Sunny’s FWAA Teardown

I got my FWAA a few days ago and had my fun with it so I’ve decided to do a tear down in preparation for future modifications I have planned. In keeping with the BLF spirit, here’s all I found:

I don’t have a clamp meter so I measured the current the emitters are pulling by using a method I recently read about where you measure the voltage drop between the mcpcb leads. I measured 3.3v which correlates to 3.1A per sst-20, totaling 9.3A with 3 of them with a fully charged vapcell H10. This little 14500 is a monster! This measurement may be incorrect though as I partially killed one of the emitters when my multimeter’s probe slid and touched the wrong thing.

Has anyone ever seen an LED die like this? My LED’s die died. :smiley:

There wasn’t much thermal paste below the honking fat mcpcb. It’s 3mm thick and is easily the chunkiest mcpcb I’ve ever seen!

Just as the mcpcb is thick, the driver cavity is shallow. I measured it to be around 3mm or less. This is a fet+1 driver, controlled by an attiny85 with the reset pin left unsoldered. First time I’ve ever seen this, maybe it saves a few pennies over the entire production run?

The driver is a double sided 1.6mm thick board which looks machine reflowed on both sides. With my limited knowledge of pcb production, I believe they reflowed the top side while the spring side passed through a wave solder machine? Actually with the spring there, that’s probably not what happened so I’m really not sure how these double sided boards are produced. I suppose the spring is probably hand placed and hot air soldered since a pick and place machine probably can’t handle a spring? Maybe one of you more knowledge peeps can enlighten me?

Regarding the cleanliness of my FWAA, it looks like the driver pcb went through a proper pcb wash cycle but they didn’t bother cleaning the flux after hand soldering the mcpcb wires. It’s a shame because the dirty stains on the mcpcb are the only ones most people will ever see, and pass judgment with.

Here are my measurements of the driver board. I reckon they’re accurate to about +/- 0.1mm. I think thefreeman might have been interested in these figures so I’ll just include them for everyone. It’s a tiny driver, really not much space for anything but passives on the spring side. The ones lumintop used look to be 0603 sized components.

And finally, the FWAA comes with a new spring in the tail switch that the Fw3a didn’t have. I’m not sure if it makes a big difference. It seems to be not very stiff, but I guess if you don’t screw the retaining ring too tightly, it can allow the switch pcb to move just a tiny bit, allowing for some looser tolerances of that signal tube?

I’m not sure how much it helps but it is a new addition (to my knowledge) not seen on other FW3A variants.

Cheers and I hope you enjoyed this quick look at the FWAA’s innards. Please excuse any errors as it’s past my bedtime and I’m quite tired.

Indeed I was interested in the measurements, thank you.
I like that this light is very small and even though I prefer side switches I find tail switches to be okay on small lights.
Although I didn’t buy it because I’m more interested in the single LED variant that is apparently in the works and if/when it’s out I’d prefer to put a more efficient CC driver in there.

If I understand the pics correctly there is no GND ring on the back (spring side) of the driver, only the switch ring, the batt- contact is only on the front ? I assume that the shelf and the tube threads are not anodised then ?

It’s good that they added a spring load to the switch tube assembly.

If the light is still disassembled I’d like to ask a few other measurements if that’s okay :
Driver cavity diameter with the shelf (likely just a few 1/10s more than the driver)
Driver cavity diameter without the shelf (i.e. clearance diameter for the components)

And also ID of the batt- tube, ID and OD of the signal tube (as it should be easier to get more precise measurements there with the calipers instead of the switch ring on the PCB)

Seems pretty cost conscious, using the attiny85 in that QFP. There is a smaller 4x4mm QFN package available, and the newer and generally better attiny816 is available in a 3x3mm VQFN.

Thnx. Really liking the new spring. My fwaa still haven’t done anything strange, maybe it works.

The spring is, apparently, what makes contact between the switch upper and outer edge and the flashlight body (non-anodized) in the tailcap.
I tried mine both with and without that golden spring and without it, the flashlight only flickers and doesn’t activate.

As for the LEDs dying, I belive it could have been a less good soldering/reflowing. One of mine had a slight “defect” too, but fortunately it didn’t die! I hope you can replace it :wink:

> There wasn’t much thermal paste

great photos, mine had even less paste, basically none under the mcpcb, and a tiny amount in the corner of the host shelf… It also had one LED that was dimmer than the others. Mine is a CuFwaa w sst-20 4000. Now with 3500k 219b… I kind of wish I had used 3000k 219b instead… first world problems… lol

output went down 15%, not a problem.

fun little FWAA, I still need to try the O ring mod for the switch

1. Yeah, that 1x osram powered FWAA would do well with a buck driver. You might have a tough time fitting it in though.

2. Nope, the only gnd contact is on the top side of the driver (not spring side). The threads and that little shelf are unnanodized.

3. 18.19mm according to my calipers (threads down to the driver shelf). My driver’s nominal diameter was 17.8mm but it had a protrusion from the pcb panel breakout tab which was 18.2mm at its widest point. Taking the driver out and putting it back was a tight squeeze.

4. Clearance for components I measured at 16.95mm with my calipers but I wouldn’t put components too close since tolerances aren’t the best. Maybe go with a 16.8mm diameter or less just to be safe?

5. Good idea on using this instead for the measurements. Signal tube inner diameter is 14.8mm and the outter diameter is 15.97mm (probably 16mm).

This does seem like a good candidate for the avr 1 series’ 3 pin flashing. Our standard 6 pin pogo adapter might just fit but it would be a very tight squeeze at best (and it would take up a significant amount of the very scarce board space).

You know more about the new 1 series chips than I do. (I’ve only read the thread) I’m interested in making a fet+1 board with a 1616 when they finally become available from back order in October.

Would a fet+1 1616 based driver work with the same passive components as our attiny85 based drivers or do you think they would need different passive component values?

Cheers, I’m a fan of your work.

I’m also interested in avr-1 because it has flash memory segments (boot, app, data) like the atmega’s. That should make it possible to reflash over i2c with a suitable boot loader. Then the board could have a JST-SH connector (if it fits) and plug straightforwardly into the existing Qwiic/Stemma ecosystem instead of having to mess around with pogo plug contraptions. It would also give 2 way communication for monitoring or whatever.

This little board would be a convenient interface device: Adafruit Trinkey QT2040 - RP2040 USB Key with Stemma QT : ID 5056 : $7.95 : Adafruit Industries, Unique & fun DIY electronics and kits

Thanks for the additional measurements.

I would use the passives DEL used except the R1/R2 voltage divider is not needed when reading VCC voltage.

I made a mistake. The driver cavity’s z-height is less than 3mm, probably around 2.7mm. I didn’t have a good way of measuring it, but putting the mcpcb into the driver cavity, it looks like the 3mm high mcpcb extends higher than the driver shelf.