Post your MT-G2 driver results here.

Tailcap is not 9 amps though :bigsmile:

Sorry, I saw this and assumed 108W at the tail… now I see the other posts. Thanks!:bigsmile:

This build is very dependent on the cells that are used, the better the cells, the better the light runs. Now the host is a cheapie, but does have more mass and size in the pill area than the “real” Trustfire X8. I real X8 pill module is not big enough for the DRY driver, so that option is out. The DRY will take 3 cells (don’t try it, will fry the emitter), and I noticed that people were posting using 3 cells on the IOS driver, so the DRY ought to be similar on three cells in series (definitely a NO, NO). I have not done that(Good Thing). Thanks for having my back relic, on this, he is much better at posting than I, but we did get similar results on the same build. I was really amazed at the build, the idea being, the the MT-G2 is very floody by nature, so I wanted a deep reflector and fairly narrow beam to contain the floodiness, and the idea worked. There is still plenty of spill in this built, but the hotspot is well contained: but quite large. There does seem to be some interesting chromatics in this build, variations of tints of yellow and white, but rather interesting than annoying. Relic can do better on the measurements of throw, but I was impressed at the throw for just 2 cells. The torch is gobbling every electron you give it, so like relic indicates: it is a very small light pumping Boo Coo amps, but seems to handle the thermals, and the MT is demanding everything from the cells, so it does get warm quickly. With the final drive being Turbo or DD, I would guess that counting efficiency and points of resistance, this is about as much light as one could expect from a torch this size at the present without things starting to go into thermal/electric failure…just my view, relic is free to add his view…, remember this is a bottom of the barrel — budget build, for this powerful of a light…at this time…

A word of caution. Do NOT use the DRY driver to drive one MT-G2 from three cells in series. The emitter will die a quick death. Quick, as in you will not get a chance to turn it off.
The IO driver is a buck driver, so it will convert the power from a 12V/1.5A level down to 6V/3A level (halves the voltage, doubles the current) in order to keep the current in check.

The other points you make are in line with my findings so far. I’m charging my Sanyo cells right now and will test output again for documenting purposes.
For about $45, this is the lowest cost and simplest MT-G2 mod I’ve seen yet. The ZY-T08 mod was more expensive (could have used a DRY driver there). It was also more work, however if it had actually worked from a 2P configuration, it would have been easy.

That is what I thought, thanks for having my back, again, relic…

OK, here are some results using fresh Sanyo ZTs charged to 4.3V
Drive current: 4.7A
Total output: 2238lm at start, 2045lm after 30 seconds (mostly battery sag, power supply shows much ‘sturdier’ output)
Throw: 19.2kcd after 30s
So it’s not bad. Would like to have a regulated boost driver to keep them lumens going steady, but you would need IMR cells for that (battery current would go over 2C as the cell drain).
You can get away with regular 2600mAh+ cells with this setup.

Pretty impressive, but like you say, it is an electron hog, and eats batteries quickly. I was quite happy with the build, do you think using the DRY in another build going 2S2P would work? Thanks again…

2S2P would work, however it might push over 5A. If you have a Noctigon and good heatsinking it should be OK. There are a lot of variables, so it is not for sure. If resistance is very low, it could be bad news. It would make for a lot better run times :slight_smile:

Direct drive off 2S 18650 gives 5A or a little less with conventional unprotected cells, none of that crazy INR stuff. But, high current means a higher Vf and more voltage sag and before not too long at all the voltage will fall below the Vf and the current & light output will start falling off.

My build with the 9A driver does right around that same 5A when run from two cells, the driver isn't regulating that at all, it's essentially direct drive since the driver wants to deliver more amps than the Vf-versus-input-voltage will allow. Adding the third cell gives it the headroom it needs to do proper regulation - higher voltage allows more current to flow. 2S2P will help reduce the amount of sag, but that just slightly prolongs the inevitable, cell voltage will still limit current long before the cells are close to being flat.

Drove one MG-T2 with 12 direct connected AMC7135’s in a DST. Gave 4 amps at the tail. Only ran for 7 to 10 minutes before I called it a night. Seemed to work fine with no issues noted.

EDIT: 350mA rated 7135’s. Used 2S NCR18650A’s charged to 8.4v combined.

EDIT2: Recounted, I had 14 7135’s. So current at the tail was about 86% of rated. I rushed the light back together. So discrepancy may be due to a high resistance point in one of the many junctions that the electricity has to cross in the DST. I’ll try to investigate this weekend.

EDIT3: Been using the above cells direct drive in the same light and getting the same current. So batteries (and probably resistance in the various connection points in the light) appear to be the cause of the lower than rated current readings.

Check at the emitter, both versions I built did only half the rated current at the LED while showing normal numbers at the tailcap, still have yet to see anyone else say they've checked the output current on a MTG2/7135 combo. One driver was a single mode only with the Vdd taken from between the two cells, other one has a normal MCU with the resistor/zener mod to drop the voltage to 'safe' levels.

Will do comfy. The light output seems about right for 4 amps though. 2 amps would be a lot of current to be consumed by those little chips. Maybe you are getting less current because you feeding a reduced voltage to vdd. I have the 7135’s directly connected to the full 2S voltage on both positive and negative.

I've done it both ways (well, 3 ways, actually). Polarity diode replaced with a jumper and another jumper direct across the empty MCU pads 6 & 8. Next version was taking the Vdd feed from between the two cells, then finally one with a functional MCU with the voltage-dropping hacks. All of them were goofy with the output current, and all went back to doing the proper rated current when retested using a single XML (and a normal 4.2v source).

7135 is either on or off, right? Only way to get less than the rated current is if you pulse the Vdd, any voltage above the turn-on point and below when it fails from overvoltage should give the same behavior at the GND and OUT pins, I'd assume. I really need to get a basic variable power supply...

Using Comfy’s 105c hack on a Q-Lite 3.04A driver with 4 chips stacked (380mA binned). 200 ohm resistor traded for the polarity diode, 4.3V zener on the capacitor. Using 2 Panasonic NCR18650PD cells I’m getting a tailcap reading of ~4.05A.

Question on testing at the emitter. Or a few questions. Test off the negative lead? Use the standard probes on the DMM? I find my DMM is way off using the probes, I have 2 short pieces of 12ga Romex that I use for Amperage testing and the numbers are then more representative of what everybody else gets. To check Vf do the same thing but in Voltage mode on the DMM? Just like testing a battery?

I’m about convinced to try it. Just hate how much it takes to solder that darn lead on with all the copper I’ve got in there!

Just ran a heat test with my IR gun, in 5 minutes it went from 91.1º-133.3º , reading taken at the lower head on the fins right at the emitter position. Heat was rapidly spreading right down into the battery tube. Top cell measured 109.9º at the head right after the run. Moments out of the light, the cells measured 4.03V down from 4.20V.

To measure emitter current, put your DMM leads in series with either the emitter negative lead or in series with the emitter positive lead.

In other words, you have to break one of the connections from the driver to the emitter contact (either the + or - emitter contact), and put one of the DMM leads on the emitter output and the other DMM lead on the emitter input pad.

Either that or get a clamp meter, then you don’t have to do that, just clamp around either the emitter + lead or around the emitter - lead.

To check the Vf, put the positive lead from your DMM on the emitter + and the negative lead from your DMM on the emitter -.

EDIT: Re. measuring current, I think that that (having to disconnect one of the emitter leads from the driver, etc. is probably why most everyone measures current at the tailcap instead of directly at the emitter.

Doesn’t require breaking connection? If the Vf is known, can the emitter amperage be reliably tracked on the charts? Or semi-reliably? I was just double checking the temperature range of the Arctic Silver Thermal Adhesive. It’s top end says 150C. With the 32mm copper star and thick aluminum shelf on the pill, and another 1/4” of copper under that, it takes a setting up around 850º on my Hakko 888 station to solder a lead on, and even then it takes remarkably long under contact before the solder softens…it doesn’t go full liquid like I’m used to seeing so it’s barely making the bond as it is. This can’t be good for the thermal adhesive.

I’ve got a narrow bevel tip on that I used to stack the chips, would this be why? Need to switch tips back I guess.

watch my back…I’m going in! :slight_smile:

I can’t believe this! I did it over and over, and it’s freakin unbelievable! I kept getting 4.40A, but wasn’t sure that was high. So I had my wife actuate the clicky to run through the modes and OMG! 8.50A!!! At the emitter! Yowza, there was some serious heat coming off that bad boy and I was almost in contact with it!

This is with 2” long 12 ga Romex wire at the negative lead on the emitter. I clamped the negative or common with hemostats, then held the positive in place ready to bail if need be.

It’s bright! :slight_smile:

I thought that I knew the answer to your question “If the Vf is known, can the emitter amperage be reliably tracked on the charts?”, but I had already been doing some testing, and now, I don’t know the answer.

The testing I’m doing is with just an XM-L emitter on a PCB. I don’t have a bench power supply, so I’m using an Accucel 6 hobby charger instead, which may be the problem.

I’m using a new DMM (Ideal Pro) for measuring the emitter voltage (voltage across the emitter) and a clamp meter to measure current. The Accucel also shows current and voltage output.

The thing that is puzzling is that as I increase the current from the Accucel (say from 1 amp to 1.5 amps), the voltage across the XM-L is dropping, from about 2.87V to like 2.7V.

I was just looking at the XM-L datasheet here (http://www.led-tech.de/produkt-pdf/cree/XLampXM-L.pdf) and that seems just the opposite from what the chart at the top of page 4 in the PDF is showing.

As I mentioned above, the problem may be the Accucel, and I know I should be using a bench supply, but I don’t currently have one.

One issue with doing that is you probably won’t be able to re-program the driver. The PWM pin on the processor chip is one of the pins used to program the chip. The polarity diode isolates the power on the processor chip from the emitter power. Without that series diode, your poor little AVR programmer will try to also power up the LED (s), and won’t be able to supply the 3+ amps. You would need to remove the 200 ohm resistor to re-program the processor.