Q8 modding

Findings with my Q8 lights:

- it’s a disconnect and not a spike or fake key press, easy to recognize: it flashes 2x as it does when powering on,

- disconnects occur with and without spring bypasses,

- disconnects occur with long protected cells and with short ones, but even on weak bumps with short cells.

Proposal to avoid power off on bumps

An additional 100 uF ceramic cap over C2 appears to be sufficient in my light. No power hickup on hardest bumps, even scratched the tailcap during tests.

bumps on the tail end will disconnect cells easier then on the side.
still, or a bike lght I would go with something with spring on both ends of the cell

Someone just needs to check it out.

A modding idea for getting springs on both ends, quite a solder job: on the tail side of the battery tube swap in shorter springs, on the driver side remove the brass ring, solder 3 or 4 short springs, i.e. the BeCu intl-outdoor springs, on the +pad and solder the brass ring on top op them.

Being realistic I can’t get my Q8 to turn off no matter how violently I shake it, only by dropping it onto my knee from 10in/25cm with the tailcap down can I get a consistent disconnect. I mount it to my bike in a way where all the force moves across the flashlight sideways and won’t cause a disconnect. I mainly ride trails, but the closest trail is pretty tore up and covered in sticks/rocks from the heavy rain over the summer and it’s a non-issue. If you’re doing riding where the Q8 losing contact is an issue then the mount is surely more likely to be compromised than the batteries. I still think that snugging up the battery tube is the easiest way to make the batteries more snug though obviously not as perfect as springs on both sides.

Are you sure? Long springs or springs from both ends are necessary when you will use any from 65mm cells or 70mm cells or 72mm cells.
I think protected cells are not common with Q8, so actual length difference is just 1-2mm. Short springs from tail+thick positive ring, nothing more needed. Soldering ring over springs is not simple, also positive ring with four imr cells is not something that I want to give move abilities (yes springs are rigid but hwo knows where this spring carried ring will slide after hard tuve screwing?)

The brass ring will be far from the side and attached to 3 or 4 springs it must be rigid enough. It does required good solder joints, perhaps drill shallow holes in the ring where the spring-tops fall in for extra security.

It is just an idea though…

If a bump disconnects one cell, it will probably shift all 4. It’s simply a matter of impact “G”, mass of cell, stiffness of spring, pre-load on spring. You can expect all 4 to move the same.

It certainly happens on my SRKs with what I would call a small slap on the tail.

In my case I want it to sail through any sort of (repeated) impact that might happen if it falls from my hand, maybe tumbling down a hillside afterwards. An extreme use-case I know.

Excellent news. I was hoping it would be that simple. Thanks very much.

Edit:

Has anyone sketched the schematic diagram yet ? Sorry, but trying to reverse-engineer it by peering at PCB layouts that may or may not be the final version, and parts lists that are well out of date, is hard work, and massive duplication of effort for anyone trying to analyse this.

I’d like to come up with a C2 value with a bit of basic analysis that I am quite sure of.

Edit: and I don’t want to mess up any of the fine tuning that was done to eliminate inductive spikes from FET switching getting to the CPU, I think there was some interaction with a series resistor added to the MCU supply, but can’t remember the details.

Edit: Flashy Mike, were your cells fully charged ? I’d like to know if your fix is still solid at minimum battery voltage, which I think is the worst case.

I think coil springs under the ring would be a very bad idea, they are bound to become distorted to the side when tightening.

I don’t think springs at that end would be a good idea at all, but if it were to be done, flat springs would be necessary. Think leaf spring on a car, 4 small ones.

Thanks for your answer Tom :beer:

I also lost a light and couldn’t find it. I could do the trip in the dark, I did it quite a few times before buying the light. But the thing was expensive! I called a friend who lived nearby to come with a flashlight, which let us find it.
Since then I’m a big fan of GITD.

DEL is the ultimate source - I thought he has schematics in his driver thread.

I'm emailing DEL and Miller now - dunno what happened but can't find the info, or the original OSHPark BLF Q8 listing.

It's all the same though - basic FET+1 design by DEL. Guess I'm so used to them, no schematic needed. Everyone copied this design. DEL added parts to improve fix issues with the ATtiny 25/45/85. He posted such great detail in the past, it's common knowledge by many of us.

Forgot PD68 did the original BLF Q8 board layout, but under the guidance of DEL. Later on, DEL started doing and posting his own OSHPark driver designs, but using the same circuit design.

Original OSHPark BLD Q8: https://oshpark.com/shared_projects/uK4XziVN

This is what ThorFire was given. Sorry, but the OSHPark listing has no details.

This is all of DEL's OSHPark listings: https://oshpark.com/profiles/DEL/page/1

This one of DEL's is pretty much closely matching what we have on the BLF Q8: https://oshpark.com/profiles/DEL/page/1

Differences are a different FET, bank of 7135's instead of 1, added a C3, and we don't populate R1 and R2. Matching ours: R3, R4, R5, C1, C2, D1. I think DEL's driver thread has some great reference info as well: https://budgetlightforum.com/t/-/44006. Check out post #3.

If you have GITD stuff on it best would be an UV flashlight to find it

If it was on a moment ago and I know it’s within several meters (like it was back then), there’s no special equipment needed. A glow o-ring like on my $7.13 Convoy would saved me from calling the friend in the middle of the night.

Is that a backup torch in your pocket or are you just glad to see me ?

Don’t know if this is useful Tom. Done from memory of prototype#3, so may not match 100% what is in the production models. It also shows the 2-resistor version of the switchboard. If anyone sees something amiss I can correct it. Also need to add the FET part number.

EDIT: FET part number added and R6 corrected to 4.99 ohm

2 is 1 and 1 is none. Always carry a spare. Even a small twisty can alleviate that lost item issue.

That looks darn good. I think our usual D1 is labeled D3 in the drawing? And the labeled D1 is the main LEDs. PD68 had the D1 labeling on the Schottky, though not sure bout the ThorFire Q8 boards.