Re: [lm-sensors] [PATCH v3 7/7] hwmon: (w83627ehf) Use 16 bit fan

[Guenter Roeck <guenter.roeck@ericsson.com>]

On Sat, Feb 12, 2011 at 01:56:09PM -0500, Andrew Lutomirski wrote:
> On Sat, Feb 12, 2011 at 12:55 PM, Guenter Roeck
> <guenter.roeck@ericsson.com> wrote:
> > On Sat, Feb 12, 2011 at 11:01:11AM -0500, Andrew Lutomirski wrote:
> > [ ... ]
> >
> > Actually,
> >
> >> fan2_min = 200 (div = 64):
> >> fan2_input = 138 (read at a different time, so may differ a little)
> >>
> >> [56013.556243] Reg 0029 = 99
> >> [56013.556286] Reg 0630 = 0046
> >> [56013.556453] Reg 0658 = 0089
> >>
> >> fan2_min = 502 (div = 32):
> >> fan2_input = 0 (read at a different time, so may differ a little)
> >>
> >> [56085.471567] Reg 0029 = 00
> >> [56085.471644] Reg 0632 = 0000
> >> [56085.471804] Reg 0658 = 0000
> >
> > Must have been blind. So the RPM register returns 0 as well. Not good.
> > Wonder what is going on.
> 
> So far, the highest number I've been able to get out of the counter
> register is 0x2e8 by setting div = 16 and playing with the pwm.
> 
> If I set div = 8, then I can't get much above 0x10b.
> 
> I have no idea what's going on either.
> 
> I tried writing 0x80 to 0x508 (which is supposed to turn the whole
> thing on), but the only bits that seem to stick are 0x7, so I have no
> idea what's going on.
> 
Here is what I got from Nuvoton.

The chip has an internal "abort time" when waiting for the next pulse from the fan.
If this abort time expires and no pulse was received, all fan count and rpm registers
will return 0. This can happen if 1) the fan speed is very low and/or 2) if
the divisor setting is too low.

This means we can not depend on a reading of 0xff to determine if the divisor
is too low. A reading of 0x0 (for 8 bit and 16 bit fan count registers as well
as RPM) can also indicate that the divisor was too low.

I'll have to think about how to proceed. Best approach might be to increase the divisor
if the fan count returns 0. If anyone has a better idea, please let me know.

Thanks,
Guenter

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors