S.E.D is disappearing

why?

Think he can?

about

flesh

ch) at,

ugh

if you'd

that?

ot

ts

I really don't know. If I were trying to design this psu but struggling I'd post what I'd worked out so far, what I was stuck with and look for sugges tions. Keith is asking us to do the whole thing for him lock stock & barrel , and being obnoxious when his assumptions see an imagined fault in what is suggested. I really don't know what his level of skill is, but either way this approach is not likely to work for him.

So far we've not had basic info (unless I missed it somewhere) such as whet her the 3v per LED is at rated 20mA or at 10A and what parameters need to b e variable over what ranges. Nor whether this is a 1 off piece of testgear or for mass production, whether it's for a cut price consumer gadget or mil spec etc etc. I have no inclination to help him at this point, he's just t rolling really, whether intentionally or not.

NT

You have $60 worth of parts just in capacitors. As a general rule it's cheaper and better performance to buy a single part that handles the job *if* it's possible.

'd

Forget it. He didn't budget his voltage overhead at all, things like the po wer supply deviating by -5%, the VF of the LEDs possibly deviating a few + hundred mV, the switch overhead, the current sensing and last and, most imp ortant, the voltage droop across his storage caps. He just runs out of head room for reasonably sized components.

There were no cost nor size requirements. If you find a single storage capacitor that will do the job, please post the link.

There were no tolerances specified so I am working with the numbers presented. Please post your solution to all these problems.

OK, you're not up to it either. BTW, are you really proposing to pull

Geez, Blobby has the first clue. Still missing a lot but at least it's a clue.

Sorta. 7V is what's left after it gets to this circuit. It's powered by something like POE, but not classical center-tap transformer connected POE (e.g. 2-wire Ethernet or FPD-Link). That 7V @250ma (max) is the input to the circuit.

That's the first step.

That's the problem. The problem is as stated, though there are other complications, like cost and temperature that were unstated because I wasn't looking for a final solutions. People were bitching about no challenges, so I thought I'd throw one out there that I thought was an interesting problem that isn't nearly as simple as it looks at first blush.

No problem, everyone. Don't bitch about political posts. No one really wants to solve a real electronics problem, rather just preen, pretending to not want political arguments. We now know where everyone stands.

Get a 10A power supply.

Most of us are happy to tackle real electronics problems, but something remarkably detail-free posted by krw doesn't actually qualify.

Krw always "knows" exactly where everybody stands. He's only got one mental model of the world, and he trusts it absolutely.

And plug it into what? The 7V 250mA source? That's the problem.

Everything that you need to do it is right here, in plain sight:

It was a perfectly legit circuit problem. And you're the guy that starts all the OT trolling posts, not krw.

How about an arrangement along the lines of:

- low power boost converter to ~12V so we have some headroom

- alumimium poly electrolytic for energy storage. A750KS567M1CAAE013 as example. 560uF, 16V, 13mR, 4.8A Irms.

- for switching a common source Nfet with a sense resistor in the source

It's pretty simple really, up the voltage with a current-limited boost converter, store it in a low ESR capacitor then use a buck converter to power the leds, so long as both converters are better than 85% efficient it's a win.

also, unlike the 7V supercap it doesn't blow any fuses when you hook it up to the supply,

Yes, boost-to-HV-then-store-capacitively-at-HV looks like the best approach. Efficiency is important with only 1.75W available for 1.2W out.

As a zany alternative I (not wholly seriously) offer an inductive storage option. Buck convert to 0.2V then store for 10ms in 200uH and let the 200uH flyback into the LEDs. A bit like the early HP handheld calculator displays.

piglet