LM317 control by PWM

[soldar]

I don't see how that circuit is supposed to convert 5V PWM into 1.25V to 30V.

Yeah, sorry, just ignore it. It was a sketch of a half baked idea which I cannot complete. At least I have been playing around and learning a bit of LTSpice so something good came of it.

[Yansi]

1) Voltage drop is important because you won't be able to get 30V out, without exceeding the maximum voltage rating of the LM358.

You won't likely be able to do that anyway, with a mains-transformer based supply.  If you add into account mains voltage variation, which is likely +-10% at most places, you have extremely small margin there, if any at all.

So back to the drawing board anyway. 

[Zero999]

1) Voltage drop is important because you won't be able to get 30V out, without exceeding the maximum voltage rating of the LM358.

You won't likely be able to do that anyway, with a mains-transformer based supply.  If you add into account mains voltage variation, which is likely +-10% at most places, you have extremely small margin there, if any at all.

So back to the drawing board anyway. 

With the LM317 on the output, the op-amp only has to swing from 0V to 27.5V to cover a 1.25V to 30V output range, so it could easily be powered from a regulated 31V power supply.

A far bigger problem is the LM317's safe operating area protection kicking in when the output voltage is set low. A centre tapped transformer with a switch should be used minimise power dissipation.

[Yansi]

Regulated dedicated 31V supply for opamp.  Then add negative bias voltage so you can lower the current limit of the 317 realiably even to short circuit operation. Then regulated this or that or that there... you end up with something pretty hilarious, as most 317 or 723 based power supplies are. Overly complex (junk)!   

[Zero999]

Regulated dedicated 31V supply for opamp.  Then add negative bias voltage so you can lower the current limit of the 317 realiably even to short circuit operation. Then regulated this or that or that there... you end up with something pretty hilarious, as most 317 or 723 based power supplies are. Overly complex (junk)!   

You get that anyway when designing a 0V to 30V PSU, whichever route you take.

If you opt for the traditional Darlington pair method you're proposing, the op-amp's power supply voltage needs to be even higher, ruling out most common op-amps or you can add a buffer circuit, but that leads to extra complexity.

[Yansi]

So why are you constantly arguing my solution is too complicated, if yours is becoming exactly the same? 

The discrete solutions would give quite more ways to implement things differently (or even better, if you want).

[Zero999]

So why are you constantly arguing my solution is too complicated, if yours is becoming exactly the same? 

The discrete solutions would give quite more ways to implement things differently (or even better, if you want).

The LM317 does things such as safe operating area and thermal protection which are virtually impossible with a discrete design and is no more complex to implement. The only extra complexity associated with the LM317 is the requirement for a current sink to provide a minimum load, which is just a resistor, two diodes and a low power transistor. This is already less complex than the Darlington pair you're proposing, before any current limiting is added.

Getting an op-amp with a sufficiently high voltage rating is a challenge whichever route you take, but the Darlington pair one is more difficult. The LM317 solution only needs the op-amp's output voltage to reach 1.25V less than the output voltage, whist the Darlington solution requires the op-amp's output to exceed the output voltage by at least 1.6V and most op-amps require a couple of volts headroom.

[Yansi]

SOA protection using discrete components is just as easy to implement.   What other thing than a current limit following a voltage drop across the pass element would you need?  Discrete transistor audio amps solve this SOA protection on common basis. 

Thermal protection may become more tricky, depending on the circuit, but again, what more then just a thermal cutoff is needed? Havin a thermal cutoff fuse in the mains input is all what is needed to protect the supply. If it can't cool enough, then it is probably badly designed or would not work anyway (as some dumbass may have covered the heatsinks, etc).

[Kalvin]

TI's LM395 "ultra reliable transistor" comes with current limiting, power limiting, and thermal overload protection.
http://www.ti.com/product/LM395/description
It is a complete IC turned into an NPN transistor with extra protection features.

[soldar]

Couple of thoughts.

The LM317 uses a resistor to set the output voltage. Could it be replaced by a FET which could be controlled by voltage?

The LM317 uses a resistor to set the output voltage. It would be easier to use a different circuit which uses a voltage reference, like a zener, and then vary that reference voltage.

[David Hess]

TI's LM395 "ultra reliable transistor" comes with current limiting, power limiting, and thermal overload protection.
http://www.ti.com/product/LM395/description
It is a complete IC turned into an NPN transistor with extra protection features.

The LM395 is a great solution to add integrated protection but they cost almost 10 times more than LM317s.

The LM317 is not ideal for reasons not discussed here yet, but it is a simple way to add current limiting, safe operating area protection, thermal protection, and buffering at a very low cost.

[iMo]

I think the OP has got enough hints on the topic he is interested in, he can elaborate and report the results. Let him work.

[Zero999]

SOA protection using discrete components is just as easy to implement.   What other thing than a current limit following a voltage drop across the pass element would you need?  Discrete transistor audio amps solve this SOA protection on common basis. 

All right not impossible,  but more complex than just using the LM317 in the first place!

TI's LM395 "ultra reliable transistor" comes with current limiting, power limiting, and thermal overload protection.
http://www.ti.com/product/LM395/description
It is a complete IC turned into an NPN transistor with extra protection features.

The LM395 is a great solution to add integrated protection but they cost almost 10 times more than LM317s.

The LM317 is not ideal for reasons not discussed here yet, but it is a simple way to add current limiting, safe operating area protection, thermal protection, and buffering at a very low cost.

Yes the LM395 is expensive!
https://uk.farnell.com/texas-instruments/lm395t/power-transistor-to-220-3-395/dp/9488278?st=LM395

Why do you think the LM317 is not ideal? The main problem I can think of is stability and the increased risk of oscillation.

[David Hess]

Yes the LM395 is expensive!
https://uk.farnell.com/texas-instruments/lm395t/power-transistor-to-220-3-395/dp/9488278?st=LM395

Meanwhile the TI LM317 from uk.farnell.com is about 1/7th the price.  The ST LM317 is less.

Why do you think the LM317 is not ideal? The main problem I can think of is stability and the increased risk of oscillation.

The LM317 complicates the frequency compensation and the -1.25 volt bias requirement for a zero volt output is inconvenient.

[Zero999]

The LM395 also doesn't appear to be that straightforward to use. It still passes a quiescent current of up to 10mA, when V_BE = 0! To get a true 0V out, a negative rail is required, but I suppose a current mirror circuit will get within 100mV or so.
http://www.ti.com/lit/ds/symlink/lm395.pdf

[tzc]

Hi,

If I connected this two ground together will it cause any problems to my circuit?
kindly give some advise
Thanks 

[Yansi]

And how would it supposed to work, without both circuits having a common reference point?