I've found a circuit that I think will be really cool to take to a final project. My goal is to eventually get it to a useable bench supply.

Currently I have it bread-boarded and seems to be working fairly well.

Here's the schematic:

I have had to substitute some parts but it still works.

Substitutions:
U1, U3 LM317, LM337 == NTE956(pos), NTE957(neg)

U2 = LM301 == LM741, tried TL070 too
-- these are the only op amps that I've tried - any recommendations that might be better suited.

Some observations:

1) D2 (LED) needs to be flipped - (to forward bias) <-- feel free to correct me here if I'm describing it wrong.

2) the voltage as measured with my hand held voltmeter shows about 10-20 milliamps difference in the positive value and the negative value. I'm not sure if this is a function of the meter's summing, or the breadboard variations or my component substitutions

More to follow as I'd like to document this project as it evolves,

Paul


Attached is a slightly larger version of the schematic that's easier to read.

 

Thought of a question: any input would be appreciated.

I want to recreate the schematic above in software and eventually layout a board and either make it myself or farm it out to a service.

Recommendations on software?

Thanks in advance!

Paul

 

Eagle and DipTrace are both nice, and free, and work on peecees and Macs. I haven't completed a project with either one, but I'm leaning towards DipTrace because it comes with a lot of tutorial videos and such to help a noob get jump-started.

BTW, your D2 LED is placed backwards.

 

help a noob get jump-started.

-good selling point for me

BTW, your D2 LED is placed backwards.

-note comment #1 under "Some observations:" in original post.

Thanks!

On edit: I forgot to ask - is it forward biased or reverse biased as drawn? I'm get confused relative to the negative leg of the circuit or does that make a difference?

 

I think as drawn you'd say that is reverse biased. It won't light up unless the negative rail voltage rises ABOVE ground, hopefully never.

 

@wayneh - thanks for the help.

 

Made a little progress on the variation in follower's voltage. By placing a small trimpot in series with R2 and pin3 of U2 I was able to fine tune the negative rail.

I also experimented more with the difference in the 741 and the TL070 by eliminating C6.

I think the LM301 in uncompensated and thus needs the cap across pins 1 and 8. What did I do with that data sheet?

 

I have some necessary mods for you, but I need to know if you can get LM317L. Note that this is in a TO92 package. It would NOT be used as a primary regulator, but for another purpose.

 

I have an LM317 but it's not the L and it's in a TO220 package. Will that work? I can look through some old salvaged parts - I may be able to scrounge an LM317L.

 

I have an LM317 but it's not the L and it's in a TO220 package. Will that work? I can look through some old salvaged parts - I may be able to scrounge an LM317L.

LM317 will work, it's just overkill. There are other ways to skin this particular cat. Let me think about it.

 

Thanks Ron H - I look forward to your advice.

 

Are you using the transformer shown? I'm asking because I need to know the unregulated DC voltages. With the 2x12V secondary, I would expect about 16V unloaded. Is that what you are planning on?

 

Ron,
It's a 12.6v-0v-12.6v transformer given a 120v input. I'll put a meter on it and let you know what I'm getting for unregulated DC after the rectifier bridge as soon I can measure it. Thanks.
Paul

 

Getting about +/-18.4V unregulated.

 

U better use two fuses for both the rails instead of just one.

 

Thanks. As soon as I can figure out diptrace, I'll recreate that schematic and add the changes.

Paul

 

OK, here's the biggest problem with the circuit you posted:
The positive supply can go down to ≈+1.25V. In order for the negative supply to follow to -1.25V, the op amp's output must go to 0V. Note that this is the most positive it needs to go, which means that, if you have the positive rail at ground, the output range must include the positive rail. Also, the input common mode range must include 0V (positive rail). This requires an op amp which is not so common, and probably relatively expensive.
If you move the op amp so that the negative supply is the master, and the positive supply is the slave, you can use an LM358 (dual) or LM321 (single), which are cheap, common op amps. These op amps have inputs and outputs which work down to the negative rail (GND, in this case). I'm posting a schematic which uses this approach.
Others may pop up with schemes that seem simpler, but I would ask them for a detailed schematic before you take off on one of them. I have given this a lot of thought, and considered a lot of alternatives, and they were all more complex or expensive than what I posted. That is not to say that there is not a better way.
The extra diodes are for short circuit protection of the regulators.
You can add fuses, LEDs, meters, output capacitors, etc., where you want them.

 

That's really awesome Ron! Thanks - I have plenty of LM358's and lots of 1n4001's so I can get this laid out right away! Thanks for the help!

Paul

 

That's really awesome Ron! Thanks - I have plenty of LM358's and lots of 1n4001's so I can get this laid out right away! Thanks for the help!

Paul

Please breadboard it first!

 

Please breadboard it first!

That's what I meant! LOL

There's a bit of a learning curve for me from bread board stage to PCB stage. I hope to manage this one to completion. (I'm bad about getting something working on the bread board then never finishing!)

As soon as we get the kids to be bed I plan to get back out to my "Laboratory" and make the changes you've come up with.

Thanks again!