linear reg with current limit

The AL5809 does thermal limit. So would a downstream regulator.

Do you mean something like this?

Yes, though I forgot you wanted the voltage controlled by a DAC. Possibly you might need to prevent current flowing into the ADJ pin of the second 317 during current limiting.

On a sunny day (Tue, 17 Oct 2017 20:39:57 +0100) it happened piglet wrote in :

I think for me it is the lead that does it. I still use 60/40.

On a sunny day (Tue, 17 Oct 2017 10:04:12 -0700) it happened Jim Thompson wrote in :

That is nice.

Also you might have to do something to ensure the minimum load current (5mA iirc) is drawn from the output, so that the voltage does not rise above what it should. A resistor across the output would do, but would cause a small interaction between the voltage setting and the current limit. A 5mA current sink at the output would be better. If your load is guaranteed to be more than minimum then don't bother.

You really should read the first post in this thread. GH

I did, but I have a short memory for some things.

I don't think the programmable aspect makes much difference except that it might be necessary to consider whether the opamp can source current into the ADJ pin during current limiting, and whether that is a problem. There are a few ways to solve it, if it is a problem at all.

Oh and you should have told me to follow the links too!

Heh.

I experimented with using an LM1117 as an amplifier, specifically how much one can abuse the ADJ pin under unusual circumstances. The answer is that you can't reasonably damage it, and the ADJ is a pretty benign load even in weird quadrants. If I use an LM317L, I'll test it to, but I suspect it's similar.

So I'm semi interested in seeing how the lm317 would work as a 'robust' pass element. (I could even get some (voltage) gain out of it.) I drew it up with the opamp driving the adjust pin and feedback from the 317 output. Would it be better to just do it open loop? And I wonder how much C I need on the output. (I'm waiting on a DK order, maybe I'll play now. :^)

George H.

Depends on the vendor, output current, and input cap.

Cheers

A 317 or 1117 is a great cheap self-protecting transistor or power amplifier. The 1.25 volt offset is accurate, so you can run them open-loop in a programmable voltage regulator application, if you like. Multiple regs can then follow a common programming rail.

Let us know if you find anything interesting about the 317.

I recently played with a few different self-protecting mosfets, and wasn't impressed.

On Wednesday, October 18, 2017 at 11:52:22 AM UTC-4, snipped-for-privacy@gmail.com wrot e:

Is that the voice of experience? Well my DK order came in so I've got to stop playing. These 317's have a national logo. I never thought about more C on the input, but indeed with 10 uF Cin I need no cap on the output! With Cin=0.1 I needed 1 uf, and worked OK to ~1kHz.

With Cin=10 uf and no Cout it worked to 100kHz! (10 and 100 ohm load though I ran out of power supply current with 10 ohms.)

It wasn't happy at all near zero volts, (~100's of mV above ground) and needed some compensation on the opamp.

Not being able to work at ground limits the usefulness to me.

George H. (Oh all built on white proto, so there's that.)

No. Think in terms of a conventional LM317 bias setup.

R1 wouldn't be there, R2 would, and there'd be a third resistor from ADJ to GND.

The way the LM317 works: If voltage across output exceeds 1.25V, output voltage _lowers_.

If voltage across R2 is less than 1.25V, output voltage rises.

Adding R1 (10 Ohms) only means that _sum_ of voltage drops across R1 and R2 must be 1.25V, _in_regulation_.

When drop across R1 reaches 1.25V (current limit), feedback collapses, effecting the current limit.

The LM395 is just a mask change version of the LM317, see...

ICE-LM195-LM117.pdf

on the S.E.D/Schematics Page of my website.

I did that analysis for Integrated Circuit Engineering (ICE) more than

...Jim Thompson

AAAARGH! Make that voltage from output to ADJ.

...Jim Thompson

Den tirsdag den 17. oktober 2017 kl. 02.56.23 UTC+2 skrev John Larkin:

LP2951 ?

drive the end of the divider to vary the output voltage

I had a hunch about that... Before the LM723 was a jellybean, it was uA723; those are still produced, in surface-mount, and not obsolete. But, you gotta ask DigiKey about 'UA723'.

Alas, it looks like, even with PNP output transistor, the common mode of the '723 error amp inputs limits you to 2V for direct inputs. In order to get usefully down to 0-6V, there's the output zener to bias the driver high (so that the current limit transistor can steal its base current). Then, if your ADC is in the 0-6V range, it'll take a resistor network biased from the VREF to pull up the ADC and sense voltages into the CM range.

These are changes to that 'figure 20' application circuit.

A resistor to current-limit the power pin of a rail/rail op amp is looking better all the time. Linear regulator building blocks are kinda... flawed.