LT3081 Parallel Different Sources

[jmoschetti45]

I have a few LT3081's laying around. I know you can parallel them for more current handling. I was wondering if I could use them in parallel for redundancy instead. Basically, 3 power supplies into 1 combined output. It doesn't matter if one or two are doing more work than the other(s), this would be purely for redundancy. I know I can just design a redundant supply, but this seems like a quick and dirty easy solution. Thoughts?

[jbb]

I’d worry about a device failing as a short circuit to ground...

[digsys]

Definitely. One of my past favourite regulators, may still have several around myself. If you check their app notes, they show you ONE method of paralleling them, pretty easy to figure out something that suits you, as it has nice control pins. Maybe use Shottky ORing diodes / "ideal diode" switches (for near zero loss if needed) etc on the outputs.
Then, depending on how fancy you want to go, maybe simple individual Isense / Vsense / switch etc etc
I've made many dozens of of production linear P/Supplies .. it's fun to play with. If you have any specific questions, feel free to ask. We especially need to know your min / max requirements / specs you're trying to achieve.

[jmoschetti45]

3.3V @ 400mA. Pretty low power application.

Maybe use Shottky ORing diodes / "ideal diode" switches (for near zero loss if needed) etc on the outputs.
Then, depending on how fancy you want to go, maybe simple individual Isense / Vsense / switch etc etc

I thought of that after I had posted. Wouldn't be hard to compensate for voltage drop with even a plain schottky diode.

[digsys]

....  Wouldn't be hard to compensate for voltage drop with even a plain schottky diode

One "cheap" trick I've used for many years (dozens products, 100s units) is - I OR the outputs with appropriate Polyfuses, then shrink-tube them together.
With outputs separate, adjust to match as close as possible, with say 75% load, then link the outputs. As ONE regulator starts to deliver a bit more current, the Poly heats up and reduces its contribution. This "trick" is used in many systems out there.
IF you have a very rapid high current switching load, you'd need to whip up a test load though. What I did was set up a series of switches with different loads and rapidly switch them In/Out and monitor on a DSO. This is really only necessary for very dynamic loads. I've tied up to 4x regs together this way.

[ogden]

Parallel config: datasheet Figure 9. You will have "redundancy" only if failure mode is "open". When one regulator fails short, output voltage will become equal to input.

I’d worry about a device failing as a short circuit to ground...

Nonissue. When you actually look at typical circuit (page #1 of the datasheet), you will notice that chip have exactly zero number of pins connected to ground.

[digsys]

... You will have "redundancy" only if failure mode is "open". When one regulator fails short, output voltage will become equal to input. .....

All easy fixable issues

[ogden]

... You will have "redundancy" only if failure mode is "open". When one regulator fails short, output voltage will become equal to input. .....

All easy fixable issues

How exactly?

[Siwastaja]

If by redundancy you mean combining multiple different, iffy input power sources (think about power harvesting, for example) to make one more reliable output, it could theoretically make sense. (Although I would just diode-or the inputs together, then use one regulator.)

If by redundancy you mean protecting against regulator failures, it won't work that way. The regulators do not just fail for no reason. They only fail if there is a catastrophic design failure somewhere, in which case it's very likely all three will fail in no time.

OTOH, if you really need super high reliability (think about life support systems), it's going to be much more difficult than just simply paralleling some regulators.

[jmoschetti45]

... You will have "redundancy" only if failure mode is "open". When one regulator fails short, output voltage will become equal to input. .....

All easy fixable issues

How exactly?

Polyfuse and some sort of clamp?

[digsys]

... Polyfuse and some sort of clamp?

Only limited by imagination and how complex / functional you want to achieve. Your selection is a definite start - and would look like this -

- In normal OR operation - the Polys provide adequate balancing (higher current > warmer poly > Vdrop > channel settles down)
- Regulator (channel) fails short - OR gate applied
- Regulator (channel) fails open - Zener / latch etc operates, Poly heats rapidly and latches. At this point you can get as fancy as you like, and it depends on WHAT failure mode you're expecting of the regulator. ie provide an over-ride Vset back to it? Kill it completely? Attempt 1-2 restarts? etc etc
The circuit I posted does have limitations, it's a start reference. If your "load" can tolerate a ~1V short term fluctuation, then you have heaps of simple options.