1. Home
  2. Electronics Design
  3. LED Constant Current with JFET

LED Constant Current with JFET

I am using a JFET with a PNP transistor to regulate the current into an LED. I picked this topology because the LEDs include green and blue with a Vf of over 3 volts and my power source is only 5 volts. I did a simulation with the current sense resistor across the gate/source of a BF861A and the circuit seems to work well.

The LEDs seem to have a pretty strong dependance between luminance and temperature and this unit will be used outdoors. Over a 50 degree temp range, the illumination of an LED changes some 40%. So I'm thinking how to incorporate some temperature adjustment too.

The threshold voltage of the JFET increases with higher temps. Since this sets the voltage on the sense resistor, it would seem to help by increasing the current at higher temps. But when I read up on the temperature characteristics of FETs it seems there are multiple effects that counteract each other. They even mention a bias point where the effects cancel. But this is not specified for the BF861A I am using, so I'm not sure how to figure this out.

--

Rick
Reply to
rickman
Loading thread data ...

A single BJT with no feedback drifts by +9%/K, whereas you want about +1%. If you apply about 8kT/e ~= 400 mV worth of emitter degeneration, you'll be in the right ballpark. The curve shape probably won't be quite right ov er that wide a range, but it should be an improvement.

Cheers

Phil Hobbs

Reply to
Phil Hobbs

I keep posting before having my coffee. 200 mV.

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
Reply to
Phil Hobbs

Take a look at some of the temp compensated LED current driver chips. Not exactly your problem, but I've been using a NSI50010YT1G for driving a string. Went from a roll my own to a SOD-123. About $0.10 each. There are adj current versions out there in a SOT-123 (I think it is) package.

Could save you board space and parts.

--
Chisolm 
Republic of Texas
Reply to
Joe Chisolm

Do you want constant light output, or constant current input, for the LED?

If you have 5V, regulated, a simple current mirror (two transistors, emitter resistors, base1==base2==collector1 connected to a program resistor to +5V, might be good. Replace the program resistor with a FET current regulator, and it's better.

An op amp and pass transistor, with a current sense resistor, does even better. Best, though, is a photosensor-based feedback system (but stray-light shielding will be important, it's not all electrical design any more).

Reply to
whit3rd

The whole photo-sensor thing is a bit of overkill I think. This is just panel and control illumination. I am thinking that a 40% variation over temperature might be a bit much. So I'd like to get that down a bit without resorting to an overly complex design. Also, there is no way to exclude outside light. In fact, if a photo-sensor is used, it would be to compensate for the outside light, more outside light = brighter LEDs.

I found a number of interesting current drivers at Motorola... opps, ON semi. None are exactly what I need and some have a fairly difficult to use interface requiring timing control.

I'm going to look at using a thermistor to trim out the temperature effects. I just thought the natural behavior of the JFET might be a rough approximation to what I need.

--

Rick
Reply to
rickman

Why not exploit a bandgap reference and use a TL431 based current source? They are so friggin cheap - online at Tayda the TO-92 is 11 cents in singles.

--


----Android NewsGroup Reader---- 
http://usenet.sinaapp.com/
Reply to
bitrex

A BJT is better--it's easily trimmable and very predictable. See e.g.

formatting link
.

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
Reply to
Phil Hobbs

Well, let's see... He has 5V and the LED is at least 3V so that leaves

2V for headroom. I think the 431 reference is 2.5V. How will that work?
Reply to
John S

PWM the shared LED supply, feedforward from a temp sensor. Or PWM the drives from software. Then each LED just needs a resistor.

But 40% is only +-20%. Nobody would notice that. So don't do anything.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

Will they work with less than 1.5 volts of overhead? I think not. Just the sense resistor will have 2.5 volts on it, no?

--

Rick
Reply to
rickman

there is a tlv431 - a 1.25v reference, but not for 11 cents

Reply to
David Eather

I think it can be done, but might require a current mirror and a few other parts you might not want to deal with.

--


----Android NewsGroup Reader---- 
http://usenet.sinaapp.com/
Reply to
bitrex

TLV431 is the 1.25V reference version.

Reply to
whit3rd

Hmm..for about 400mV, use a flower power DCT (translation, Germanium transistor with base tied to collector).

Reply to
Robert Baer

Wouldn't a resistor work well? It will deliver slightly more current when the LED is hot; maybe enough to be visually acceptable.

The only catch is that a specific resistor is needed for each LED part number. The Vf for blue/violet/white LEDs is all over the place.

--
I will not see posts from astraweb, theremailer, dizum, or google 
because they host Usenet flooders.
Reply to
Kevin McMurtrie

You want 200 mV with zero TC, though, so that the PNP's TC_Vbe dominates.

My usual uses for LEDs with stable outputs is in lower-tech sensors, e.g. a pulse-ox or something like that. Lowish source noise is a win there. If this is an array of display LEDs, I'm inclined to agree with JL--use a PWM or just fuhgeddaboudit.

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
Reply to
Phil Hobbs

I found a solution that is too large to write in the margin of this page.

--

Rick
Reply to
rickman

You just need to hit the "refermat" button. ;)

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
Reply to
Phil Hobbs

Lol! That was a good one!

I found the LM234, a rather old temperature sensor that can be configured as a constant current regulator all in one. A PNP is used to boost the drive current and an N channel FET for a shut down. A PWM might be lower cost, but it will require a temperature sensor and programming. I'd rather just leave out that additional complication for this product and let the user focus on using the display.

When the product reaches 100k production I'll do a respin. lol

--

Rick
Reply to
rickman

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.