Visualize current

I'm not aware of a physical material that does this, but it is standard fare for field solvers -- most all of them will create pretty, color-coded 3D displays of current densities and animate them for you, if desired.

Isn't it what a LED does? react to current and emit light proportionally to the current that goes through it? Granted, it might not be very linear, but it does give a good indication...

I did an experiment once with an LED. People were asking, "So, what happens to the light as the current decreases?"

So, I took an LED, a BMF cap, and a wall-wart, and a resistor that gave me 20 mA steady-state. I disconnected power, and let the cap discharge through the LED. (just a plain vanilla RS-type red LED). From 20 mA down to 10 mA, there was very little observable difference in the brightness, but from 10 mA on down, the light level was virtually linear with the current, albeit the log response of my eyes probably introduced some kind of artifact that made it look "linear."

All the way down to zero, with the room lights off, the shades drawn, etc.

So, from 0-10 mA, I'd say even a jellybean LED is linear enough to be useful.

Cheers! Rich

I have a Tadiran 3v, 3 a-h battery, a 1M resistor, and a good green LED soldered together. It should still be visibly glowing for 20 years or so. This LED is visible in office light at around 100 nA. I've been meaning to try it in the dark, adapted. That might work at a few nA.

John

Visually, maybe. LEDs have horrible nonlinearities at low current. That's why optocouplers are so nonlinear.

Cheers

Phil Hobbs

On a sunny day (Fri, 24 Jul 2009 17:33:54 -0400) it happened Phil Hobbs wrote in :

Depends on how low I guess. Optos are great, and quite linear. LEDs are actually quite linear at normal currents. I have used LEDs optocouplers for audio driven by an opamp current source, as line isolation in phone line interfaces. You could even buy the thing in one chip: MOC5010 Good enough for audio.

You may have a little difficulty applying this to a PCB but it's neat non-the-less

just below the "links" section.

H.

With audio, you're using Class A bias of your LED, so you don't see the problem. Try making a current transfer ratio plot of your average optocoupler down to microamp input currents--there's a very pronounced toe.

Cheers

Phil Hobbs

However, you can get optocouplers with two phototransistors in there. Drive it with an opamp, place one of them in the FB path, and voila. Really nifty devices. I remember almost jumping for joy when I first found out about those. They aren't even very expensive.

No offense, but I can refute this. I set up an experiment on the lab bench, with a 4Nsomethingorother, to see how linear its transfer ratio was, and almost fell out of my chair when my graph was practically indistinguishable from a straight line.

Once again, 0-10 mA was virtually linear. My intent was to use it for isolated feedback in a PS we were trying to design, but after spending about 60 grand, the boss dropped the project.

Thanks, Rich

OK, you got me here. In my experiment upthread, I never tested the setup to currents that low.

Thanks, Rich

$60k on a SMPS design? Sixty?

I suspect that you failed to look down at low currents. Unless you took, say, 10000 data points. If you really have an opto that's linear to say, 1 part in 10**3 down to the low microamps, I'm not offended, I'm delighted. I want some!

Cheers

Phil Hobbs

Well, I wasted about three months, on engineer pay, plus everybody else's time I wasted, and materials and crap.

Maybe it was only about thirty, but the boss decided it was too much, and I didn't really have the proper training anyway.

Or are you saying that sixty is cheap? ;-)

Thanks, Rich

Thirty is still a lot for one switcher.

No, it's not cheap. For this much money my clients get a pretty complex design done at my office ;-)

But I do make sure that things that others can do for less are done by others. For example, layouts are either done at the client or we use a layouter from my network. I do the nasty areas and then he lays out the whole board. Same with prototype stuffing, there we hire technicians to do that. Of course, this does not include EMC lab time and other agency approvals.

Yikes! I guess not. They'll quickly eat that much and do crap work for it.

What? Layouters? If you guys need a really good contract layouter drop me a line.

No, "EMC labs and other agency approvals".

Yep, they sure can slurp up a chunk of the budget. Most of the time that's because people call in consultants after they got a bloody nose at the lab numerous times, instead of before. Earlier this year I had such a project, probably half a dozen EMC lab sessions, failed every time. Then I redesigned it, sailed straight through. Margins to the point where the guys at the EMC lab thought their analyzer input stage had broken. This isn't meant as bragging, it's just that everyone has their own specialty and it does not make sense to try to go it alone. Just like I do not do my own tax returns even though I probably could. Instead I let my CPA take care of that because he specializes in tax law.