Hmmmmm, that is interesting.
Kinda like “damned if you do & damned if you don’t”. That is a tough place to be put in……… :weary: … 
Personally I have a solution to that problem….
But, since it would not even be close to clearing the “PC Bar”; I will refrain from sharing it……… :smiling_imp: … 
I can live with it, but I would expect Simon/Convoy is going to hear from people who want a refund or new light because of the moonlight malfunction.
Sadly you might be right. that would be a shame too in my book…… considering the apparent reason it’s like that. :person_facepalming:
I just have to say this……….
You people who write THIS STUFF simply amaze me!!!
Thank you and hat’s off to you………. 
… :+1:
Heck, those that can even understand it impress me…… 
Possibly, but hopefully this is just a one-off situation. And while I like having the option of a moonlight mode… how often do you pick up a light the size of a C8 and say “I want an almost indiscernible amount of light from you!”?
This world takes all kinds of people! Some of us code all day long, some drive tractors, other save lives. We’ve all got an amazing role to play! (not to down play TK’s great work - code can be a work of art sometimes!)
Very true! That’s why it’s really not a big deal. 100–20–1% gives me enough flexibility for a light this size.
Notice that the line starting with a value of one specifies it’s for an FET driver, not the 7135 regulation chip driver. Different beast.
TK sent me the code she was sending to Simon, I flashed it to a C8 for JDub as well as a different code set to a few other lights (mistakenly, sorry Jared!)
She very thoughtfully has the code set up for those who would like to tweak it to their own drivers, both FET and 7135 lines are present throughout, all one must do is choose which style they are using.
Please contact Simon through the store. I’ll mention it to him as well but I know he will want you to contact him directly. This is the first issue I’ve heard of but this is all completely new. Hopefully it’s just a fluke. Any problems will be resolved properly I’m sure. Simon would not ask you to “live with it”.
Well said ‘gchart’……. :+1:
Good news, seems like it might in fact be a fluke. Just did a quick test of my 2nd new C8 that I had left at the office, and it doesn’t show that behavior at all. Moonlight seems to work just fine on my 2nd one.
That is good news!!! :+1:
In all the hundreds of drivers I’ve built using the ATTiny13A MCU, I finally got one that glitched. It “shifted gears” unreliably, randomly, with no rhyme or reason. Built a new one, next MCU out of the sleeve, works fine. So apparently there are random fluctuations in the components that allow for slight differences and this sometimes trips us up when we’re using parameters that are right at the edge to begin with.
I also once got a driver from a known high quality dealer that had a sense resistor glitch. Even the little simple sense resistors can seemingly be off by a significant margin once in a while. It’s been said before, if something CAN go wrong, it’s gonna find ME! 
Thanks for the update, gentlemen. Glad it seems to be an isolated event.
And my apologies for calling TK’s code into question earlier. I was wrongly assuming that what was published on launchpad would be the same as what she sent over to Simon. :person_facepalming:
FWIW and until TK comments, PWM of 1/255 should work fine with this firmware.
(IIRC the PWM frequency is adapted to the mode: Lower frequency for lower modes. So the 7135s actually have time to consistently turn on fully even with very low PWM values.)
Biscotti firmware works fine on all my flashlights with 6*7135 and 8*7135 drivers (s2+, s3, c8, old MTE with P7 led)
You may be onto something here. As I said, I haven't actually flashed Biscotti onto anything yet, so I haven't a clue. But Biscotti does adjust the prescaler and between Fast & Phase for the lower modes, so it very well could work with a PWM value of 1/255.
Edit: removed some misinformation to hopefully avoid confusion.
Level here is a number of element in RAMP_FET array. So all works correct.
Ugh, good call. Serves me right for posting to forums while at work… read thru the code too quickly :person_facepalming:
Naw, I mentioned that issue (questioned about it, actually) back in msg #81 here. It’s a fairly well-known issue, or so I thought.
I think the problem is that when you’re parallelling 8 7135s, having to power up each one at once, all from a single µC’s output pin, there’s just a limited supply of current that pin can source, so the “capacitance” of all of them taken together needs time to “charge up” to +V, and if the width of that pulse is just too narrow/short, and that bank of 7135s “charges up” just too slowly, well, nothing happens.
If the bank of 7135s needs 2 pulses to get going, with a pulse-width of 125 counts, they’ll see 123 pulses worth of power. 81 pulses, they’ll see 79. 23, they’ll see 21. 5, they’ll see 3. 1, they’ll see… 0!
Not quite exactly accurate, but a reasonable facsimile of what’s happening.