Bench power supply designs -- series pass transistor or IC

[Krakonos]

Hi!

I've been recently thinking about linear bench power supplies (with adjustable current limit), and found out there is pretty much a shortage of designs based on a series pass transistors and opamps. It seems a lot of designs prefer IC of some kind, be it LM317 or LT3080. Any ideas why? Here is what I come up with:

1) Thermal protection included in the IC.
2) Short circuit protection possibly included in the IC.
3) Transistorphobia.

Anything else I missed? I did some experiments and found a simple circuit with LM358 and a beefy BJT works just fine, even on breadboard, for low currents (up to ~1A tested).

It surprises me even more if the supply is controlled by a micro, as the thermal protection could be implemented in software, the same goes for short circuit (if the hardware is able to withstand the short for at least a fraction of a second).

[Kalvin]

- Stability?
- Transient response?
- Regulation accuracy under different load conditions?

[tautech]

Hi!

I've been recently thinking about linear bench power supplies (with adjustable current limit), and found out there is pretty much a shortage of designs based on a series pass transistors and opamps. It seems a lot of designs prefer IC of some kind, be it LM317 or LT3080. Any ideas why? Here is what I come up with:

1) Thermal protection included in the IC.
2) Short circuit protection possibly included in the IC.
3) Transistorphobia.

4) Simplicity
5) Reliability

[MosherIV]

I've been recently thinking about linear bench power supplies (with adjustable current limit), and found out there is pretty much a shortage of designs based on a series pass transistors and opamps.

How hard did you look?

Try "diy op-amp psu", easier to switch to images and just look at the schematics that come up.
Admittedly, most of them are difficult for beginners to understand.

I alway refer to 1 particular project on the web because I think it is the easiest to understand.

  It surprises me even more if the supply is controlled by a micro, as the thermal protection could be implemented in software,

I work as a professiinal embedded software engineer with experince in electronics - I would never recommend relying on a micro to protect from thermal overload. It is easier and more reliable to build in hardware thermal shutdown. By all means, use the micro for thermal monitoring.

the same goes for short circuit (if the hardware is able to withstand the short for at least a fraction of a second).

Again, the micro is not good enough to protect from short circuits.
If the PSU design is correct, the current limit should protect the PSU from short circuits indefinitely.

[Gyro]

Why not split the difference and try an LM395 "Ultra Reliable Power Transistor "? It includes current limiting, SOA protection and thermal limiting.

http://www.ti.com/product/LM395

It's an old Nat Semi design that's stood the test of time. 

[Seekonk]

I have a big bag of those 395 and they are real but any decent PNP with a 317 should be pretty robust.

[m98]

Hi!

I've been recently thinking about linear bench power supplies (with adjustable current limit), and found out there is pretty much a shortage of designs based on a series pass transistors and opamps. It seems a lot of designs prefer IC of some kind, be it LM317 or LT3080. Any ideas why? Here is what I come up with:

1) Thermal protection included in the IC.
2) Short circuit protection possibly included in the IC.
3) Transistorphobia.

4) Simplicity
5) Reliability

6) Lower part count
7) Lower total BOM cost
(Well, only relevant for commercial designs, but still...)

[Krakonos]

Whoa, that's a lot of responses!

I think that some of you misunderstood the question. I'm thinking about adjustable voltage and current psu, just like a regular bench power supply, and I don't see how using an IC would lower your BOM or simplify your circuit significantly, as I think most of the troublesome circuitry still needs to be built. I guess it removes the voltage regulation part, but you still need to measure current and handle the limiting, which seem to be the hard thing, right?

Also, I'm not a trained EE, didn't mention that; though I'm not afraid of complicated circuits and actually enjoy reading them and learning different techniques. Sometimes it's hard though, if the intention isn't clear. I looked around quite a bit, but noticed the transistor circuits that were not switchers are pretty scarce. Many of the designs (both simple and complicated) use an IC as the power stage.

The LM395 looks pretty good, I'll bookmark that!

[Seekonk]

There are never ending posts from people wanting to build 317 power supplies.  Wanting to use a 317 as a current limiter is pure goofiness.  That is looking at everything as a black box.  In itself a 317 incorporates a power transistor op amp and reference which is a great start but to get a truly functioning power supply a lot of parts need to be added.  If adding an opamp, at least add a negative supply for it so the supply can get down to zero.  Then it is two op amp sections.  One that senses voltage and the other to sense current.  These are ORed  together.  Either one can drop the supply voltage till conditions are met.  You might as well use discrete parts for all that.  It will take quite a while for that to seem reasonable to you.  I still have this longing for a three terminal part that can do everything, but snap out of it quickly when it comes down to design.

[MosherIV]

I looked around quite a bit, but noticed the transistor circuits that were not switchers are pretty scarce. Many of the designs (both simple and complicated) use an IC as the power stage.

Yes, because using something like a LM317 is dead easy to build a simple CV only PSU.
If you want both CV and Current Limit then it starts to get harder quickly.
There are examples out there if you look but then you have to know what to look for.

I don't see how using an IC would lower your BOM or simplify your circuit significantly, as I think most of the troublesome circuitry still needs to be built. I guess it removes the voltage regulation part, but you still need to measure current and handle the limiting, which seem to be the hard thing, right?

Correct, using a dedicated Voltage Regulator in a circuit where it is trying to control both the voltage and measure and limit the current becomes self defeating. In other words the circuit becomes so complicated, to over come what the voltage regulator device is doing, that it is easier to design and build a voltage regulator circuit from scratch. If this was not the case then most commercial linear bench PSUs would be built around some Voltage Regulator device but they are not, they are built around a voltage regulator circuit based on op-amps.

You can do it with just transistors but I could not figure them out, I am not that good with transistors. The op-amp based circuits I understand.

[MosherIV]

FYI

This is the PSU design that I often refer people to
http://www.microsyl.com/index.php/2010/03/31/bench-power-supply-0-25v-0-5amp/

It is divided up into clear sections that are easy to identify their function.

[rsjsouza]

Another example of linear-only power supply is the ubiquitous Mastech HY3005 (a schematics is shown in a post of mine here).