Hi everyone

I've had success with my 12v ATX supply but now I'm in a position where I need a more powerful regulated DC supply with an adjustable range of at least 30-50V @ 20A max. This is my first power supply from scratch so I figured that linear would be simpler to understand and construct than a SMPS. Having searched through the forums for LM317 high current supplies I have found some very helpful threads, mainly this one

LM317 thread link

So after viewing other circuits and consulting the LM317HVT datasheet I have come up with the following design which ideally would give me a maximum voltage in-out differential of 60V. At such a high differential only a small current <100mA can flow through the regulator so at any voltage out value I need to limit the regulator current flow to 100mA.



T1 is a 240/50V(x2) toroidal @ 1kVA with both secondary's wired in parallel, BR1 is undefined at the moment and C4 is the smoothing capacitor with a higher enough value to assure <10%ripple @70.7Vp 20A load. The bypass circuit I have used is basically the one from the TI datasheet with a few adjustments.

1)BD744C is replacing the 2N2905 as it can handle the higher voltage, my worry here is that it has a significantly higher collector cut-off current than the 2n2905.

2)The 3x TIP73 mentioned in the datasheet seem to be unavailable so I chose the BD743C complement to the BD744C. So hopefully this will distribute the max 20A over all 3 NPNs and the base current should be kept down to about 270mA and the PNP base current requirement about 40mA. Assuming R2 is there to limit this current to the PNP then if I estimate I have 68.9V max from 70.7 Vp after FWB(1.2V) and R1(0.6V) voltage drops I get a resistor value of 1.8K (2.8W diss.).

3)Diodes have been added to protect from capacitor discharge across the regulator.

Now the problems I have come across are as follows:-

1) I am unsure as to the value of R1 as I need to limit the current through the regulator at any output voltage. I have read this discussed in the thread linked above however I couldn't reconcile it with my design as it required the output voltage from the regulator to be less than half the input or the maths didn't work. (link post #27)

2)R4 is only described in the datasheet as "minimum 30mA", not sure as to it's purpose but I have set it to draw a minimum of 30ma across the range of output voltages with a max power dissipation of around 10W. Not great not sure why I need it.

3)V1 and V2 are used for voltage adjust and I was aiming for a coarse and fine setup but have I got the ratio and configuration setup correctly for this application?

4)When using the 3xNPNs in parallel configuration do I need to add additional base resistors?

5)Would it be advisable to add additional circuitry for inrush current limiting and short protection?

Note about the load: The load this is primarily designed for will be a Royer oscillator induction heater. Ideally I would like to have 30-60V output so I can experiment with inputs but if a wider voltage range is attainable this would be preferable.

Thanks for taking the time to read my design everyone and if you can help or see any improvements/faults you wouldn't mind sharing I would be really grateful. xD

Thanks all

P.S. I have attached datasheets for quick access.

 

If you want a maximum of 100mA in the regulator, then set R1 to 7ohms, that will give a voltage drop of 0.7v at 100mA to turn on Q1, but you have no output protection if the other transistors fail and go S/C.

 

This is a pretty common circuit, very robust and stable. Good choice. I do see that you are lacking emitter resistors for Q2,3,4. I would start with .1 ohm. This will make the transistors share the load more equally.

Do you use LTspice? It would be an excellent program to optimize this circuit.

 

If you want a maximum of 100mA in the regulator, then set R1 to 7ohms, that will give a voltage drop of 0.7v at 100mA to turn on Q1, but you have no output protection if the other transistors fail and go S/C.

Thanks Dave I was hoping it was going to be that simple and with a nice low resistance @140mA it'll be a nice low 0.14W dissipation. Good call on that diode too, thanks I could of lost those NPNs when I switched the oscillator off. Do you think I'd be better off with 1 diode here or multiple in parallel with the NPNs?

This is a pretty common circuit, very robust and stable. Good choice. I do see that you are lacking emitter resistors for Q2,3,4. I would start with .1 ohm. This will make the transistors share the load more equally.

Do you use LTspice? It would be an excellent program to optimize this circuit.

Thanks lestravelled I've seen some of your posts in previous LM317 threads which have been very helpful I hadn't thought of adding emitter resistors to distribute the load but this was a concern I neglected to mention. The only worry I'd have with adding 0.1ohm resistors is that if the 20A is evenly distributed that would be 5w dissipated across those resistors, would I be able to get the same effect with a lower resistance value? I do have LTspice but I haven't had much experience with it so I'm switching between Circuit Wizard (not great but simple) and MultiSim (great but complicated lol). Does LTspice have a model for the LM317HVT?

Thanks guys

 

@Marcus2012
You would actually dissipate 13.3 watts total (4.4 watts each) on the emitter resistors at 20 amp output. That number is nothing compared to the wattage that Q2,3,4 will be dissipating.

 

@Marcus2012
You would actually dissipate 13.3 watts total (4.4 watts each) on the emitter resistors at 20 amp output. That number is nothing compared to the wattage that Q2,3,4 will be dissipating.

Yeah you do make a very compelling point I have been subconsciously neglecting this whole time. That this is never going to be a 100% efficient supply and I will have to accept losses in conversion. Is there a lower resistance threshold at which point this negative feedback becomes ineffective?

 

Hi everyone

I've had success with my 12v ATX supply but now I'm in a position where I need a more powerful regulated DC supply with an adjustable range of at least 30-50V @ 20A max. This is my first power supply from scratch so I figured that linear would be simpler to understand and construct than a SMPS. Having searched through the forums for LM317 high current supplies I have found some very helpful threads, mainly this one

LM317 thread link

So after viewing other circuits and consulting the LM317HVT datasheet I have come up with the following design which ideally would give me a maximum voltage in-out differential of 60V. At such a high differential only a small current <100mA can flow through the regulator so at any voltage out value I need to limit the regulator current flow to 100mA.

View attachment 99415

T1 is a 240/50V(x2) toroidal @ 1kVA with both secondary's wired in parallel, BR1 is undefined at the moment and C4 is the smoothing capacitor with a higher enough value to assure <10%ripple @70.7Vp 20A load. The bypass circuit I have used is basically the one from the TI datasheet with a few adjustments.

1)BD744C is replacing the 2N2905 as it can handle the higher voltage, my worry here is that it has a significantly higher collector cut-off current than the 2n2905.

2)The 3x TIP73 mentioned in the datasheet seem to be unavailable so I chose the BD743C complement to the BD744C. So hopefully this will distribute the max 20A over all 3 NPNs and the base current should be kept down to about 270mA and the PNP base current requirement about 40mA. Assuming R2 is there to limit this current to the PNP then if I estimate I have 68.9V max from 70.7 Vp after FWB(1.2V) and R1(0.6V) voltage drops I get a resistor value of 1.8K (2.8W diss.).

3)Diodes have been added to protect from capacitor discharge across the regulator.

Now the problems I have come across are as follows:-

1) I am unsure as to the value of R1 as I need to limit the current through the regulator at any output voltage. I have read this discussed in the thread linked above however I couldn't reconcile it with my design as it required the output voltage from the regulator to be less than half the input or the maths didn't work. (link post #27)

2)R4 is only described in the datasheet as "minimum 30mA", not sure as to it's purpose but I have set it to draw a minimum of 30ma across the range of output voltages with a max power dissipation of around 10W. Not great not sure why I need it.

3)V1 and V2 are used for voltage adjust and I was aiming for a coarse and fine setup but have I got the ratio and configuration setup correctly for this application?

4)When using the 3xNPNs in parallel configuration do I need to add additional base resistors?

5)Would it be advisable to add additional circuitry for inrush current limiting and short protection?

Note about the load: The load this is primarily designed for will be a Royer oscillator induction heater. Ideally I would like to have 30-60V output so I can experiment with inputs but if a wider voltage range is attainable this would be preferable.

Thanks for taking the time to read my design everyone and if you can help or see any improvements/faults you wouldn't mind sharing I would be really grateful. xD

Thanks all

P.S. I have attached datasheets for quick access.

And a REALLY BIG heat sink. I've never considered anything that big before. At what point do you consider gas cooling (Freon, CO2) or refrigerated liquid (silicon) cooling. . I have seen military systems using such things.

 

And a REALLY BIG heat sink. I've never considered anything that big before. At what point do you consider gas cooling (Freon, CO2) or refrigerated liquid (silicon) cooling. . I have seen military systems using such things.

lol yeah I was originally thinking of buying 200mm x 100mm x 40mm, threading it and bolting it to the back or side of the unit and I'm not entirely certain that'll be big enough . I'm not sure at the moment but I think a big external heat sink would probably preferable to an internal one with forced air.

 

A fan blowing across a heat sink does wonderful things.

 

lol yeah I was originally thinking of buying 200mm x 100mm x 40mm, threading it and bolting it to the back or side of the unit and I'm not entirely certain that'll be big enough . I'm not sure at the moment but I think a big external heat sink would probably preferable to an internal one with forced air.

The math of heat sinks that large is beyond me but off hand I would say the size is about right. Yes, external.

 

By the way what is R4 doing and why is it so important it be at 30mA minimum? is this a dummy load to ensure a minimum current draw across the regulator?

 

lol yeah I was originally thinking of buying 200mm x 100mm x 40mm, threading it and bolting it to the back or side of the unit and I'm not entirely certain that'll be big enough . I'm not sure at the moment but I think a big external heat sink would probably preferable to an internal one with forced air.

You can't get there with 3 BD743C's. They are only good for 90 watts each even if you can keep the case at 25C.

 

You can't get there with 3 BD743C's. They are only good for 90 watts each even if you can keep the case at 25C.

Good spot, I've just come to this conclusion myself I'm looking at 150W+ NPNs at the moment.

EDIT: Looks like I'll now be using either a TO-3 package or a TO-247 package was they will most likely be the most efficient at dissipating power. TO-3's look like they are the standard tried and tested method but I have never used these.

 

Good spot, I've just come to this conclusion myself I'm looking at 150W+ NPNs at the moment.

EDIT: Looks like I'll now be using either a TO-3 package or a TO-247 package was they will most likely be the most efficient at dissipating power. TO-3's look like they are the standard tried and tested method but I have never used these.

What does this brute do? Maybe there is an easier way. How good does the regulation need to be?

 

What does this brute do? Maybe there is an easier way. How good does the regulation need to be?

These NPNs will be a bypass conduit for most of the current but the voltage drop across each NPN will match the regulator V in-out differential. As it stands, to get a 30V voltage drop I couldn't draw more than 3Amp through each transistor without power dissipation going over 60W. In order to get the most efficient range for between 30-50V I'm thinking I may need to get a 240/45V(x2) instead of the 50V(x2) transformer. Then my Vp would only be 63.63V lowering my differential by 7V.

I'm looking at the MJ15003G NPN TO-3s at the moment, they look brutish enough to handle what I need I am thinking I may stick with the same PNP though as the most that will ever be dissipating is 60W @ a 60V differential the Lm317 won't be doing much work either (6Wish).

 

These NPNs will be a bypass conduit for most of the current but the voltage drop across each NPN will match the regulator V in-out differential. As it stands, to get a 30V voltage drop I couldn't draw more than 3Amp through each transistor without power dissipation going over 60W. In order to get the most efficient range for between 30-50V I'm thinking I may need to get a 240/45V(x2) instead of the 50V(x2) transformer. Then my Vp would only be 63.63V lowering my differential by 7V.

I'm looking at the MJ15003G NPN TO-3s at the moment, they look brutish enough to handle what I need I am thinking I may stick with the same PNP though as the most that will ever be dissipating is 60W @ a 60V differential the Lm317 won't be doing much work either (6Wish).

The problem is when you want 30 volts at 20 amps. At that point there is about 40 volts and 20 amps in the transistors.
I make that 800 watts.
A switching regulator might be better, but the voltage is kind of high and inductors for it might be expensive. The other option might be a pre-regulator to lower the voltage when the output is low.

 

The problem is when you want 30 volts at 20 amps. At that point there is about 40 volts and 20 amps in the transistors.
I make that 800 watts.

Yeah I know , unfortunately this fact is something I'd have to accept. I can slightly improve it with a 40V transformer to minimized to volt drop at 48Vout (my "new" maximum output voltage) to 3-5V across each NPN. Of course I'd still only have about half my max current at a 30V out without popping the NPNs but I could live with that. But........this is where I ask if this is cost effective when compared to buying a variable supply?

 

Yeah I know , unfortunately this fact is something I'd have to accept. I can slightly improve it with a 40V transformer to minimized to volt drop at 48Vout (my "new" maximum output voltage) to 3-5V across each NPN. Of course I'd still only have about half my max current at a 30V out without popping the NPNs but I could live with that. But........this is where I ask if this is cost effective when compared to buying a variable supply?

What are you running with it?

 

What are you running with it?

Royer oscillator induction heater but I was hoping to use it in future with similar projects.

 

Royer oscillator induction heater but I was hoping to use it in future with similar projects.

For that you probably don't need to regulate it. You could just use the big filter caps.