Bench Power Supply dilemma: linear or switched

Projects General Guidance
1

I am building myself a compact bench PSU, controlled by an Arduino.

Basically I'll need up to 5A, and voltages between 1.25V and 18V.

Instead of buying a 110V-20V transformer, and start the project from scratch, I decided to start from an old HP notebook switched power supply, that gives me 19.5V, and up to 4.7A (totallying about 91.6W). I'll settle with 4.7A.

My build will start from there, and here is my dilemma: I have both a LM338T (linear regulator, 5A max) and a LM2679T-ADJ (switched, 5A max) regulators, and I am undecided on which to use. I have a list of both pros and cons for each approach, which are:

LM338T:
Pros:

  • Fewer component count (easier to build and maintain)
  • Short-circuit protect in the output
  • More rugged
  • Less chance of screwups in component selection and build, since I'll etch the PCB myself
  • Does not generate HF noise and has a flat output curve
  • Can reject the ripples from the notebook switched PSU

Cons:

  • Too much heat
  • Huge dropout voltage (up to 2.5V, so I'd have to settle for 17.0V max)
  • Components are easier to calculate and source

LM2679T-ADJ
Pros:

  • Generates less heat since is switched
  • I'll learn a lot in the process, since I never worked with SMPS before and I've been reading a lot from Texas Instrument on last few days
  • Can get me more than 18V, up to 18.5V, I think. I may not need that, but I like to have that little bit more!
  • Can limit the output current internally, although I am already taking care of that with a ACS712 (5A) connected to the Arduino

Cons:

  • More complex build: requires more components, that need to be very well specified
  • All bench PSUs are linear, so I'll take a chance building one that's entirely switched
  • I have no idea how it will behave having an already rippled input
  • Although it is adjustable, I don't know how it will behave on a variable setup, since the calculations for minimum and maximum voltage output in my desired range are quite different
  • HF noise

Thinking about it, the LM338T seems to be the best choice, but I am really concerned about heat, since I don't plan on adding a fan to the case, but I can add a huge ass heatsink to the LM338. I keed remembering Dave's (from eevblog) findings on the problem with his Rigol DP832, that was resetting because a LM317T was getting too hot.

What do you guys think?

2

The LM338 plus heatsink might be 10x the price of a buck converter from ebay.
The LM338 will also be 10x the size.

A buck converter lowers the voltage, but also increases the current.
19.5volt/4.7A in could be 5volt/15A out.
Leo..

3

Does it need to be varied often? Cheap switch modes online are not designed for regular adjustments. You could replace the onboard potentiometer.

Weedpharma

4

Wawa:
The LM338 plus heatsink might be 10x the price of a buck converter from ebay.
The LM338 will also be 10x the size.

The IC itself is less than half the price of a LM2679T-ADJ. I have lots of heatsinks that I've stripped off from old hardware. Actually I have a box full of heatsinks, of several sizes and shapes.

Size is really a concern, since the heatsink takes quite some space, but on the other hand I'll save on footprint, since it requires less components to function properly and is more forgiving on the design. The LM2679, OTOH, is not very forgiving when it comes to board layout.

weedpharma:
Does it need to be varied often?

Not really, but it needs to be variable. It's not like I'll be playing with the adjust knob all the time, but in every use of ut I might need a different voltage or current or both.

Please keep in mind that I already have all (or most of) the parts for both projects, but I'll choose only one of them. I'll use only through-hole components, since I'll design and etch the 1-sided PCB myself.

I am still trying to get some 1N5825 schottky diodes for the LM2679 regulator, in case I choose to use it.

5

I have not checked the datasheet of the LM338T carefully but with 19.5V input I very much doubt it will provide 4.7A at a low output voltage such as 3V. That would require 78W dissipation into the heatsink.

6

http://www.ebay.com/itm/DC-DC-CC-CV-Buck-Converter-Step-down-Power-Supply-Module-7-32V-to-0-8-28V-12A-HC-/261997095157?hash=item3d003e68f5

I can't build a 12Amp stepdown supply for that price.

Add several of these ones for fixed voltages.

Leo..