Oshpark Projects

I’m up for it Mattaus if you want to send me the brd and sch files….I can tweak what you have and then send em back to you to post to OSHPark…this way the revision stays sane

In reference to MRsDNF’s post #1046 - I made an error on my layout for the v13 contact-ring/adapter boards. I knew better, so it was just an oversight on my part. I will post a new version very soon, but not tonight.

Dan @ OSHPark had this to say on shipping:

Credit goes to WarHawk for fixing the boards up. I've removed the old 17/20DD boards and repalced them (and added a 22) with VERSION 1.0 of them. I've started fresh because the revision numbers were going all over the place. I'll need a BOM provided for these as well. I will then add them to the description.

https://oshpark.com/shared_projects/IHvO85FY

https://oshpark.com/shared_projects/SNtRM4Vs

https://oshpark.com/shared_projects/D2dJbbBm

Cheers,

- Matt

Nice Matt! Like'n the rev # - hopefully easier all around to track. Thanks Warhawk and of course, comfy! comfy has been amaz'n on this mod, plus I really like we now have other FET options.

I’m not going to keep beating on it, but as I said in comfychair’s thread I do not think that this is the ideal layout. My understanding is that C1 should be as close to MCU pin 8 as possible for best decoupling. Since we are putting C1 before D1, that means that by proxy D1 should also be as close as possible (to allow C1 to get closer).

comfychair clearly demonstrated that the board works as laid out now in the new v1’s, but that doesn’t mean that we won’t run into a scenario later where it’s a problem.

D1 is pretty dang close to pin 8, isn't it? It's closer than in the revision using gate resistors, the diode no longer has to move aside for the gate in/out vias. C1 physical location no longer matters as it's no longer filtering pin 8 directly (there's now a diode in the way), it's filtering battery voltage instead, which is where the big spike used to be.

The old boards with longer traces and jumpers all over work fine with the cap relocated, I am having a hard time imagining a scenario that would bring about anything like the issues you got with the old circuit.

Let me throw this at you: doesn’t that thought process apply to almost every decoupling cap? (“Just filter the supply voltage and there is no longer a need for the decoupling cap/caps!”) Yet they are still used and necessary.

EDIT: and I could have been more clear about D1 - I don’t care where it’s at. Getting C1 closer to pin8 is all I care about, D1 just happens to be in the way! So D1 has to get closer for C1 to get closer.

Probably totally irrelevant, but I’ve not had any issues in some 20 lights I’ve built with these 17mm drivers. None. All work perfectly and almost every one is maxed out which is what they’re for, right? Since the extra resistors came into the picture I’ve just been building them all with the pull-down and gate resistors both, no issues anywhere. And I’m using a 200 ohm gate with the 10,000 ohm pull-down (I think that’s what the pull-down is, going on memory which is known fallible)

I don’t take measurements and readings and oscilloscope pictures, I assemble and use. Probably missing something somewhere but my stash of FET’s is dwindling, again, and everything I’ve used them in works. No complaints and very happy to have had the opportunity.

So, flawed or whatever…I still say thanks to all involved in making these happen! :bigsmile:

I don't understand this stuff well enough to see why C1 has to be in any particular spot in relation to pin 8. So I'm not saying it doesn't matter, I'm saying I don't understand why it matters. If D1's cathode were physically touching pin 8, and C1 tee'd directly off the anode to ground, how would it be different? It's still filtering the input to the diode and not the output. What would the positives of effectively zero-length traces be, compared to negatives of the trace lengths on the latest version?

I'm really not trying to argue you out of your position, just trying to understand what the improvement would be and/or what could go wrong.

Yeah, but... the MCUs are still running at or over 6 volts, and the signal to the FET is really nasty and leaves the FET in the bad 'not really on and not really off' state for far too long while switching. The gate resistor only softens the spike when the FET turns off, which has the side effect of not spiking the MCU and causing it to shut down. It makes them function, but doesn't cure the initial root cause - that being the arrangement of diode+capacitor which creates an unintentional boost circuit.

Fair enough. Unfortunately… I don’t understand my position that well myself. As you know, my understanding is that we want to put capacitance very close to sensitive but low-draw devices. The idea, as I understand it, is that in the event of a “problem” in the power supply the PCB traces provide enough resistance and/or inductance to allow the cap to smooth power for the nearby device “before” other factors can affect the cap.

If I’m way off base I hope some of our more knowledgeable folks will swoop in and point us to good explanations!

A lot of what you guys write I just get the gist of but don’t fully comprehend. Is it possible to run the traces differently on the 17mm drivers so that a momentary switch can be used on the tail of a flashlight and not run a lead or wire alongside the battery? I would like the ability to use a momentary switch on the tail of a light using the FET, Star OS without making a jumper to the switch. In my imagination it’s possible but I lack the knowledge or skills to imagine how or write it out for Oshpark. Before I get too into this, is this possible to do?

Well with the Small Sun ZY-A629, the driver plate at the front of the flashlight (where our drivers currently sit) is a simple contact plate with a resistor bridge to ground, the actual driver that controls the flashies and stuff are on the clicky end. It would be possible to design the clicky end to be the “driver” by having the electronics be on that end, but the architecture (space for batteries/springs) would be incredibly difficult from flashlight to flashlight.

Imagine a C8 (or many of our favorite lights) that was end cap driver driven

Sorry for the derail

For that hardware/firmware, you just have to ground pin 2 to control it. The switch can even be on a remote 2-wire cable. On some it would be easier than others (ZY-T08, with batteries side by side, leaves space for a wire down the middle). Doing it in a way that the tube can still unscrew from the head without messing up the switch connection is the part where it gets not so straightforward.

I know this has been covered but there are currently several different threads on the subject and its almost impossible to find info, as is with the new design with the CAP on Batt+ to GND and no gate resistor does the AOD510 FET work properly?

What is the [what looks like an 0805 pad] marked 4 connected to MCU pin2?

Also on this board where would the zener diode go?

This is like quantum mechanics, the more I learn the less I know. :nerd_face: My goal is to plug it all together in something like a convoy S2 or Sinner’s copper 18350 host. How thin could the wires safely be? They don’t carry the full amps that the light will be using if I understand correctly. Then if I run a wire to the switch through the body, how do I change the battery of a single cell light without having to remove the wire?

I guess we need Matt to have success with his little switch. That would probably be the easiest. (For me)

Yep, those are the obstacles you'll have to figure out and the solution will be different for different lights. And on some it may well turn out to not be possible (or practical).

Resistor at pin 2 is for the off-time 1uF/10v capacitor. Zener diode would go anywhere on the trace from the D1 pad (resistor instead of diode, on a zener-mod driver) and MCU pin 8. Most convenient would probably be to put it between one of the vias and to ground.