[lm-sensors] ADT7463

[lm-sensors] ADT7463

[Paul Aviles]

David, I emailed Margit, but have not received a reply.  There must be
something goofy with this and I cannot point to it. My inclination will be
that something in the hardware, as the software has not changed for the
results I have. 

This unit has a ADT7463 chip on it. I did a test to see if the BIOS based
fan and temperature controls were working right. 
The fans that fails is #3, one of the 2 CPU fans. Fans #1 which is left to
#3 is working perfectly. So is fan #1 in the back of the unit.

I changed the pwm for fan #1 and set it to 255 and rpms go up to like 13K.
The temp of the cpu goes down and so the other two sensors for local and sys
temp. I had it running like this for about 5 min and I was expecting fan #3
to go down in rpms as a result, but nothing. It stayed at the very same rpm.

Then I ran fancontrol and all 3 fans respond to it, with the exception that
with the other 2 fans, the moment you change pwm it has a direct
correlation. I normally have pwm set at 64 or 32 which is more than adequate
for temperature control. I can set fan #3 to pwm of 3 and it does not come
down from 9400 rpms or so. If I do this to the other 2 fans they will
probably stop of come very close to it. Is like the response curve for pwm
is not the same for fan #3 and the other 2 fans.

So, I don't know what is causing this, but is very irritating the noise this
units make.  Could sensors be causing some corruption like with the
thinkpads in some memory or register?

Is the fact that the other 2 fans work perfectly and #3 does not an
indication of bad hardware?

Regards,

Paul Aviles

[lm-sensors] ADT7463

[David Hubbard]

Hi Paul,

> David, I emailed Margit, but have not received a reply.
I think you will need to include lm-sensors at lm-sensors.org in your
message to get past spam filters, etc.

> This unit has a ADT7463 chip on it. I did a test to see if the BIOS based
> fan and temperature controls were working right.
> The fans that fails is #3, one of the 2 CPU fans. Fans #1 which is left to
> #3 is working perfectly. So is fan #1 in the back of the unit.
>
> I changed the pwm for fan #1 and set it to 255 and rpms go up to like 13K.
> The temp of the cpu goes down and so the other two sensors for local and sys
> temp. I had it running like this for about 5 min and I was expecting fan #3
> to go down in rpms as a result, but nothing. It stayed at the very same rpm.
>
> Then I ran fancontrol and all 3 fans respond to it, with the exception that
> with the other 2 fans, the moment you change pwm it has a direct
> correlation. I normally have pwm set at 64 or 32 which is more than adequate
> for temperature control. I can set fan #3 to pwm of 3 and it does not come
> down from 9400 rpms or so. If I do this to the other 2 fans they will
> probably stop of come very close to it. Is like the response curve for pwm
> is not the same for fan #3 and the other 2 fans.
>
> So, I don't know what is causing this, but is very irritating the noise this
> units make.  Could sensors be causing some corruption like with the
> thinkpads in some memory or register?

No. The thinkpad bug was an eeprom on the i2c bus. That's not related
to your hardware monitor.

> Is the fact that the other 2 fans work perfectly and #3 does not an
> indication of bad hardware?

That's what it sounds like. If you want to take the time to test it,
we can design some electrical tests for the outputs of the fans.
However, it may require soldering, and if you short the fan pins, you
can damage the motherboard, so I will only make suggestions if you
think that's what you want to do. If you can get the motherboard
replaced, that may be the best thing to do. The ADT7463 is not being
corrupted by the linux driver.

David

[lm-sensors] ADT7463

[Paul Aviles]

David, I got another board from Tyan and the same crappy problem. One thing
I found today by playing with fan1. The connector has 3 cables. Looking it
from one side are blue, red and black. If you disconnect the fan obviously
sensors detects it and flags an alarm. I then separated the red cable from
the fan molded connector and used a regular hard drive power connector (4
cables, yellow, black, black, red) and connected them both. I think that is
+5V. That way it goes from the power supply directly to the fan bypassing
the motherboard. The other 2 cables, blue and black are still connected to
the fan molded connector and to the motherboard.

This spins the fan around 6K rpms and makes sensors still see the fan.
However, you cannot control it.  I don't know exactly what this means, but I
can only guess that either something funky is going on with the adt7463 or
with some of the pwm circuitry maybe in another IC or resistor.

Your thoughts?

Thanks

Paul


-----Original Message-----
From: David Hubbard [mailto:david.c.hubbard at gmail.com] 
Sent: Monday, December 18, 2006 7:34 PM
To: Paul Aviles; lm-sensors at lm-sensors.org
Subject: Re: ADT7463

Hi Paul,

> David, I emailed Margit, but have not received a reply.
I think you will need to include lm-sensors at lm-sensors.org in your message
to get past spam filters, etc.

> This unit has a ADT7463 chip on it. I did a test to see if the BIOS 
> based fan and temperature controls were working right.
> The fans that fails is #3, one of the 2 CPU fans. Fans #1 which is 
> left to
> #3 is working perfectly. So is fan #1 in the back of the unit.
>
> I changed the pwm for fan #1 and set it to 255 and rpms go up to like 13K.
> The temp of the cpu goes down and so the other two sensors for local 
> and sys temp. I had it running like this for about 5 min and I was 
> expecting fan #3 to go down in rpms as a result, but nothing. It stayed at
the very same rpm.
>
> Then I ran fancontrol and all 3 fans respond to it, with the exception 
> that with the other 2 fans, the moment you change pwm it has a direct 
> correlation. I normally have pwm set at 64 or 32 which is more than 
> adequate for temperature control. I can set fan #3 to pwm of 3 and it 
> does not come down from 9400 rpms or so. If I do this to the other 2 
> fans they will probably stop of come very close to it. Is like the 
> response curve for pwm is not the same for fan #3 and the other 2 fans.
>
> So, I don't know what is causing this, but is very irritating the 
> noise this units make.  Could sensors be causing some corruption like 
> with the thinkpads in some memory or register?

No. The thinkpad bug was an eeprom on the i2c bus. That's not related to
your hardware monitor.

> Is the fact that the other 2 fans work perfectly and #3 does not an 
> indication of bad hardware?

That's what it sounds like. If you want to take the time to test it, we can
design some electrical tests for the outputs of the fans.
However, it may require soldering, and if you short the fan pins, you can
damage the motherboard, so I will only make suggestions if you think that's
what you want to do. If you can get the motherboard replaced, that may be
the best thing to do. The ADT7463 is not being corrupted by the linux
driver.

David

[lm-sensors] ADT7463

[David Hubbard]

Hi Paul,

As usual, I'm squeezing too many things into too little time, but I'll
throw out some ideas. Apologies for not suggesting more in terms of
measuring the PWM peak voltage / PWM frequency / duty cycle without an
oscilloscope -- obviously you could do all that easily using the right
tools. The PWM frequency is the most non-obvious contributor to fan
control problems. (Since the duty cycle is directly controllable, and
voltage problems are usualy catastrophic, needing a board
replacement.)


> David, I got another board from Tyan and the same crappy problem. One thing
> I found today by playing with fan1. The connector has 3 cables. Looking it
> from one side are blue, red and black. If you disconnect the fan obviously
> sensors detects it and flags an alarm. I then separated the red cable from
> the fan molded connector and used a regular hard drive power connector (4
> cables, yellow, black, black, red) and connected them both. I think that is
> +5V. That way it goes from the power supply directly to the fan bypassing
> the motherboard. The other 2 cables, blue and black are still connected to
> the fan molded connector and to the motherboard.

I wonder if the fans you are using have a higher current rating (or,
equivalent, higher wattage) than the ADT7463 or Tyan board are
designed for. You'd need to consult the motherboard manual and ADT7463
datasheet, but if no information is there, it could still be a
problem.

> This spins the fan around 6K rpms and makes sensors still see the fan.
> However, you cannot control it.  I don't know exactly what this means, but I
> can only guess that either something funky is going on with the adt7463 or
> with some of the pwm circuitry maybe in another IC or resistor.

The ADT7463 datasheet should show a pinout. With an oscilloscope, you
could compare the voltage on the pin of the ADT7463 and the voltage at
the fan. That would indicate if any other circuitry is involved.

If the voltages (and waveforms) are pretty much the same, I still
wonder if the ADT7463 can put out the amperage / wattage to drive your
fans.

If it's not a problem driving the fans, you might need to change the
PWM frequency. This might be as simple as writing to the correct file
in sysfs. Or you might need to write support for it in the driver. I
don't know if the ADT7463 can even change its PWM frequency, but it's
a normal feature of PWM controllers, so I expect it can.

I was thinking of a diode-capacitor setup that would allow you to
measure peak voltage with a voltmeter, in the absense of an
oscilloscope. You could measure the PWM frequency using your
soundcard, if you have a good audio recording program (e.g. audacity
will work) once you are certain that the voltage range is reasonable.
You'll want to cut the voltage down with a voltage divider (a few
resistors) first, then route the signal to an input on your soundcard
(there may be a CD audio input you could attach to). None of the
components are very expensive (total price would be about US$0.25),
and it sounds like you're willing to work with these boards to find
out a solution. An oscilloscope would really be the best way.

I haven't looked at the ADT7463 datasheet, so these are random guesses
that might help you with the troubleshooting. Further, you'll want to
be really careful putting probes on your motherboard. You could damage
it (shorting out the ADT7463 would probably break important
motherboard functionality) ... proceed at your own risk, etc., etc.

I'll keep up on this thread and reply when I can,
David

[lm-sensors] ADT7463

[Enrico Ng]

I have a JVC Mininote XP-MP7431 laptop.  It is pretty old but I know
that I can control the fan using speedfan in windows.  I'm having
trouble getting my sensors working in linux.  I am using ubuntu.
When I do "sensors-detect" it does not detect anything.  I've tried
loading lm85 but that does not make any difference.  I'm wondering if I
need to recompile the kernel to add support or something.  I'm not sure
whats already in there as I can't find the .config.

I'd appreciate any help.
Let me know if you need the output from any commands, I'm unsure about
what to attach.

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

Re: [lm-sensors] ADT7463

[Hans de Goede]

Enrico Ng wrote:
> I have a JVC Mininote XP-MP7431 laptop.  It is pretty old but I know
> that I can control the fan using speedfan in windows.  I'm having
> trouble getting my sensors working in linux.  I am using ubuntu.
> When I do "sensors-detect" it does not detect anything.  I've tried
> loading lm85 but that does not make any difference.  I'm wondering if I
> need to recompile the kernel to add support or something.  I'm not sure
> whats already in there as I can't find the .config.
> 
> I'd appreciate any help.
> Let me know if you need the output from any commands, I'm unsure about
> what to attach.

Hmm,

That chip should be detected just fine, can you please try this version of 
sensors-detect:
http://www.lm-sensors.org/browser/lm-sensors/branches/lm-sensors-3.0.0/prog/detect/sensors-detect?format=raw

(All you need to do is safe the one file and make it executable)

And include the output of running this in your next mail?

Regards,

Hans

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

Re: [lm-sensors] ADT7463

[Hans de Goede]

Enrico Ng wrote:
> On Thu, 01 Jan 2009 15:24:38 +0100
> Hans de Goede <hdegoede@redhat.com> wrote:
> 
>>
>> Enrico Ng wrote:
>>> I have a JVC Mininote XP-MP7431 laptop.  It is pretty old but I know
>>> that I can control the fan using speedfan in windows.  I'm having
>>> trouble getting my sensors working in linux.  I am using ubuntu.
>>> When I do "sensors-detect" it does not detect anything.  I've tried
>>> loading lm85 but that does not make any difference.  I'm wondering
>>> if I need to recompile the kernel to add support or something.  I'm
>>> not sure whats already in there as I can't find the .config.
>>>
>>> I'd appreciate any help.
>>> Let me know if you need the output from any commands, I'm unsure
>>> about what to attach.
>> Hmm,
>>
>> That chip should be detected just fine, can you please try this
>> version of sensors-detect:
>> http://www.lm-sensors.org/browser/lm-sensors/branches/lm-sensors-3.0.0/prog/detect/sensors-detect?format=raw
>>
>> (All you need to do is safe the one file and make it executable)
>>
>> And include the output of running this in your next mail?
>>
>> Regards,
>>
>> Hans
> 
> Here is the output, and I've attached some other output you may find
> useful.
> 
> In /proc/acpi/fan, I have one fan but it only displays on/off state
> In /proc/acpi/processor/CPU1/ it has info  limit  power  throttling
> In  /proc/acpi/thermal_zone/THRM/ it has cooling_mode
> polling_frequency  state  temperature  trip_points
> 

Ah, ok, so you've no (visible) smbus adapters, and acpi thermal management, 
that means that the BIOS has hidden your smbus adapter, this can be worked 
around, but this is a *BAD* idea, as this means that ACPI is using the smbus, 
and we thus shouldn't touch it.

The fact that speedfan under windows does allow you to touch your hardware 
monitoring is not really relevant in this case. We can drive / exposue your 
hwmon IC, but we *really* don't want to.

Regards,

Hans

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