DQG Fairy power upgrade

CRX · 2019-10-22T16:47:42.000Z

Shit.

How did you go about swapping the LED? The pads are quite easily ripped off these things.

The pads on the board have limited trace connections to the SMD components, especially the lower right LED connection.

The diaphragm button can be finicky too, clean underneath with good connections through the PCB?

Luxeon v2 has the little circular nik out of the centre pad at the negative side.

Agro · 2019-10-22T19:01:09.000Z

Hot air for a while, then grabbed the led with tweezers to see if it will go off. It did, I didn’t use anything I’d consider “force”.
I’ll try unsoldering the led later, maybe tomorrow.
Thanks CRX

CRX · 2019-10-22T20:16:24.000Z

I would check this connection first.

Agro · 2020-01-21T21:25:01.000Z

Only today I got the energy to work on the light again.
So…I checked connections again.
It turned out that I soldered it OK but I misunderstood negative connection - it is not on the outer ring but on the pad next to 3R6 resistor.

So I put it in the pill - it worked. But it was loose. How did DQG keep it in place? Was the insulation ring enough to hold the parts together? As of now - it is not.

While screwing the pill into the body - it moved. And the 3-legged component fell off.
I took my iron and soldered it back, looks very ugly but OK, at least I didn’t blow it away with hot air.

So…it lights up. Thanks again CRX.

But it is a moonlight-special, very dim single mode. That’s not news - I modded a broken light because I was afraid that it wouldn’t survive my modding and I didn’t want to care too much. But now I would like to enjoy this warm Fairy….
Any hint on what would it take to repair it for good?

Also, the driver falls off as soon as gravity encourages it to do so. What would be the best way to keep it in place?

CRX · 2020-01-21T21:37:36.000Z

A little solder paste on the outer trace, insert into pill and use a jet lighter on & off to heat the side of the pill with components facing up and don’t cook it or blast the LED etc.
That’s what I do anyway.

Agro · 2020-01-21T21:45:29.000Z

A little too permanent for me. I thought about silicone glue…and that’s what I’ll probably end up doing. But only if I can fix the driver, otherwise it doesn’t really make sense.

CRX · 2020-01-21T21:53:23.000Z

Well, it’s only traces, pads, resistors, LED and that switching component so what did you break?
And remember that ground trace needs to connect securely to the pill so glue on that might not be the best.

Agro · 2020-01-22T07:44:42.000Z

CRX:

Well, it’s only traces, pads, resistors, LED and that switching component so what did you break?

And solder joints.

You say it’s a switch?
OK, I think I understand how it works now.
Batt+ goes to 3.6 ohm resistor and then to the LED+.
Batt- and LED- there are 2 parallel paths. One is with 51 ohm switch the other is with a FET.
Once the FET gate gets power it reduces resistance by a lot.

I think I’ll be able to measure the circuit now.

I wonder why did DQG use a FET at all? Couldn’t it work with just 2 resistors in parallel, 55 ohm connecting the first pad to LED+ and 3.8 ohm connecting the second pad to the LED+?

CRX:

And remember that ground trace needs to connect securely to the pill so glue on that might not be the best.

Good point. Maybe I should get some conductive glue. I’ve heard that they tend to be quite fragile, unlike solder. Whether too fragile or just right…I’ll see.

Agro · 2020-01-22T20:23:50.000Z

There is no high-and-low, just a single mode.
Measured 0.5 mA at fresh start, quickly falling.

Checked cell voltage - 3V.
OK, let’s put it in the charger and in the meantime try a 26350 that I got at hand. 26 mA, too much to stare at the emitter.

I measured everything I knew how to measure. Importantly, I checked that gate gets continuity with the central pad.
I don’t know how to measure whether the FET is fine…I thought there might have been some short, so after optical inspection (which didn’t reveal any) I measured continuity. My DMM measures there is between all pairs of legs. Is that OK?…

CRX · 2020-01-24T13:41:23.000Z

I doubt it should have continuity between all the legs.
I don’t like messing with SMD’s, electronics etc, never have so don’t know much.
The way I see it, you should at least have the normal low mode when a cell is connected if the components are all ok, a partial short or fault in the 3 pin switching component sounds likely.

So you should have a direct path through the 3R6 - LED - 510 - Ground when the low mode outer cell side trace is energized.
Then when the centre pad is energized it causes the switch to connect the LED negative directly to the ground ring bypassing the 510 so only the 3R6 resistor is in play giving high mode.

So you’re getting the exact same very low output on high or low mode now?
I would remove the 3pin component and see if the normal low mode works as normal, I think it’s busted if you’re sure everything else is ok.

DB_Custom · 2020-01-24T14:17:52.000Z

A2 is a small FET, I’ve stacked another before for better power or changed it out to a better one. I have seen these little three legged FET’s do up to like 5.7A off an 18650 before, so it could be a cell limitation more than anything,

And yes, you must maintain a solid ground or high mode is likely not going to work.

Edit: As I recall the original D80 driver used two of the little FET, I’ve built a lot of D80’s with our FET+1 drivers and had plenty of the stock driver to rob parts from…

CRX · 2020-01-24T14:59:55.000Z

Right so it’s basically acting as a switch when powered up?
What is the purpose of the 473 - 47kΩ resistor?

DB_Custom · 2020-01-24T15:11:04.000Z

I’m not sure but it could be to smooth the switch, prevent spikes?

It’s connected to the thermal pad and the 510 resistor (through the thermal pad) so it would appear to be for the low output side. I’m no electronics guru either so I can only surmise.

Agro · 2020-01-24T15:23:49.000Z

Low output? Not gate?

DB_Custom · 2020-01-24T15:53:34.000Z

I am not sure, my memory confuses things and the more I look at it the more confusing it gets. A result of 20 years of concussions, sorry.

We need someone more in the know concerning electronic schematics to give this a glance.

CRX · 2020-01-24T16:01:02.000Z

Yeah, change the title to FREE PORN! or CAT PICTURES!
But seriously maybe change the title or start a new thread about these things to catch people in the knows attention, could end up with a lot of useful info on how they work.
Shameful really that I haven’t learned about this stuff yet but I know the rabbithole will be deep and never ending once I start so have been avoiding firmware & drivers so far.

Agro · 2020-01-24T18:05:48.000Z

I removed the FET and measured the board. There’s continuity between all pairs of pads….
which is not surprising after the fact.
Orange an blue have a direct connection with a 47k resistor.
Blue and violet have a connection with a 51 ohm resistor.
Orange and violet have a connection through blue.

I think that the FET is broken. But I’ll try to reinstall it, maybe there was some soldering issue.

DB_Custom · 2020-01-24T19:52:28.000Z

I googled it and there are good descriptors of how to test the FET, lost me pretty quick but that’s easy to do these days…

Agro · 2020-01-25T09:01:17.000Z

DB Custom:

I googled it and there are good descriptors of how to test the FET, lost me pretty quick but that’s easy to do these days…

Indeed.
I set out to do it, took the FET and started measuring. Test 1 - probably failed. Not sure, the tiny MOSFET would move under the slightest touch. I turned to find something to hold it. When I turned back…the MOSFET was gone. Did my t-shirt touch it during the turn and swipe it off the table? I don’t know. Anyway - I failed to find it. I think I’ll need to ask somewhere else to find a replacement….

DB_Custom · 2020-01-25T13:34:06.000Z

Digikey has them, it’s an SOT-23 MOSFET. They’re pretty cheap, from what I saw perusing the list looking for what to expect power wise.