I think TK’s comment on initial boot makes sense(also the only part I understood, oof!) since with shtdn tied to pwm the boost ic wouldn’t be able to turn on to waken the mcu in the first place. Logic trap.
bicycle/fish. To smooth out PWM you’d use a big RC circuit, but as you said the cap would be ridiculous.
IF we had the space, and we don’t, we could attempt to bias the FB pin using an op-amp and a small voltage source based on an RC circuit which was fed with PWM from the MCU. EDIT: this is something I’d like to play with in 17mm though.
Good enough, heh. 
Keep these explanations up wight and you may have me understanding whats going on.
Bumping a good thread. Anything new?
Hi RBD
Thanks for asking.
I’m still on this project and will be prodding further when I wrapped up HQ10D, UniversalBreakoutBoard and BLF SRK FET V3.
Just as time permits.
Short status here:
HQB15 works fine
After what the scope guys brought up I will keep the 2 resistors in the design
Cell gets drained very quickly without LVP although the led is off.
I strongly doubt our usual Lvp will work in 1aa config (no voltage for mcu to regulate anything.
I even have doubts that lvp by cutting pwm to the fet will work at all, as the boost circuit is not cut (i want to test that).
I’m about to make the 17mm version based on the SS24 diode, hoping to make it easier for others to try this if they want to.
CU
Is there any type of component which would be open by default, but which could be closed to block current when it receives a signal? (then would stay closed until power is disconnected and reconnected)
If so, something like that could be used for LVP. Put it before the boost circuit and it could maybe cut the power on command. I don’t know if something like that exists though.
Yes, perhaps we can use pin4 of the boost controller (“SHDN”)
In 2aa config switching to gnd will shutdown the pam2803
I can connect mcuPin3 with pamPin4
@Toykeeper, can the mcuPin3 deliver Gnd as output and how (which Register do I set how)?
Voltage can still be measured via mcuPin7—- probably only 17mm Board
In 1aa mcu wont work in SHDN as mcu needs >1.8V
Random confusion:
1, Where is the parasitic drain, in the boost ic?
2, I thought that 13A Vref was somewhere south of 1.1V so B+ straight to pin 7 without R1, just a high value R2. Since the boost ic works down to .9V the mcu will still be awake as long as it needs to be for a nimh(min 1V).
3 re 1, How much drain does the boost ic represent on its own(quiescent current?
It sounds like this is an issue only for a light left switched on and forgotten. Complete shutoff might not work but a slow beacon mode might be appropriate(always wanted one of those without all the other blinkies that usually come with it).
1. I believe so. Looks like the boost circuit is constantly trying to boost when the eneloop regains some power, then dropping again under slightest load
DMM always shows rising and falling voltage at the eneloop
2. vref is not the point yes its 1.1v
Attiny needs acc to datasheet 1.8v to work
3. in SHDN pam2803 needs 1yA when pin4 is pulled to gnd (datasheet pg 3)
My idea for 2aa is to measure voltage with mcuPin7 and switch the pam2803 with mcuPin3
Give me 24 hours and I come up with a oshpark Board
Yes it’s only needed for Lights left on
Drains eneloop within a day it seems, Rough
Well, i could live with a dead eneloop once in a while
Most boost driver dont have LVP anyway
Looking at HJK’s review of the nanjg 110 running off a single AA Eneloop and referencing his run time tests of that cell it looks like the stock driver can just stay above the 1.1v mark for most of the run at @ 2A draw so it should be able to keep the LED and mcu powered at least that long. It seems like serendipity that Vref is .1V above where we’d want to stop anyway so it makes a good LVP when B+ drops below Vref. With the boost ic functioning down to B+ = .9V the mcu should still be powered sufficiently to at least signal low voltage via the fet even if it can’t completely shut down the driver. It doesn’t look like this would work for primary cells too well as their discharge curves aren’t nearly as flat.
I get the feeling we’re not quite in sinc on the Vref vs mcu power thing. The mcu is connected to boost ic (out) so as long as the boost ic has .9V input then the mcu should have more than 1.8V and be awake, right? If LVP is triggered at 1.1V(light starts unique to LVP blink mode) then we have ~ .1V warning before the cell drops to 1V and another .1V before the pam2803 can no longer power the led and mcu but continues to drain the cell.
@RBD
I think you are right with all you said, my worries are set after the timeframe you described, thus after the shutdown in 1AA mode (= no more boosting and Vbat below 1.8V).
Then MCU is unpowered and I have not the faintest idea whether mcuPin3 (in my idea connected to pamPin4) will keep the state we need it to be (like Low/GND/Zero…), when the MCU is not powered enough to function as the controlling unit.
I’m pretty hopeful we can make it shutdown at a reasonable voltage, question is: will it stay shut down? Unless anybody can tell us, we have to try.
I still can only guess whether we can toggle the SHDN of the PAM2803 with Pin3 (PB4) of the Attiny anyway…
My idea is something like
DDRB |= (1 << PB4); // set Pin3 to output
PORTB &= ~(1 << PB4); // turning off Pin3
This works very well with the AMC7135 for a nice triple channel driver, so this would be my first try here.
Thanks for the clarification HQ, that’s why I suggested a dedicated low voltage output mode instead of true shut off but if you can figure it out it would certainly be worthwhile. If your successful in toggling shutdown and keeping it down after the mcu is off, how do you reverse that once the cell is recharged.
With this (yet to be tested) design the firmware needs a line to set mcuPin3 to “High” at startup anyway, otherwise the boost circuit will not start.
So when power is restored, pamPin4 should be toggled high that way. Like everytime the clicky was pressed in normal operation when the cell is still good.
Care for some testing, anyone?
Just asking because Oshpark boards need 4-6 weeks from ordering until they are at my doorbell.
This asks for some patience for each new generation of boards.
HQB17B v2
17.6mm single sided boost driver for ATtiny13A
Just to make it easier to build, I based it almost completely on the Fasttech boost driver
This was not possible on 15mm, so this one is 17mm size (oversized to 17.6mm)
Designed for LVP by
- measuring voltage via ATtiny pin2 and R4, R5
- toggling the PAM2803 via its pin4 (SHDN) by setting the ATtiny pin3 (PB4)
Based on the Fasttech-driver components (PAM2803, L1, D1, C1, C2, R1, R2, R3, FET)
Link removed: I mislabeled R2 and R3 (swapped them).
MCU: ATtiny13A SSU
C3: optional decoupling capacitor for MCU
OTC: optional offtime capacitor
R4, R5: voltage divider resistors
.
I have some nanjg 110’s from Rich and I think a few other smaller single mode boost drivers from KD but not that one and no way to flash the mcu. Phooey.
That was close, huh? 
I can order some FastTech boards and Oshpark pcb’s and maybe even swap the parts even though I don’t have a HAR gun but then I’d need to send them off to someone who can program them. Maybe if I take the inductor off last and install it first I can use the connected pads to reflow the solder under it. Anyone in the SF Bay Area do flashing? (Please not the raincoat kind)
No sweat.
I really can’t recommend hand-soldering on these (I use a simple Youyue858D+)
And I’m still not convinced this LVP thing is really truly worth the effort we need to put into this.
I’m about to finish the 17mm board for a simple FT-driver swap plus ATtiny13A, that might be a better and easier starting point.
As far as the 15mm line goes, I’m still happy with the HQ15B v2 approach. I got them as 0.8mm board already, but the fab shifted the soldermask. So before I re-order them, I’ll take the opportunity to fine-tune them and then, in 4-6 weeks …
Mostly I’ll use any 15 mm version for presents/gaws for people that don’t use rechargables anyway. It would be nice for nimh cells though. At a minimum a low voltage signal could be done.
How much does the voltage from the buck ic drop as the cell(s) discharge? I know with one cell it falls right out the gate.