Like this.

Remove the ground symbols and you are okay.

 

Like this.

You have to redraw the circuit and connected the first ground together and labels as GND1, and the first output labels as V1 or Vo1, the second ground labels as GND2, and the second output labels as V2 or Vo2, the third ground labels as GND3, and the third output as V3 or Vo3, of course you can't connected all the ground together.

 

I'm still thinking of trying something with an optocoupler... I have a bunch of these,
and maybe I can learn something new with them.

Below is the LTspice simulation my take on using an optocoupler to control two (or more) LM317's from one voltage control.
The control voltage could be generated from a pot of course.
The op amp drive for the optos provides a voltage to current conversion by putting the input diodes in the feedback loop to avoid the non-linearity of the LED input voltage versus current.
The pots (U6 and U7) are to allow matching of the two outputs due to variations between LM317s and the opto gains.
It shows one supply with common grounds, which you would need to separate for your application.

 

This looks good, and I have the parts. Of course, all the grounds are separated... I have dozens of transformers, not a problem.
Each of the four LM317 units will have separate grounds, and the 5v control (or 10v, if I use four LM317, & four optocouplers, etc.

I like U6 and U7. I just knew this scheme would be wonky and needing trim.

I'm thinking to us 4N28 optocouplers, as I have a bunch.

Below is the LTspice simulation my take on using an optocoupler to control two (or more) LM317's from one voltage control.
The control voltage could be generated from a pot of course.
The op amp drive for the optos provides a voltage to current conversion by putting the input diodes in the feedback loop to avoid the non-linearity of the LED input voltage versus current.
The pots (U6 and U7) are to allow matching of the two outputs due to variations between LM317s and the opto gains.
It shows one supply with common grounds, which you would need to separate for your application.

View attachment 114872

 

The 4N28 has lower transfer gain than the 4N25, so the value of R1 has to be reduced to increase the LED current for the same output current.
Below is the circuit with 4N28's.
I also added a 2.5V reference voltage and a pot for the voltage control.

 

Interesting

 

crutschow, it looks promising, I have those parts, thank you.

What is the voltage at the terminal of R2 and the Op Amp?
Is there a 100 ohm resistor on the wiper of the 10k pot? Sorry, I'm a little
soft on nomenclature/spice.

The 4N28 has lower transfer gain than the 4N25, so the value of R1 has to be reduced to increase the LED current for the same output current.
Below is the circuit with 4N28's.
I also added a 2.5V reference voltage and a pot for the voltage control.

View attachment 114887

 

Will temperature fluctuation of the optocoupler cause any problems, or will the op amp compensate? I'm assembling this
dead-bug style on a heatsink.

 

...........
What is the voltage at the terminal of R2 and the Op Amp?
Is there a 100 ohm resistor on the wiper of the 10k pot? Sorry, I'm a little
soft on nomenclature/spice.

The voltage at the junction of R2 and the top of the pot is 2.5V from the U9 shunt voltage reference.
The opamp input voltage varies between 0V and 2.5V as determined by the wiper position.
No, there is no 100 ohm resistor. It's just a 10kΩ pot.
The w/100 is the wiper position (0 to 1 equals 0% to 100% wiper rotation) for simulation purposes.

Will temperature fluctuation of the optocoupler cause any problems, or will the op amp compensate?

The op amp will compensate for any input diode voltage change with temperature but not for any change in the input/output gain.

 

Thanks.

What is the voltage at R2/V+ ?

And the voltage at U3/V+ ? I guessing at least 5v for the diode drop thru the two optocouplers,
or 10v, as I am using four optocouplers...

The voltage at the junction of R2 and the top of the pot is 2.5V from the U9 shunt voltage reference.
The opamp input voltage varies between 0V and 2.5V as determined by the wiper position.
No, there is no 100 ohm resistor. It's just a 10kΩ pot.
The w/100 is the wiper position (0 to 1 equals 0% to 100% wiper rotation) for simulation purposes.

The op amp will compensate for any input diode voltage change with temperature but not for any change in the input/output gain.

 

As shown all V+ are the same voltage, generated by V1 (arbitrarily 25V).
If you use a separate voltage for the op amp and reference then it should be at least 12V.

 

Okay, I think I understand it. I have enough information to complete the build.
Let's cross our fingers and see.

As shown all V+ are the same voltage, generated by V1 (arbitrarily 25V).
If you use a separate voltage for the op amp and reference then it should be at least 12V.

 

It's been over a week, so I thought I should post something.
The good news is that crutschow's optoisolation circuit is working!

The 10k pots are perhaps too low of a value, and anyway, I should
have used 10-turn pots, but only had three of those on hand. 1-turn
trimming potentiometers are very finicky.

My supply of 4n28 opto isolators had a wide variation. From 10 units,
I was able to match four that I could dial into a low of ~2v, and a high of ~22v.
If I do this again, I'm going to want to start with a double handful of these.
I just got lucky to find 4 of these that were close to same spec. from such
a small sample. I had several that were functional, but way off the mark for what
I am trying to do.

I am learning. I shall post more about this when I have the series lash-up done.
Soon I hope to find out how many volts it takes to let the magic smoke out of a LM723/LM317/etc.

Mucho thanks!

 

The 4N28 has lower transfer gain than the 4N25, so the value of R1 has to be reduced to increase the LED current for the same output current.
Below is the circuit with 4N28's.
I also added a 2.5V reference voltage and a pot for the voltage control.

View attachment 114887

I tried several potentiometer values for U8, from1k to 47k, and they all worked the same:
all the adj. was in the last quarter turn.

I replaced R1 with a 1k ten-turn trimmer. At 668 ohms for R1, I have a smooth range of
adjustment on U8, a ten turn 10k pot.

All for now.

 

is there some other way I can do this?

use a set of gears on N independent pots?

Or a mcu reading off one pot and controls N independent digital pots?

if all fails, use mind power.

 

Or a mcu reading off one pot and controls N independent digital pots?

As long as the digital pots are optically coupled for isolation.