Oshpark Projects

Try the pull-down…

Maybe not a bad diode, but it is a different diode, right? (than the one you use on the 105C)

Supposed to be the same diode, or at least functionally interchangeable. http://www.digikey.com/product-detail/en/ZLLS410TA/ZLLS410CT-ND/1767188

OK, progress, of a sort: the old standby version of luxdrv works properly with the diode removed & replaced with a jumper.

edit: Of course, the voltage monitoring is off without the drop from the diode, it gives 13 blinks for a cell with only 4.1v. But at least this is getting somewhere.

So all it does is remove the reverse polarity protection otherwise

I wonder if a switching diode would be better

http://www.ebay.com/itm/like/121300037210?lpid=82

See my edit above about the voltage detection. Also, it can't be programmed while the jumper is in place. The LEDs turn on when the clip is attached & plugged in. So obviously this isn't a solution, just a good clue as to what's wrong.

Does anyone know what the exact diode is on the Nanjg 105C? I have though I had finally figured it out about 6 times by now, still no closer than when I started, looks like.

If my memory is correct, I have seen the diode with S4 marking on it. That is 1N5819.

I tried a couple of different SMD resistors between the MCU and GATE, 100 ohm and 200 ohm, they both work and function fine. Did some searching and reading (still over my head) but from what I could gather this should ease the load significantly on the MCU. Definitely worth a shot for those of you who are having trouble.

For testing purposes I just bent up the leg and soldered the 0805 SMD resistor between the pad and FET leg. In future boards pads could easily be placed for a gate resistor.

I was testing this on the BLF17DD board BTW.

Are you using the same diodes as me, and did you have issues like mine that went away with the resistor?

e: In other words, is this a workaround to let it work with the wrong diode? If so it'd obviously be better to just switch to the correct diode.

I think it's the same one, which is the one from Digikey marked "41". It worked fine for me without any tweaks at all (without the gate resistor) using STAR 1.1 set to 0x21 (phase correct).

I wanted to test this out because DBCstm has had a few of these FET drivers die, including one that I built for him. Smoked the traces and on another one where he bypassed the traces the MCU may have died. Was thinking that our extra capacitors were just a bandaid for the true problem, which was excessive inrush current load when the gate is being charged. This is just a SWAG, I haven't taken any measurements with my fancy HF DMM or anything like that.

Don't mind the solder blobs, I've had stuff on and off a few times on this board. I went "tombstone" style.

Adding the resistor didn't break it, but that's different than confirming it fixes something.

I can't find that 1N5819 in the right size, the one on the Nanjg is definitely the same size as the SOD-323s I have here, is there yet another cryptic package size that's the same as SOD-323 but using a different name?

What I'm saying is that it may or may not fix your problem Chris, but it's worth a try; either way I feel like it will probably increase long-term reliability of the driver. Several have died of unknown causes that shouldn't have died and we have ruled out a few things that we thought were killing them, this is the next step in my mind.

The fact that jumpering over the diode fixes something tends to tell me that there is a power issue and maybe that we're asking too much of the MCU. Adding the resistor will decrease the peak load on the MCU. Try it and see what it does, it can't hurt. I think that we might have been right on the edge of working without the resistor but that doesn't mean it's a good thing.

1N5819 has different sizes of packing. SS14 is the size of SMA. I have this data sheet on hand.

I haven't heard anything about MCUs dying, who's had that happen? I don't think even a short to ground on PB1 will kill them, will it? I know I've done that before and they always worked again after it was fixed. I can't recall ever killing a tiny13. I've physically broken them, damaged pins and stuff, but never fried one.

Regardless of the outcomes here I will endeavor to add gate and pull down resistor pads to the boards tonight. Note the original intention of these boards was to make a straight swap product with the 105C. Just add the right FET and away you go. This is why the fairly basic but still fundamental resistors were forgotten about. It's two added components that people need laying about. It also doesn't help that the boards appear (for the most part) to work without them!

I don't think they are needed. I believe the attiny has an internal pulldown on the output pins, and if they aren't working right with the wrong diodes that's no reason to add other stuff. With the right diode, the existing boards work fine.

Hey guess what

If you do that, it will then not only change modes reliably but will also work with fast-PWM firmwares. The one into the gate might not be needed, I tried it first by itself and it wasn't right (tried 10 ohm first, no change at all, then 100 ohm and it was better, but still not right), I didn't try yet with only the 10k pulldown.

Still though, why does it need these resistors, when the same circuit with the same components doesn't need them when on the 105C board?

So what have you got there? When I normally use FETs I place a 100ohm resistor between the MCU and the gate of the FET, and a 100k resistor from the gate to ground.

100 ohm inline with the gate, 10k to ground.

It works with nlite, and the other fast-PWM build I had done based on nlite with some minor changes. Moonlight isn't as low as it was with 7135s, but I don't have source code for either one to make any changes to that. luxdrv still makes the switch whine, but modes change like they should now.

I don't know for sure what the default PWM speeds are in the STAR versions, and I don't know them well enough yet to go changing stuff. For instance, in v1.1 Momentary, at the top it sez

#define F_CPU 4800000UL

but then farther down,

    // Set timer to do PWM for correct output pin and set prescaler timing
    TCCR0A = 0x23; // phase corrected PWM is 0x21 for PB1, fast-PWM is 0x23
    TCCR0B = 0x01; // pre-scaler for timer (1 => 1, 2 => 8, 3 => 64...)

I thought fast PWM had to have the CPU at 9600000? Makes my head hurt, I dunno. Anyway, whatever that one is, it works like that on the 46mm FET driver, using the same parts and no gate or pulldown resistors... :tired:

Oh, Matt, if you're going to go back and do some revisions, I have a suggestion for the SRK driver. The selector pads for the switch input are a royal pain, what do you think about something like this instead?