Current source / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Current source

Move Thread

LAN_403

Dr. Dark Current

Mon Oct 27 2008, 11:31AM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

I needed a current source and remembered something about JFET transistors. So I looked it up on Google and surely they can be used and the current can be varied with just 1 resistor.
That sounds cool, but the problem is that the current is fairly constant only down to ~2V, then it starts dropping off quickly.
What are other ways of a simple implementation of current source, which can produce relatively constant current down to at least 1V?

Dr. Slack

Mon Oct 27 2008, 01:20PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

An ordinary bipolar will get down to a lower voltage, approaching VCEsat. All bipolars will run out of steam in the last volt, but some go lower than others, so it's worth trying several different types, IIRC the bcx17 is quite good as a low voltage current source, and some of the very low VCEsat Zetec transistors.

There are several ways to do it, with various performance and complexity

1) emitter to supply, collector to load, base to a large resistor trickling in a small current. Simple and lowest voltage, but temperature and device dependant, as the output current is the highly varaible beta * base current. Not the pro's way of doing it!

2) base to 1v (or so), select a dropper resistor to develop 0.3v (or so) at your output current connected to the emitter. As it uses -ve feedback, it's temperature stable (relatively) and consistent from device to device, but you lose the extra voltage across the feedback resistor. Needless to say, more feedback volt drop makes a more stable current source,

With the right transsistor, either should be still working to some degree with only 200mV VCE

(Does anybody remember a very old D. Self audio amp design, where he used a red LED to bias the base, the aim being that the led tempco matched the transistor tempco, to get current independant of temperature, but that loses 1v to the feedback resistor)

3) you can drop less voltage in the feedback resisotr, say 100mV, and use an opamp to sense the voltage and control the pass transistor, to get very stable current and very low voltage operation, but now the parts count is increasing rather beyond your single resistor.

so you pays yer money ...

4) It's a bit cheating really, but if you have access to a super-voltage, then you can connect the base to the normal supply rail, shove current at its emitter from the super-voltage, and the collector will relay this current down to the load, and the compliance will go *beyond* the rail (VBE ~ 0.7v but VCEsat ~<0.5v). This is the common base or cascode connection. "Shoving current at its emitter" need take no more than one resistor to a constant voltage, as the emitter voltage sits pretty much constant at rail + 0.7v. You sometimes get a super-voltage for free, if regulating a higher voltage down to the 5v or whatever you need for the rest of your circuit. As the higher voltage is unregulated, you may need to use a current source to get a reasonable stability of current into the emitter, but at least you have a reasonable voltage drop to play with.

5) Staying within the rails, there is a nice little device that we use at work to bias up a current supplied RF amplifier from a low voltage rail, the BCV62A current mirror. We use it with 250mV drop on the feedback resistors, with 15mA control and 75mA output. The data sheet typical VCEsat varies from 75mV at 10mA output to 250mV at 100mA output.

Dr. Dark Current

Mon Oct 27 2008, 02:02PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Thanks for the reply.
This is for a LED flashlight so the current should not be dependent on the supply voltage, so BJT+ resistor is not really ideal.
Option 2) sounds interesting but it has the additional drop (which could or could not be problem).

What about a MOSFET current source? Is the constant current achievable down to low Vds?
(+Vs in my case is 7-10 volts, from a discharging battery)

Dr. Slack

Mon Oct 27 2008, 02:53PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

A MOSFET has the potential to get down to the square root of bugger-all VDSsat. However, a fixed bias on the gate will give a very unstable current with FET threshhold tempco, and the zener tempco itself will give unstable current via the high gm.

With 7v available, you could go for the opamp and a very small source resistor option. Put perhaps 100mV across R1 at your rated output current, or even 50mV, choose R2/3 with D2 voltage to set that, or make R2/3 a potenetiometer to set the output current. The 324 amplifier (LF358 IIRC for singles) senses at ground and will run down to less than 5v rail, though IIRC it cannot get the output all the way up to +ve rail which may be needed depending on the threshhold of the FET. If that's the case, then there are Rail2Rail amps available. D1 is the LED load

It may be worth checking for stability, I suspect the amp will dominate the dynamics.

Thanks, you've just helped me design my own LED flashlight.

Steve Conner

Mon Oct 27 2008, 03:15PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Many LED flashlights use a switched-mode regulator based on a buck converter with a current control feedback loop. That can inherently get down to almost zero voltage drop. I've seen people take a LM2576 switching regulator chip and bodge it to an op-amp current sensor to achieve the current control.

The LMC6481 is a rail-to-rail op-amp that might be suitable for this or Dr. Slack's circuit. The A grade has low enough offset that you could use a really small current sense resistor.

My LED conversion of a 3C Maglite uses the latest uber high tech "regulator", a 2.2 ohm resistor

LithiumLord

Mon Oct 27 2008, 05:40PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Check Zetex ICs - they have many LED drivers.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob