Building a new +/- adjustable/selectable supply

[JimmyM]

A long while back I build an LM317 based adjustable bench supply. Nothing fancy. No current limiting or anything like that. Now I have a big Mastech HY5020 for the big stuff and still use the old LM317 based unit for small stuff. But I'm messing around with Op-Amps now and would like to build a +/- supply.
My thoughts are...
20V center tapped transformer, 1 Diode per transformer leg.
An LM317 & an LM337 (I wouldn't really need a lot of current for op-amp stuff)
Filter caps, etc.
Question: To reduce ripple on the regulator inputs, would it be better to use a C-L-C filter instead of just a Capacitor?
If I want just selectable outputs +/-5V, +/-9V, +/-12V, etc I can just use a 2 deck rotary switch to change the Adj-to-ground resistor value (trimmers on each of course to allow fine tuning and matching).

But what if I wanted to make a fully adjustable unit? Would I have to use a 2 deck pot? How could I ensure matching of the output voltages?
Thanks, guys. I know you folks have built something like this before.

[ejeffrey]

You want a total of 4 diodes to make a bridge rectifier.

You probably want a 30 volt center-tapped transformer (i.e., 2x 15 volt AC windings). This will give you a rectified voltage of around 20 volts after the diode drop, which is plenty to get +/- 15 volt supplies.  If you only want +/-12, you can use a 24 VCT transformer.

You only need a single deck switch, all of the resistors can be connected to the adjust terminal, and then the switch connects one of them to ground and the others float.

LM317 has great ripple rejection at 100/120 Hz, so you don't really need more than that.  If you do anything I would use an CRC filter on the input side.  Even a couple of ohms provides a lot of attenuation of the higher harmonics that the regulator isn't so good rejecting.

[Zero999]

http://www.silicontronics.com/index.php?action=ezportal;sa=page;p=7

[JimmyM]

This is EXACTLY what I was looking for!
I even have a uA741 already. as well as a couple of bridge rectifiers.

Brilliant guys. Thanks a lot.

[ivan747]

If I want just selectable outputs +/-5V, +/-9V, +/-12V, etc I can just use a 2 deck rotary switch to change the Adj-to-ground resistor value (trimmers on each of course to allow fine tuning and matching).

But what if I wanted to make a fully adjustable unit? Would I have to use a 2 deck pot? How could I ensure matching of the output voltages?
Thanks, guys. I know you folks have built something like this before.

You can have he contact switch toggle between fixed voltages and one variable mode using a dual gang pot instead of a fixed resistor. I have though about modding a small supply with fixed voltages (3, 4.5, 6...) to do these on one of the switch's positions. The problem is that it won't fit.

[alm]

Pay attention to what will happen during switching, especially if it's a break-before-make switch. Depending on how the resistive divider is connected, the output will either go to zero or to max.

[ejeffrey]

One way to avoid that problem is to use a resistor chain from output to ground, and have the switch select which tap to connect to the adjustment pin.  In this situation, only the adjust current (50 uA IIRC) flows through the switch.  Put a capacitor on the adjustment pin.  Then, with a break-before-make switch, the capacitor will hold the adjust pin voltage at an almost constant level (rising slightly due to the adjust current charging the capacitor).  With a make-before-break switch, which rotary switches commonly are, you simply get an output voltage somewhere between the two switch positions.  In either case, this has the advantage of eliminating errors due to the switch contact resistance since only a tiny current flows.

The disadvantage of this technique is that the LM317 is designed to have a constant 240 ohm resistor from output to adjust.  This ensures that the minimum load current of 5 milliamp is always met.  You have to take care to make sure you meet this requirement or the output voltage will rise under low loads.

[ivan747]

Thank you  very much, Hero999

[ivan747]

Pay attention to what will happen during switching, especially if it's a break-before-make switch. Depending on how the resistive divider is connected, the output will either go to zero or to max.

Will an LC low pass filter remedy it? I know it's not worth the money, just wondering 

[ejeffrey]

Pay attention to what will happen during switching, especially if it's a break-before-make switch. Depending on how the resistive divider is connected, the output will either go to zero or to max.

Will an LC low pass filter remedy it? I know it's not worth the money, just wondering 

Maybe going to 0V (or 1.25 V) is OK, but a switchable power supply that goes momentarily to Vmax during switching is not worth having.  You are guaranteed to destroy circuits with it.  The way to solve it is to find a way to buffer the adjust pin during switching.  If the adjust pin is disconnected during switching, a simple capacitor will do the trick.  A make-before-break rotary switch is a tolerable solution -- at a certain point the two ground resistors will be in parallel and the voltage will drop, but that is usually better than an overvoltage.

[ivan747]

I have a very cheap power supply and I noticed it spikes when changing the output voltage. Mod time!