LM317 Variable supply - Feedback/Suggestions appreciated.

[Anonmis]

So I have decided to attempt in making my own variable power supply. I would like to start off with something a little more "simpler" and work on more advanced ones in the future. I have watched the power supply videos on here as well as several different ones on youtube. I have also done a search for LM317 on the forums to see what has already been covered. I have drawn a design and would appreciate if I could get feedback whether or not this type of design would work. (Please see attached).

Questions that I would appreciate feedback on:
- Am I able to use the same LM317 input voltage on the Voltage inverter input? (same node).
- Would this design work or what are some issues that it can possibly have?
- Is a center tapped transformer better, why or why not?
- What are somethings I should pay closer attention to?
- Verifying that I am connecting this transformer right..... My reference in this case would be the - of the rectified voltage.

Feel free to answer any questions or provide any feedback that may be helpful. Thanks everyone.

[Strada916]

See app notes on LM317 I'm sure I have seen two diodes in series between ground and adjust variable resistor. To make it flot 1.2v above.

[Anonmis]

I have checked the app notes before, but I checked them again now just to make sure. I am not seeing the configuration you mentioned in the datasheet, unless I missed it somewhere lol. They do have a 0-30V regulator design, but this uses a negative supply which in my design I am using the voltage inverter to create that.

That is my quick look over though, I will go back and look at the datasheet a little closer.

[Thor-Arne]

Strada916: If my memory serves me well; the diodes on the adj pin is used to compensate for the drop in a pass transistor, usually used with fixed voltage regulators like the 78XX series.

Anonmis: Your charge pump should be connected to the AC side of the bridge like they've done here.

[stratogazer]

Which charge pump are you using? The 7660 family would work but watch the input voltages. My design is similar using 3 charge pumps in parallel to reduce output resistance and increase output current. I'm using the ICL7662 by maxim.

[Anonmis]

Thanks for the suggestions, I am still working out a couple of different designs to see which works best. I might try the one you mentioned.

[Paul Price]

-2V bias for LM117 the hard way.  If your power supply range is below 15V you don't need the 12V zener and 220 ohm resistor.

[kt315]

Paul, could you explain please, how do you get -2 V out of it?

[ELEKTRICK GASCHIEF]

Paul, could you explain please, how do you get -2 V out of it?

The red LED is acting as a 2 volt zener diode!


My take on getting the lm317 down to zero volts was to use the ICL7660 to generate  the negative supply, though I think the output impedance might be too high, I think the 555 and charge pump might be better.

I thought that maybe a -ve lm337l regulator could be used post charge pump to get a more stable negative reference.

Another concern with charge pumps is the amount of ripple on their outputs, of course a regulator like the lm337l or something with a TL431 should clean that up, it would be a shame to mire the lm317's otherwise excellent ripple performance with a noisy negative reference.

[kt315]

Curious. Thanks.

[c4757p]

Whatever happened to just grabbing a negative rail off the transformer? Use a center-tapped transformer, two-diode full wave rectifier for positive and another diode for negative. No need for a center tap either if you don't mind half wave. That 555 circuit is awfully complicated just to drop 1.25V off the output...

[digsys]

Option
https://www.eevblog.com/forum/beginners/zero-volts-from-a-lm-317/msg227118/#msg227118

[rdl]

Use a center-tapped transformer, two-diode full wave rectifier for positive and another diode for negative.

Please excuse my lack of knowledge, but doesn't this basically waste half the output capacity of the transformer for no reason other than generating a couple of milliamps for the negative reference?

[codeboy2k]

Your knowledge is correct,  it will waste half the transformer's output capacity if the only reason for using a centertap is for the purpose of generating a small -1.25 V bias for the negative supply.

The center tap as described by that post is best used for an equal, bi-polar supply of + and - voltages. It gives a full wave rectified voltage at + and - the transformer's half-winding rating.

If what is wanted is a single output positive supply capable of being adjusted down to zero, then it's much better to generate a small -1.25V bias for the Adj pin using one of the many methods already described (half-wave, charge pump, zener, shunt-regulator, etc).

[c4757p]

Use a center-tapped transformer, two-diode full wave rectifier for positive and another diode for negative.

Please excuse my lack of knowledge, but doesn't this basically waste half the output capacity of the transformer for no reason other than generating a couple of milliamps for the negative reference?

You're both misunderstanding what I was suggesting - more like this. No capacity is wasted - both half-windings are supplying positive current. One winding is also supplying some negative.

[mariush]

Use a center-tapped transformer, two-diode full wave rectifier for positive and another diode for negative.

Please excuse my lack of knowledge, but doesn't this basically waste half the output capacity of the transformer for no reason other than generating a couple of milliamps for the negative reference?

I never claimed it would be very power efficient, or cheap, just that it's possible.

But c4757p is right, it's possible to use a bridge rectifier and a diode to have both full current dc and a negative rail ..

[codeboy2k]

You're both misunderstanding what I was suggesting - more like this. No capacity is wasted - both half-windings are supplying positive current. One winding is also supplying some negative.

yes, I misunderstood.  I somehow thought you were talking abot a 4-diode full wave bridge and a center tap, thus bipolar. The 2 diode full wave and the  half-wave negative rail is really all that's needed; there's plenty of info in this post for anyone interested to do it right now, in many different ways, and when they do or do not have a center tap.

That negative rail in your example can be filtered, zenered to -5.1V or so, LM337 to -1.25 volts, output filtered, and bob's ur uncle.

This thread should be sticky.  Or at least in the wiki.

[Anonmis]

Everyone has been giving very valuable information which I will definitely be trying out