LM317 as constant current, gang ?

Thread Starter

russpatterson

Joined Feb 1, 2010
353


I need a ~2.13 Amp constant current device in order to test some batteries. I got some surplus LM317's for cheap and would like to gang them up as shown in the schematic.

I figured the 4.7 and 5.1 ohm get me 2.44 ohms and that should give me ~510 mA. Four of those get's 2.05 Amps plus one with a 10 ohm for 130 mA for a total of ~2.17 amps total. Using the formula from here http://diyaudioprojects.com/Technical/Voltage-Regulator/

The results were spectacular (if you like the smell of resistors burning). It worked fine with one but something went amiss when I ganged them up. Should this be able to work? Maybe I have some bad LM317's for $.30 each? Does tying the inputs and outputs put a possible higher voltage on the output than the input and kill the LM317?
 

Attachments

Wendy

Joined Mar 24, 2008
21,929
How accurate must it be?

Transistors make good constant current sources. If you don't have to ground the load there are some really easy ways of doing it with op amps and 0.1Ω resistors measuring the load to ground.
 

Thread Starter

russpatterson

Joined Feb 1, 2010
353
I just want to simulate ~25 or ~26 watts of LED's running at 12V. It doesn't have to be perfect but I'd like it to be within 5% or so. What does the circuit look like for that?
 
Last edited:

Wendy

Joined Mar 24, 2008
21,929
This is one of my favorites.



I'm not sure about it's stability though. The resistors need major tweaking too. The op amp version, which I have yet to draw, should be a lot more stable.

The design shown above will go to about 600ma max. For 2.4 A max use about 3 more diodes in line with CR2 to increase the range, and drop R3 down to 470Ω and R2 to 200Ω to 500Ω. Temperature variation will cause drift.

As for the op amp circuit, I'll post it when I draw it.
 
Last edited:

Thread Starter

russpatterson

Joined Feb 1, 2010
353
Thanks for posting, I don't think I have heavy duty BJT transistor around, that I can attach a heat sink to. I've got plenty of MOSFET's but I'm sure the circuit would be different for one of those. I'll order some parts and try it out.

I'm still interested to learn if it's possible to gang up the LM317's.
 

THE_RB

Joined Feb 11, 2008
5,438
Russ, possibly the best option would be to use one LM317 and a large pass transistor, like in the LM317 datasheet (the circuit for a high amperage supply). That principle will work just as well for a constant current hihgh amperage supply as it does for a voltage regulated high amp supply.
 

Wendy

Joined Mar 24, 2008
21,929
True enough, the only problem I see is anything there is a BJT involved temperature is going to shift the BE dropping voltage, which is the stability issue I was talking about.

Here is the circuit I was talking about using an op amp. This circuit does compensate for the BE voltage shift nicely...



It has the following problems though...

1. Replace Vcc for a regulated voltage on R1 for much greater stability.

2. Add a capacitor on the plus input of the op amp for a soft start.

3. It is possible the op amp will not have enough drive for Q1, if so replace Q1 with a Sziklai or Darlington Pair. It will drop an extra 0.6V, but radically decrease loading on the op amp.

4. Speed of response adjustment is almost entirely dependent on the op amp speed.

5. You will need an op amp that will go to ground on the input, something like the LM386 or such. Someone suggested a much faster model for this, but I've forgotten the part number.

6. R2 need not be a high wattage, you need to calculate what your requirements are. By being such a low ohmage value it can take more current than might be apparent at first.
 

Ron H

Joined Apr 14, 2005
7,014
Desired load current is ~2.13 Amperes; he wants to simulate a number of LEDs.



12v lead-acid batteries, so basically a range of ~11v to ~14v.
It was not clear to me that LEDs were being used as a test load. Even if they were: How were they connected? The actual load is important.
I missed the 12V post the first time around.
 

SgtWookie

Joined Jul 17, 2007
22,210
It was not clear to me that LEDs were being used as a test load. Even if they were: How were they connected? The actual load is important.
Well, he wants to simulate 2.13A going through LEDs, apparently in some form of array. He did not mention the Vf of the diodes nor what the specific configuration of the LEDs will be.
 

Ron H

Joined Apr 14, 2005
7,014
Well, he wants to simulate 2.13A going through LEDs, apparently in some form of array. He did not mention the Vf of the diodes nor what the specific configuration of the LEDs will be.
A good start would be to tell us the load he was using when he smoked the resistors.
 

Thread Starter

russpatterson

Joined Feb 1, 2010
353
Thanks all for the replies. I will look in the LM317 data sheet for the high amperage circuit. I want to build a circuit to be able to simulate any number and size of LED's. I guess the fact of what load is being simulated is irrelevant.
Sorry for the confusion. I will use a 1 ohm sand resistor as the actual load.

I want to discharge 12V AGM batteries in order to test charger tweaks.

The 4.7 Ohm resistors where the one's that burned up.
 
Last edited:

THE_RB

Joined Feb 11, 2008
5,438
True enough, the only problem I see is anything there is a BJT involved temperature is going to shift the BE dropping voltage, which is the stability issue I was talking about.
...
No that's not the right circuit Bill, the one i was talking about is shown below, it uses the LM317 for regulation and the PNP pass transistor is on the high side. Regulation is independent of the transistor characteristics.

Normally this is used for larger current voltage regulators, although there is no reason it cannot be used for constant current regulators too.

The 2.2 ohm 1W resistor sets the crossover point at about 0.6/2.2 = 300mA so the LM317 passes 300mA and the rest of the current is passed by the PNP transistor. Current, (as before) is set by the resistor between the LM317 ADJ and Vout pins, so that 1.25v on that resistor gives the desired current.
 

Attachments

Top