Good job answering your own question. Now you see why we don’t do it
What are the dimensions of SS34? If they fit, and it seems like they would, I’d consider a PowerDi5 package or Powermite3. Take a look over here for some suggestions on PowerDi5, and take a look at post #47 over here to see a Powermite3 diode in place on the common LD-29. Note that PowerDi5 and Powermite3 are the same size.
That’s a great start on the diagram - there’s nothing to laugh at. It’s a horrible, tedious process that few of us love.
The upper left pin your reference almost certainly hooks up to QX9922 I keep saying QX9922, but I mean QX9920! Anyway, it almost certainly hooks up to Pin 1, the one marked with the dimple.
When you talk about resistor values, I’m not really following (eg I refuse to continue doing math ;-)). Let’s simply refer to set currents and cut out the middle man. In terms of whether it would be OK without potting… who knows, why not just pot it and be done? If you are afraid it will fail, “trying it” without potting it won’t help it not fail. It will simply make the driver more likely to fail. I’m not certain what you are referring to when you say thermal putty. Are you talking about ImA4Wheelr’s reversible potting recipe?
Yes, the data you posted earlier was exactly what I was looking for. It showed pretty much what I expected. I don’t see where the driver ever gets past 75% efficient and it quickly dives once you take it past 3A output. At 3.59A it’s 65% efficient. That’s a pretty massive loss in the driver, ~6W. Actually while we’re on the subject… you’d better pot that driver Seriously though, I don’t think it’s a good idea to use the driver at that drive current unless you can get the efficiency up.
When you do your spreadsheet you can simply add a column of Pout/Pin to get efficiency.
I haven’t digested your entire post yet, but I don’t understand something. The PowerDI5 package looks quite different from the SS34. The SS34 has only 2 pins/legs, one on each end of the chip. The PowerDI5 package looks like it has 3 pins, or more like a tab on one end and 2 pings on the other end. So how would a PowerDI5 packaged component replace an SS34?
Link doesn’t work. Try Digikey’s “shorten URL” function (near the top, chain link icon) or use the Advanced Post Editor when you post the normal link.
That said… as you can see, I ask for help when selecting diodes. That’s why I tried to point you towards one I’d already looked at extensively. The diodes discussed in the thread I linked are certainly available at Digikey. I’m not saying that they are the best or even that I know they’ll work, just that they are the best thing I can suggest to you right now!
Anyway I can’t come up with much to check at the moment.
I’d check out the modes chip a little more. One pin should hook to a supply voltage - in this case that would be hooking to the side of W8 with a line on it I think. Another pin should hook to real GND I think. Another pin should hook to the QX9920. And another pin I think will connect to that brown capacitor next to W8.
Mostly that stuff might give us a lead on the identity of the modes chip, or whether it’s feasible to replace it with something like a SOT23-6 Atmel ATtiny10. Granted you might need to drum up a little more interest in this driver for anyone to want to work on that… IMO efficiency would need to get to 75-90% at 3A for this thing to get really interesting. Maybe try one of the SOT23-3 FETs the BLF15DD folks have been talking about, I dunno.
I guess that it’s more correctly “thermally conductive putty”, i.e., putty-like stuff that is able to conduct heat. Hank at IOS use to include a couple of small blocks with some of the drivers he sold (maybe still does).
I missed your post #133 above. I’ll check the FET from the BLF thread:
EDIT: FYI, as you probably guessed, I’m more interested in trying to do what you talked about in your last paragraph above, i.e., to try to increase the efficiency.
I think the reason that I actually like blinky modes is that when I’m testing drivers, it sometimes the only way to tell if the driver is actually in use, vs. a short and going to direct drive, because on the bench, it’s sometimes hard to tell low mode from high mode.
Maybe if I actually USED my lights, I’d hate the blinkies, but for testing drivers, they’re very helpful :)!
I assumed that AOEC was an Alpha and Omega Semiconductor FET since it started with AO. Now I realize that I was probably wrong about that. Who knows what those markings indicate, but I think it’s a fair guess that replacing the FET is a good move. May as well replace both of course.
My test LED is on a little (20mm diameter x 25mm tall or so) copper rod. I sit an LED light bulb diffusing dome over top of that. It’s generally fairly easy to tell the modes apart, but things do still look very bright on either high or a short… so I know what you mean.
I know that I keep going off-topic on this thread, but despite wanting to fully paramerize this driver, I’m really “feeling the need” to put one of these in a small/tiny host :(. I was messing around with the one with the R100 stacked on the R200, with an MT-G2, and got to 6.2V @ 2.99 amps on the emitter (according to the el-cheapo laser-guided thermometer I had, the emitter got to 197F, and the driver to about 98F, so things would get a bit toasty :laughing:!
Wherever you put it, the driver will need to be heatsinked.
No thermal protection, no LVP, no turbo stepdown… sounds like a pipebomb ;-). And I’m not saying I wouldn’t do it. :-p Just take the usual precautions. 1/2 twist will fully lock out a Roche F12, so that’s maybe an option for 2x18350 & an MT-G2.
FYI, I haven’t abandoned or given up on this yet. I’m still waiting for the new FET and diode. In the meantime, I’ve been trying to get one of the drivers (the one with R200 and R100 stacked) into a small light, but the pill was too short to accommodate this driver with the toroid (the toroid was just a tiny bit too tall), so I’m back to waiting for the parts.
I’ve been thinking that when the parts come in, I may pull the toroid off, so I’ll take pictures if I do that.