[WIP] 15mm PAM2803 w/ ATtiny13A rough layout / possibility

U mean this one? https://budgetlightforum.com/t/-/37354

thread de-hibernating… :slight_smile:
I’ll dig up what I got so far.
But allow for a minute or two.

.

Edit:
Every now and then I prodded along with this driver, but with other matters at hand I did not find the energy to post all my ups and downs. Just hoping to lure @wight into this again…

0:)

.

.
So this seems to work.
It’s Fasttech SKU1127407 (comes in blue now) with mods.
MCU changed to Attiny13a (which was pre-programmed)
Some MCU-traces cut (Pins1/4/5)
Rewired: Pin6 to Fet-Gate (red wire)
MCU+ is already connected to V+out
Rewired: MCU- to GND

MCU- of the original driver is connected to Fet-Source/PAM3. This did not work with the Attiny, it wouldn’t change modes.
Rewired MCU- to GND (black wire) and voila: 3-mode boost driver that did survive several 1AA and 2AA drains.

Still don’t know what the 2 resistors (103 / 18D) really do…

.

.
-Unfortunately this did not work out…
Modes do change, but output is flickering in all modes.
(just no time for debugging atm… :frowning: )-

((( EDIT: Rebuilding it did work, see posts further down )))

.

Not that one Tom E. It was a low voltage driver that was nearly completed.

I think this is the same driver on CPF.

Would be nice to see that Fastech driver stripped and a schematic of the oshpark board.

@HarleyQuin - That looks like a bunch of good work.

  1. As you pointed out way back (nearly a year ago :frowning: ) in post #23, there is clearly a problem with my BRD (actually with my schematic). My schematic is very poor, but I have only myself to blame. I don’t think that I confused PAM2803 with PAM2805, but maybe I did? Who knows. I didn’t actually create PAM2803 a part in Eagle, I used another part with the same footprint. Maybe that’s how I got confused (all pins are labeled incorrectly of course ;) ). In any case, you are correct the output cap should be elsewhere.
  2. Looks like you’ve done all the hard work for us.
  3. It’s too bad we can’t do better than 370mV, but at a glance you do seem to be right! With that in mind this inexpensive diode seems to be a good selection for the project.

I went ahead and ordered 3 fresh donor boards from FT.

Does anyone in the USA have a few of the schottky diodes to spare? I don’t want to mooch, but I also am not ready to place an order with Mouser/Digikey at the moment. I’d be happy to trade something to cover the parts and postage (mosfet(s), 7135(s), PCB (s), etc).

Dang Alex - you mean the standard diodes we use on our regular Atmel based boards? Just got in 100 from Mouser, so I will gladly send you a few. Can't get them in the mail til tomorrow AM, but don't you dare try to send me anything for payment .

Please pm me - I'd love to help you out - least I can do...

Ah, no. I should have been more specific! I mean the ones HarleyQuin refers to in post #35. This is a diode which is part of the boost circuit such that the whole LED current will flow through it. Our “normal” protection diodes do not handle a high enough “IF” (forward current) to be used in that role, despite being generally the same type of diode. I appreciate the offer all the same though.

OK, no prob . Didn't check the entirety here what's goin on... Think I'm 0-2 , I'll just hang on the side...

I have a spare boost circuit from a thorfire tg06 (AA
/14500). The inductor top is chipped but I think the coil is intact. Would that have parts you could use? If so, send me your address and it’s yours.

I saw that a few months ago, what a great thread, one of the best I’ve came across at cpf!

Don’t sweat it. I know that if you’d had what I needed you’d have gladly sent it over. It appears that I left quite a bit of wiggle room for that diode in my own layout. I think that I have some much larger diodes on hand here with similar forward voltage drop. It’s not really the same thing but it’s probably enough to get things started if I can’t scrape up any BAT60A for the moment.

As for the link to the wrong thread - let’s face it, we are interested in both topics! I want a high-quality single-lion boost driver for 6V LEDs as well, it’s just as separate project. I’m glad you linked to it.

Thanks Gunga. If it’s PAM2803 (learn to identify w/ #10 here) you may want to hang onto it so that you can build one of these yourself. :wink: Since I’ve already got 3x of the donor driver on order from FT the only thing I’m really looking for ATM is that specific diode mentioned above. Once we can insure that that diode performs as expected and nail down another thing or two the layout may change. (Since that diode is so small it allows for changes.)

This person is awesome. They seem to be what most of us dream of being… designing and building drivers, doing firmware, turning flashlights, knurling, etc… all the while documenting it with pictures. I’ve jumped into the thread a couple of times since it was linked in this one - lots of great work to see.

Thanks, lots of heavy reading but I plowed through some this morning then looked at the images. What I had in mind was not so ambitious but while trying to generate a 2803 eagle part was mulling this over. With the device done I’ve started on a schematic for two boost ic’s. Not on at the same time but connected to 13A pins 5/6 and pam switch pins(simple on5 or on6, no pwm). Should be room for that if all the caps are on the bottom. Might even be able to put the shottky in the middle of the spring pad. As you say, there’s probably not room for the complexity of the dqg driver on a 15mm driver and I’d prefer to to maximize efficiency by avoiding pwm. Is there an inductor device in eagle libraries the correct size or do I need to make that library part too?

wight wrote:

Does anyone in the USA have a few of the schottky diodes to spare? I don’t want to mooch, but I also am not ready to place an order with Mouser/Digikey at the moment. I’d be happy to trade something to cover the parts and postage (mosfet(s), 7135(s), PCB, etc).

Those 3 amp jobbers HQ alerted to? I have some that I picked up off Ali. Haven't used any yet. You're welcome to a handful. Just PM my your address and I will get them in the mail tomorrow.

No problem. I wasn’t really trying to suggest that your idea was as complex as the DQG driver or that it needed to be. Make sure you’ve got space in the light for all of that stuff you want to put on the bottom. I don’t think that all AA lights which use 15mm drivers have space for components/springs/etc on the bottom, right?

I’m right there with you on avoiding PWM, although I don’t know how much it’s going to improve efficiency…

As far as an Eagle Part, what I’ve got on my (untested!) PCB is POWER INDUCTORS SMD CR Series with “CR32” marked for both the Package and the Value. Looks like it’s from the stock “inductors” library. I don’t know that it’s the right one or a good fit for the harvested component; I think that the Eagle part is around a 3.5mm or 4mm inductor. I just put mine on big pours for now so that it can be fixed with scraping. I sure do mention scraping in a lot of my driver threads… :wink:

Thanks ImA4Wheelr, PM sent. I’ll be frying PAM2803’s before we know it. :wink:

Thanks Alex, I saw that part # but the eentsy picture in the library window doesn’t help much. As you say, as long as it’s close then good enough for now. I really have no idea what the best way to proceed is just figured on throwing ricks until something breaks. Hopefully several approaches can lead to one good one.

Wight, really happy to have you back in.

.

This picture is closest to a ‘stripped’ version of the FT driver I have at hand. It’s the exact driver I did modify, pic taken while in the process of swapping MCUs. What you can’t see is the connection between pin4 and pin6 under the PAM2803.

.

.

There is the other thread where we discussed the boost driver and @ImA4Wheelr had drawn a schematic of the FT driver. Just borrowing, hope that’s ok.

My successful Mod was this:
[EDIT: new picture, first picture was missing the connection between MCUpin6 and FetGate (the PWM signal).

.

.

Thanks for stepping in :beer:
The BAT60A is quite easy to source in Europe but plays hard to get overseas.

I had briefly tested the BAT60A on my modded FT driver as well and it worked.

.

This is the layout I used for the board, presumably following the circuit of the modded FT driver.
Including R2 and R3 (well, it had worked with them…).
C2 and C3 are in parallel, C2 as close to PAMpin5 as possible and C3 as close to MCUpin8 as possible
R and C (0805 or 0603) are almost all reduced in size.
L is 42mm and the part selfmade (part is only tDocu, pads are simple layer 1 current paths). Eagle inductors are scarce.
Overall size is 15.4mm to ensure a GND ring of about 15mm.

Well, FYI:

.

.

HarleyQuin, that’s a pretty clear pic. Looks like they’ve bumped down from R150 to R120 for a small increase in output? Probably just whatever was cheapest at the time but R12 is the right component since FT lists it as an 800mA driver.

HarleyQuin and I recently talked about what the two “extra” resistors are doing in there. I noted that they were not part of the application circuit and HQ pointed out that they are present on the working design sold by FT. The resistors in question are the two 0603 seen in post #57. According to HarleyQuin’s circuit the 33k one connects gate to GND (through Rsense) and the 150k one connects Vout/Vcc to GND directly. The 33k one seems to be a pulldown resistor for the transistor, but the 150k one… no idea really.

The following is probably another case of “whatever is cheapest” but I noticed that on one of the no-MCU versions FT sells there is clearly an R150 for sense and an R150 connecting LED- across where the missing transistor is. It’s hard to see on the other no-MCU listing, but I think that one shows a zero-ohm jumper. Again, I seriously doubt that any of that is relevant.

FWIW what both no-MCU examples do show is that these two resistors are not populated (333/33k and 18D/150k) when the MCU/transistor is not populated! Maybe the 150k is a bleeder to help ensure clean mode changes? This is something that RMM just pointed out in another thread when the topic of using an LDO in a clicky came up. I feel decently confident on that now, that’s probably what we are looking at.

Come to think of it, does HQ’s recent pic showing the FT board with MCU removed show that the MCU ground is on the sense side of Rsense? (eg both the pulldown and MCU GND are not connected directly to GND?) Is there any significance to that?

This might be due to differences in the MCUs (Attiny vs PIC?)

When I first swapped MCUs I connected pin4 of the Attiny13a as it was with the original MCU, not directly to GND but to the other side of Rsense. That did not work. Modes would not change (I briefly wrote it in Post#57). So I furthermore cut the trace to MCUpin4 on the board and connected MCUpin4 to GND and the driver worked.

Thanks HQ, that gives me a bit more confidence to proceed. I doubt I’ll be as creative as you but it will at the very least waste a few days.