Op 29 mei 2008, om 14:42 heeft Peter 'p2' De Schrijver het volgende geschreven:
Hi,root@beagleboard:/sys/class/hwmon/hwmon0/device# cat temp1_input 96 does anyone from TI have a hint if the chip is really that hot, or that we need to scale the value?Maybe it's in Fahrenheit?That was my first thought, but looking that the omap3530 TRM (spruf98.pdf) on page 836: ADC code Temperature °C From To 0 -40 -40 32 -0.7 0.8 64 44.9 46.3 96 89.3 90.6 And that maps to the adc_to_temp var at line 69 from http://source.mvista.com/git/gitweb.cgi?p=linux-omap-2.6.git;a=blob;f=drivers/hwmon/omap34xx_temp.c Even if it didn't map, the raw ADC value is only ±6 °C off from the real temperature around that point. So the adc value the kernel reads gets translated correctly to °C, so if we think it's incorrect, the kernel might be reading the ADC wrong.The driver basically does what's described in figure 7-16 of theomap3430 TRM. Ie, rise SOC, wait until EOCZ rises, lower SOC, wait EOCZ goes low, read result. The timeouts are taken directly from this figure.So I can't really see what the driver would be doing wrong, unless obviously the hardware doesn't work the way the TRM describes :)
Some interesting observations: root@beagleboard:~# cat /sys/devices/platform/omap34xx_temp/temp1_input 45 root@omap3evm:~# cat /sys/devices/platform/omap34xx_temp/temp1_input 80The beagle is using an ES3.0 omap3, the evm an ES2.1 omap3, so I suspect something got fixed in between :)
regards, Koen
Attachment:
PGP.sig
Description: This is a digitally signed message part
