The issue with having only one current limit (due to the internal protection) can also be dealt with by using a second LM317 as a current regulator, before the LM317 voltage regulator.

Sure and that only throws away ANOTHER 3V OF HEADROOM (voltage drop) meaning the transformer voltage has to be higher and waste even more power.

I'll yell you a secret I learned working first at fairchild Semiconductor and then 20 years at National Semiconductor: those "application circuits" on the data sheet are mostly a joke. They basically paid us to come up with as many different circuits as we could and being ridiculous, a gross waste of money, etc was no deterrent.

The LM317 is a very good product for what it is: a three terminal voltage regulator.

A lab supply IT AIN'T.

 

I have a great respect for your experience Bountyhunter, and I agree with your point that building a higher performance PSU can be very worthwhile.

I think we have different motivations here; you want the OP to build a high performance PSU and I would be happy to see him just build something simple and reliable that works good enough. Ultimately the choice is his and not ours.

 

Thanks guys for the replies.
bountyhunter thank you for your contribution, i would like to know the performances of your discrete psu (if you build it do you have pics/comments?) and how do you get the 12-20V bias voltage?

 

I've modified my design, i will use only one LM338K because of voltage drop, the current limiter is done with a npn and 2 shunts in the return path : used this attached schematic found on a datasheet i will merge it with my current design (my features are :-1.25V for 0V min voltage,soft start,shutdown).

Current limit is set at ~1.2A with R3=0.5 and R4=2xR3
tested on proteus simulation, works great has expected.
Need to build it for real.

 

Hi Bountyhunter, sorry to wake up this thread but I want to build the supply you showed.

There are a few parts in the schematic PDF that have no value, can you clarify how these values are determined?

thx
Ancel