LED Boatguy, yes, that is correct. It’s still a good idea to measure the voltage with a DMM since lab power supplies are often not terribly precise, but the general approach is exactly what battcheck is intended for. 
(well, at least my lab power supply isn’t very precise… it consistently reads about 0.1V low)
Thank you. Ya, the Fluke always has the final say, though they are only a couple hundredths of a volt apart.
Built my first one of these today! I must say it’s very tricky soldering one of these buy hand!
Hey guys! Quick question for you. I ordered 12 of these to build new version Convoy C8's as the retaining ring doesn't clear the 7135s on the 105C. Of course swapping two 7135's from back to front is much easier than transferring everything over to one of these boards, but it's a fun project (although soldering all of these components can be quite the pain). Anyways, the first one I built is having off-time issues.
Every build I've done with the 105C and a CAP_THRESHOLD of 130 has worked great with maybe a 1 second off-time. This one is more like 5 seconds. I haven't tried swapping out the capacitor to see if it somehow went bad (it's the same 1uF like I always use), but has anyone had this issue before? Could it have anything to do with board differences, or is it most likely a bad component or a bad connection somewhere?
I’m not very good at hardware, but from what I’ve heard many offtime-calibration issues are due to the resistor values. Could the resistors perhaps be swapped, or different types than usual?
I had to change the offtime cap values quite a bit on another custom-built driver. Instead of the usual (and nice) values around 130, it ended up with a threshold of about 250. I needed to dump out the raw ADC values with offtime-cap.c in order to even get in the ballpark at first. Even after sitting off overnight, it reads out a value of about 40.
If you’re not able to get the OTC to work as desired, you can still switch to the memory decay method. I made a version of STAR specifically for this, changing as little as possible to make it work that way:
http://bazaar.launchpad.net/~toykeeper/flashlight-firmware/trunk/files/head:/ToyKeeper/STAR_noinit/
Jonny, I've never built this particular board, but I've built thousands of others with off-time capacitors on them and I've never had to change the OTC values that much. Is it possible that you got a 10uF cap. in place of the 1uF?
I'd try another board just to make sure it's not one of the components before you end up chasing your tail.
i think this may be another area where wight just did things “differently” with unintended consquences. With this board, and a few others of his design, I have to use an OTC value of about 245 to get it to my liking. That’s with a 1uf cap.
There is a short discussion about it related to it on page 3 of his 22mm 7135 board.
Ahh, it might be the diode bypass feature which I haven't looked into yet (to even understand what it does). So I'm not going insane then. I'll play around with values around 240 for the OTC. Thanks guys!
Well, if that were the case then all of my FET drivers (unless zener modified) and the BLFxxDD "1.0" drivers would also need higher OTC values, but they don't. I use the same OTC values across the board. All of those (except my zener mod) have the capacitor before the diode instead of after.
Hmm, maybe I did swap the resistors by accident and that somehow effects it? It's just odd that wight said he noticed the same issue...
https://budgetlightforum.com/t/-/26993?page=2#comment-718398
That is interesting. Sometimes even with these simple drivers something as simple as changing trace length and proximity to other components can have unexpected results.
I’ve been playing with the 105d boards in the new Convoys, and they work well with the memory decay method for off time switching.
Interesting. On the newer Convoy C8 hosts I've built the ring cleared the 7135s without any issues. Did they change things again?
This is a picture that I took when I listed the new Convoy C8 host.
You can see that there is a ton of room with the 105D:
The ones I got Saturday look like that, lots of room with a 105d board, I haven’t tried putting a 105c in to them.
I don't have any 105D's, but my 105C's have the 7135 all of the way up to the edge of the board :( I can just move the two that are and stack them on the other side, which is much easier than swapping all of the components.
Sorry, can someone point me to a description of what "memory decay" is? Does it make the whole off-time cap stuff obsolete?
If you only want short/long presses and don’t mind the threshold being hardwired to about 0.6s, then yes, it can obsolete the off-time cap stuff. But if you want other thresholds or more than one threshold, you still need the OTC.
This thread describes the method in more detail:
I hope it’s okay that I made a version of STAR_offtime which uses memory decay instead of the OTC. People were requesting it and it was an easy change, so I made it.
Awesome, thanks for the link! That is an elegant solution.
[quote=ToyKeeper] I hope it's okay that I made a version of STAR_offtime which uses memory decay instead of the OTC. People were requesting it and it was an easy change, so I made it. [/quote]
Of course!
So I’m having a problem I’ve never had before. I have a driver stuck in moon mode i think? I’ve checked all connections and can’t find any shorts, but the driver will not change modes. I don’t think it’s a bad MCU, because lvp is functioning correctly.
Any ideas?
Check your OTC cap (if you're using one).
Which value for R1 und R2 you prefer when doing zener build? Thanks!