Hi,

i already found a few circuits but i really need it to to be small.

Similar to this

Any suggestions?

Thanks!

 

hi K,
How many LED's in series and what are their operating voltages?
E

 

It's not for leds but for electrolysis.

 

This works with 5VDC in, and outputs 16mA or so. 2 BJTs, 2 resistors. It doesn't matter how much current the LEDs want to draw, all they get to fight over is 16mA.

 

How accurate does the 15mA have to be?

 

3 parts. This is a simple as it can be. Output current changes with temperature, expect only 10% accuracy.
R1 = 39 ohms
R2 = Vee x 1000
T = 2N4401, 3904, 2222.

ak

 

If the supply varies over a 5V to 30V range, here's a variant which keeps the current within a 13.8-15.8 mA range over a 10C to 30C temperature range:

Q2 will dissipate ~400mW when the supply is 40V, so will probably need a small heat-sink.

 

i'd say between 14-15ma (as accurate as possible)

I can also test under 30v (5v-20v), in term of space the heatsink is an issue.

Thanks for the suggestions so far.

 

See

 

What is the maximum voltage drop across the electrolysis load?

To minimize the ≈-0.3%/°C temperature drift due to Vbe variation, and also current change with voltage, you can use a TLV431 programmable reference in place of the current-sense transistor.
It has a much lower temperature coefficient and higher voltage gain than the transistor to significantly reduce the current variation from both temperature and supply voltage change.
It's main disadvantage is that it drops 1.25V across the current sense resistor R1 whereas the transistor drop is only about 0.7V, thus increasing the minimum operating voltage by about 0.55V.

LTspice simulation below:

 

See

 

So you gain a little extra stability by using J1 as a constant-current source in place of a resistor and adding R2?

 

https://www.diodes.com/assets/Datasheets/AL5809.pdf

15 ma, brutally simple.

 

So you gain a little extra stability by using J1 as a constant-current source in place of a resistor and adding R2?

I get stability more than 10 times better than in your simple scheme. Although I expected better stability. Perhaps this is due to different models of reference voltage sources

 

I get stability more than 10 times better than in your simple scheme. Although I expected better stability. Perhaps this is due to different models of reference voltage sources

So the question is, how good is good enough for the TS.

 

Bordodynov this seems accurate enough for what i need!

Sensacell i'll definitely look for one, thanks for bringing this.

Thanks again to everybody for their suggestions!

 

Linear technology LT3092

http://www.linear.com/product/LT3092

 

3 parts. This is a simple as it can be. Output current changes with temperature, expect only 10% accuracy.
R1 = 39 ohms
R2 = Vee x 1000
T = 2N4401, 3904, 2222.

ak

show off

 

The LT part is sort of 3 parts too. Bypass cap(s) might add more. Real estate might still be about the same.

 

When I was in school, Tektronix was on campus recruiting, and came with a short quiz with some real-world problems. One was how to implement a 30 uA current source in a scope (early 1970's, solid state chassis with a CRT). The answer they were looking for was a single 10 M resistor to the regulated +300 V plate supply.

ak