S.E.D is disappearing

Ah, at least something I can make sense of. I'm half (or 90%) talking out of my ass 'cause I've never designed a switch mode PS. How about a buck converter, where you use the LED's in place of the return current diode... I know there are going to be head room issues. (LED's in parallel is probably a bad idea.)

George H.

Huh, that's sounds something like my silly idea... (small minds think alike.. no insult intended. :^)

George H.

Any way the LEDs can be re-wired as two 3V loads in parallel?

Not necessarily silly. See "Self-Inverting-CD-Ignition.pdf" on the S.E.D/Schematics Page of my website. I first ran this for automotive ignition systems in 1968. ...Jim Thompson

Exactly. Now, how do you do a "low power boost converter" that will always be low current (and guaranteed to start up)? Remember, the input current can never be above 250mA.

I don't think that much capacitance is needed (@12V) but may have muffed the arithmetic.

BTW, how do aluminums take high dI/DT? At 100Hz, heating isn't likely to be a problem but are there other failure mechanisms?

That's a really good start but there is nothing limiting the input current and a lot of the stored energy is being wasted in the linear current sink (I'd have to do the calculations to see if you haven't blown the input current budget). Replacing this linear current sink with a buck, was my plan.

Cool! But a 200uH 10A (Isat) inductor with a low enough DCR may be a little large.

Yeah, there is nothing to this circuit design stuff. Everything you ever need is in one place, online!

Sure, as long as the current through them is reasonably well matched.

Yes.

No one needs engineers, when they have DigiKey, right?

It can be done all linear, but it's very tight and needs big caps. Sounds like it would be better to change the specs.

Same as no one needs architects, when they have bricks.

Den mandag den 11. september 2017 kl. 16.45.13 UTC+2 skrev snipped-for-privacy@notreal.com:

sepic?

560uF @ 12V ~= 40mJ

330uF @ 12V ~= 24mJ

I think it is in the ball park

Well that's when I would have to start paging through the 10,000 boost converter / LED driver chips to see if there is one that can be told to have a suitable input current limit or a soft start pin that can be fiddled with...

I think it's about right. 10A for 200us is 2mC. Q=CV, so DeltaV = Q/C =

I don't know anything about them really, just that they are supposed to be low impedance and that I have never seemed to actually need one.

The *rms* current is well within the rating of course. Why do you think high dI/dT could be bad specifically? It's not so enormous considering the AlPolys are meant for high frequency SMPS use AIUI. The mosfet drive can be adjusted for slower switching if it's a problem.

Yes basically what jason proposed I think.

The linear drive gets more "efficient" if the 12V is reduced to approach the 6V LED terminal voltage. So you could use higher capacitance and reduce the boost voltage.

(I was wondering about some kind of servoed system along the lines of the above, it depends on the LED current curves and how much the output is allowed to vary during the pulse.)

Thanks,

Remember, a boost converter has a direct path from input to output, so soft-start isn't possible. Lasse's idea of a Sepic is good (I'm not a PS expert and custom magnetics are frowned upon but all is possible).

I probably muffed it. I'd come up with ~100uF.

It's just an unusual application for AL's. AlPos would probably be a better choice.

Yes.

I picked the project (the illuminator is a very small piece of the project) up from an engineer who recently left the company. His opinion was that the output had to be quite uniform. I don't see why but I may not have tripped across the reason yet. Basically, this is the light source for an IR camera. The LEDs are on during the integration phase of the sensor (no reason to illuminate as the sensor is being scanned). It's a global shutter device so it's any change in illumination shouldn't show up in the image. Wild variations may change efficiency, though.

Den mandag den 11. september 2017 kl. 19.32.53 UTC+2 skrev snipped-for-privacy@notreal.com:

sepic shouldn't need anything special, just two inductors or one of the numerous duals

The issue is not the circuit problem, it's the absence of adequate info & other senseless behaviour problems by keith.

Maybe you need new glasses.

NT

It would eliminate the need for 2x smpsus.

NT

Now it makes sense. He doesn't know how to current limit an smpsu or what issues lytics have with high di/dt, and felt unable to say so.

Next time don't use that as a reason to pick fights with people & insult them. It makes you look like a prat.

NT