I should have taken a couple pics of the full beam pattern - there are a few artifacts. I got multi-LED SMO reflector SST-40's that look better, but the power/throw of this flat LED is probably more susceptible to the artifacts, yes - hard to tell unless you can compare them side by side. But also, dunno if they are really noticeable outdoors.
Convoy 3x21A SFT-40 6500K first impressions, info, mods
I love threads like this, Tom explains things well.
Simon makes a fantastic host, always did, just think the electronics and attention to detail lacks a bit:
- of the stock 18 AWG wires, for some reason black was shorter than red - doesn't seem to be a logical reason for this
- the red wire had a good slice in it, but seemed to be in a safe place - probably occurred on the tight passage/drilled hole through the shelf. Could have been bigger and smoothed out the sharp edges.
- the wires are a type of low strand count so they are stiffer than what we typically use and the silicone coating is thinner, compared to Hank's or Turnigy branded
- doesn't give the priority to the MCPCB contact surface -- such a critical thermal junction, must be flat, smooth, and use a quality grease (falls short here)
Granted the shelf and MCPCB requires some labor, QC, and/or better machining, I'm sure costly, but using a cheaper grade wire and thermal grease? I think the few cents extra here is worth it.
For the host, the threads, anodizing, finning, designs, optics have always been above this price range.
I totally agree with you. Convoy overall is excellent but always lacking in some way. I hope he improves so these deficiencies can be addressed. Don’t forget, he is making money, we are buying flashlights. He has a bigger stake in this than we do.
Post #2 above updated with the full mod completed.
great review, thanx
sft40 totally suck in matter of tint. it is even worse than sst70.
what is advantage of SFT40 over XHP35 HI? the last one looks way better in matter of color
I'm very mixed on the opinion of the tint, same as the pics show. The real tint to my eye seems to match closest to that last pic, but if you look for the green, you can see it, but it's subtle. I would say it's more green at low output that high output.
The XHP35 HI is great with availability of warmer tints, but it's difficult to work with because of the 12V requirement and high Vf for a 12V LED. You really need a boost driver even if you have a 4S battery setup, as DEL found out with the BLF GT driver. I think the SFT-40 will out throw the XHP35 HI as well and I'm pretty sure the dated XHP35 HI is less efficient when driven hard than the SFT-40. I think the XPH35 HI is based on 4 LED's comparable to the XP-G or XP-G2 - it's based on older designed LED's.
It very much looks like a linear driver just like their 5 and 6A driver, bridging the two 20mΩ sense resistors will result in full direct drive output with no modes. You can decrease the sense resistors in order to achieve a higher output, assuming that the mosfets can handle the additional thermal stress.
Ok, interesting. I wasn't sure if those were sense resistors because they are directly in the Batt- path before the FET's. Yep, that's what happened when I bridged them - modes were all messed up, or rather the ramping was about useless. Thanks thefreeman! Wish I knew more about the various LED driver circuit designs.
This is an example of a linear driver :
The op-amp would be what looks like a SOT-23 package at the top right of the board, the Vsense trace starts under the left R020 resistor, goes to a resistor (R8) then the op-amp.
The PWM signal out of the MCU goes through a RC filter to smooth the signal into a voltage, probably a two stages RC filter actually seeing the number of passives next to the op-amp, and then a voltage divider to set Vsence.
So you have I=Vsense/Rsense and since you know I=17A with Rsense =10mΩ you can calculate what Rsense you need if you want 24A or something else.
Dang, a little too late for me - driver stripped, but good to know, specially if others are interested in tweaking more output. Does this mean there's no noticeable PWM's on all output levels?
No PWM indeed.
Ah yes somehow I missed the modding part in the second post, with the components removed we can clearly see it is indeed a linear driver.
Thank you Tom E for your very detailed and comprehensive review. After reading it I realized that 4x21A SFT40 is not able to overcome the 4X18A SBT90.2 so I really believe that I will not buy it instead I am waiting for news for a 3x21A SBT90.2 that I would buy with my eyes closed.
I swapped them 3x21A SFT40 and 4X18A SBT90.2, I was writing and thinking about more ...
I don't know if anyone did lumens/throw measurements on the 4X18A SBT90.2, but based on his listings the SBT90.2 is probably out throwing the 3X21A SFT-40 but can't say what the lumens are, but thinking they are roughly the same. His posted specs on lumens are not accurate but thinking the throw #'s are fairly accurate. This makes sense because it's easier to measure throw. He also doesn't state what the amps draw is of the SBT90.2, so can't compare that, but I suspect the SBT90.2 is drawing more amps.
The 3X21A triple reflector could have been designed better for throw by making the cups wider and having more overlap.
But certainly the beam's tint of the SBT90.2 is better.
Ohh - tech reviews:
4920 ANSI lumens mentioned, and throw measured at 367 kcd and 337 kcd.
damn.. this is kind of disheartening to hear, I was looking for every reason to buy the 3x21a sft40 over the 4x18a sbt90, but looks like I would go with the 4x18a. Of course, I need to figure out how to piggy back a driver like that. I think if i can figure most of it out, except what to take off the original driver (did you remove everything?)
also if you don't mind me asking, what driver did you use for the piggyback? Or what driver is a good candidate? (Fet w Anduril and flash pad)?
Yes - removed all parts from the stock driver. Driver I used was designed by a BLFer Quadrupel, uploaded to OSHPark, then I reflowed the parts. It's a FET+1, forget if 20 or 22 mm in diam. Actually it's the driver he designed for use in the Amutorch E3/E3S.
For driver, 20-22 mm is a good size. FET+1's are easy to work with. Flash pad for pogo pins don't work out well unless it's exposing the pins on the top (SMD) side. I was gonna use one but on the driver PCB, only the backside pogo pin holes were conductive, on the top side they were covered over.
awesome, thanks for the info! I have something to consider now
Very nice modding job Tom. What cpu is on the board you used? Is it pulled from another light, or something you built? Do you know what the old cpu was? It looks in your pic like they sanded off the markings!
I wonder if you’ve tried TK’s very latest ramp code, from just yesterday or so. See the current thread about it: better low modes with dynamic PWM
It is incredibly clever from my perspective.