Thanks for nice comment - it was all possible because you make excellent tests of the emitters!
Also, looking forward to your test of the XP-G3
18 A on a XP-G3 triple sounds insane, just about ~5000 LED lumens in theory. I want to build a triple. I was tipping towards the 219C, simply because I thought that an âaverageâ triple 219C build would reach about 10-13 A - at which currents I can live with the downsides of the 219C compared to the XP-G3. But if it will be difficult to limit the DD draw to âjustâ 13A or less with the nichias, I canât justify using the 219C.
What do you think? Average soldering, triple 219C and H17F driver (sub-optimal FET), would that stay around or below 13A? I guess I can adjust the draw current by changing the wire AWG and the 18650 cell.
My XP-G3 triple is making around 3400 lumens but they are 90+ CRI LEDâs and not top bin. I could see over 4000 lumens if they were top bin 70 cri.
You can limit the current on a FET driver by limiting the duty cycle of the FET, I limit it to around 75% with a single 219C in order to keep the current in check.
Although like you said cell selection and wire size can play a big role as well. With 24AWG wires and say a GA cell it should be around 12-15A I would guess but it would drop quickly as the cell depleats. It could even be lower then that with 24awg, I usually use 20awg so I donât have a lot of experience with thinner wires.
I just built a 4000K 219C D240 triple (from Kaidomain) in a S2+ with thin 28AWG. On a fully charged, 1 year old Panasonic NCR18650FF the light pulls 6.25A. both springs bypassed with 18AWG wires, the light is quite usable with this setup. Nice tint as well.
edit
driver is FET only driver by wight, with a NXP FET PSMN0R9-25YLC
I ordered a quad 219c 4000k high cri = ) from a member here,
For my M1, I am hoping to get some serious amount of high quality light, with mtn max fet driver, and a 30Q
Apologies, yes - I misspelled the battery. ncr18650PF.
At this point I didnât drill a bypass hole for the bat+ cable through the driver yet, will do that later with another driver (because 18AWG makes the spring almost incapable of any compression and that is an issue). tailcap spring is bypassed with 18AWG and goes through the board to the switch contact (needed to put a piece of plastic on the other side of the switch so it would be level again).
I just built a few triples today, 2 X6âs and an S8.
The Triple 219Câs with 26awg wire still pull ~15A with a high drain cell. Also 26AWG IS NOT recommended. You can get away with it on current batteries because the current drops before the wire can melt but tested on my bench power supply it started smoking after a few minutes at 15A. I would not go thinner then 24AWG with the newer cells that are coming out.
Now if I use a GA instead of a high drain cell the current drops to a bit over 10A. This is still WAY too much for the tube lights and even far too much for the X6.
Even with an old laptop pull it was still doing around 9A (which is far too much for that cell, it was quite warm when I pulled it out).
Now simply reducing the duty down to ~75% will drop the current to a reasonable level. Or simply donât go all the way to turbo until the voltage drops some.
I also built an XP-L HI triple and it only pulls 12A by comparison and puts out more lumens but the CRI is far worse.
The XP-L2 HI if it ever comes out would be amazing. I want to try slicing the dome on the XP-L2.
I have 5 XP-L2âs on the way with 4 destined for a quad. Not sure what to do with the 5th one. Perhaps Iâll try slicing it⌠but what to put it in? Perhaps a C8?
I am thinking about putting 3 XP-L2âs in an M6 and when I get a Q8 I think it will also get the L2 treatment.
C8 sized reflectors look pretty good with the L2, sliced it should be interesting if nothing else. Plus you need at least a C8 light to handle the heat.
If I like how the beam is, perhaps a Convoy L2 with a FET, single cell tube and a high drain 26650 would make for a nice pairing. And would have enough mass to handle a bit more heat.
Ugh, Iâm going to need a bigger flashlight shelf.
I wonder if software could do this reducing automatically? That would be great for these high amp lights, wouldnât it? Say for voltage above 3.85 high output is %75 duty cycle. And for under 3.85 it goes to %100 or whatever combination of voltage and duty cycle one might prefer. Could this be implemented into the code with a simple if/else check in the loop?
Edit: speaking of the fet channel only. However, there could be a percent of duty cycle check against the fet no matter the mode group ramp level.
As long as there is voltage monitoring this should be rather easy to implement.
Iâd go for something a little more advanced myself. A little formula that lowers the output the closer to 4.2V it gets. It wouldnât cost that many extra bytes.
Hmm? This would allow fet drivers to power a single 219c cri with out over driving it and also adding longer output regulation as the battery drains. I donât think Iâm familiar enough with code to handle figuring out how to do it though. I would need some help.
+1
There is no other way that using buck converters.
All next leds will have lower and lower Vf. All next cells will have higher voltage and lower internal resistance. Unregulated fet drivers will burn leds too fast, neither with or without PWM.
LD-3, regulated no pwm. Linear not buck. Uses sense resistor to monitor current and adjusts gate voltage on an FET to maintain constant current. Problem? Proprietary UI so no reflashing, no bistro, no guppydrv.
With 1v difference it will burn or step down very fast. Such boards can heatspread 2-3W, with 1v difference between cell voltage and Vf it should giveaway same number of watts as current (6A=6W, 9A=9W, 12A=12W). And all this processes with ~75% battery using effincy.
The L2 would pair very nicely with it, just a more pricey host for a build. I built an XHP35 L2 recently with silular heat and while it does get hot it is manageable.
I have wanted to get a current compensation loop or âsoftware regulationâ setup going for the TA drivers for some time but since TK is MIA and no one else has really been able to or willing to work on bistro nothing has come of it.
It would not be hard, simply use the existing voltage check and table and then bias the output PWM based on the cell voltage. It would take a bit of tweaking but would work good enough to give you a reasonable regulation via PWM and software.
This is wrong way. Dont forget that you can kill led from big current without overheating. Just one day you will have such combination of led and cell that would give too much pick current, neither with or without pwm.
While you canât PWM 2S voltage and it work with a 3V LED the way things have worked out so far PWM current limiting works fine for all the LEDâs currently on the market.
All of the LEDâs that would be killed with too high of a current have a high Vf and can not pull enough current from a single cell to damage themselves (they also do not need the PWM limited, it would be simply for lower modes to have some âregulationâ).
All of the new LEDâs that have low Vf and need the current limited are tough enough to not die even with extreme amounts of power. I have tested the 219C to over 12A and the XP-L2 to 15A, both were still working after the tests. So it is highly unlikely that you will damage the LEDâs by PWM them with current LED or cell tech.
Down the road as things improve that might change but for now it is a good option.
Sadly we have no other options for most of the lights we like to build. buck and boost drivers are too large and expensive for what we do in most cases and do not offer enough current for most of us.
Linear drivers can not handle the latest gen emitters without overheating.
The only thing left is PWM an FET driver. It is not ideal but there are no other options.
