Safe, as long as the cells are matched - brand/size and age. Since they are in parallel, 2 cells will drain quicker than 3 and maybe get slightly lower output #'s.
The SIRA20DP looks quite good as well, both are very high quality choices.
The real confusion for me is why this FET is having issues as it is very over-rated for the task.
Those amp ratings on FET's is deceptive - perfect world type of specs. One spec to look at is the amps mounted on a 1" x 1" FR4 board, more real world for us. The Infineon is like 40 amps, the SIR178DP is 84-100 amps, and the SIRA20DP is 65-81 amps.
Another spec that I think is important for our use is RDS(on) - the SIR178DP is the best of these 3 on that as well. There's a few more impressive specs there, though I'm not that knowledgeable if they benefit us in our application. The max amp rating though is insane! Other ones I looked at like +100 amps have other issues, like slow speeds or high resistance.
Hhmm, maybe we should look at the gate input carefully. It's gotta be something borderline because some drivers don't have the glow, others do. I should restore the Infineon on the proto and do some heat tests.
I took mine apart. Unfortunately, my wires weren’t quite long enough to flip the driver over for desoldering, so I could only take some multimeter readings.
1. As Tom reported out, it’s definitely leaking through the driver somewhere. I also powered my driver while floating in the air without the retaining ring, and I still got glow, meaning there’s no significant leakage path through the MCPCB/body.
2. I cleaned the driver with IPA and still got glow, so I don’t think it’s a flux issue.
3. The drain-source voltage for both big and little FETS is 1.825 V when the driver is powered with light off. Gate voltage is right on 0 V.
Here’s what’s interesting, the big FET’s datasheet posted by Artiet earlier indicates a max zero gate voltage drain current of 0.5 mA (room temp, V_DS = 20 V), which is surprisingly high, imo. Now, I know 1.825 V is much less than 20 V, but what if we’re really leaking tiny amounts of current while the FET is in depletion?
R_DS(on) is the most important one I can think of for flashlight applications, but even 1 mOhm vs 0.5 mOhm isn’t really a deal-breaker. At steady state, we operate well within the current and thermal envelopes for the FETs mentioned, and transient current spikes have their own, higher limits published in the datasheet. I believe most FETs we’re interested in are also plenty fast with respect to switching time, especially if the PWM is only at 10 or 20 kHz. I would imagine that cost and footprint (and now maybe also max zero gate voltage drain current), matter way more.
EDIT: I Do have the "warm and cozy" (trademarked by TomE) LED glow. It is So Faint that even when I went into a dark room this afternoon and cupped my hands around the reflector I couldn't see it, it literally had to be pitch dark outside for me to tell.
some initial takeaways of this light:
-this light is beautifully machined, balanced in terms of weight, and size
- I do not need a handle (xl- xxl glove size hands), it is comfortable to hold and not too big
- the battery tube machining is good for grip, even though I didn't think it would be
- i don't care the LED glows, i absolutely enjoy this light, and have no desire to fix it unless it is unbelievably easy and I can do it, the light is so good I could care less about the led glow. The switch glow is 10x brighter anyway
- this thing THROWS, so it is my first large head size thrower and I'm totally impressed! Amazing
- the heat management is AMAZING. Probably the second most impressive thing to me. I had it in turbo for 2 minutes, it didn't get hot. It's so lightweight I was weighting for the moment but it didn't happen. My astrolux mf02sv2 gets hot in about 25 seconds on turbo and VERY hot in a minute.
- there is no "blue" to led in turbo. Both my mf02sv2 and k1 sbt90.2 turn noticeably blue in turbo, a clear tint shift from the high mode. This light does not tint shift from top of ramp to turbo, or it is so m8niscule I can't se it. I was impressed bc it looks brighter than my mf02sv2 also. That's awesome!
- if there is a negative about this light I haven't found it yet. I was running it for a half hour. It's awesome. The price is awesome, the finish is awesome. The heat Mgmt awesome .
- very happy! Thanks wildtrail and TA and luxP, tomE, and everyone else involved
That zero gate voltage drain spec is much higher than the SIR Vishay family of FET's we/I use - SIR800DP and SIRA20DP, and also much higher than the Infineon FET I have and am using: the BSC009NE2LS5, https://www.infineon.com/dgdl/Infineon-BSC009NE2LS5-DataSheet-v02_01-EN.pdf?fileId=5546d4624bcaebcf014c096c7c152310
Here's an explanation: https://www.tek.com/support/faqs/how-can-i-test-mosfet-zero-gate-voltage-drain-current-my-curve-tracer, not sure I understand it all but this could be the possible cause of the glow. This spec of this particular FET model is way off from other known good FET's, so maybe?
Someone more savvy/knowledgeable of FETs could probably know for sure.
In fact the Infineon I use (bought from Arrow), model: BSC009NE2LS5 appears to have better specs than the variant in the WT90, model: BSC009NE2LS5I. It's better in amps and RDS(on).
Just looking at the zero gate voltage drain, the BSC009NE2LS5I's is 500 times higher than the BSC009NE2LS5 (500 uA compared to 1 uA at room temp)!
Hhhmmm..........
Thanks, very good info and I agree with you, as I was alluding to earlier I have a feeling that the FET is getting a bit of leakage and that is leading to it leaking a bit of current to the LED.
I will admit that since I have not really been active in the flashlight world the last few years I am a bit behind on the latest FET’s so I stuck to the last “best FET’s” recommendations I had.
The SIRA20DP does indeed look superior I was simply not aware of it (also the driver specs were finished over 1.5 years ago, was it even out back then?).
You are correct that the R_DS is most important for flashlights followed closely by the voltage required to get it. Some FET’s need 10v+ to get the low resistance values but all our lights work off 5v. There have been a few cases where lower resistance FET’s performed worse in our lights due to that.
As I said above, the SIRA20DP is a superior FET, I was simply not aware of it 1.5 years ago when this driver was designed (or until you mentioned it honestly).
Have you tried installing it onto the driver to see if it glows?
Mentioned already above somewhere - replaced the FET with a SIRA20DP and no glow, though did notice it glows after turning the light OFF, but it fades to completely no glow in a few seconds.
Lexel sticks with an Infineon (think different model than I use, not sure). And the SIR800DP is used on many lights as well, thinking this was from your recommendations.
I have been using/recommending the SIRA20DP probably since 2018. Here it dates back to May 2017: https://budgetlightforum.com/t/-/45831
Here's maybe my first post about them in Oct 2017: https://budgetlightforum.com/t/-/46241/1032
I do get excited by a new good spec FET 
lol, sorry having trouble keeping track of everything.
I can have them update the FET for any future drivers produced but that won’t change anything for the existing drivers and honestly not worth them messing with it since it has no real effect except OCD for some people.
Anyone that is annoyed by it though knows the fix, swap out the FET for the SIRA20DP.
Yes, the SIR800/400 were what I used almost exclusively up until the GT4 when I decided to “upgrade” to the best FET people told me about at the time lol.
I like new and better FET’s as well, you get more gains now days from them then you do from LED’s :person_facepalming:
Gains with no cost I like, and it's all about efficiency - whether FETs, LEDs, AR lenses, or even solar panels.
Recently looked into home solar from one vendor, and I'm not excited al all going with panels that are 14-15% efficiency when 22% panels are available now, and not much more in cost. Of course you have to weigh all the factors but too many pushy sales goin on that gives me the feeling of scam.
Not real up to date on the latest solar tech, we have too many trees to make it practical here, considered it many years ago.
I know that the efficiency is always climbing and I agree, better off paying a bit more now for a higher quality panel. Course as I have gone through life I have come to realize that this applies to almost everything.
Better off spending more and getting something good now then having to re-buy it later when the cheap option breaks. I generally can’t afford the high quality options so had a LOT of cheap stuff break only to realize later that it would of been cheaper to buy better at first.
That said, I also shy away from the tippy top performance items as they generally have a heavy early adopter / technology tax that doesn’t make sense. I find the best is generally 1 or 2 steps down from the best of the best.
FET’s and LED’s are nice in that they are cheap enough you can just get the best and not worry about it lol.
I had to be in a pitch dark room to find out mine has the cosy warm glow as well.
Hard to catch on camera.
Picture is a bit over exposed, in reality you have to try very hard to see it.
Your sideswitch looks to be quite bright, mine is barely visible in daylight, even indoors.
At night it is OK though.
I thought it was bright too, I mean I like it just struck me. I didn't adjust it.
something wierd happened thought- maybe someone can let me know what I maybe missing-
about a month ago I bought (3) 40t3's for the wt90. I charged them and they've been sitting in the plastic case. I threw em in my WT90 yesterday and played for a half hour or so. I electronically lock the light out everytime I'm done using it. I have the DIMMEST led glow, bright switch.
I went to use the login again this morning at like 4 am and noticed it was not as bright as my mf02sv2 and also my WT90 wasn't getting warm at all, after 4 minutes turbo. I did 3-click batter check and got 3 bolts on my 40t. I threw them on my Xtar Vc4s and this afternoon they charged 3800mah (meaning they were at like 200mah when I put them on the charger this morning??)
now, there is completely the chance that i never charged them when I received them in the mail last month, and I hope that is what happened. But if not, there is something really wierd going on here, right??
I threw in my 3 (fully charged ) p42a this morning, and made sure i knew they were at 4.2v.
im going to track the drain of this set, see if it happens again? Just thought I'd mention it. Thanks!
Lets hope you forgot to charge them!
If they drained in a day from a full charge something would of got really hot, that is a lot of power to dump into the light without it being lit. I am guessing you forgot to charge them but keep an eye on it this time around to be sure.
TA - will Do, thanks for that info. You are right, I didn't think about that. I keep in in a small tool bag with plastic covering foam, something probably would've melted a little maybe.
yokiamy - I agree!
im keeping an eye on the P42a, and will throw the 40t back in too in next few days. Thanks for the responses.
The flashlight arrived today, just got it in my hands.
For what I see it works perfect, (will check for the «warm and cozy glow».) 
Machining and all external is also very nice.
Everything has been said already by others, and I just want to support their statements that this is a very nice flashlight.
KUDOS to WildTrail for the making. 
Thank you TA for all “for all the fun” you must have had with bringing this together for all of us.
Also thank you *to all others *for supporting with testing of the light, writing the manual, etc etc. (no name mentioned, none forgotten.)
Hope everyone else is happy with their light.
(Now I will just have to wait a few months before I really can start testing out the light.
That is the downside of living in the land of the midnight sun. :-))
For what its worth… I received my light about a week ago. I quite like it as well. Build quality seems very good.