LED Constant Current with JFET

The ambient light could vary by 3-6 orders of magnitude. And the users eyes will adapt. So 20% is essentially invisible.

We commonly use four different LEDs on a front panel, red, orange, green and blue. We don't try particularly hard to set the currents, and often use one quad r-pack to set all four. Nobody notices.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

And if you have a spare op-amp in that chip and need some (-) voltage on the Vee side, you can buck boost the Vee leg via tranny and the op- amp with a simple relaxation oscillator and a zener diode for the buck boost and regulation.

Basically a self generated - rail for the chip :)

I've done it... so I could get a dual rail using a single

9Volt battery but with a common grd..

In some cases, if the circuit already has a reliable pulse operating, you can use that for the reference..

One could also use a blockig osc but then you need to do a tapped inductor or a transformer.

Talk about being cheap!

Jamie

I'm sure they don't.

--

Rick

It's been my experience that, in a given line, of colors red through blue, the greens perform terribly, easily 6dB less. Dunno what it is about GaP, but it sucks. You have to go out of your way to select the 'high efficiency' GaN-XYZ based ones, that are 3V but still green (I forget why, if it's phosphor doping or something). So your Vf's are 1.5-2V for red and yellow, and 3-3.6V for green and up, and all the brightnesses are about right.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

Phosphide LEDs are much better behaved in instruments, though--the shapes of their emission spectra are almost perfectly independent of drive current. That makes it easy to extend the dynamic range of simple instruments by adjusting the source brightness.

Nitride LEDs walk all over the place--badly enough that their spectral slopes pull the effective peak wavelengths of even narrow interference filters--so you have to do all the work on the detector side.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

We use a series of Osram surface-mount LEDs, straight and right angle, on our instrument panels.

I did this, based on eyeball judgement of brightness:

These LEDs are essentially ohmic above about 4 mA and have zero TCs at roughly 5 mA, but we usually run them at lower currents. As blinky test indicators on the surface of a PCB, we run them at a maybe a hundred uA so as not to blind people probing around.

People tend to find blue LEDs to be bright/annoying, so if the higher Vf reduces the current through a resistor, that might be good.

The luminous intensity specs on the Osrams cover about a 3:1 max/min range. Eyes are pretty forgiving, since everything looks fine on our panels.

LEDs keep getting better over time, which has gotten us some complaints about over-bright LEDs, especially blue, but that seems to have settled out in the last few years. We used to run SiC blues at 50 mA, and we run the InGaN parts at maybe 2.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

The green GaN LEDs are my favs. I use them for just about all indicators. I like the color and they're very efficient.

The only time I've used a current source to drive LEDs was green GaNs from a LiIon battery. The battery varied so much that I had to regulate it somehow and a bipolar/FET current source was the best and simplest. The GaN LEds say they're 3-3.3V, but they work fine well below that.

..or call it a marginal solution?

urrrrrghh.... ;)

Den onsdag den 17. juni 2015 kl. 17.04.30 UTC+2 skrev John Larkin:

these are cute:

thought 10mA is a bit much

-Lasse

The current regulation isn't really impressive - about 7 mA at 1 volt, eventually hitting 10 mA - but it's good enough for LED indicators. There are probably apps where this is better than a resistor.

The LND150 depletion fet at Idss is an option. They are usually pretty close to 1.6 mA. 1 volt is about the half-current point.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

YES! They are VERY good constant current devices, and can be used over a rather nice current range. There are others in that family useful for other current ranges.

I recently breadboarded up a circuit ("charge sensitive" integrator preamp for a photomultiplier tube) which uses a chain of LEDs and a current-setting resistor to create the bias/reference voltages for its transistors. I wanted to be able to run it on DC supplies available "in the field" (anywhere from 9 to 15) with predictable performance, so I replaced the current-setting resistor in the design with a 2N3819 (G-S shorted) that I selected out of my bag of spares to give me the desired amount of current in the bias chain.

Seems to work nicely... the circuit's draw from the DC supply is quite constant over the voltage range I tested.

The problem with jfets is the typically huge min/max Idss range, a problem for production. It's 2 to 20 mA for the 3819. The LND150s seem to be very repeatable. BF862 is pretty good.

As a power supply bleeder, I sometimes use two depl fets, one to ground and a smaller one through the "power" LED. Pull the plug, and the LED stays on bright for a while, then winks out.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

They are made for a different purpose than what I need. I wish to correct for the nearly 50% change in luminance over temperature. These devices are for preventing thermal runaway. The temperature coefficient is negated between the two apps.

--

Rick

Did you look at that .asc file I posted? It does what you want, using one P NP and one resistor per LED (plus two more to set the bias for all). Costs less than half a volt of headroom.

Cheers

Phil Hobbs

That's pretty simple. No reason not to use that. Digikey says the

2N5087 is an obsolete part. Any special reason for using it rather than a jelly bean jar part like the 2N3906 or something similar in surface mount?

Thanks for the circuit. I see now it is fighting fire with fire. I figured there was something very simple like this.

--

Rick

Nothing special except that it comes free with LTspice. Any jellybean will work fine.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

To get a 2N3904 to work with the temperature step I had to add the parameter temp={tm} as you had in yours. I thought you could just use the variable temp in the .step statement and it would work without the parameter on the Value2 line. Do I misunderstand that?

--

Rick