LM1085 DIY Driver

[rhd]

Hmmmm, but how do you regulate the amount of current the laser diode gets?

Diodes will just pull as much current as you can supply them with, until they die.

 

[anselm]

If you are able to regulate the voltage with enough precision and stability, in my understanding
it will limit the current as welll. (Excess) voltage pushes current across resistance (LD).
Say, FOR EXAMPLE, a red's diode Vf at 400mA is 3.3V.
Supply it with EXACTLY 3.30V and you may see MAYBE 300mA being pushed through,
likely depending on the batteries internal IR, length and diameter of wire used, etc.

Supply it with 3.35V and you might just get those 400mA you were looking for.

I just pulled these numbers out of my ass , so don't rely on them!
I'd be happy if one of our resident electro-wizards could either tear my post apart or confirm it.

 

[HIMNL9]

The problem with this system, is that actual ICs used in voltage mode (with no current feedback), have a too much high tolerance, for be used safely for LDs.

Following your example, if you get 300mA with 3.3V and 400mA with 3.35V (indicative values just for the example), you have a 25% of current variation with 1.49% of variation in the supply voltage (the 5mV) ..... now, suppose to use a common LDO, that may have, like, 10mV of load regulation tolerance at half of the rated current (it's a very common value, this one), and you may end in a fluctuation of 200mA on your LD, without any possible regulation of it (cause your regulator, in voltage mode, have no current feedback, so no possibility to regulate it)

Is for this reason, basically, that all the LD drivers works in current mode with a Rsense for the feedback, or in alternative in power mode with a photodiode feedback, but not in pure voltage mode.

 

[rhd]

Agreed ^

Just take a look at the vF on most PIV curves. It's a terrifying thought to be relying on such a flat curve for regulation of anything.

 

[Cyparagon]

The voltage regulation is very good though. Even better with a constant input voltage like batteries. The main concern is the diode drawing less voltage as it heats up, which is why you Mr. Larry suggests a large heat sink AND setting the voltage where it drives the diode slightly below your optimal current when you initially turn it on - to allow for this increase.

 

[rhd]

(...) Even better with a constant input voltage like batteries.

That wasn't my traditional understanding of how batteries supply voltage. They're not constant, are they? They decrease voltage as their capacity depletes, do they not?

 

[Johnyz]

About that heatsink thingy you was talking about, Flexmod has an anodized thus non-conductive heatsink. LM1085 has not, but you can always use a Mica sheet (used for decades when attaching transistors to heatsinks). They cost $0.1 from my local electronics store.

 

[jib77]

Sekisui thermal tape is also good. Thermal tapes and sheets are only good as long as you are joining two flat surfaces though.

Here is the results of a test with my LM1117 driver setup for 430mA with a 3V dummy load @ single li-ion voltages:

4.2V = 430mA
4.1V = 420mA
4.0V = 410mA
3.9V = 400mA
3.8V = 380mA
3.7V = 360mA, really starts dropping after this.

 

[Johnyz]

Man, that LM1117 thing is just excellent.

 

[Wolfman29]

I just tried sticking an 18650 in my hotlight host to see if the single li-ion worked for the LOC. No good :\ It only produced a dim LED, and that was a fully charged li-ion.

Hell. It even started dropping in power after my two 10440s dropped below 3.5V each. But maybe I just got a bad driver.

 

[jib77]

I just tried sticking an 18650 in my hotlight host to see if the single li-ion worked for the LOC. No good :\ It only produced a dim LED, and that was a fully charged li-ion.

Hell. It even started dropping in power after my two 10440s dropped below 3.5V each. But maybe I just got a bad driver.

Look @ post #13 ... anything below 3.6 and your cell is almost out of juice. Sounds like you have crappy cells.

 

[Wolfman29]

Oh, no. The cells are in series, so it adds up to 7V, and fully charged they are 8.4V total, which is perfectly fine. (Protected Ultrafire).

However, with *one* 18650 fully charged it couldn't power my LOC.

 

[purevw]

FYI-

Great price on 1085 ICs in the TO-252 package - 100x for $33
AME1085AMCT AME Low Dropout (LDO) Regulators

The regulator that came up in the link shows maximum 7V input with 1.4 dropout. Thought price seemed to be unusually low. A Texas instruments (National Semiconductor) LM1085 at Mouser lists for over $200 for 100, which is 37V max at 1.3 dropout if I remember correctly. Please note that the TI regulator is a TO-220 package.

 

[rhd]

The regulator that came up in the link shows maximum 7V input with 1.4 dropout. Thought price seemed to be unusually low. A Texas instruments (National Semiconductor) LM1085 at Mouser lists for over $200 for 100, which is 37V max at 1.3 dropout if I remember correctly. Please note that the TI regulator is a TO-220 package.

And that explains why these didn't work well for me.

Okay, I have a question about TO 252 based 1085 regulators. I got a bunch of them recently (the ICs) and used them on a straight forward PCB - much like a Mohgasm I guess.

Well, unlike TO263 or TO220, these things basically need immediate heat sinking for almost any high current use.

Problem, is that even fully heat sinked (via the top of the case, which is all you can really heatsink if it's mounted on a PCB), these guys have thermal problems a minute or so in. I've never experienced the same issue with TO263 ot 220, which seam to handle heat much better, even when heatsinked on the "wrong" side.

Thing is, I must be doing something wrong, no? I can't imagine that anyone would use divers like the Jib77 or Mohgasm if at 1.8A they could only last a minute or so.

I don't really *need* to solve this, but I bought a bunch of the ICs from Kiyoukan when he was in a bind, and I'd hate to waste them.

 

[Ghostchrome]

And that explains why these didn't work well for me.

Okay, I have a question about TO 252 based 1085 regulators. I got a bunch of them recently (the ICs) and used them on a straight forward PCB - much like a Mohgasm I guess.

Well, unlike TO263 or TO220, these things basically need immediate heat sinking for almost any high current use.

Problem, is that even fully heat sinked (via the top of the case, which is all you can really heatsink if it's mounted on a PCB), these guys have thermal problems a minute or so in. I've never experienced the same issue with TO263 ot 220, which seam to handle heat much better, even when heatsinked on the "wrong" side.

Thing is, I must be doing something wrong, no? I can't imagine that anyone would use divers like the Jib77 or Mohgasm if at 1.8A they could only last a minute or so.

I don't really *need* to solve this, but I bought a bunch of the ICs from Kiyoukan when he was in a bind, and I'd hate to waste them.

Yeah, I have issues with a 1085 based driver set to 1.8A as well. It starts strobing maaaybe 30 or so seconds in. I don't think there is good enough heat sinking on it, and it's Arctic Alumina'ed to the pill.

 

[rhd]

Yeah, I have issues with a 1085 based driver set to 1.8A as well. It starts strobing maaaybe 30 or so seconds in. I don't think there is good enough heat sinking on it, and it's Arctic Alumina'ed to the pill.

I *think* it's a fundamental flaw in the concept of using a PCB-mounted TO-252 version of the IC. The bottom side of the IC, with the thermal pad, is heatsinked to the PCB itself. You rely on arctic silvering the WRONG side of the IC (the epoxy/plastic top) to the pill/heatsink/host.

I just don't thinke TO-252 ICs have enough:
- Surface area to dissipate the heat to the air
- Volume to absorb the heat in the first place

However, I'm stumped because I feel like if this was the case, we'd be hearing similar complaints about lots of the drivers out there. The Jib 1.8, the Mohgasm 1.8, etc. And we don't - or at least I've never heard anything negative about those drivers.

I've never (until now) wanted to use a TO-252 IC. I've always suffered with the larger packages, as a trade off for knowing they could handle themselves. But obviously people use 252, and *must* be having some luck with it, otherwise I wouldn't see it used so much.