Battery Witness on

The battery witness is turned on when the engine starts, this is like this for more than 20 minutes, it goes out and load normally, alternator was checked and it is correct. The vehicle is a Mazda 323 F year 2001

The alternator has its regulator? or is the computer regulates it? If it is the computer check the ground or battery cables, (check the battery to see if it is inflated or boot X the top) and check connecting the positive terminal with the switch in off to check some current consumption

Bata, the battery has approximately 1 year and this OK, the alternator has its regulatory box, what I have investigated is that the CPU comes out a signal for the witness and that the CPU sends signal to the alternator to work. I want to check the CPU pins to see that they are not wet or with corrosion, the doubt is that I do not know where the CPU is in this vehicle. Now the fault is rare because once the alternator passes normally.

the alternator well? you're sure? When this happens it is due to dirt in the guides of the brushes that are stuck and when it heats down and make contact or the collector of the alternator (where they touch the brushes) it is worn and the brushes Apolla where it should not and does not make contact, there is a pump and it gives them to work.

Rare things are seen every day, but I work to say that there is no vehicle that the alternator has something to do with a switchboard. Imagine for a moment that the regulator or diode plaque is broken, UUUUUUF central to the garbage.

Try this, put a housing mass of the alternator to chassis. and put a fat cable from the alternator to + direct battery and tests to see if the failure continues.
If you continue, disconnect the escitation and put a positive cable with a bulb in medium, if that bulb does not turn off alternator.

When solutions tell us how I have gone.