[MOD] Armytek Wizard Pro XHP50 to 4x Nichia E21A

Flashlight Modding and DIY Parts
21

Here’s the test result:

Being an R9080 the 4xE21A is dimmer than XHP50. But lower forward voltage makes for longer regulated and total runtime. At max current the difference was almost 0,5V. I adjusted the correction factor for lumen output because I forgot I moved the input hole to the side of my integrating box.

Armytek Wizard Pro WW
T1: 755 lumens
T2: 1394 lumens
M1: 26 lumens
M2: 138 lumens
M3: 333 lumens
FF1: 0,05 lumens
FF2: 0,36 lumens
FF3: 2,56 lumens

Armytek Wizard Pro + VR16SP4 (4 x E21A R9080)
T1: 591 lumens
T2: 1104 lumens
M1: 43 lumens
M2: 116 lumens
M3: 266 lumens
FF1: 0,095 lumens
FF2: 2,47 lumens
FF3: 4,4 lumens

As you can see from above. In lower current those E21As are more efficient but not so at higher current. In FF modes, the white paint also boost output. I didn’t believe my numbers too. So I opened a brand new Wizard Pro and rechecked it, yes it is. Unless I got two Wizard Pro with lower than average FF modes, then it is what it is.

- Clemence

22

Thanks for all the testing!
The output is rather low, but that way to be expected. The normal Wizard with cw XHP50 does 1800otf Lumens in T2 and 900 in T1 acccording to Armytek (yours seems to be a bit dim, even if ww).

23

The WW XHP50 Wizard Pro doesn’t actually achieve the OTF lumens Armytek claims. Zak reported it in his review last year, and my ceilingbounce shoebox results also agree with Clemence’s results to within the margin of error.

24

The drop in lumens is well justified and totally worth it for such a beautiful clean beam with 9080 LEDs vs the stock 70 CRI OEM.
Well done Clemence!

25

As promised:

- Clemence

26

Thanks! I believed you before ;).
Do you have a lux meter? My interest is in the higher modes, 1-4h.
If you don’t have one, you could also do a video outside against a wall from a few meters distance.

27

I’ll do it later this week. I also forgot to connect the LED switch wire. You can see it in my video, the button didn’t lit up, doh! :person_facepalming:
Yes I have a lux meter (TASI-632A), that’s how I did the measurement. The cd must be much lower since the quadtrix E21A has very broad beam, almost as wide as the spill. This is perfect for any close works and around trekking but not for runners like you.

EDIT:
I used Djozz standardized flashlight he sent me. My test result also not too far off those of Maukka’s and Zak’s. I don’t know how AT test their flaslights. IMHO it’s almost impossible to get 1800 OTF lumens using current driver and optic set. The bezel alone eats up ~5% off the output, AR coated glass ~2, TIR optic ~20.

- Clemence

28

Could you also do it with a protoype pCB if you have one lying around with LEDs on it?

29

UPDATE 180220:
4x E21A (6V + resistor mod) runtime

4x E21A on VR16SP4 in 6V 2s2p config (2x sm403 R9080 & 2x sm503 R9080)
Resistor changed from 0,05 Ohm to 0,025 Ohm (2x R050 stacked in parallel).

Without resistor mod, the output in 6V was roughly 70% (800-ish lumens) of the 12V (4x E21A). The output in T1 & T2 was perfectly fine but Main1, Main2, and Main3 modes became too low to be useful for daily uses.

I tried to get almost the same output as in 12V by further reducing the current sense resistor to 0,016 Ohm (triple stacking) but didn’t like the short regulated runtime for just a little extra output. Bypassing the current sense resistor made the driver behaved very similar to direct drive with almost little to no regulation (OEM 3200mAh battery). 0,025 Ohm is the best overall (output - runtime - temperature).

With 144AM R9050 and 0,025 Ohm resistor mod the max output should be very close to 1100 - 1200 lumens. But I prefer the E21A for it’s higher CRI and better efficiency at lower modes.

It’s interesting that the 6V mod doesn’t do much to add more regulated runtime. With 0,016 Ohm resistor, the output and runtime were closely matched the 12V, both using 4x E21A. Based on earlier observations on E21A R7000 behaviour: it’s very efficient up to 3A despite being a domeless, closely matched 144AM R7000 domed performance. I strongly suspect the problem is in the driver design. This driver is just too small to handle more than 15 watt output. Even the OEM 12V XHP50 can only deliver regulated max output for a brief minutes (under unrealistic water cooling). Bob McBob did a test at freezing 2,5C and the result still far from the water cooled one.

My suggestions for sensible high CRI R9050/R9080 modification outcomes are:

- 6V without resistor mod for longest regulated runtime. About 700ish lm (4x E21A) or 800ish lm (144AM) OTF output, dim Main modes, and extremely dim firefly modes.

- 6V with 0,025 ohm resistor mod for long regulated runtime. 971-1000 lm (4x E21A) or 1200 lm (144AM) OTF output and ideal Main and firefly modes. Additional benefit: doesn’t need to to use high drain battery - more extra capacity potential. Re-potting can be tricky, Loctite Stycast is a good option.

  • 12V + 4x E21A without resistor mod for better regulated runtime than OEM XHP50. 1000-1100 lm OTF output, slightly dimmer main modes (Main1 is slightly brighter), and much brighter firefly modes.

EDIT: forgot to include the output

Armytek Wizard Pro WW
T1: 755 lumens
T2: 1394 lumens
M1: 26 lumens
M2: 138 lumens
M3: 333 lumens
FF1: 0,05 lumens
FF2: 0,36 lumens
FF3: 2,56 lumens

Armytek Wizard Pro + VR16SP4 (4 x E21A 12V)
T1: 591 lumens
T2: 1104 lumens
M1: 43 lumens
M2: 116 lumens
M3: 266 lumens
FF1: 0,095 lumens
FF2: 2,47 lumens
FF3: 4,4 lumens

Armytek Wizard Pro + VR16SP4 (4 x E21A 6V) + 0,025 ohm resistor
T1: 523 lumens
T2: 977 lumens
M1: 32 lumens
M2: 91 lumens
M3: 217 lumens
FF1: 0,079 lumens
FF2: 0,74 lumens
FF3: 1,72 lumens

- Clemence

30

Another boost gain using proper optic: 1400 OTF lumens in R9080! 300 lm gained over stock TIR optic.
As the LED die size increase, TIR optic selection becomes more critical.
4x sm653 D240 E21A R9080 are certainly outputs lower than an XHP50 CRI 70, but combined with proper optics the OTF output is dramatic.


I measured the output of 4000K (D220), 4500K (2x 3000K D200 + 2x 4000K D220), and 6500K (D240). Then I interpolated the lowest possible number using the flux bin to get approximation of 2000K D160 output. The flux bins adhere closely to the measured output. D240 E21A gets much brighter than XHP50 at lower current.

Armytek Wizard Pro WW (OEM)
T2: 1394 lumens
T1: 755 lumens
M3: 333 lumens
M2: 138 lumens
M1: 26 lumens
FF3: 2,56 lumens
FF2: 0,36 lumens
FF1: 0,05 lumens

Armytek Wizard Pro WW 4xE21A (2000K D160 - 6500K D240)
T2: 903 - 1400 lm
T1: 475 - 758 lm
M3: 205 - 330lm
M2: 86 - 145 lm
M1: 18 - 50 lm
FF3: 2,5 - 5,4 lm
FF2: 0,5 - 2,8 lm
FF1: 0,02 - 0,095 lm

- Clemence

31

I’d like to understand why did you gain that 27%, but I fail so far…
How did you modify the optics?
Do I understand correctly that the emitting surface doesn’t fully fit in the lens opening? I guess opening is what you call aperture?
Assuming yes…there are just a fairly small corners that are under the flats. Even if light from them went straight through the optics without any loss…that doesn’t seem to be 27% of LES…and the OEM TIR doesn’t black it out, it should pass most of it forward as well.

32

I’m also wondering how this is happening. Unless my TASI 632A light meter messed up with the reading. Member Bob_mcbob got the same improvement much earlier than me. He got whooping increase too using the same optic I used on his Tiara and Wizard. But his gain was not as dramatic as these E21A.
Measured many times and got consistent results.

Yes the LES doesn’t fit inside the lens opening (aperture). The optic sits at 0,025mm above the LEDs. The other logical answer is the narrower emission angle of E21A.

- Clemence

33

Clemence, that Armytek Wizard Pro w/ 4xE21A beam looks fantastic! Can you sell them modded?

34

Currently (several hours ago) I only have three of them, but they’re all sold out before even hit the store. I have incoming Tiaras waiting at the custom.

- Clemence

35

I wonder how the non-pro perform. I don’t know why there aren’t the normal version review and mods

36

I just recently modded a non-pro Wizard because I want to use 3V LEDs. The process of modding it is the same.

The biggest difference are the missing LEDs under the button cap. This lead to them using the main LED for the low-voltage warning which I find dangerous when you are using it for running for example (it starts to blink on and off a few times).
The other difference is the U.I. It doesn’t have the two mode groups in addition to the turbo, only one.

37

I am too using the non-pro version so i know it characteristic, but i want to know does it have pot in the inside, the real lumen, does it stand true for what the company say it is. How is its efficiency, ….things like that. Will you share your modding work pics on the wizard noob :D?

38

The driver is potted. This is the same for all Wizards. I don’t know the efficiency, but it’s probably fine. Only the the XHP50 Wizard is inefficient when using a 12V LED.

Here are the pics of my Wizard XHP50 mod. The only difference to the non-pro Wizard is the LED (text is in German). Everythink else looks identical.

39

[quote=The_Driver]

I already saw this and didn’t reconise it was you :). Still a little upset for not seing the nonpro version mod making process.
Everything is identical like the driver, the thermal sensor and so on? So because the nonpro version use the 3V LED so the efficiency is ok right? God what a relief

Ich freue mich sehr dass ich nicht die pro version gekauft habe :partying_face:

40

The older (non pro, non ribbed) version driver is the same as the newer Tiara except for the UI. It’s more efficient than the XHP50 version especially in Turbo mode but not as bright. The real output is in 700ish lumens for the older XPL.

- Clemence