LM334 end of regulation

[TorchBoy]

I've been playing around with various current control circuits and have gradually moved toward the "Higher Output Current" circuit in fig 8 of the National datasheet for the LM334, with a Luxeon K2 placed in the customary spot. It all works quite well, and I'm testing with three and four AA NiMH cells.

One problem - I've noticed that when the cells go flat the current through my LED drops. In other words, the circuit eventually goes out of regulation. Surely that's not going to be good if that causes Iset to go over 10mA. How is one supposed to wind the thing up (ie, know to turn it off) when the battery starts to flatten? Is this why Doug put the 390 ohm resistor in series with the LM334 for his world torch circuit? Does it prevent the LM334 from sucking too much current in such a situation?

(I don't think the exact spec of my components will have much bearing on this, so I haven't gone into the details there, and I'm still changing and tweaking them anyway.)

 

[TorchBoy]

Thinking aloud here - how about a 2V yellow LED in series with the LM334? It would get really bright if the white LED draws less current and the LM334 draws more current. And it would blow instead of the LM334 if it drew too much current.

 

[TorchBoy]

I think I've found the answer. http://poc.purdue.org/docs/caving/LED_Headlamp.pdf has the circuit (same as the one in the National datasheet except for the placement of the second resistor) and some handy-looking equations.

R1 = 0.064/I
R2 ~ 300/I
Where I is total LED current
C1>1µF
The LM334 is a 3-terminal adjustable current source. It operates to maintain 64mV between its R and V- terminals (hence across R1) and does so by passing current into the V+ terminal and out of the R terminal. Assuming adequate cell voltage, the LED current is the beta of Q1 times the R2 current. R1 sets the LED current which is 64mV/R1 for reasonably high Q1 beta.
Hence R1=64mV/I where I is the desired LED current. R2 is selected to regulate the base current to an appropriate value when the battery voltage drops to the point that the circuit falls out of regulation. Under this condition the LM334 will draw all the current it can through R2. C1 prevents the circuit from oscillating and any value from 1 to 10µF works well.

 

[evan9162]

or you could use a MOSFET, where you don't need any significant amount of current at all. With a MOSFET, the 334 will only be drawing uA of current at any point.

 

[Doug S]

:lolsign: I thought the text in the first 3 pages of that linked document looked familiar. Upon closer inspection I recognized that it had been lifted from an article that I had written it for a printed publication back in 1998!

I think I've found the answer. http://poc.purdue.org/docs/caving/LED_Headlamp.pdf has the circuit (same as the one in the National datasheet except for the placement of the second resistor) and some handy-looking equations.

 

[TorchBoy]

Sadly the low voltage MOSFETs around here are quite expensive. I picked up a lower cost one a couple of days ago but for 1A it wants over 4V on the gate. Might be doable with four AA cells under load. I'm still playing around with components - still a whole week to go before my caving trip. (And I'll take a soldering iron with me. )

Doug, why no resistor on the stability capacitor in that circuit? And is 1µF OK for ~1A?

 

[Doug S]

Doug, why no resistor on the stability capacitor in that circuit? And is 1µF OK for ~1A?

Actually using a series RC combo instead of just a capacitor will be more stable over a wider range of conditions. I checked my notes on this circuit and it appears that with C greater than 0.047uF and R between 100 and 470 ohms it works well. Circuit should work at 1A but remember that the main transistor has to dissapate the difference between the battery and LED voltage [except for the 64mV across the sense resistor].

 

[TorchBoy]

The transistor heatsink didn't even get warm with 4 AAs, which I thought was a bit strange since it was dropping over a volt at over an amp (ie, over one watt). When I tried just a dropping resistor it got too hot to touch for a similar wattage dissipated.

I realised this afternoon that the MOSFET I got on Monday probably won't do for the LM334-regulated circuit. It's an n-channel one which I got for trying out a different circuit, based on two NPN transistors. For the LM334 regulated circuit I presume I need a p-channel MOSFET?

 

[evan9162]

Yep, you need a P-channel MOSFET

I've "perfected" using a MOSFET transistor with the '334 (and built many regulators with this setup, all work perfectly) - check out the thread here:

http://candlepowerforums.com/vb/showthread.php?t=75204

The pass transistors that I've built these with really aren't too expensive. Mouser has the '6021 (D2PAK, higher power and current) for just USD $1.23, and the lower power SOT223 part ('434) is only USD $0.73 each.

You might check out to see if these wouldn't be too expensive to get to NZ from Mouser.

 

[TorchBoy]

I had a look around a couple of NZ electronics outfits' web sites last night, and it seems that p-channel MOSFETs are very rare in NZ and quite expensive. A NZ$22 one looked like it had good specs but that's the cost of two whole Cree XR-Es (incl shipping)!

The specs on that MOSFET look great but Mouser only has courier shipping, which means it would be less expensive to just buy the NZ$22 MOSFET here. Break-even point would be three of them. At least they would arrive before my next caving trip. :laughing:

What package does it come in? The datasheet they linked from their site shows TO-263AB. Surface mount? Heatsinkable? But you mentioned D2PAK?

 

[evan9162]

D2PAK = TO263AB

It's surface mountable, but doesn't have to be. You can straighten the outer pins with pliers, and the center pin (drain) is long enough to solder to. Then you can thermal epoxy the tab (also drain) to a heat sink of some kind, just make sure to isolate the heat sink (or isolate the tab from the heat sink).

Too bad shipping costs so much.