This kind of information is great, where can we find all the components for the lowest free shipping price.
I hadn’t even searched ebay for this stuff before djozz build write up and he mentioned it, i just had assumed wrongly that it is mouser/digi-key/farnell and the likes that you find such specific parts.
I built one of these drivers using the original PCB and the parts list below and put it into a triple nichia 219 micro mag. I'm getting 4.7 amps at the tail with an Efest IMR 14500. The light maintains high mode for 5-20 seconds on a freshly charged cell then blinks three times and goes to a light level somewhere between medium and high. Anybody got an idea what's going on and how to fix it?
R1 = 22k
R2 = 4.7k
C1 = 10 uF cap
D1 = SOD323 schottky diode
MCU = ATTINY13A flashed with n-lite by RMM off a QLITE, moon activated
FET = PSMN3R0-30YLDX
Zener and OTC are unpopulated.
Brian
That’s your low voltage stepdown of course.
HKJ - Efest IMR14500 V2 700mAh (Red)
If after monitoring your voltage during the process you find that stepdown is happening at too high a voltage you may increase decrease the resistance of R1.
EDIT: oops, corrected w/ strikethrough text.
What do you suggest I change R1 to, a starting range would help.
Thanks,
B
I suggest you measure what voltage your driver is stepping down at.
You could also switch to a firmware you can modify, so you can set the voltage thresholds yourself.
In case it helps, I have a firmware which will blink out the raw ADC value for voltage. It doesn’t currently do this under load, but that would be easy to change. The point is, you just have to turn the light on, measure the cell voltage with a DMM, write down the value it blinks out, and you’ll have the exact value you need to transplant into STAR or some other firmware. (or at least a few values along a curve, from which you can estimate the actual values needed)
http://bazaar.launchpad.net/~toykeeper/flashlight-firmware/trunk/files/head:/ToyKeeper/battcheck/
3.47v
Is your firmware set up for that 223 at R1 or is it still thinking there’s a 1912 there? That might well be why.
That’s not why. In that case the stepdown would happen at too low a voltage, not too high. EDIT: I was wrong about that. Sorry DBCstm! 
That said, nlite is a hex from DrJones. It’s precompiled for the reverse polarity protection diode to be in series with the 19.1k/4.7k divider. The 22k/4.7k combo is specifically intended to mitigate the effect of removing the diode from in front of the voltage divider. That’s what this driver does.
Reducing the value of R1 is still the way to go, but if kyfishguy’s numbers are accurate it appears that DrJones setup the stepdown to happen at a much higher voltage than STAR for example.
That is the voltage under load?
(Because if so, this really doesn’t add up to me. DrJones’s standard nlite firmware should be looking for something like approximately 0.55v to start stepping down. The 22k / 4.7k combo should not give such a low voltage until around 3.1v at the battery.)
Yes,
That is the voltage at the battery under load. I'm thinking I'll just start over with new components and a different MCU and see what happens.
I made a mistake in my post above. Now edited w/ strike through and some follow up text.
If you are unable to flash the MCU you should just drop R1 until you get the cutoff you want. I’d try 19.1k based on your measurements so far. If your measurements are not accurate then my suggestion will also not be accurate.
As these guys have noted, you have another problem if you have the R1: 22K and R2: 4.7K resistors in there and it is stepping down at only 3.47v. For a sanity check, it is good to measure what the MCU pin is seeing, as that will rule out all variables except for the MCU itself. As wight noted, it shouldn't start stepping down until around 0.55v on pin 7.
The 22K R1 instead of the 19.1K was to compensate for the different driver layout so that the standard firmware ADC values could still be used.
Ok, I tried the 19.1k resistor and it runs fine. I ran the voltage down to 3.28 which amounted to a couple minutes run time. I had to stop there because the pill was outside of the host and getting very hot. I'll do some further testing tomorrow to see how low the voltage will go before stepdown. It has become a certainty that I've got to gear up to flash firmware if I'm going to continue playing with this stuff.
Thanks guys,
Brian
Definitely do your follow-up testing please, that resistor should make cutoff voltage way too low.
… is it possible that you don’t actually have nlite on that MCU?
I’m sure you’ll be much happier with the ability to flash.
I don't know it's n lite but it is a 3 mode, no disco with solder selectable moon mode, memory, and mode order. I'll update with test results.
Wight,
I tested low voltage stepdown with the 19.1k resistor installed. Stepdown occurred reliably at 2.8 volts.
Brian
My opinion is that as long as you discontinue use as soon as stepdown starts you’ll have no issues. Looking back at the NLITE section on DrJones’ site it appears that there is no low voltage cut-off, only a single stepdown. (Whereas STAR repeatedly steps down in an effort to do a ‘battery stretch’.) Understand that your battery has exceptionally little capacity left at 2.8v, so you will have no significant early warning.
So after acknowledging these important limitations, this sounds fine within the scope of what you can currently do for yourself. Clearly my recommendation would be different if you could flash.
Further thought leads me to believe I got the MCU discussed above off a q-lite from IOS. Actual investigation confirms the presence of solder select disco modes. My bad.
To actually test NLITE on this driver I built a second driver with a different MCU. This one came from RMM about a month ago on a 4x 7135 driver and I had him flash nlite onto it. I built the driver with 22k resistor first and got the first stepdown at 3.21 volts. There were several steps noted to progressively lower light levels but there was no associated flashing. I substituted a 19.1k resistor and didn't get a stepdown until 2.7 volts. That's a long way of saying NLITE works fine.
Cool, thanks for the additional info. Definitely useful stuff!
Your measurements appear to show a variance from the specified value on either the 22k or the 19.1k resistor. No big surprise there. For example, 1% tolerance on a 19.1k allows for the wide range of 18909 to 19291 ohms.
To knock the 22k resistor down you could piggyback a high value resistor such as 220k to 330k. Ideally that might net a 2.96v, 3.01v, or 3.04v stepdown respectively for 220k, 270k, or 330k. (this is based off of your 22k stepdown measurement, not your 19.1k stepdown measurement)
I scrounged a 330k resistor and piggybacked on the 22k. Stepdown voltage was 2.95v. I'm putting the light back together now!
Any thoughts on a piggyback resistor value that might reduce the stepdown voltage from the 3.47v I was getting with the 22k and q-lite combo? The 2.8v I'm getting with the 19.1k is a bit low for comfort.