That was my initial impulse, but I wanted to ask the source directly. AE sent me an e-mail earlier with it at the top of the list. I have bought many Convoy lights on AE over the years, so I guess the algorithm just saw the keyword in the title and thought I would want to buy this too. Otherwise, I never would have seen it.
I just meant before stepdown occurs.
SFT25R at 8A lasts over 10 minutes
SFT42R at 8A with a 8-10A 30 sec burst is 5.5 minutes. It runs hotter, unless the 8A driver is doing something weird with the SFT25R.
Either my lumen tube has an issue with measuring the L21B with an SFT25R or the 8A driver isn’t able to drive it at 8A. I only measure 1550 lumens and that’s what I measured with the T6 at only 5A.
2 Thanks
Thanks, I see what you mean!
I’d say this is the more probable cause. The SFT25R’s Vf is extremely high (more than 3.8V) at 8A; for a non-high-drain 21700, voltage sag alone suffices to bring it under, not to mention losses in the circuit.
3 Thanks
For the sft-42r to run hotter than the sft-25r driven at the same power, despite converting far more electricity to light, makes no physical sense. I suspect the sft-25r is driven at lower power - around 5A-6A.
2 Thanks
Just had a look at my lumen graph of both 100% and 50%, the output at 50% (4A) is still way below the T6 with 5A driver. It’s very strange, both on L21B and C8+ with the 8A buck.
Graph of L21B and T6 with SFT25R
2 Thanks
Could you explain? Is the T6 output in the graph at its 100%?
Yes, the T6 is at 100% 5A.
It’s strange that even with the 8A buck set to group 8 with a 50% level (4A), that the output is way below the T6 at 5A
1 Thank
I did try a 5A buck in the L21B when I first noticed this, the output dropped to 1300, so I have no clue what’s going on. Is it possible that these buck drivers have an issue driving the SFT25R?
Swapped in an SFT40 with the same drivers and they performed as expected. It’s only pairing an SFT25R it underperforms.
2 Thanks
Well, this test I just ran was super interesting. The output dropped way below the usual 35% on other drivers. The temp remained at 50°C, no issues with heat here, the output reduced whenever the temp increased to 51°C.
35% = 2250 lms
8 Thanks
How hard is it to design a driver that can be set to various currents? Something like a 20A limit that has an option to set the current from 3 - 20A. Then the levels are always spaced out at the same 12 groups steps, like 0.1, 1, 10, 35, 100.
That would be super useful.
4 Thanks
I think that pushing more than 6A in SFT25R didn’t have much sense because You don’t get some big rise of lumens, it only heats more. With one cell it is not easy to long drive SFT 25 with 8 A because of high VF.
Maybe driver of L21B has bigger voltage sag inside it than AA/14500 driver of T6. Ir will be ideal to measure voltage which come to led to see what’s happening there.
I think that ideal host for SFT 25 is 2 serial cell light with buck driver if You want to push it longer time with stabilization on its limit about 8A.
I like very much T6 and for me it give useful amount of light and throw even on first 1% mode, on 10% mode it is about 2 hours battery life but on 100% battery life is very short. I will try to make on lathe longer barrel for T6 and use two serial Vapcell H10 but with paralell connection, so I can get longer battery life. I will connect battery with wires parallel and wrap it, so I can charge it with magnetuc trustfire charger. New barrel will have bigger inside diameter so this homemade 2x14500 - 14100 battery can go inside. It can be made with Vapcell F15 because with parallel connection current from battery will be smaller than its limit of 3A but I have only one F15.
Very nice data thank you. All signs of functional sensor-based thermal regulation. The sustained output is super impressive versus my Wurkkos TS30S Pro.
Convoy L8 with SBT90.2 officially on my radar. 
Review by BLF’s @Sirstinky on 1lumen HERE :
I think Simon mentioned some time ago that these stepdowns exist to avoid driver damage at high currents.
So making the driver even more powerful and limiting maximum current to the point where it won’t self destruct is good. I’ve seen that one Fenix light can sustain it’s “turbo” till dead battery, and that can be achieved by underdriving everything for better efficiency
1 Thank
The SFT-25R has a really high Vf, connecting it directly to a 4.2V PSU and the LED draws ≈5A. So really it doesn’t even need regulation and can be direct drived from 1 cell. So I would expect it’s output to drop quickly on a buck driver as the battery will drop below 4V really fast.
I think you would get better output with a boost driver but not sure what the maximum Vf is on the 25R yet.
1 Thank
Yes, he did say that in 2023, but that was because this particular component was different. The driver manufacturer swapped it because it was expensive or hard to source when a new batch was required. The chip would fail if it held 100% for too long.
Everyone complained so Simon had it reverted back in the next batch. This affected all boost and buck drivers.
The 20A buck in my above graph doesn’t show this early stepdown behaviour either.
2 Thanks
Huh? That’s not L8…
Sorry for the confusion. Should read:
Convoy L8 with SBT90.2 officially on my radar.
Below is runtime of my TS30S Pro with SBT90.2 reviewed by Sirstinky on 1lumen for comparison with the runtime of L21A above.
I mentioned L8 (instead of L21A) only because I prefer side-switch and USB-C charging.
Are you sure L8 got the same driver as L21A? Because it used to have the unregulated one.
https://zeroair.org/2022/02/10/convoy-l8-flashlight-review/#Power_and_Runtime
I am relieved to see this. Thank you so much for measuring and sharing.