Pete Lefferts LED current source

Never heard of that before ;-)

...Jim Thompson

LEDs were used in some audio power amps for biasing if you want a commercial application.

An interesting side note, when the high efficiency LEDs hit the market, some chips had difficultly driving them due to the higher "on- voltage" (forward bias). That is, you drove them with some chips, and the light was dimmer due to the current source running out of compliance. When I did the MAX7219, I put in more headroom to handle these higher voltage devices.

I had the opposite problem. We use blue led's on module front panels as VMEbus activity indicators. The early Cree SiC parts neeeded 50 mA, so I drove them with a couple of 74F38's, 5 volts, 27 ohms. As the blue led's kept getting better, my customers started complaining about being blinded by them. We're down to about 4 mA now.

Which brings up another issue... for later.

Gotta go.

John

There is some misunderstanding here. I wasn't asking about an LED

+Vs >------+--------+ | | D1 --- .-. \\ /~~> | | R1 ----- | | 1K | '-' | | | | | |

That may be the first _published_ use, but I thought up the notion of using an LED to bias a transistor (not for a current source, but still...) on my own. I'm sure that others have, too, and very possibly before 1975 (but certainly not before the invention of the LED!!!).

Yes, that is exactly what Lefferts said in the reference cited. So the xstr Vbe then gets subtracted from the LED's Vf of 1.6-1.8V, leaving a low TC drop of ~1V across the emitter R.

Walt Jung

That may depend on the particular LED used -- I'd check if I were going to use anything other than 33 year old red LEDs.

Now wonder how I do that in an IC ?:-)

...Jim Thompson

Of *course* it depends upon the particular LED, and the xstr, and the relative currents, etc., etc. All of this makes the setup less than high precision. But, nevertheless, it is one that is high on utility, and low on cost. Pennies in fact.

Tim, that's fine if you did something similar prior to Lefferts, but it wouldn't help anyone else then or now, would it? That's what publication is for, the sharing of ideas. All I was trying to do with this query was to establish whether or not anyone had published the concept before him. It was/is a clever trick!

Walt Jung

Oh, I didn't do it _prior_ to Lefferts. In 1975 I was still in grade school. I could put a transistor circuit together from a plan, but I couldn't design one myself.

I was just picking nits with your wording: You said "did anyone _do_", making the academic assumption that if it hasn't been published it doesn't exist. I was making the military assumption that if it's sticking out of my chest, it exists and is important regardless of whether there's a paper on it.

Why, clearly you demand that the customer use a process that'll let you build an LED into your otherwise all-silicon device.

What could be easier?

On Tue, 21 Oct 2008 16:32:42 +0000, Walt Jung wrote: ...

Hmmm... I don't see where I said "did anyone _do_".

As for the relative value of publication, that should be obvious to anyone who doesn't want to re-invent the wheel. And, an existing article on how to do some circuit trick is a clear path of help. There might be

Walt Jung

People were using anything they could find to function as simple lower voltage references. Leffert's recorded observations re temp compensation is what is actually important about this publication.

Obviously others will have used the component for similar purposes previously, simply because it seemed to work in simpler biasing applications, but considerations for cost, batch/long-term stability, noise and mtbf of optical components in general will have restricted serious use then, as they continue to do, today.

RL

"I am trying to establish who and when first used".

"used". Not "published".

I understand the value of publication, but it is different from use.

After all, folks were using stone tool _long_ before any articles on how to make them came out.

Yes. Exactly my feelings as well. Personally the original post read as if Lefferts is due to be enthroned/celebrated/placed in a pantheon of fame, as some guru circuit hero. Any possible pretenders to the throne were being weeded out prior to the celebrations. The bit that galls me is the general assumption throughout the academic world that any idea (no matter how trivial) belongs solely to person who first -publishes- it. Hence those who were using the same idea years beforehand, or who assigned no great value to it, or had no means or channels of publication, are to remain in vile obscurity. On the other hand, the staff scientist who is paid to play with new devices and can write up their work in house literature and maybe even patent this idea using house lawyers, receives all future credit and will be accorded due deference and a session of forelock tugging whenever their name is quoted.

It doesn't depend much on anything except the temperature and doping level--the current dependence is really weak. (IIRC you can get up to about -3 mV/K at picoamp bias currents.)

I've used the LED/BJT follower trick often--I first saw it in probably

Forward-biased diodes have a noise temperature of 150K at room temperature, so if you drive it reasonably hard, you can make a really quiet voltage reference this way. (I'm sure there'll be some raised eyebrows in this group over a 'voltage reference' with 150 uV/K drift. Take it up with Maxwell's Demon.)

Cheers,

Phil Hobbs

You could do it in SiGe...but you'd have about a 400 mV voltage reference. 'Course you could put 3 in series.

Cheers,

Phil Hobbs