@YorkshireDave No problem. What threw me was the period after every word. Seemed like you were pounding your fist with every word. Oh well. I'll get over it. - - - I'm over it.

Still, there's not much more I can offer.

Stay well.

Tony. Are you able to explain to me how a 317 pre regulator works pls? Also, would you be able to explain how to calc what pot value I need for voltage adjustment for a 317? Ive read the data sheet but somehow it just doesn't go in! TIA ;-)

 

how a 317 pre regulator works

Ref post #15, it keeps a constant voltage drop across the output regulator as determined by R1 and R2.
Thus it shares the total voltage drop between the two regulators.

how to calc what pot value I need for voltage adjustment for a 317?

Here's a simple way--
The LM317 maintains a constant 1.25V between the Vout pin and the ADJ pin.
This means there is a constant 1.25V across R3, and thus the R3 current is also constant at 1.25V / 120 = 10.417mA, independent of output voltage.
The output voltage is then determined by the sum of the voltage drop across R4, due to this current (since negligible current goes into the ADJ pin), plus 1.25V.
So, for example, to get an output voltage of 15V, the value of R4 = (15V-1.25V) / 10.471mA = 1.32KΩ.
So you pick a pot resistance for the maximum output voltage you want.

Make sense?

 

If you don't want to work t out, have a look at this table.
http://www.reuk.co.uk/wordpress/electric-circuit/lm317-voltage-calculator/

 

@crutschow explained it better than I could have.

R3 & R4 form a voltage divider. As Crutschow explained, there's a reference voltage internally constructed. When the output goes higher, so does the sense voltage at the junction of R3 R4. If the sense voltage is higher than the reference voltage then the regulator cuts back on the voltage output. When the voltage falls, the sense voltage drops below the reference voltage and the regulator UPS the power to maintain the voltage. Between upping and dropping the voltage is a balancing point where the regulator is outputting the desired voltage.

Calculating for R3 and R4 is easy enough. It's just a voltage divider. You can use a pot or two resistors. But if using a pot - make sure you have a minimum and maximum limiting resistor in circuit so you don't stress the 317 accidentally by turning the pot to full up or full down.

 

Ref post #15, it keeps a constant voltage drop across the output regulator as determined by R1 and R2.
Thus it shares the total voltage drop between the two regulators.
Here's a simple way--
The LM317 maintains a constant 1.25V between the Vout pin and the ADJ pin.
This means there is a constant 1.25V across R3, and thus the R3 current is also constant at 1.25V / 120 = 10.417mA, independent of output voltage.
The output voltage is then determined by the sum of the voltage drop across R4, due to this current (since negligible current goes into the ADJ pin), plus 1.25V.
So, for example, to get an output voltage of 15V, the value of R4 = (15V-1.25V) / 10.471mA = 1.32KΩ.
So you pick a pot resistance for the maximum output voltage you want.

Make sense?

Thank you so much for being so patient and understanding ;-)

So, the 317 needs a setpoint of 1.25v. Its output voltage can be varied using the pot so that you get that 1.25v at the op voltage you need? In other words, you have a pot, subject to minimum & maximum values so as not to kill the 317, so that AT the output voltage you need, it feeds 1.25v back into the 317 to tell it it's at 'setpoint'? By jove I think I understand! Thank you (and Tony) for clarifying.

So re the pre reg. With a supply of say 70v, do you set that up (work out the resistances as above) so that there is say 35v drop across it? Does that then mean that you only have an adjustable output from 3-32v or is it 3-67v?

 

OOF! How are you dropping 70 volts down to the safe input range of the 317? I hope not by a voltage divider using resistors.

 

OOF! How are you dropping 70 volts down to the safe input range of the 317? I hope not by a voltage divider using resistors.

Remember, std 317s can cope with a maximum drop of 40v. I'm using the HV versions which cope with 60v so should be fine. Oddly, I cannot find anywhere where the max applicable voltage is, only the max drop across...

 

So re the pre reg. With a supply of say 70v, do you set that up (work out the resistances as above) so that there is say 35v drop across it? Does that then mean that you only have an adjustable output from 3-32v or is it 3-67v?

Neither.
It's 1.25-32V.

 

Your Transformer will be wired like this circuit, so you could use the bridge rectifier and smoothing capacitors , you will need to drop the voltage down to 40V to input into the LM317 or LM337 , with the ideas posted. Ideally you need PnP pass transistors for extra current, or if you use LM337 you can use NpN.


View attachment 203776

Tony, your 30-0-30 is not what the OP said. With 40-0-40 the output voltages will be 56V off load.
The circuit is however the starting point. If I coudl suggest, I'd drop the 317 and use a discrete circuit to provide up to 56V on each side. It is possible to design a + and - voltage regulator using 2N3055's without needing the PNP complement.

 

Neither.
It's 1.25-32V.

I tried to sim the tracking pre-regulator mentioned in post 15 and compare it to a (non-tracking) pre-regulator. In both cases, the pre-regulator is set for around 33-34V, although with the tracking pre-regulator it allows a much wider final setpoint.

I freely admit I have no idea if I did this right (I know I can trick myself by setting up bad simulations - I've done it before!) Does this simulation look realistic? Are both of these circuit variations legitimate?

 

So, if I'd like the widest output range (with my LM317HVs & input of 70V) I'd have to modify the pre-reg to drop 10V so I'd have an adjustable OP of tween 1.25 & 57v. Having investigated further, I understand a mod using diodes in the cct means you can achieve 0v - but that's an aside...

I'd like to use the LM317s as I have bag full of 'em ;-)

 

I was wrong.
The output does go to the maximum with a tracking preregulator.

I'd like to use the LM317s

You can.
With a slight tweak to the value of R2, the two LM317's start out sharing the voltage drop (purple and white traces), and then at the maximum output (green trace), the drop across both of then is only a few volts.
So the maximum drop across either one is never more than about 35V under normal operation.

But you must take care not to short the output, as that will put 70V across U5.

 

I understand a mod using diodes in the cct means you can achieve 0v

That's only if you have a negative voltage source available.

 

You will have to take into account the dissipation of the regulator.
So, if you have 60V across the regulator, the max current you are able to draw will be reduced, and more so with lower output voltage set.
And the power supply needs to be able to survive a short circuit.
Make sure you add all the protection diodes as per the app note in the data sheet.
A switch mode tracking pre-regulator may be way to go, or just get a different transformer.
I think a variable supply with such a wide voltage range is not really needed.
Maybe a 1.25V to 13.8V is ok. Then have fixed 5V, 12V and 24V supplies.

 

Great news thank you.

In terms of dissipation, I assume it's greatest at low voltage & high current so both must be heat sink mounted.

One last thing. Where does the emitter of the 2n3055 connect? The input of the 1st or 2nd 317?

 

I don't know if this is helpful in this thread, but the pre-regulator circuit intrigued me, and I have one more weird idea inspired by it. When I first saw the tracking pre-regulator, I thought it was going to split the voltage drop (and therefore the dissipation) between the two regulators somewhat evenly, but that's not how it works at all. So, I decided to try to come up with a variation that does.

I can't get them perfectly balanced, but I've found that adding one resistor, and changing the value of two others, allows you to split the wattage between the two regulators in a more balanced way. Any potential benefit from this would be largest in the middle of the range when the tracking pre-regulator would have one regulator doing almost all the work. For example, I simulated both versions of the circuit with a 100mA load on each, and there's a point when the output voltage is at about 36V where the tracking pre-regulator circuit has one device dissipating 3.3W and the other doing very little. This new load-splitting circuit has both regulators working almost evenly at that point, both below 2W dissipation. So maybe you can run more current for the same amount of heat sinking with this arrangement (or use smaller heat sinks for any given current draw,) especially if you're needing to run in the middle voltage ranges.

Like many of my ideas, I've only played with this in simulation, and I may have inadvertently violated some design rules (like if the LM317 had min/max feedback resistor values, etc.) So, I'm floating this idea out there as something for others who know better than me to evaluate, not as a vetted solution.

 

I don't know if this is helpful in this thread, but the pre-regulator circuit intrigued me, and I have one more weird idea inspired by it. When I first saw the tracking pre-regulator, I thought it was going to split the voltage drop (and therefore the dissipation) between the two regulators somewhat evenly, but that's not how it works at all. So, I decided to try to come up with a variation that does.

I can't get them perfectly balanced, but I've found that adding one resistor, and changing the value of two others, allows you to split the wattage between the two regulators in a more balanced way. Any potential benefit from this would be largest in the middle of the range when the tracking pre-regulator would have one regulator doing almost all the work. For example, I simulated both versions of the circuit with a 100mA load on each, and there's a point when the output voltage is at about 36V where the tracking pre-regulator circuit has one device dissipating 3.3W and the other doing very little. This new load-splitting circuit has both regulators working almost evenly at that point, both below 2W dissipation. So maybe you can run more current for the same amount of heat sinking with this arrangement (or use smaller heat sinks for any given current draw,) especially if you're needing to run in the middle voltage ranges.

Like many of my ideas, I've only played with this in simulation, and I may have inadvertently violated some design rules (like if the LM317 had min/max feedback resistor values, etc.) So, I'm floating this idea out there as something for others who know better than me to evaluate, not as a vetted solution.
View attachment 204091

Love that you've been inspired to experiment
A couple of questions if I may.
1 whats happening at the junction of the resistors & the adj input to the 1st 317? How do the voltages work out to deliver the 1.25v?
2 as per my prev post where do I connect the emitter of the 2n3055 to dissipate more current?

Just wanted to thank everyone for your help

 

Love that you've been inspired to experiment
A couple of questions if I may.
1 whats happening at the junction of the resistors & the adj input to the 1st 317? How do the voltages work out to deliver the 1.25v?
2 as per my prev post where do I connect the emitter of the 2n3055 to dissipate more current?

Just wanted to thank everyone for your help

I was sleepy and not thinking clearly last night. It was just a vague concept that seemed interesting. I think I understand it better this morning with fresh eyes, and I've made some adjustments.

In the version above, you can see the node labelled "Adj1" is the junction of the voltage divider between the 70V supply and the output of the 2nd LM317. This divider splits the difference between the two voltages evenly. The regulator always works to maintain 1.25V between adj pin and out pin, so the output of the 1st LM317 is constantly 1.25V above the mid-point that would have been my ideal target. It's never quite where I wanted, but the slope is right and it's pretty close. I've dramatically increased the value of R7 so that the output of LM317 number 1 no longer plays a large role in controlling the voltage at the adjust pin - it's primarily controlled by the input and output voltages. This provides a more balanced and consistent sharing of the load between the two regulators, although there's still a constant offset in terms of power dissipation between them.

In order to cancel out most of that offset, in the next version of the circuit (above,) I tried adding diodes between the supply and R10. You can see in the graph that the diodes drop the supply side of the voltage divider by a few volts, which compensates for the 1.25V offset the LM317 maintains. This balances the power dissipation even more than the previous circuit, although I'm not sure how much this slight improvement matters.

As I said in earlier posts, these are all just brainstorming concepts. I'm having fun tinkering, but I can't promise this would work properly in reality. It would take careful study of datasheets and some real-world verification before I'd trust it (or someone who understands this stuff far better than I looking it over and giving their blessing!) Also, it occurs to me that I've done all of this simulating with steady DC inputs and steady current draw. It would be interesting to see how much my added resistors impact stability and noise rejection. For all I know I've inadvertently defeated much of the regulator's stability by adding resistors that provide a complete bypass path around them. I'll have to do some more sims with noise injection at supply side and varying load demands...

 

Love that you've been inspired to experiment
A couple of questions if I may.
1 whats happening at the junction of the resistors & the adj input to the 1st 317? How do the voltages work out to deliver the 1.25v?
2 as per my prev post where do I connect the emitter of the 2n3055 to dissipate more current?

Just wanted to thank everyone for your help

I apologize - I didn't really catch what the plan was with the 2N3055, so I can't answer that one for you.

I do have an update on my silly wattage splitting scheme. Further simulation showed that it was prone to severe oscillations in many scenarios. I experimented with adding resistance and/or capacitance in a number of places with no luck, and finally found a spot that seems to help. As before, I can't promise anything as far as real world performance, but with the addition of C1 in parallel with the lower resistor "Rpot2" (it was a trim-pot in some versions of my simulations,) the circuit seems to be stable now.

I've added a few extra voltage sources to the sim, one simulating 60Hz ripple, and the other simulating a burst of high frequency noise. The regulator appears to provide stable output regardless of these varying voltages, and it provides an mostly-balanced split in dissipation between the two regulators, regardless of output voltage setting. I don't know how useful this circuit would be in the real world, but it was a lot of fun to experiment with!