Yeah, it works out of the light, other than timing and voltage being off. I think I got the 22 value off your oshpark description some time ago, but it says 19.1 here. I understand it throwing timing off. I don't really understand it throwing voltage off that much though. I cleaned up the r2 soldering just to be sure.
so 4.7/(4.7+19.1)*4.2 is .83 V with 22 it's 0.74V Not a 3x difference. Something else may be wrong still. I have to wonder if 1.1 is just a pegged response, like the code says for i = 1 to int(voltage) { flash} , written so as to always hit the loop once anyway.
Yeah, something is up with the voltage reading, it is usually about .2v off with the 22k.
I started out using 22k but swapped back to 19.1 due to the OTC.
The latest parts list is always in the OP here. Too hard to update all the oshpark boards.
I'll check voltages later. Either the voltage on the sense pin isn't what it should be, or the attiny is busted. I was too quick to dismiss the voltage thing and put it back toghether, but yeah, I think that's really not right regardless of 19 or 22. In fact, it seems like the divider isn't hooked up at all. Like R1 is disconnected.
or better yet, like the attiny pin is not connected well.
This is very possible. I would reflow the entire driver before you do much more, it has fixed things more often then you might think.
I have no idea what the command should be. Could you list the individual command for each one.
The easiest thing would be to simply rename the file you want to flash to bistro-texas-avenger.hex and copy it into the AVRdude folder.
Then simply run the flash25bistro.bat again. It will flash the renamed Hex instead of the default version.
I think this was the thread that we discussed throttling FET’s to control current to an XHP35?
I just tested 4 XHP35 HI in parallel with a simple “oldschool” FET driver using the NXP FET. It was run off 4 30Q’s freshly charged, with 22awg wiring that was a little longer than necessary. .
I programmed the driver with many levels to see when I needed to cut it off for 10amp total draw (2.5A per emitter). To my surprise, on 255 pwm it netted ~8.5amps total. That could certainly be increased with better wiring and a better FET, but these emitters might not need as much throttling as we thought when used in parallel.
Okay, I put each one in a separate folder with the right name. Then renamed each file to bistro-texas-avenger.hex. I just drag the one I want from its folder and drop it into the "avrdude Simply run flash25bistro.bat from command line" folder. when its done I just drag the hex back to its folder.
It worked. Thanks for getting me to this point. That was way easier than I expected.
There's still a lot more that I want to figure out. Like those calibration files. What, where, when and why?
Can you measure Vf at the star while on? Maybe your connections are worse than you think.
Perhaps a little later. This test was done essentially floating above my bench top, so I only dare having it on for a second
From the OP
This is the first link provided, and the only hex there is from 2015. It seems to be standard bistro, which by the way is what I'm using at the moment. I wouldn't think that should be an issue for testing other than not fully utilizing the two 7135 channels, no?
Maybe it's better to remove this confusing link?
Yep, I ran a detailed test last week that showed much the same thing, I tested both PWM and constant current and compared it in the light box to see the efficiency change.
I am waiting to post the results until I get some more XHP35’s in that KB is sending me for some more testing.
The long and short of it though is that the XHP35 is tougher then we thought. It does handle PWM from an FET driver and it is surprisingly efficiency down to around 50% duty.
The low modes are the issue, while it will handle the FET down there it is a lot of extra stress and efficiency drops the lower you go.
I will post up all the details once I test these other XHP35’s.
I can say that 7135’s don’t work with the xhp35 though lol.
For you th ONLY thing you care about is the hex files and AVRdude. This is what 95% of people will do and why I desired it to be setup like it is.
The code, calibration files and all that other stuff is strictly for those that want to edit the raw code and compile it themselves (FAR FAR FAR more in depth and skill needed). It will do exactly nothing for you to mess with them unless you also recompile the firmware.
I have these measurements and will be posting them once I test the rest of the LED’s later this week. The final Vf for a single low bin C4 90+ CRI xhp35 HD was around ~15.xV @ 3A subtracting wiring losses.
It looks like TK moved the tripledown into a subfolder: ~toykeeper/flashlight-firmware/tiny25 : files for revision 217
Although that is standard bistro, my edit of it is posted down lower and it is what I recomend you using. That could easily explain your issues. Try flashing the hex from my download.
Nope, my issues were R1 was shorted. Just had to probe it out and it was obvious. As you said though, simply zapping it with the hot air for a few seconds fixed it. Now it all works perfectly other than the calibration being off and OTC is a bit slow. Probably just over a second to get a middle click. I can live with that. I'll get some 19.1k resistors though.
Now that I'm setup I don't mind playing with software but probably won't have a chance for a couple of months anyway.
I saw trippledown there too and guessed that might be it, but there was no hex, so I kind of said what the heck and flashed the regular hex. It does work fine too and for this light's use, I probably won't even miss the extra 7135's much, but I'll reflash it at some point.
Great, now that the hardware is working flash the Texas avenger version of the firmware to take advantage of the 3rd channel.