Review: UltraFire T90 SST-90 3x26650 Flashlight from Wallbuys (Picture Heavy)

Review: UltraFire T90 SST-90 3x26650 Flashlight

Relic's Rating: ★★★★

Wallbuys recently added this light to their inventory and after some debate I decided to order it. Wallbuys shipped it the next day and it arrived at my mailbox 11 days later.

Link to product

The package included the light and three 18650 adapters. The strap and accessory bag was missing.

Update: I contacted Wallbuys via PM regarding the missing items and Cherry replied within an hour informing me that the missing item was retrieved from the supplier and will be shipped before they break for holidays (I just checked and it has shipped with tracking number). I'll add my review of the strap and bag when it arrives.

The exterior packaging was good, a cardboard box covered in bubble wrap and completely sealed with tape. However, inside was an unfortunately familiar story. The main body and head were in a bubble bag next to the extension tube. The flimsy bubble bag tore and the parts were banging into each other. the result is a few knicks and scratches in the ano. My Tr-3T6 from DX was in a similar package. I believe this is how they come from the manufacturer, and the sellers do not open the box to repack the items. This would be a good idea, Wallbuys and any other sellers listening. Before you ship a light, open the box and wrap the parts in another layer so they cannot get into each other. It might save you from a few returns. I will not return mine based on some scratches, but others might.

OK, on to the review.

My first impression when I opened the box; this thing is huge! I mean huge! Let's compare to a familiar light, the HD2010:

The HD2010 barely reaches up to the T90 body! And from the front:

(OK, I put the Tank007 E09 in there for fun)

There's a significant jump in reflector size over the HD2010.

On the other end is a monster of a tailcap. Even with the length and being top-heavy, this light will tailstand on a flat surface:

Compared to an SK68:

Completely useless fact: The T90 tailcap weighs the same as an SK68; just to add a little more perspective. :)

Overall, the machining is very good and the ano is decent. There are a few scratches from shipping, and I would expect a few more from normal use. It does not feel like HA-III grade ano. The threads are OK, trapezoidal and fairly smooth. The tailcap threads are good, well lubed and feature double O-rings. Other threads were dry but smooth once lubed. There are dual O-rings on the main body tube, however this light does not support a two cell configuration. O-rings are in place at each joint, however two locations are inadequate meaning the light is splash-proof only as shipped.

Moving on to the tear down:

The head is so big it breaks down into three pieces (the reflector cone is removable too). An unprotected 26650 is shown on the right.

Update: Feb. 8 2013: I just received my TF Flame Protected 26650 cells from Fasttech and can confirm that they will not fit. not even close. I cannot even get the tailcap to the first O-ring. With one TF protected and two TF unprotected I can just get it to close over the last Oring, but I didn't want to try going for completely closed. It may be possible by modding the driver cover so the center portion is removed allowing the battery to directly contact the driver spring. That may be just enough space to allow these cells in. For reference, the protected cells are 68.5mm long and the unprotected cells are 66.6mm long.

Tearing the head down further (I got lazy on my shots from yesterday, no light box):

And the rest:

The head is a one piece with integrated pill. This makes emitter access a pain, but I think it's a good thermal design decision. That's a 40W emitter; there's a lot of heat to deal with. Based on the ample heat sinking, it should come as no surprise that heat is very well managed here.

Looking closer in the emitter end of the head reveals the first issue; workmanship in this area is not the best.

I'm not sure what the heck happened here. Whoever put this one together either had some bad luck or really needed to go. There are two things wrong here.

First, the screw on the right stuck up too much, so what to do; get out the die grinder and shave the head down. OK. Must have run out of low profile screws that day.

The second problem; the LED was not reflowed on the center of the star. Of course, the star is perfectly centered in the head, so the end result is an out-of-center emitter. To make matters worse, that's Fujik under the emitter. I suspected this when looking at the initial beam pattern.

Ultimately I did reposition the star so the emitter was centered (see mod in first reply for details). After this, the beam looked a lot better.

here is the front after centering (sorry, no before shot):

OK, so at this point, I'm starting to get worried about what is on the other end of the head; the driver. let's take a look:

The driver is press fit into place with a brass ring and plastic cover (not shown, see below). This made it difficult to remove. With some prying and poking it popped out.

My first reaction was, "That's it? This is supposed to drive an SST-90?". My TR-3T6 has a way larger driver. Based on my testing, physical size isn't everything; this is a very good driver.

At stock it pumps seven Amps into the SST-90. This is the highest I've seen. Most reports of the TF X6 driver are around five Amps at stock.

Efficiency was very good too at 86%. My spirits lifted after these test results; there is hope for a resistor mod (Check the first reply, you will not be disappointed).

before I forget, here is the driver cover, Battery side then driver side:

The driver has a rather small spring that contacts the center piece in the cover. I added some copper braid to deal with the relatively high battery current (especially after mods). The tailcap was also improved dramatically with a little copper braid. At stock, I got 260mV at 3A. After the braid, it dropped to 64mV at 3A. The switch itself seems to be very good.

I used three unprotected 26650 cells in this light. I do not have enough protected cells to determine if they will work. It seems a bit loose with unprotected, so there may be hope. I may add a nice short strong spring to the driver cover to reinforce the cell contact.

Indoor Beamshots:

Here are some beam shots on a white'ish wall from 1m away. Camera settings were f/8, 1/30s, ISO 200. The black dots are 16" apart horizontally, 12" apart vertically. The spill is good for this kind of light; low enough to not distract, bright enough for close-up visibility. This reflector is quite smooth (one minor imperfection not affecting the beam) and there are no visible rings. With an emitter this size, there is a large four-petal flower for a hotspot. The hotspot is quite large (maybe slightly larger than the HD2010 spot) which I find more useful than the tiny C8 XR-E spots.

High:

Medium:

Low:

There is noticeable PWM in Medium and Low. Medium is about 50%, while Low is about 10%.

Dimensions:

  • Overall Length: 354mm
  • Bezel Diameter: 79.4mm
  • Head Heat Sink Diameter: 62mm
  • Neck Heat Sink Diameter: 54mm
  • Body Diameter: 35.6mm
  • Tailcap Ring Diameter: 46mm
  • Tailcap Diameter: 37mm
  • Reflector Inner Diameter: 67.5mm
  • Reflector Outer Diameter: 73mm
  • Reflector Depth: 51.8mm
  • Reflector Emitter hole Diameter: 12.4mm
  • Lens Diameter:73.5mm
  • Lens Thickness: 2.45mm

Weights (without batteries):

  • Overall: 768g
  • Head: 498g
  • Neck: 62g
  • Body Tube: 76g
  • Extension Tube: 65g
  • Tailcap: 66g

Performance (stock, 12.6V supply):

  • Light Output: ~1200 lumens
  • Beam Intensity: ~50kcd
  • Power In: 12.6V * 2.54A = 31.9W
  • Power Out: 3.86V * 7.1A = 27.4W
  • Driver Efficiency: 86%

Power Source Options: 3x26650 (recommended) or 3x16650, unprotected cells only. Protected cells are not likely to fit.

Switch type: tailcap, reverse-clicky

Modes: High, Medium, Low, Strobe, SOS

Mode Memory: Short-off mode memory (while on, quickly half-press the switch to move to the next mode)

Thermal Management:

Timed measurements were not taken before the mod (I just couldn't wait). On a 20 minute run at ambient of 19C with no air flow, the head quickly heated up to 38C and slowly continued to 42C before the test was stopped. Measurement of the emitter star immediately after turning off gave a temperature of 43C. Thermal management is very good; no surprise there. there's almost one pound of aluminum just in the head. Even with that, external temperature starts to rise within 15-20 seconds of turning on. Heat is conducted away from the emitter very well.

Driver thermal management is also very good. The large brass ring does a decent job of pulling heat away from the PCB. Still, the driver FET and diode get quite hot after a five minute run (FET measured at 95C, diode not far behind) in open air. Any resistor mod should be paired with a heat sink on these critical components (see my mod entry below).

Conclusions:

I like this light. I had some ups and downs when I first started looking at it (I like the look (and it's huge), poor workmanship in emitter area, excellent driver). After fixing the emitter alignment (hopefully I just got a bad one) I like the beam pattern. To be honest, I didn't get a remarkable increase in throw after alignment (It went from 46kcd to 50kcd), so a minor misalignment is more of an aesthetics thing.

Relic Recommended, especially for modders.

Thanks for reading! searchID8934

Mod 1: Driver Resistor Mod for 9-11 Amps
Disclaimer: Every light will have minor variations the could result in big changes in the drive current. I recommend you measure your emitter current before and after making mods. Do not rely on my test results alone; they may differ significantly from yours.
OK, time for some fun. I was pleased to see the stock power and efficiency numbers from this driver. Those results told me that this driver can be pushed quite a bit further. I went a little too far before settling on a number I’m comfortable with. It begins with the driver.

On a stock driver we find three resistors on two pads. You can see one R200, under it is one R082 and to the right is another R082. These are all in parallel, giving a resistance of 0.034 Ohms. This supplies the emitter with a respectable seven Amps. However, we have a nine Amp emitter, so…… yeah we need to get to nine Amps (at least :wink: ). That 22AWG wire will have to go too.

First attempt: R082, R082, R100, R100 (0.0225 Ohms). I did this by removing the R200 and adding two R100 resistors. In hindsight, I should have known this was going to go too far. Running from a bench supply set to 12.6V, 4.3A, and it was hitting the limit. I cranked up the current and it went to around 4.8A in. The emitter current was 11.3A!!! Time to turn it off. I will note that the driver was handling it for at least a few seconds, for anyone who plans to mount the emitter directly on copper. This was too much for me…

Second Attempt: R082, R082, R100, R200. (0.0254 Ohms) From stock, add one R100 resistor. This was closer; 10.4A to the emitter, however still a little too high for my comfort.

Third Attempt: R082, R082, R200, R150. (0.0277 Ohms) From sock, add one R150 resistor (since I didn’t have one, I used two R330 resistors, equal to R165). Using this combo gives 9.7A of emitter current. This is where I wanted it. If you have the R150, I’d expect something around 10.0A.
Power In: 12.6V * 3.67A = 46.2W (up from 31.9W)
Power Out: 4.07V * 9.69A = 39.4W (up from 27.4W)
Efficiency: 85% (down from 86%)
Nice! The emitter power jumped by 13W and the efficiency is virtually the same as at stock.
But what about the real numbers; lumens and throw:
Light Output: ~1464 lm (up from 1200 lm)
Beam Intensity: ~73kcd (up from 50kcd) Updated, I found that measuring at 1 meter was creating an error, maybe due to my lux meter (max 50k lux). I don’t trust the original 50kcd number either, most likely closer to 57kcd. To verify, I measured at 5 feet and 12 feet, both came within a few percent of 73kcd. Not bad.

Now, what about the added heat in the driver. The LED is fine, there’s lots of heat sinking for it. The driver MOSFET and diode were getting quite hot at stock output levels (likely why they limited it to 7A). I needed to come up with something to get the heat away from the critical components. There’s lots of space in the driver chamber (you could bury a Pharoah in there). What about adding a heat sink?

That pesky toroid is in the way. Gonna have to move it. Off it comes.

There, now there is room to add a heat sink. I just glued it on with Fujik which should hold it fine. The two large black components on the right were not flush with each other, so one heat sink would not sit good. I solved this by sanding them down until they were flush (they already had the part number scraped off).
Here is the result, with the coil added back on, standing up this time.

And from the other side:

Depending on your heat sink the resistors may be covered. You may want to mod them first. This heat sink was just something I had lying around and fit quite well. If I get a chance to, I will measure it as well as the driver and its chamber.

One other change was getting rid of the 22AWG wire and using 20AWG instead.

Thermal management (in progress):
table(table#posts).
|Time (m)|0|1|2|3|5|10|15|
|Temperature (°C)|18.6|25.0|31.0|34.6|40.0|51.2|59.0|

After 15 minutes, the head is uncomfortable to hold. I would limit use on high at this duration for the preservation of the driver. The light output dropped to 1300 lm at the 15 minute mark.
Cell temperatures, 39, 31, 27. The addition of a neck heat sink really helps keep heat away from the cells.

Outdoor beamshots: (Added Feb 5, 2013)
My comparison unit was the trusty HD2010, driven around 4A, ~45kcd. It was rather cold and the air was somewhat clear. The target is a stand of spruce trees.
First up: 150m (title above picture)
Control - 150m

UF T90 - 150m

HD2010 - 150m

Next: 210m (title above picture)
Control - 210m

UF T90 - 210m

HD2010 - 210m

Finally: 260m (title above picture)
Control - 260m

UF T90 - 260m

HD2010 - 260m

Considering that before the mod these two lights were fairly close, this was a decent improvement in output. Those trees are a rather dark green, only the dead one out in front is a lighter target.
Note: on my photo editing machine these shots look close to what I seen live. On my other machine they look a little darker. They may only be good enough to compare to one another, not to other shots taken at other times. All shots taken at f5.6, 4s, ISO 400.

This was a successful mod for me on this light. I like the results and will probably leave it like that for a little while anyway. :wink:

Mod 2: Emitter Swap: MT-G2
This mod has its own thread, here

Very informative, so I will know what to do and not to do before I get mine. In view, of what is available now from Cree, how would this compare to a newer emitter torch, of the same value, in your estimation? Would you recommend it to other forum members, in view of the new Crees? Can’t wait for you resistor mod, and resultant output.

Ooohh this looks interesting. Thanks for sharing! I’m looking forward to the updates. I wonder how this thing will stack up to other SST-90 throwers.

The emitter is not efficient compared to newer ones. Until MT-G2 came along it was the most powerful single emitter.
I got it as a mod host, and I have more in store for it than a resistor mod :wink:

Nice review, would you mind measuring the OD of the body tube for me?

This is a clone of the X6 with 26650 support then?

Is it better than the x6 output wise?

It is there, body diameter.

From what I read on here and DX, the TF X6 driver pushes about five emitter amps, so this would drive 40% harder. I think the output is slifhtly better, but the X6 reflector might be better, I haven’t seen it though…
I think this driver is better for modding (details coming soon) and 26650 support is a big plus. The larger body tube pairs up better with the huge head.

Nice review relic! How does the throw compare to the HD2010? Any beamshots?

sounds like you’re thinking of a resistor mod AND mt-g2

if it lives, I’m guessing 2x overall output, same throw

:open_mouth:

Was dragging my posts because of the thousand post giveaway… Excellent light, excellent review… I am curious as to the variations in performance between the two. The ones that use 18650 & 26550… Any ideas?

It outthrows the HD2010 stock, and with the mod really outthrows it. No outdoor shots yet, it’s been snowing just enough to make it not possible.

I think the die size of the SST-90 and the MT-G2 are similar, so I’m expecting (hoping) the hotspot stays the same size, therefore, more throw… we’ll see :wink:

I can’t compare, but based on reading here and there, the T90 might perform slightly better (higher emitter drive current, more capable driver). I’d like to compare the reflectors, since that is where the most important difference w.r.t throw will be. Looking at the reflector sold on DX I think they are identical (and it is reported that this is the one in the X6). The dimensions match my measurements.

Thanks relic38 for the review. That thing is a real whopper. Any chance of a shot in your hand (hands)?

Here’s a not so great in-hand shot.

holy crap - that thing is 14” long!

resistor mod? Are you going to push 9A? :slight_smile:

Check the first reply for my resistor mod.

And, yes, I am pushing 9 Amps. You could push a lot more if you wanted to. :wink:

Thermal test just finished, results added to mod entry. She runs a little hot, but with some airflow, it should be OK. This was with it just sitting on a desk.

Wow. Your hand shot shows just how big that monster is. Well done on the driver mod. You really know your way around these things. A GPS would not even show me around on electrical thingys.

wow… bravo!