LM317 Voltage Reference not constant

[Benm]

There seems to be a school of thought here that the big output is needed to protect the laser diode from transients (don't understand it much, seems like a big cap would cause more problems than it would solve, but the more experienced guys here do it, it works, so copy-cat it).

The output capacitor has a specified minimum by the manufacturer, and should be placed to prevend unexpected oscillations and such. Same goes for the input cap unless the wires to the battery are very short. National semiconductor advises a minimum output cap of 1 uF, which is probably something to stick by.

 

[123splat]

Ben,
Other guys here on the forum have suggested 10 mikes and someone even used 100 mike. I don't see it myself, but if it works for them, oh well... But your comment about sticking with the manufacturer's recommendation makes good sense (altough, I've been playing with the 117 series regulators for 25 to 30 years, not for lasers, and very seldom put the output cap on for batt applications with constant loads, never seen oscillations,, probablly NOT a good practice, but has been servicable for me).

I checked out your BJT current source in your signature link. That's the first reference that I've seen here on using a real constant current source for a driver. Wonder why nobody else has suggested/tried it?? (LM317 or the like is easier, i quess). What do you think about using a MOSFET configured as a constant current diode, or a current limiter (gate to drain resistor)? Be interrested in hearing your thoughts/experiences.

 

[GT273]

So are you saying that my 47 uF might be a bit much or is the 1 uF just a minimum value? Should I consider changing or is it good the way it is? I don't feel like the 47 uF has been an issue but I haven't received my diode in the mail yet.

 

[Benm]

Ben,
I checked out your BJT current source in your signature link. That's the first reference that I've seen here on using a real constant current source for a driver. Wonder why nobody else has suggested/tried it?? (LM317 or the like is easier, i quess).

I'm not really sure why this design isn't used more often. It offers a number of benefits over the 317 design, but one downsite is that it does not have internal protection of any kind, so if you make it dissipate too much you will blow up the output transistor.

What do you think about using a MOSFET configured as a constant current diode, or a current limiter (gate to drain resistor)? Be interrested in hearing your thoughts/experiences.

I've seen those, but those are limited to a small maximum current (10 to 20 mA or so, depening on the mosfet), you can just rig the gate to the source. They're great for driving 5mm leds from variable supply voltages, but otherwise not very practical.

There are many approaches to constant current drivers though, and i plan on publishing some more. Personally i like the opamp based designes best, since you can set the current by a potmeter that doesnt have to carry all of the power, and you can do modulation as well.

 

[123splat]

GT, You ought to be OK with the 47 mike, yes, more than spec, but others here have used much more and not reported any problems.

Ben, The low current issue is with constant current diodes (basically FET's with the drain shorted to the gate). I thought maybe using a higher current device, like a MOSFET, would provide better currents. I have not been able to find anything in any of my text books and databooks mentioning it. I'll have to try a sacrifice MOSFET and see what happens. My quest is for a CC source with minimum parts count and minimum voltage drop.

 

[Benm]

Unfortunately the size/power of the mosfet has little to do with the current it sets at. Basically all mosfets can act like current sources by themselves within certain limits. In the simplest case this works by shorting the gate to ground/source, givig Vgs = 0, and a current that depending on the exact fet is something like 10-20 mA.

Interestingly, this current limiting effect also persists for Vgs voltage above zero. You could, for example, raise the gate voltage 0.6 or 1.2 volts over the source voltage using a diode a bias resistor. This will result in a higher current source, but with a bit more voltage dropout before it reaches full current.

 

[123splat]

Yeah, that figures... DRAT!! Nice conservation though.

Gt, Didn't mean to threadjack. It really was on topic,, sort'a....

 

[Benm]

Don't think it matters that much, there are so many topics on 317 drivers out here aldready.

It should be noted that both mosfets and bipolar transistors are current sources by default. On a bipolar, the base current gets mulitplied by the transistors gain (hfe), something typically between 60 and 250.

The mosfet does the same, but instead of multiplying base current, it has a curve between gate-source voltage and output current.

Both are typically used as switches though, where the input current and voltage respectively are high enough to saturate the device.

 

[123splat]

Depending upon which region they are operating in.... ha ha

Sorry, couldn't resist. Thanks for a stimulating conversation.