Linear 5V reg with adj.current limit

Oooops 12V fixed.

Most of the ones that use external pass transistors also have an external resistor to set the current limit, but it's not very accurately controlled.

Eg. MIC5156YM

Current limit 35 +/- 7 mV, so if you used a 0.35 ohm resistor, it would limit between 80mA and 120mA at Tj = 25°C, and would change quite a bit with junction temperature as well.

Best regards, Spehro Pefhany

An LM723 with a pass transistor. I've found it in lots of benchtop supplies with adjustable voltage and current limits. The datasheet has examples for a starting point.

But the current limit voltage is the BE turn-on voltage of a transistor, and so varies dramatically over temperature (this is what Spehro was alluding to).

I'm not sure if the best way to do this isn't to build it up from op-amps and a pass transistor, at least if you want a precise and stable current limit as well as a precise and stable voltage set point.

The world really needs programmable current limiters and current-limit-programmable voltage regulators. It's amazing how few there are. The current limiters are mostly fixed-function stuff, like for USB or PoE.

You could make it yourself, but it would take a bunch of parts.

If there's a negative supply handy, an LM317 is a good fit--use an OC comparator like an LM393 or 319 to pull the adjust pin down to -1.2V. (To the OP: Watch that the output cap doesn't blow up the regulator when you do this.)

Cheers

Phil Hobbs

Here might be a starting place, with adjustable voltage output voltage _and_ current limit....

-- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

Bingo!

That's the way it's done. Lots of linear "bricks" still use that old

I designed a power supply recently, and was surprised that nothing better than a 723 was available.

Yes, it's like the Willys Jeep and the Land Rover. Modern linear regulators are often ill-conceived in that they do not allow access to internal stuff like switcher chips do.

You could roll your own using a Cypress PSoC. They have analog building blocks in there and AFAIK are the only affordable analog programmable device out there. It's not against the law to abandon the digital stuff in them but they do cost around a buck or more, which is a problem for many of my cases. The performance of the analog parts is not much to write home about but one can play tricks like automatic offset compensation because the uC in there is now essentially free. Once you've got a design going you could file it away as in-house IP and use it over and over again.

The problem with "modern" parts like that is (1) all the work you have to do to program them and (2) they only work at low voltages and (3) they will be obsolete before the 723 is!

Which is why I like the 723 :-)

Plus nothing can beat the 17c price tag:

But if I were tasked with designing a linear supply that had accurate current limiting over any real temperature range, I would not use the

Putting that aside (and it is a quibble, unless you have some need of accurate current limiting), it's still a nice chip.

Yeah, it sure ain't a precision circuit and compensating for the Vbe tempco would be a bear. I wish they had put a 2nd error amp in there like the TL494 has.

Coaxing a TL494 into linear mode mode via a PWM trick would be cool. Maybe run one output into an RC lowpass and drive the gate of a power MOSFET with that. But hand out sick bags before the design review ...

It's pretty much the only one out there, at least in the sub-20c class.

you can use the diode trick to lift the common :)

Jamie

Or just put two in series with the output, inside the feedback loop. But in an off-line supply, it's usually easy to generate a low-current negative rail, or else use an LM395 instead of the 317. (Same die, different metallization.)

Cheers

Phil Hobbs

You need to go below ground to stop it, _not_ above.

That might just work... from input, LM317 connected as current source, value equal to current limit, followed by LM317 configured for 5V.

Yep, My analysis for Integrated Circuit Engineering, 42 years ago...