【convoy】L21B SFT40 3000K Ra=95 available now

The output current value of 2.3A is already the limit of this driver. We can consider using FET drive, the current can reach 8A.

FET boost driver? AWWW YEAH

Yes please!

As far as I understand, what Simon tries to say or means here is he can probably ask the ∅17/20/22mm SST-40 drivers manufacturer to make a special run of ∅22mm 6A SST-40 drivers with an additional FET turbo mode, bypassing the sense resistor :???: and driving the MOSFET gates with full cell voltage to allow as much current as possible to the emitters. This, or a driver without sense resistor and PWM in all modes :| but the highest (like the usual FET drivers other flashlights have) which, in my opinion, is less desirable to people who likes linears, regulation and overall dislikes :P PWM.

Sorry, I didn’t make it clear.
I mean, keeping the original mcpcb unchanged (3LEDs in parallel)

Thanks the Lord and praise agnelucio, he just published a well cooked modding review of the ∅22 mm Convoy driver: Convoy XHP35 Driver Analysis / Testing / Schematic. As I suspected, the boost driver has lots of modding potential! agnelucio made it work pumping more than 8 A of current to 2S emitters (XHP50/XHP70, for example).

Don't get me wrong Simon, but I knew this MP3431 driver could go much above 2.3 A or 30 W, in fact it can do 50+ W. :THUMBS-UP:

For hard-modders the path is to replace the VLDO and also change R2 to lower the low voltage cut-off. 3 V cut-off is too high, you cannot fully drain a cell by cutting that high even if at low current, much less if you run powerful high drain modes. In addition to this the flashlight has springs, switch and other stuff which causes voltage drops. Therefore, in no way a 3 V cut-off is right.

Thank you for your kind words.

I should point out that the stock driver does work well, and is (in my opinion) quite well designed for XHP35 operation.

It is configured with an input current limit of 10A, so Simon is correct, the driver was *designed* to handle ~30W.

It can be pushed to over 50W, as I have shown, but I'm not sure this is necessarily a good idea for Convoy to implement as stock. Here's why:

At stock 10A input current test, driver loss is approximately 2.5W, which is fairly easy for most lights to dissipate.

At 20A current, most of the loss is governed by I^2*R conduction loss. 20A is 2x 10A, so we expect loss to go up by a factor of 2^2 ie x4, so 2.5*4=10W.

And voila! At 20A input test, driver loss is approximately 10/11W. 10W driver loss is not sustainable for more than a few seconds without very good heat transfer.

The stock driver performs well up to the 10A limit, and I fully understand why Convoy chose to design it this way for most customers.

The rest of the design is generally good, with one or two exceptions.

If I could suggest any changes to the design, they would be:

  1. Use a 2.5V LDO instead of 3.3V (improves regulation at no added cost)
  2. Take LVP voltage from before the regulator. (prevents LVP being dependent on LDO dropout - no added cost)
  3. Increase the input current limit slightly to maintain high mode regulation at low battery. (improves low battery regulation at no added cost)

XHP35 needs 35.5W in high mode, but 10A limit causes output to drop below 3.5V battery voltage.

In high mode at low battery voltage: 35.5W / ( 3V * 85% eff ) = 13.9A, so an input current limit of 14A would fix this problem.

From MP3431 datasheet, this would require changing ILim (R14 in my schematic) to 20k - no added cost.

A driver for sbt90.2 and single cell L2, L21a or L6 would be great.

For L2 and L21a, you could use any 20mm or 22mm (respectively) DD FET driver. For L6 you’d need a buck driver.

With the proper driver, I think L6 with its large reflector, aggressive heat sink and 2S battery configuration could be a great host for the SBT90.2, but I don’t know of any readily available buck drivers that can do anything close to the 20+ amps you’d want for this LED.

Acebeam K30-GT has a fantastic 26A buck driver. I’d love to see an L6 with SBT90.2 and a similar driver.

how about installing SBT90.2 in 4X18A?

If you are able to source the SBT90.2 5700K that would be great.

The main reason I don’t want to use SBT90 is that its cost is 4 to 5 times that of CREE XHP70.2, but the output luminous flux value is not as high as XHP70.2.

1 Thank

It is true but you should follow the hype;-)

The advantage of the SBT90.2 is the actual emitting area is much smaller than the XHP70.2 resulting in greater candelas per lumen. That’s great for a high-power thrower, but I agree that the price makes it a bit of a niche part.

[quote=Simon Mao]

Oh yeah...

Skylumen did it


I would be much more interested in seeing SBT90.2 in the M26C or L21A. I really want a thrower with that emitter than isn’t super heavy and large like the K30GT or larger searchlights, and isn’t crap quality like the FT03S.

Yes, both true but the SBT90.2 is far far better than the XHP70.2 for throw, and that's why we are very interested in it.

The 4x18A would be a great host for a SBT90.2:

  • can use a cheap FET based driver
  • you can get plenty of amps out of 4P 18650's
  • the 4x18A is chunky enough to last a while on high amps, ok, maybe a short while on max/turbo of 20+ amps
  • can mount the SBT90.2 on a hefty size copper DTP MCPCB (I hope so anyway) - the bigger the better!

The 4x18 has some weight in it, might have even better thermal dissipation than a L2/L21A
But it does not need thermal stepdown at 55 DegC, better to use a programmeable driver, anduril would be nice…

The 4x18A isn’t a budget light (or what I consider one) at $60, and adding a SBT90 plus a driver revision and whatever else would be needed (changing battery configuration) would add even more cost. Its doable, and would be a probably low demand item. I guess it’s up to Simon whether to tackle it. Would be interesting to see how cheap he could sell it since the Convoy lights are in my opinion some of the best for the money currently.