I’m just guessing, but looking at the screenshot of the code he took in post #41, it looks like he has one MCU output going to the 7135 (the original output, I think) and then added code to have a 2nd output to the FET Gate input.
In the screenshot he has two arrays, amc7135modes and fetmodes, and I guess the numbers in the arrays correspond to the modes vs. PWM frequencies for each of those inputs?
So, 1st mode, has:
MODE 7135 FET
= =
off 0 0
1 5 0
2 255 0
3 255 56
4 0 255
So do you see the logic there?
In mode 1, 7135 PWM is 5, and FET PWM is 0, i.e., only current via the 7135 and at low PWM.
In mode 2, 7135 PWF is 255, and FET PWM is still 0, i.e., still only current via the 7135, probably at max, or 350mA.
etc.
Something like that anyway :)…
So, it seems like what he (Major) did was:
1) Remove all 7135s except one of them
2) Add the FET, wired to MCU pin 5
3) Mod the minidrv firmware
From the reviews, it’s unclear exactly what exact version, but the ones in their pics on their website shows a small cap on the battery side/spring side of the driver. Assuming that’s the case, can someone tell me if I can move that cap on to the emitter side of the driver after I clean off all of the 7135s and have added a FET? And, if so, exactly where (since the pics from comfy’s original post/thread are gone now)?
It looks like you’re in the USA. Have you thought about getting some oshpark boards instead? For those in the usa it’s easier than hacking up nanjg boards imho.
The nice thing about using the NANJG board is I don’t have to go buy the different parts (including the MCU), but I just did one where I am also burning firmware and it’s a pain because at least the way I do the FET, it gets in the way of the SOIC clip, so I have to desolder when I want to re-burn firmware. Does the SOIC clip fit on the MCU on the oshpark boards without removing anything?
I accidentally shorted + - of this driver on upper emitter side with reflector (one layer of kapton tape for isolation on reflector bottom is not good enough :) )
So whole flashlight smoked, battery got hot(lasted probably for 10 or more seconds), i got aware when I saw smoke. After that accident:
- switch died
- driver remained working on one mode(high only). I will throw it away but it survived :)
- Led survived
- Panasonic NCR18650PF survived without any issue?? Lito kala says it is fine.
So I just wanted to post my bad experience here. This is good driver and panasonic is one great cell. No explosion and it still works like a champ.
All that current of the short went through the FET, was it the FET that smoked? Or the wires?, If that current was over the FET-specs it may just be permanently shorted. Perhaps if you just swap the FET, the driver lives again. :-)
The attiny did not receive an abnormal voltage, so I would not expect it to die. The only abnormal thing that has happened to the driver is a large current through the FET.
I agree. if the only problem a reflector short (bypassing LED), the FET, cells, and driver traces would take the hit. Are those protected cells you were using? The protection circuits may allow the cell to work to some degree, but they can mess with mode swiching when they get messed up. I think Ohaya had an issue like that recently. You might want to try different cells, but I would suggest doing an autopsy on the driver first. Makes sure nothing melted and shorted.
You shouldn't need to pot these DD drivers to keep them cool. They don't generate much heat. Maybe you potted to protect wires from vibration?
Yes I potted to save wires from high centerfire recoils. I use AK-47 C (high, mid, low mode). This is the only one of 50 that failed by my mistake. I just wanted to report that to you guys to be more careful than I in building or modding :)
Bats are Pana NCR18650PF (there is no protected version of this cell on the market). I put this cells on discharge test year ago and now this happened and it survived :) . I don't know how but obviously it has some kind of internal protection even without added pcb's.
I will have to scrape fujik of driver to see what happened.
I managed to redo fujik treatment. Everything was in place although artic alumina under FET melted during fan de soldering. It seems that FET died since it looks and smells smoked but I did not want to bother replacing with another one. Curious thing is that driver still works on one weak mode.
With this test I found that FUJIK can withstand heat better than artic alumina when we do plastic to pcb gluing like I did with FET before leg and wire soldering.
For metal to metal I don't even doubt AA is far better than Fujik.
So if curing time is not issue I suggest that you put fujik under small FET. Than wait for 24hours for firm fujik cure in warm room, and after that you can finish driver leg soldering and adding +- wires.
You made me just recall something. I once modded a Nanjg 105C similar to what you did above and it only had High mode too. Turned that the FET Drain Pin (Big Tab across the back of the FET) was touching the inside of the pill and thus taking ground straight to the LED - Lead.
EDIT: I must be having a brain fart. Is the Gate soldered to Ground and the Source to the PWM feed? That can't be right. The OP shows the same setup though. I thought these N-Channels in this package were bottom-left pin = Gate and bottom-right pin = Source (when looking from the top). I must have something mixed up.
I don't know how to properly express myself but everything is connected as Djozz said. Like this:
I just scrape solder mask of those 4 vias and fill vias with solder for even better + contact to spring side.
My situation(one mode only) happened after shorting +- upper emitter wires with reflector ( i had only one layer of kapton tape on the bottom of reflector as protection and while I was pressing reflector +- wires somehow managed to dig hole into kapton tape, and then smoke out happened :) )
My drivers are potted so no way anything contacts with pill.
I haven't had any driver failure so far and they are in use for more than 3 months by various people.
It is dirt cheap and it works. It works even when smoked :)
I would not say this driver is winner over my diy Djozz driver guys.
My driver has High - mid - low without memory effect so it always starts on high. And I would not change that for new banggood driver unless they made similar modes.
Yes BG driver looks good but modes does not sound good to me.
Edit: - diy driver has reverse polarity protection
- It has a kind of low battery warning protection (light drops on about 1-2% at 2,8V of battery)
- 4.3A current draw of single ncr18650PF and all that stays in nice regulation for about 10 minutes on high and than another 10 minutes of 90-80% power, and then another 10-15 minutes around 80-50% output and then after some time it steps to low light warning protection which works for about 1 hour until it drops to 2,5V battery voltage.
- It is safe from start to complete battery discharge. So for all driver creators please try to make simple fet driver without memory, and user interface and with High - mid - low mode. Why? Try to sell flashlight to hunters and military guys and they'll say that they want instant full power action without any user interface.
Driver makers please try to make this mentioned version of driver and I guarantee that it will be your bestseller. You have nothing to loose.
and TK’s A6 firmware.
