For whatever it's worth, attached is the schematic for a lab power supply I built about 15 years ago and still use for powering my op amp experiments. It's a dual-tracking supply with symmetrical +/- outputs adjustable from zero volts up to just over +/- 15V, and has fixed current limiting at 65 mA which is adequate for the stuff I need to power. It also has a fixed +5V output for the odd logic chip or two.

 

Worst case at +/-1.25V data sheet showing this 1.25V using two digits
Using LM317 and LM337 and no tracking. The data sheet page 6 table 7.5 Electrical Charecteristics:
Line Regulation %/V see (2) below states that: line regulation is expressed as the percentage change in output voltage per 1-V change in input.
( I think that 15V LM317 would typically be 0.01%/13.75V = 0.1375 and same for -(LM337))
having a range of polarity difference +/- 0.005V and maintaining 0.0015 V would cover the line regulation however a few op amps projects may need tighter regulation. Going beyond the limitation of your meter's ability to verify would be silly to build super precision things without ultra precision equipment.
https://www.ti.com/lit/ds/symlink/lm317.pdf

 

Waiting on DK, my order still not shipped? Found this on eBay and ordered one and put it together. LM317 with a 110/12V XFMR that actually puts out 15.7VDC without load. Not too bad for the ~$14 and a couple in series will work. Single turn V setting pot is too touchy so replacing with a multiturn. Still going to build the one I planned to but these will be added to my pile of stuff.
https://www.ebay.com/itm/293513256925

 

The DK order showed up unexpectedly. Never received a shipping notice? I thought I had everything but came up short on the 5k 1% resistors. Turns out they are not so easy to get. Ended up having to order them in 3W size. Oh well... I could have cobbled them up by putting some in series.

 

DK tends to do that. They ship immediately without notifying you. That's a problem for me because whenever I receive components from DK, import taxes must be payed. And those taxes usually vary depending on the components' country of origin. And so I have to have a fair amount of cash and change with me to be ready to pay when the package arrives.

The courier only accepts cash at the moment of delivery, or one can pay online in advance. But the latter is not possible if one doesn't have the precise shipping information.

 

They surprised me when I ordered the resistors ~9 PM last night. I not only received the order confirmation EMail but also the shipping and tracking number. It was only a couple of resistors so I opted for 1st Class Mail and they sent it right out. I didn't expect them to ship until after the weekend.

 

@SamR I used two 12v dc power supplies to supply +/- 12v to get going quickly at home. Theres some high frequency noise but if you’re using 741 or similar it won’t matter. Of course you should always use bypass caps for opamps.
I cut off the plugs and soldered two pin headers to the ends and marked polarity.

 

Breadboarded this design. Liked it because it only needed one pot to adjust Vo. Can't get it to work. Does not adjust V equally. When V matched at max, on min can be 0.6 V difference.

Going back to the National Semiconductor typical. Also shown as typical on several other manufacturers PDFs.

 

Does not adjust V equally. When V matched at max, on min can be 0.6 V difference.

Do you have any load on the supplies? In theory, a 100 ohm current set resistor should satisfy the minimum current required for the regulator to function. In all the years I've been using LM317, I've only had one that required more like 20mA before it would regulate.

 

Good point, no I did not have a load on it. Will go back and put a resistor across the +/- Vo and see.

EDIT: Put a couple of 220Ω across the rail for ~50mA load and the V would not even adjust. Circuit analysis for this is above my level so other than looking at it and thinking what a neat idea it's out of my arena. I have done several DC power supplies from XFMRs over the years, both fixed and adjustable, but not a dual +/- Vo. Using 2 pots to set each output separately is really not that big of a deal so looks like I'm headed that way now.

 

Circuit analysis for this is above my level so other than looking at it and thinking what a neat idea it's out of my arena.

R1, R2, R4, and R5 determine the accuracy of the tracking. The parallel combination of R2, R3, and R4 drop the voltage that is added to the voltage across R1 and R5. The ground reference between R4 and R5 is what makes the regulators track.

Put a load on both supplies and post measurements for the voltages on the regulator terminals.

 

Tried it again after removing the 10uV pots. Now it is adjusting but... 0.08V delta @ max and 1.1V delta @ min. Definitely, time to go dual pots. This is with the 220Ω from both rails to gnd.

 

Tried it again after removing the 10uV pots.

Those caps are to prevent noise from affecting voltage. Not sure why they'd need to be so large.

Now it is adjusting but... 0.08V delta @ max and 1.1V delta @ min.

Post the voltages at the regulator pins so we can see what's going on.

 

Getting +/- 9.52 out of the rectifier. This is with a 12V CT transformer and will eventually be using 28V CT. I have a pair of 1000uF right after the rectifier so don't really think it needs the 10uF and the charge in them will slow the adjusting? Surprisingly, at the inputs to the regulators, I get -9.57 & +9.56.

I'm not paying attention and lost sight of what it was that I really needed. An adjustable PSU from ~ +/-16V to +/-10V. It really doesn't matter what it does below that.

Put the 28V CT XFMR on. Now +20.06V & -20.16 then on turn down +7.96 &-10.60. So it is not good enough. And the pot is warm.

 

Getting +/- 9.52 out of the rectifier. This is with a 12V CT transformer

That must be an unloaded voltage. The regulator will drop 2-3V, so at what voltage do the regulators stop working?

Put the 28V CT XFMR on. Now +20.06V & -20.16 then on turn down +7.96 &-10.60. So it is not good enough. And the pot is warm.

What are the voltages on the regulator terminals? Maximum current through the pot will be around 20mA, so it shouldn't get warm.

 

-20.72 & +20.90

 

at what voltage do the regulators stop working?

Don't know... rated 40V max

 

-20.72 & +20.90

What are the voltages on the other terminals?

Don't know... rated 40V max

I was referring to the regulator dropout voltage not being satisfied, causing the regulators to stop regulating. The dropout voltage is current dependent.

 

After swapping the XFMR, the 220Ω 1/2W load resistors were getting hot. 20V/220Ω=91mAx20V=1.8W So obviously too hot!
Replaced w/ 1kΩ 1/2W and 20V/1k=20mAx20V=0.4W
10V/1kΩ=10mA and the minimum load current is 1.5mA from the PDF, so good, and 10mAx10V=0.1W also good. So good for the load.

But now no turndown on the LM337. 317 works fine. So... Swapped a few 337 regulators and still no turndown.

I'm thinking that for this to work correctly, the adjust curves on the regulators must be identical. Which I doubt they would be? I would hope for 10ths of a volt but... Since it is for experimental use, just how close does it need to be? I'm certainly not doing published research.

Letting it run awhile... The load resistors are barely getting warm as is the 2A DB207 rectifier, but nowhere near hot. Everything else feels cold.

 

Vout=-21.92V @ the rectifier and & Regulator Vout(max)=-20.13. So I've got ~1.45V overhead from Vin to Vout(max) on the regulator so there may be dropout. At that delta V @ room temp it would be a bit shy. The curve looks @ room temp to be ~2V minimum. Preferable 2.5-3.0V. The 317 can handle it but not the 337 even though they have the same curve. Odd but it is what it is.