Something isn't right, ~2V offset. Build 2. Discovered in build 1 that what I thought was a 5kΩ multiturn was in fact a 2k. 270Ω resistor in the LM317 adjust circuit is now cool and adjusts. The 337 tracks but is off by about -2V so when I set the +V to 15V, -V is about -17V. Swapped the 741 out and still the same. Next will be to swap the 337 if no one has a better idea. I did check the 10k resistors and they match and the 4.7k is on spec. I did leave off the output protection diodes and output transient spike tantalum capacitors on this second build.

LM317 on right.

 

I remember some time ago building an op amp PS from scratch as sort of a hit or miss approach.
I tried different configurations. Spent a lot of time did'nt get the best results, but it was ok for a lot of op amp projects.
I liked this video because it shows improvement step by step using a 500mA 12-0-12 and gives some expectation.
the oscilloscope method that shows some progress in reducing sharp edge and ripple.

 

Well, I am using 2200uF smoothing caps on each leg (I did upsize them on build 2 although not really needed) and regulators so the line on the scope is flat... And only using a 180mA +/- 28V XFMR (~24V max out of the regulators unloaded except the 1MΩ of the DMMs) since I really don't need much to supply a few opamps. I'll check out the video. Thx.

 

I have been over and over it and can find nothing wrong and still a 2V offset. What could cause an offset? Is it not tracking correctly? Is there an error in the resistor sizing of the tracking circuit? I removed the pair of 10kΩ and double-checked the replacements and they are correct. Ran the Variac on up to 120V and the offset dropped to 1.6V. It has to be the opamp tracking. BINGO! It is the opamp. Started swapping them out and they are 2 different lots. One lot is 1.6V differential and the other lot is 0.12V differential and I can live with that. Turned it down to +/-10V and 0 differential. Bad lot of 741s and yes I have some on order from DK. Yes, they work but are definitely off-spec. Problem solved. Thanks for all the input and help guys. I now have a working opamp voltage tracking dual voltage power supply working! Now I need to learn the math behind the theory,

Thx, Sam

 

One lot is 1.6V differential and the other lot is 0.12V differential and I can live with that.

That's still pretty crappy. I just breadboarded a tracking dual supply using an LM317 and LM358. The voltages track within 0.01V.

Bad lot of 741s and yes I have some on order from DK.

Were these some low priced parts from Ali Express/Amazon/eBay? If you're buying new ones, why not get something a little more user friendly? Even the lowly LM358 will have fewer limitations.

Are you tossing the bad parts you come across?

 

@dl324 Hadn't thought about using a different model opamp and I have several on hand to choose from that I put in my stock of parts. At this point, all I was looking for was a dual voltage power supply to use for studying opamps as they are new territory for me. So mission accomplished. Yeah, I'm starting to get a small drawer of "bad" parts. They actually test as "good" on my chip tester and do function as an opamp, just not accurately on-spec. I am slowly beginning to understand the true value of REAL parts after being bitten by BAD ones a few times now.

 

I'm starting to get a small drawer of "bad" parts.

If it were me, I'd throw them away.

They actually test as "good" on my chip tester and do function as an opamp, just not accurately on-spec.

Most testers are just going to do a DORA (Dead OR Alive) test.

I am slowly beginning to understand the true value of REAL parts after being bitten by BAD ones a few times now.

If you value your time at all, buying cheap parts is expensive. I'm retired, but won't waste my time testing questionable parts unless they're expensive enough to make it worth the time it takes.

Here's a no frills dual tracking supply that you shouldn't have any problems analyzing:

To operate with inputs of +/- 20V, you'd need to use some zeners so you can use asymmetrical supplies for the opamp. I tested with +/- 16V and that was pushing the LM358.

If you can't figure out what the opamps are doing, post questions. Otherwise, I'll leave it to you to test your understanding.

For clarity, capacitors are not shown.

Hint: The voltage divider isn't what you might think at first.

 

I understand the basics of it but not quite there on the analysis of it. Very interesting using the opamp to control the flow of the 3906 to set the voltage. But aren't you giving up some of the advantages of the regulator such as short circuit protection and high load overtemperature protection? I guess it all depends on just what you are using it to power. I'm putting it into a todo file to work with later as I do find it very interesting.

 

But aren't you giving up some of the advantages of the regulator such as short circuit protection and high load overtemperature protection?

That's why I described it as no frills. The only thing you might really need is current limiting. That's easy to add on after you understand how the existing circuit works.

EDIT: You may also want to use a power transistor with a heat sink. At low voltage and 100mA, it would be dissipating watts.

I understand the basics of it but not quite there on the analysis of it.

Regarding analysis, just break it down to the building blocks.

1. R3, R4, R5, and R6
2. IC2A, R6, and R7
3. IC2B and Q1

 

@dl324 @OBW0549 @AnalogKid
In particular, I want to thank you guys for your guidance and input on this project. It wouldn't have happened without you and I sincerely appreciate the help! Now I have to add a meter and put it in a box. Always something more to do...

 

Photos of the finished unit ...

 

Photos of the finished unit ...

See #121

 

The circuits discussed so far have one thing in common - the negative output tracks the positive output, but the positive output does not track the negative output. That is, if you load the + output until its output voltage sags by 50%, the negative output tracks this change and decrease by 50%; but if you load the negative output similarliy, the positive output does not track the change. But wait ...

Over in audio-land there is a two-opamp circuit for a balanced line driver that is truly balanced. It is an all-solid-state equivalent of a transformer output. If you load it asymmetrically in either direction, the amplitudes of both signal phases remain equal. IIRC the circuit always has output resistors, such as 604 ohms, so applying this in a power supply design might not work. But it sure is tempting for a *truly* tracking dual supply. Hmmm...

ak

 

See #121

Terminals, meters, knobs, box - the one you're gonna sell on ebay and get rich.

ak

 

... the one you're gonna sell on ebay and get rich.

... or rather the kit that should've been offered in Popular Electronics back in 70's (or early 80's ... you choose) that was dreamed and wanted by every geek and nerd in those days.

 

I've built several LM317 PSUs but never needed a negative power supply so the LM337 was new for me and voltage tracking unheard of before now. Since my intended use is to supply power for the opamps to free up my 2 linear bench PSUs to use for the circuit I don't forsee loading the -V more than the +V. But I am in new territory for me with opamps so there may be some things I don't anticipate. I do understand that it is the 317 leading the 337 around so to speak. Using the opamp voltage follower is far better than what I was running into with setting each regulator independantly. I also have the 2 independently adjustable PSUs I built from eBay kits so I can bias the opamp supply with different voltages if I need to get to 0V and the opamp can't quite get there with one side grounded. All in all, it was a great learning experience and I look forward to learning more about opamps.

 

I'm putting it into a todo file to work with later as I do find it very interesting.

I changed R3 in the voltage divider so the positive LM358 supply could be 10V so the maximum supply differential will be 30V. If you limit the maximum output voltage to 15V, you can reduce it further.

 

I got the 741 op-amps from DK and installed one into the power supply and had a surprise, it got worse! Going over things and noticed that the original circuit was designed for a LM301 and had pins 1 & 8 connected by a 0.001uF capacitor. For the LM741 it should have been pins 1 & 5 so I added another capacitor from pin 8 to pin 5 so either op-amp could be used. I also put 1kΩ load resistors across the + and - rails and the differential went down significantly and was steady when adjusting from 15 to 10V. The differential was ~10mV with the load resistors installed and when I put the TI 741 in from DK it increased to ~ 20mV so I took it out and put the original chinese sourced one back in as it was giving better results.

I have used the power supply several times already @ +/- 15V. I can adjust it if needed but for what I've been doing 15V is fine. One circuit had a comparator feeding an integrator which in turn fed a differentiator so it was supplying power to 3 741s at once and did a fine job of it.

 

The differential was ~10mV with the load resistors installed and when I put the TI 741 in from DK it increased to ~ 20mV so I took it out and put the original chinese sourced one back in as it was giving better results.

You're multiplying the offset voltage by the open loop gain. You can put a 10k pot between the offset null pins, connect the wiper to the negative supply, and adjust the pot to null the offset voltage.

 

I actually did that, not on the PSU but between the rails with a pair of 10kΩ and a 1k trimpot on a breadboard it supplied, when working with a Wein Bridge Oscillator. And when I peaked the sine wave on the scope the pot's wiper voltage was exactly 0. So yeah the 2 10kΩ resistors between the rails used to set the LM337 are 1% and close but they are not exactly matched. I can live with a 10-20mV differential. And if I need better I know where to go to fix it. All I would have to do is add a 1k trimpot between the 2 10kΩ resistors in the PSU and feed it's wiper to the LM337 adjust circuit.