard to mount.

[profdc9]

So I built up my power supply board.  So far so good.  I need to get a bigger heatsink, fan, and enclosure.  The next version has a few changes not on this version, including bigger pads for single diodes for the full wave bridge, capacitors, and TO-247 as well as TO-220 transistors.

If there are any other suggestions, let me know.



Dan