Dual channel 0-15V 0-500mA power supply

[kallek]

Hello everyone!

This is my first post in the Forum! I want to share you my power supply project progress. I have been using basic Lm317 power supply which i built four years ago, but i want capability to limit current when testing new circuits.

Specs:

- Two independent channels, 0-15V, 0-500mA
- Linear, TIP122 darlington series pass transistors
- Passive cooling, heatsink on back panel
- 40VA 2x15VAC toroidal transformer that I had already
- Zener references
- Lm324 OpAmp for sensing voltage and current, and driving CV/CC leds
- Low side current sensing, seven 10 ohm 0,6W resistors in parallel
- Analog panel meters for voltage and current readings, two per channel
- Switch to short output to set current limit
- Load switch
- Single turn wirewound potentiometers

I have already etched and assembled the circuit board, and it looks like working fine. Schematic is not ready yet, but I will share it when I have finished it. Transformers voltage is slightly low, but not too bad, no large ripple on 15V at 500mA with 3300uf smoothing capacitors. I intend to use toggle switch as load switch. I know it have that contact bouncing effect, but is it too bad practice if it does not overshoot? Next step is get sheet metal for case, and wait panel meters to arrive.

Thank you for advises and comments!

[David Hess]

- Two independent channels, 0-15V, 0-500mA

Were you going to make the channels floating from each other?  This had the advantage of allowing them to be combined in parallel for higher current, series for higher voltage, or series for a bipolar tracking.

Linear, TIP122 darlington series pass transistors

10 watts is in the range where integrated regulators which provide built-in thermal protection could be used as pass elements.

Lm324 OpAmp for sensing voltage and current, and driving CV/CC leds

I would not get hung up on using a quad operational amplifier.  Some of the better designs take advantage of separate amplifiers with different supplies for each output or the same output.

Low side current sensing, seven 10 ohm 0,6W resistors in parallel

Why require low side current sensing?

Single turn wirewound potentiometers

I prefer good quality single turn potentiometers for coarse and fine control but both methods have their virtues.

[kallek]

Thanks David Hess!

Were you going to make the channels floating from each other?  This had the advantage of allowing them to be combined in parallel for higher current, series for higher voltage, or series for a bipolar tracking.

Yes, I will use them just floating or connect to ground binding post. I assume that this is possible with two separate windings? That is true.

10 watts is in the range where integrated regulators which provide built-in thermal protection could be used as pass elements.

I have to think that.

I would not get hung up on using a quad operational amplifier.  Some of the better designs take advantage of separate amplifiers with different supplies for each output or the same output.

Why require low side current sensing?

I have these on hand and layout was easy to make. With high enough value current sense resistor accuracy is about 2mA and current can be set down to under 20mA. I choose low side sensing because then diff. amplifier is not deeded.

I prefer good quality single turn potentiometers for coarse and fine control but both methods have their virtues.

It could be nicer... Good quality pots are not cheap so I try with single pots with big knobs.

[Kleinstein]

The LM317 can be used as a kind of protected transistor at the low power. However it may cause a little more voltage lost - so it could be a problem in this example with a relatively low transformer voltage.

I like the choice of low current and voltage for a beginner's project  . so less of the magic smoke to escape.

With separate windings one could use the 2 supplies in series / floating. However the maximum permissible voltage between the 2 parts may be limited.

[MarkF]

Pictures broken...

[Zero999]

Pictures broken...

Yes, the original poster should attach them here, rather than some image hosting site.

[kallek]

Thanks, tried to fix them.

[MarkF]

Thanks, tried to fix them.

They're okay now.

[exe]

Nice build! Can we see the schematics, please?

[David Hess]

Why require low side current sensing?

I have these on hand and layout was easy to make. With high enough value current sense resistor accuracy is about 2mA and current can be set down to under 20mA. I choose low side sensing because then diff. amplifier is not deeded.

Nothing requires high side current sensing to use a difference amplifier; the error amplifier for controlling the current can operate at the high side directly if the control voltage is referenced to the high side which is trivial when poteniometers are used.  A simple example is shown below.

[kallek]

Thank you. Actually, I used that ti circuit to design my own. I just put current sense resistor below negative terminal, because my opamp can go to negative rail but not to positive. I will show my schematic when I have drawn it.

[Zero999]

Why require low side current sensing?

I have these on hand and layout was easy to make. With high enough value current sense resistor accuracy is about 2mA and current can be set down to under 20mA. I choose low side sensing because then diff. amplifier is not deeded.

Nothing requires high side current sensing to use a difference amplifier; the error amplifier for controlling the current can operate at the high side directly if the control voltage is referenced to the high side which is trivial when poteniometers are used.  A simple example is shown below.

The LM101A is now expensive and hard to get hold of. Use the LM201A instead.

What's going on with the diode across C8? I thought polarised capacitors can withstand short reverse voltage transients, so I doubt it's to protect against that.

[David Hess]

Nothing requires high side current sensing to use a difference amplifier; the error amplifier for controlling the current can operate at the high side directly if the control voltage is referenced to the high side which is trivial when poteniometers are used.  A simple example is shown below.

The LM101A is now expensive and hard to get hold of. Use the LM201A instead.

I just use that schematic as an example.  The unique thing about the LM301A at the time was that its input common mode range includes the positive supply which made it almost uniquely useful in high side sensing applications.  A lot of old JFET parts would also work.  Today many rail-to-rail input parts meet that requirement or a slightly higher supply voltage for the operational amplifier could be used.

Incidentally, there is some advantage to putting the current sense resistor directly adjacent to the output capacitor although this could be done on the low or high side.  If AC feedback is taken before the current sense resistor, then the increased ESR of the output capacitor as seen from the error amplifier provides extra phase lead to increase stability.  Here R15 provides the same function but at the cost of increasing the high frequency output impedance.  The old Tektronix PS503 type power supplies do this with the current sense resistor on the high side but without the AC feedback part.  Some modern low dropout integrated regulators which can use ceramic capacitors do this internally.

What's going on with the diode across C8? I thought polarized capacitors can withstand short reverse voltage transients, so I doubt it's to protect against that.

Q2 could potentially sink 2 amps of current when pulling down a large capacitive load into the negative supply which is unlikely to be rated for that so the diode is required to avoid damaging C8.  Q2 and associated parts could be left out if that function is not required or undesirable.

[exe]

Q2 could potentially sink 2 amps of current when pulling down a large capacitive load into the negative supply which is unlikely to be rated for that so the diode is required to avoid damaging C8.

Interesting, so a capacitor can be damage by too large current through it? I had this question is a long time ago, all I could find was that average current through the cap should be within its rating, or it may overheat. But no information that, say, shorting leads of a charged capacitor can damage it. Or charging it too quickly.

[Kleinstein]

The problem is more that the negative supply may not be able to provide 2 A of current. So if the supply than goes into sinking mode this could bring the negative supply to positive values and thus reverse the capacitors.  This would be an unlikely scenario and many electrolytic caps do survive a little reverse voltage for a short time. The diode would just limit that voltage.

In simulations I found ot tricky with the shunt at the collector side of the power transistors. This kind of slows down the transistor and thus can help with instability. There is nothing really bad with a low side shunt, especially with an OP that works at the negative rail.

[kallek]

Update to this project. Two days ago I got four of these nice Kyoritsu panel meters. Yesterday we bought 1mm and 1,5mm thick sheet metal for metal project box from local metal store. Today I started cutting and drilling.

[kallek]

Here is the schematic. I might have made mistakes, so please give your suggestions.

EDIT: R3 should be 4.7k, not 470.

[xavier60]

Make provision for compensation components for the CC op-amp.
A series resistor feeding pin 6, also a speed up diode across this resistor. 
A series resistor and capacitor from pin 7 to pin 6.
The high value of R7 will cause some extra CC op-amp input offset.
It might be better with D8 across the B-E junction.

[xavier60]

Actually I bread-boarded the CC loop and found that compensation isn't really needed. There is a short burst of ringing when the load is applied due to lead inductance. I don't expect this to be a practical problem. I would still put a resistor in series with pin 6 just for input protection.

[kallek]

Thank you xavier60, some good points!

I am not very experienced in electronics yet, and not very good at english, so there are couple of things i have to clear.

I somehow understand the purpose of resistor feeding pin 6, but what that diode across it do? How I should bias it?

I put D8 there to prevent Q1 turning on if pin 7 can not go complitely down. But is it something to do with switching speed? Did you mean reverse biased diode across B-E junction?

Because I made the board before understanding to ask help, I will now experience with it and modify it if necessery. I will use your tips to do things better in the future. I really appreciate you sharing your knowledge!

Edit. Thank you xavier60 for your time! So you only recommend resistor or resistor and diode both?

Kalle

[xavier60]

Thank you xavier60, some good points!

I am not very experienced in electronics yet, and not very good at english, so there are couple of things i have to clear.

I somehow understand the purpose of resistor feeding pin 6, but what that diode across it do? How I should bias it?

I put D8 there to prevent Q1 turning on if pin 7 can not go complitely down. But is it something to do with switching speed? Did you mean reverse biased diode across B-E junction?

Because I made the board before understanding to ask help, I will now experience with it and modify it if necessery. I will use your tips to do things better in the future projects. I really appreciate you sharing your knowledge!

Edit. Thank you xavier60 for your time! So you only recommend resistor or resistor and diode both?

Kalle

The diode across the resistor to pin 6 isn't needed now.
While you were posting I did another experiment. I set the unreg rail to 20V and put a 2.2K resistor from unreg to the output of the CC op-amp. The op-amp's output was only able to pull down to 1.2V with about 9mA load. Add to this the 0.7V drop of D9 increases to about 1.9V, so D8 is needed where you have put it and may not lose enough voltage to allow Q1 to turn off fully.
My idea was for the op-amps to be able to pull some current out of the Base of Q1 to help it to turn off more quickly when the power supply is suddenly unloaded. A diode anti-parallel with D8 should help.
Have you tried bread-boarding the circuit so that you can further investigate these issues?

[kallek]

That confirms what I have experienced. Did you have that 2,2k pull-down resistor too? If I remember correct, I measured something like 0,7V from pin 7. I have modified circuit since last bread-boarding, but I will do it and further investigate it.

Kalle

[xavier60]

I didn't have the 2.2K to ground. And I'm using an LM358 which might be a bit different.
Using a 1A schottky for D9 will also help.

[Alti]

LM324 does not have input current bias compensation so you have to make sure input impedance of both input nodes is equal. The LM324A has 0.1R at one node and 470k at the other.

As reference voltage, you use zener diode. Get a TL431. A single one will do.

Do not connect voltage feedback before ammeter but after it.

I'd put a fast 0.5A fuse and a varistor just after feedback tap.

[xavier60]

Another surprise. Pulsing the output with 1A load caused a very small 50mV overshoot with 100uF output capacitance and a D44H BJT.
It still would be a good idea to add the anti-parallel diode in case the BJT  leakage rises too much with temperature.