Skyray driver problem - help..

Strango, you’re da man!! I unbridged R2 and it worked!!! :beer: :student: I also ran a copper braid to the “-” terminal and braided the springs. The lux output went from original 750 to 950, which I think is pretty cool considering the simple mods. Even though LEDs are not Cree they have a nice warm white tint, so I’ll just leave this light as is. Here’s a few “after” pics…


Urw.
P.S. Do not forget to add thermalgrease.

Got the same exact driver in my recent 4X clone from Amazon. I got mine cranking out lumens -- like 4,000 lumens now, and stock was 1600 down to 1300 in 30 secs, with the stock driver!!

You need to restore R1 and R2, hopefully you didn't damage anything. The key amps bump method with this driver is simple:

  • See the crazy squiggle trace going from batt- (grnd) to the power input pin of the FET? Solder a jumper wire over the squiggle trace. Think I used 20 or 22 AWG teflon wire.What I did was scrape the silk screening off the trace where it's really close to the outer ring, then soldered the wire from that newly made pad direct to the FET pin, lower right pin.
  • add wire bypass's to all 4 springs. Drill holes thru the tail MCPCB so the bypass wires come out the top, then bridge those bypass wires direct to pad where it returns - by the screws I believe.
  • The totally crap fake LED's had to go, so I reflowed XM-L2 U2 3C's I happen to have laying around
  • do what you can to increase thermal path- I added thermal grease to where the MCPCB contacts the ledge of the housing

This is mine:

Big wire then added jumping from top pad to lower pad for measuring current.

Sorry, don't have the pic showing the jumper wire on the driver uploaded. Will post later.

I not say anything Tom. :smiley:

wow… I’ve never seen springs get modded like that. Thanks for sharing your pics, but I don’t really feel like reflowing LEDs since I’ve never done it and from what I’ve seen online it’s not that simple. It’s pretty crazy though that you got 4k lumens out of that driver… :open_mouth:

Ok. Sorry, didn't see strango's post details and your follow up, so seems ok, but could do a little more if you want. I replaced the LED wires with 22 AWG as well

Here's the missing pictures with more details of my mod.

This is the 16 AWG wire all set for tail amp measurments using a clamp meter:

This is actually a 26 AWG teflon coated wire (think I said 22 AWG or so earlier), but your braid probably works bout the same:

Inside, shows the white thermal grease I applied around the shelf:

Exact #'s I recorded for stock SecurityIng 4X light, guessing 6A-8A:

lumens: 1,632 @start, 1,350 @30 secs (17.3% drop)

After Full mods and UCL lens, reading of 14.0A on a clamp meter:

lumens: 4,352 @start, 3,958 @30 secs (9% drop), 35.5 kcd throw, measured from 5m (377 meters)

Tom E, have you considered an idea of replacing that 2SK4212 FET to smth with lower Rds ?
From datasheet (http://documentation.renesas.com/doc/DocumentServer/D19564EJ1V0DS00.pdf) I see it has Rds=14mOhm. Seems not very bad. I’m not sure if it worth to be replaced or not.

And have you recognized that unknown MCU ?

I wonder if it can be easily replaced with attiny13A and some FW with tuneable mods and higher PWM frequency in low/mid ?

Yea, thought of using a better FET and looked into replacing the MCU, but Atmel's are pin #4 grnd, pin #8 Vcc, while PIC's are exactly opposite I believe. So easy way to tell is check the grnd and Vcc pins usually easy to find. If matches to an Atmel, possible to replace it then. Think this one is PIC configured, so not easy.

Even for the FET improvement, 14A that I'm at now is pushing this really poor MCPCB. Probably higher amps will start kill'n LED's, even the better XM-L2's I got on there now. DTP makes a huge difference, starts to show measurably by 2A per LED or less even, and pushing over 3 to 3.5A per LED, you start running at risk. That's for a single LED, but this setup for 3, 4, or more LED's could be worse because the heat of one effects the others. Lots of nasty things can occur. I had an old C8 with the old good DD driver, like a HD2010. Dropped in a good cell, and the solder under the LED started melting, and I tightened the bezel, it shorted out the LED by moving on the pads and blew out.

Dale did this recently with a 12 LED J20 - LED's started dropping at high amps, high output. Not so much the LED's themselves is the problem, but inability to pull the heat away from them effectively. Direct Thermal Path MCPCB's solve all that.

Ok, I see.
Will be at home in few hours and post photos of custom pill for this torch. Mine is 3 Led, and as I use it with stock mcpcb I think 3-3.5A per LED will be OK.
Btw what is the relation between Hi and Low mods here ?

Good guess is they PWM low mode, based on the simplistic design. Looks like R2 is hanging on the FET gate, but if you put a scope on the signal going into the FET gate, you should see the actual PWM rate they use, which is probably relatively low for what we use in our custom drivers. My 85 based boards do like 15 KHz (can go as high as 30 KHz), 13A's can do a min of 4 KHz.

I suspect the PIC or PIC clones out there are cheaper tham Atmel's, because I rarely see an Atmel in a budget light.

Sharpie - yes, all good. Stock MCPCB good up to a certain point, but even so, not so efficient, so you are getting less lumens for same amps, and if you are doing 10A across 3 LED's, 3.3A or so per LED, hhmm, maybe 20-25% or so? Dunno - think we charted this stuff in the past, but also very dependent on specific MCPCB's. Think my biggest bottleneck were those fake LED's - wow, there's some bad ones out there...

Guys, would you be willing to make some outdoor beamshots of your modded SRKs (maybe with a few other lights for comparison) I think it would be pretty cool if we could see them in action side by side.

Sorry, not easy for me at all.... Not much open spaces around me here on the island.

Tom, going back to your pics, what is the purpose of soldering springs to the back side of the plate? Have you compared reading of that mod to “standard” spring braiding?

Well, I'm not soldering the springs, I'm just continuing the heavy gauge bypass wires thru to the bottom of the plate. This results in 100% total elimination of board traces for carrying the signal. No - didn't measure those specific path differences, but it just makes sense and eliminates the unknowns of the specific flashlight design of spring pad traces. Some may be excellent, some very poor. The guys who designed could care less we are using 4-5 times the current they thought it would be carrying, so who knows what they did there.

We just eliminate all doubt with this method. I used to use braids and found they were breaking. You have to be very careful in soldering them, and use better quality braid. I prefer 22 AWG or preferably 20 AWG wires, if they fit ok. Sometimes I'll use 24 AWG for small springs. Never heard or saw a wire bypass breaking.

Years back when I was cranking out Super Shocker mods, I got to a point of bypassing every trace on the battery carrier, and I did increase amps, measurably.

Came home an decided to do the same simplest mod.
Led wires - 18AWG
Forgot to measure current before and after but looks a bit brighter now (still worse than stock roche m170 with “green” driver ).
I wonder if I have the same fake cree leds (LB ?).

Powered by 4 protected UR18650ZY
With that custom pill body gets hot fast.

Ambient: 21°C
HI mode (measured with infrared thermometer in area near sidebutton):

  • 30sec: 24
  • 1min: 27
  • 1min30sec: 30
  • 2min: 32.5
  • 3min: 36
  • 4min: 38
  • 5min: 40 (warm)
  • 10min: 50 (hot)

3LED

Body has been (dont know correct word for this sorry :person_facepalming: , extended? tneh grinded) for beter contact with pill

Al pill, sits pretty tight

Interesting. Yea, been using DC-Fix on some light myself. Didn't think of trying it on a SRK - interesting... My 5X SRK clone has crazy ring patterns in the beam.

I work with an EE doing PIC designs on battery powered assemblies, an add-on product we sell. Didn't look myself into it, but he said he built in power saving sleep mode support in the firmware.

strango - you might have real XM-L2's there, not sure. It's easy for me to spot the XML fakes which is what I had, but the XM-L2 fakes I dunno.

I gess No.
with more leds current through 1 led is lower.
so efficiency lm/W is better
See example attached

With less LEDs you have more W consumption and less lm output for the same total current.

Wait... Where is this app you got there? Pretty cool, actually, but where?

Is this real or predicted some how? Didn't think CREE would show XM-L2's up to 9A

lumens/watt is where multi-leds have gains, for sure.

All the best for you Tom E
that screenshot is from LedCalc - LEDs comparison (http://ledcalc.fonarevka.ru)
a looot of LED types.

There is an offline app for windows available as well but it has no english interface.

I believe it is not predicted but based on afficial pdf-s

9A is total current for X*LED in parallel.