const-current thing

Fig 2 is pretty gross. Lousy headroom design. Sheeesh! ...Jim Thompson

-- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

Current-regulator diodes aren't new. Siliconix was making 'em a decade or more ago (just gate-source shorted JFETs).

J500 series and CR200 etc. were part numbers.

A grounded-base transistor with emitter resistor to negative supply is cheaper and can do a wider range of currents than a fixed diode.

On a sunny day (Tue, 25 May 2010 10:28:42 -0700) it happened John Larkin wrote in :

These dat 20 mA is a bit high for LEDs used as indicator. I have some green ones now at 3.3V with 33k (!) in series, the LEDs drop 2.38 and 2.44 V, so that makes .92 and .86 V over the resistors, giving a current of about 28 uA, good visibiliy in bright room light, JUST not blinding when you look directly into it. ftp://panteltje.com/pub/low_current_LEDs_img_1964.jpg The blue one is on 10% PWM with 180 Ohm in series on 3.3 V. That current source product may not sell. Resistors are cheaper.

They appear to be aiming this at automotive applications where you have a big difference in voltages depending on whether the engine is operating or not-- especially important when you subtract a series string of LED Vfs from it. Limited application area, but huge potential in terms of picoacres* of silicon sold.

• 5mils x 5mils ~= 4 picoacres.

On a sunny day (Tue, 25 May 2010 14:17:55 -0400) it happened Spehro Pefhany wrote in :

If I was to design foe in a car, I would have the electronics on a regulated supply. Output from a micro driving a LED would NO WAY go to an unregulated source.

Interior lighting would be adjustable, color and brightness, so same story.

For the headlights 10 mA is not enough. Small silicon farmland if you ask me :-)

especially important when you subtract a series

According to the datasheet it's been around at least since Nov 2009 so it must be selling. Vishay makes the SST series.

I have a night light that's a Tadiran lithium cell with an Avago green LED and a 1.5M resistor. It should last at least 20 years.

If you need more light, you can hold it in your mouth and bridge the resistor with your tongue. Hands-free adjustable dimming!

John

On a sunny day (Tue, 25 May 2010 12:13:05 -0700) it happened John Larkin wrote in :

In yoiur mouth? How about this tooth inspection LED? ftp://panteltje.com/pub/tooth_inspection_LED_img_1970.jpg hehe

Figure 5 is what has me worried. A 15% current decrease occurs over seconds.

Great idea, no cigar for the implementation. Still, it's a lot cheaper than the LTC thing.

Cheers

Phil Hobbs

If you only need 10 ma and can tolerate using a resistor the LM334 is much better - lower voltage overhead, much higher impedance and most other specs better too. (That's why I suggested it for the low current, low noise , low drop-out project)

I didn't get to that ;-)

That's down right shameful. So much for pulsed operation.

I do stable CMOS mirrors all the time... 3% is trivial.

...Jim Thompson

It would be good for gross stuff, like pushing current into a zener to power some opamps floating on a power rail. It could also be a protective current limiter.

Laser-trimmed precise 2-terminal current limiters would be great, in the 1 mA ballpark. Or 3-terminal, programmable with one resistor.

John

That's a classic jfet thing. Kind of sloppy and expensive.

John

What's that negative resistance all about?

Agreed. OTOH you can do a lot with an LM317L and a resistor.

Cheers

Phil Hobbs

Those are fairly horrible below 10 volts. You're way better off with an LM317L or LM334.

Cheers

Phil Hobbs

Noise?

Digikey would not have over 37,000 (as in thousand) of these in stock if they didn't sell.