LM723 Based Power Supply

As mentioned in the comments, the base drive transistor is reversed, the current limit has no fold-back and therefore isn't safe if you have a short circuit or very low impedance load, and you won't be getting the most out of your pass transistor because you're driving your regulator IC and base drive transistor from the same source as the collector of your pass transistor.

I'd suggest you get a copy of the ARRL handbook, which has a number of power supply designs presented along with explanations of _why_ things are done the way they're done.

The datasheet has some example designs that can give you a place to start.

Fold-back can be very tricky to implement and keep stable _and_ get reliable start-up.

I prefer thermal or current-limit-induced snap-off, requiring a reset.

I don't know why people insist on using such crap as the 723, when you can do much better with a TL431 and a few discrete transistors. ...Jim Thompson

IIRC there is also an LM350. Much easier but it won't go below 1.25V. Another problem is power dissipation. You probably need some circuitry to switch the secondary transformer turns.

Probably because of all the circuits out there based on the 723, and the seductive power of a "one chip does all" application (never mind that it's a "one chip does all in a mediocre way")

The 723 does have a buried-zener reference, though, so it's probably a lot quieter than a TL431. I might have to switch to that as a jellybean quiet reference if the LM329 goes away!

Cheers

Phil Hobbs

If it needs to be 'proven', you need to BUY, not build.

There's a half dozen tricks that will help: the input power can be transformer-tap-selected to lower the waste, or the pass transistors can be substituted with regulator ICs (which have inbuilt thermal limiting and won't burn up), or you can use two regulators with different voltage ranges, and diode-isolate so they share the output pin.

The easiest thing to do, though, is just to make a brute-force DC power supply, and plug it into a Variac when you want to adjust it. That works, it's efficient, and the Variac (variable autotransformer) comes in handy for lots of other things.

You might try the LM338K adjustable regulator, 5 Amp, built in thermal and current limiting. Adjusts from 1.2 to 30 volts with a few parts -

-Bill

Even with an infinite heatsink, this part cannot (there's a graph in the data sheet) manage 2.5A if the in/out difference gets as high as 25V. Reference accuracy is only about +/- 5% over temperature, if that matters (the uA723 is better).

Have a look at

Several power suppplies there.

whit3rd wrote in news:604e1c03-7fe6-4ba7-a59d- snipped-for-privacy@e14g2000yqe.googlegroups.co m:

For such small power pretty well any cheap bjt or fet in say a to247,TO-3PN or ISOPLUS247 is fine you'll need two maybe three for the pass device. Notice I didnt say T0-3 and similiar because heatsinks are more widely avilable for the other packages. A $20 or less heatsink should be good enough.Mount them on something like this

I was easily dissapateing 120W useing two T0-3PN fets on a heatsink from a microwave oven with a $5 42CFM fan. The case temp was at about 90C.The heatsink was less then half the size of the one in the link.

The only time you are going to dissapate 82W would be in the event of catastophic failure. The 723 has foldback protection if it fails the supply should be fused anyway.

A 160VA or better variac is pretty pricey though and bulky, they are definetly useful though.

The tracking pre-regulator design on the data sheet might (p12)

And how much can a 2N3055 dissipate in the same TO-3 package with an infinite heatsink? Probably depends on ambient temp? You could use a center tapped xformer to select either full or half voltage input to reduce heat. There should be some 12VAC (17 volt peak) battery charger transformers using the SCR idea. I think they use a center taped transformer so they only need 2 diodes. If you use a bridge and a switch, you can select 17 volts peak or 34, which should be about right for an adjustable 30 volt supply. Maybe DigiKey has something that would work with a center tap?

-Bill

Enough power (spec sheet says 115W), but the safe-operating-area limit for continuous 2.5A output is only 20V, at 25 C (i.e. with infinite heat sink in a warm room). That's called the 'second-breakdown' limit.

It's a good, cheap transistor, but it'll take several in parallel to do this job, and/or a series transistor to take some of the voltage-drop burden.

A laughable claim!

At least use MJ15001 for starters. 2N3055 is old and should be forgotten.

Tim

"Tim Williams" wrote in news:i9mv68$udl$ snipped-for-privacy@news.eternal-september.org:

Whats wrong with a P-FET. $8 AT Digikey 3 pages of 300W plus pacakages to choose from.

IXYS makes a linear line of fets too,nice. ;-)

The next high power regulator I make is going to be N-FETS for the pass device; much larger selection, cheaper heatsinks with a larger selection and no stability concerns at least compared to a PFET or PNP.

On a sunny day (Wed, 20 Oct 2010 16:04:19 GMT) it happened Hammy wrote in :

Has anybody mentioned switchers in this thread? Seems so much 'cooler'.

Jan Panteltje wrote in news:i9n445$khl$ snipped-for-privacy@news.albasani.net:

He wants 0-30 out I've seen variable output flybacks in that power range but they cant vary that widely.

This one here is 18-24V @85W out.

You could use an SMPS on the front end to step down the line voltge. That way you can save the size and expense of a 50/60Hz transformer. Then just use an op-amp based regulator driving a fet pass device on a brick heatsink for your variable output.

Given the size of some of the higher power 300W plus benchtop power supplies today they have to be SMPS's at least the front end.