Excellent job. I hope I can afford it when released…
Complete, ready to run!
It was interesting soldering the QFN chip, I’d never done one before.
I also moved the cap on the bottom up (I made this change already in the design but I wanted to physically move this one, electrically it’s the same spot, just a slightly different location to be in line with the resistors) then changed out the resistor’s to 100k, they’re also 0805 but they definitely fit on the pads better. All that’s left is to wait for the host!
Next version (v3.2.1) boards ordered.
driver porn!
Allright guys first light using one of these is up and running!
This is my SolarStorm T4 (just got here at 4:30 this afternoon). It’s running 2.1A to each emitter, on high white with HCRI mode enabled it does 990Lm at 30sec (the HCRI option adds in some RGB to the main white levels).
I have found I’ll need Everett to take this chip back and increase the minimum PWM, the lowest the emitter will light (.6Lm) is several steps from the lowest PWM level or the ramp (also low RGB and low flashes don’t work either).
I’m waiting for my updated boards to get here and I’ll be sharing the OSHPark files once I verify I didn’t mess something up.
I also received my 17mm_RGBW_Clicky boards today, being the same exact circuit just with a different MCU I expect it to be working and that’s my code so I can adjust the PWM levels as needed.
Driver installed on stripped factory driver
Running the lowest white level that had RGB added in with HCRI mode on
A little higher
Well that was stupid…
Note to self, in applications such as this MOUNT THE DRIVER WITH THE RESISTORS UP!
?? Short?
No I made sure there couldn’t be any shorting out, just made it a giant pain to access the resistors.
You know its a PITA when it only takes doing it one time decide to rip it all apart and rewire it lol.
Does anyone have an opinion on if trying a higher value resistor might be able to get the low PWM levels to light? I tried going lower, made it all the way to 30k (from 100k) with no change and it still working the same, no better no worse but below 30k and it would go sporadic switching modes in its own. I also tried jumpering it with tweezers and that shut the light down instantly so I didn’t do that again. Currently I went back to 100k. Maybe I should try going higher?
What is the 100K for, exactly? Can you post a schematic?
The lowest emitter pwm value is going to be a tug of war between the period, or frequency, value and the pwm duty cycle register value. With the solar rechargeable contest light I ended up using a 1Khz frequency for my direct drive to make gains on a broader range of PWM values. Wasn’t concerned with the lowest possible pwm value with that driver. With the 105C linear ramp driver, I think it was a 4.5kHz? pwm freq., and an OCR0B value of 4 for the lowest duty cycle value.
First off, I do not know how the code is doing the mode switching, so the following tip may or may not be relevant. But, I’ll try one last time to get you to add a 0.1uf cap between the SW+ and ground, and then mum’s the word for sketchy mode switching.
There’s no need for a debounce cap on the switch; the firmware runs a very robust debouncing algorithm that requires 10 consecutive samples spaced at 1ms to match before it registers a level change, so the ~3.3ms time constant you’d introduce with a 0.1uF cap wouldn’t have an effect
tterev3;
OK then, just checking. Some switches are really lousy, or bouncey. Using a cap can be a quick check when you are unsure of the software.
Absolutely, it’s good advice in many situations. I try to avoid it in this case because the switch pin is also MCLR used in ICSP so it could slow down transitions and affect the programming process
I like using MCLR as a switch input on the low pin count devices too. Always nice to have an extra gpio handy. Ya, so no cap on MCLR is a good idea. If I used dfn packages, I might have been able to figure that out before hand.
Well good and bad news, bad news is I had a clamp slip while I was working on it and thanks to the lack of a polarity diode it fried the chip.
Good news is setting it aside gave me time to build my RGBW_clicky light (however the only host I could come up with was a huge POS with a non serviceable switch) and that’s running well. The moonlight red works but the lowest steps of the ramp don’t but that’s my code so I can change it as needed.
Thanks for the suggestions guys but tomorrow morning when I go yell at the post office to find my SL-3 reflector I tried to send comfy (they lost it still in Columbus!) I’ll also be sending the light to tterev3 so he can have it in hand to determine the lowest possible PWM level, gonna use 50k resistors (I tried going up today, it helped a lot but not all the way then I killed it). I figure I’ll have him do it for 50k then should I have issues (with other emitters or anything else) in the future I’ll have a buffer to be able to add more resistance then.
Pic’s / video later, I’m freaking exhausted, I’ve probably spend close to 30 of the past 48 hours working on these two lights (plus I built 8 other AVR drivers for people / builds).
Oh one last thing, while he’s changing the other stuff he’s gonna turn off police strobe (why do I need that when I have pukelight) and make the locator beacon run from the UV channel so I can hook the switches LED up (used for a stock battery indicator) to the UV enable pin to be the locator!
Edit: Something interested I noticed, the SolarStorm switch uses a 4 pin plug that is a fenix connector, not like the same type of connector, the very same connector on all my dual side switch fenix’s, same PN and all.
This T4 is a nice light (tho I admit I didn’t so much as load batteries in it stock, driver looks like any other basic multi-emitter / SRK clone). I wonder what other parts could possible be sourced from / made at the same place that makes fenix parts. I also just got an SC02 and while its a very good quality light this T4 in downright impressive, I’m even 99% sure its type III ano!
Edit edit: I don’t care about the missing lowest ramp steps, I’m leaving this together as is for atleast 24 hours and showing at least one of my friends / clients before I tear back into it incase I break something again even tho I have more than enough spare parts to build 5 new ones lol.
Are the 50K/100K being used as gate resistors?? :O
Yes, sort of, but they play an additional role in that CC circuit... You could try to bring them down to 1k or even 470R, that should brighten up some lower modes I think.
Yes they’re a gate resistor and they definitely seem to be very important.
DrJones are you confident in those values? If so I’ll gladly give them a try but as I was experimenting yesterday moving down in resistance ~10k at a time the light stopped working all together at 40k but if you think its worth trying a 1k I’ll give that a shot.
I realized something last night, by now I’ve got several hundred bucks wrapped up in these RRBW projects (around $500 at the moment and more money to be spent before I have even the first light ready to ship), I could of bought one of every tier-one brand RGBW light on the market with that money…
Totally worth it tho!
I'm not sure if you will have the most toys when you die, but you sure have the best ones! 
Great work!!
Hey Dan, I have a spare Supfire A6, I would not mind building a quad optic with Dsche board and rgbw xpe2's in that. So are you going to sell build-up rgbw non-electronic-switch drivers with that great UI separately??
I would still love to see a schematic, to try and figure out why in the world anything would need (or even function with) such high gate resistance, and also what exactly is going on with the pairs of FETs(?) that appear to be running in series. :|