On Tue, 3 Jun 2003 20:09:43 +0200, Stef <s.mientki@MAILBOX.KUN.NL> wrote:

> thanks Diego !
>
> It's well documented
> and I even believe I can read some words in Spanish now  ;-)
>
> I've a few questions/remarks:
> 1. do you wait long enough, after the discharge pulse, before starting
> the measurements ?

Code line #629 (v089):

      Carga_ON;
      DelayBigMs(tc);
      Carga_OFF;
      DelayMs(4);
      LeeVal();
      [...]
      Descarga_ON;
      DelayMs(td);
      Descarga_OFF;
      DelayBigMs(tr);

(tc=800ms, td=41ms, tr=135ms)

Charge at 1A is on for tc, 4ms rest time, then measurement. I made some
tests and change de LeeVal() to the end of the pulse train (800ms charge,
4ms rest, 41ms discharge, 135ms rest, 20ms while measurement) with no
visible changes in the result.

> 2. why not sample as fast as you can, to increase the signal to noise
> ratio ?

The noise is not such noise. It seems to be made by the battery itself. The
battery voltage is not noisy while it is not in the charging process (i.e.,
during discharge). DonB4t ask me why :-)

Quoting what David VanHorn said in an old message:

"Your readings are likely distributed along a slope, where the battery
voltage is changing rapidly. I've seen this before. The steepness of the
slope will depend on what the recent activity is, and how long ago it was."

> 3. as far as I can see, you calculate the slope, between 2 consecutive
> charge pulses,

Well, it is a butterworth filter. The one I coded mix the last two readings
and the last output value from the filter.

> whereas Galaxypower did it by linear regression over 12
> charge pulses (I've put this note on a web page, see:
> http://oase.uci.kun.nl/~mientki/PIC/Projects/Reflex_charger.htm

Will check this too. I tested some types of filters and the one I use give
me the best results. But I never tested a linear regression.

Cheers,
Diego.

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body