From: Jerry Van Baren <gerald.vanbaren_at_smiths-aerospace.com>
Date: Tue, 8 Feb 2005 13:57:28 -0500

Matthias Klumpp wrote:
> Hi Jerry and all others,
>
> # -----Ursprüngliche Nachricht-----
> # Von: rockbox-bounces_at_cool.haxx.se
> # [mailto:rockbox-bounces_at_cool.haxx.se]Im Auftrag von Jerry Van Baren
> # Gesendet: Montag, 7. Februar 2005 03:34
> # An: rockbox_at_cool.haxx.se
> # Betreff: Re: More aggressive Rockbox battery charging
> #
> # I updated the patch and the ajbrec.ajz file.
>
> <snip>
>
> # Matthias: please try the new load.
>
> <snip>
>
> The test is now completed.
> CONGRATULATIONS - I guess that's it :-)))
> Don't know the original english quotation, but it should be something like
> this:
> "a short step for Jerry, but a big step for 'Rockbox-hood'...." ;-O))
>
> Ok, here's what happened/what I did:
> - V2.4 flashed
> - your 2nd .ajz rolo'ed
> - Deep Discharge = off
> - Trickle Charge = on
> - Battery cap. = 2000 (as the batteries are)
> - Bat's previously with Archos FW charged "fully" until it displays
> "charged"
> - Test started at 18:30
> - Bat. symbol static at 100%
> - Remaining time showed 10h25m
> - at (more or less) 20:00 (I was not seeing it live time), bat. symbol
> started to "run", rem. time still 10h25m
> - then it run during the night playing always the same dir (boring job ;-))
> - I found the JB this morning at 7:30 still playing
> - it did NOT got stuck this time
> - bat. symbol still running, rem. time was 6h25m
>
> So this is nearly 12h of playing while charger is connected.
>
> If you need some more information or any other kind of test I should do,
> please let me know (at least this would be my way of contribution).
>
> How it will go on ?
>
> Will there be some more improvement, like Jörg purposed (?):
> - "over thinking" the necessity of setup options "deep discharge" and
> "trickle charge"
> - immediately start charging after ON
> - measuring input voltage
> - estimation of input current
> - adaption of duty cycle
> - trickle charge for ever when deltaV fails
>
> Is there a plan, when your patch will be available in the daily builds ?
>
> Thanks
>
> Matthias

Excellent, now that we know the problem, the solution shouldn't be too
far behind :-0.

I was running the charging numbers and thinking about this a bit while
you were running tests.

Jens Arnold strongly recommends that we not exceed 0.05C for the trickle
charge rate (e.g. 75mA for 1500mAH batteries, 110mA for 2200mAH batteries).

If the AJR is being used, we need to increase our charge current by
100-230mA, depending on backlight and HDD state, so that we supply the
operating current as well as charging current.

FAQ A12: In our testing we found the following results:
HDD off, backlight off, idle 94 mA
HDD off, backlight off, play 97 mA
HDD off, backlight on, idle 129 mA
HDD off, backlight on, play 131 mA
HDD on, backlight on, play 230 mA
HDD on, reading, backlight off ~ 600 mA
HDD spin up before read max 800 mA

FAQ A10: Here is how changing the input voltage changes the charging
current:

voltage charging current (limited only by the AJB)
  7,0V 10 mA
  7,5V 30 mA
  8,0V 50 mA
  8,5V 70 mA
  9,0V 140 mA
  9,5V 250 mA
10,0V 330 mA
10,5V 350 mA
11,0V 350 mA
11,5V 350 mA
12,0V 350 mA

Looking at the schematic, it appears that the MC34063A is used as a
constant current supply, but it achieves constant current only if it has
sufficient input voltage (>=10.5v). Assuming we can read the input
voltage (Jörg says we can), we should be able to set the proper duty
cycle based on the combination of input voltage and AJR state
(on/off/backlight/HDD activity). Of course, we would

Estimating current based on the voltages we can read remains TBD,
although I'm fairly confident it can be done with sufficient accuracy
(we don't need a whole lot of accuracy). I sent an email yesterday
trying to understand what is read with ADC channel AN7 but nobody's
answered it yet.

Now I (we) need to roll all the above into a revised charging algorithm.

gvb
_______________________________________________
http://cool.haxx.se/mailman/listinfo/rockbox
Received on 2005-02-08