[lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Max Baker]

Hi,

Newbie question here.  I have a Jetway J7F4K1G5D, w/ a VIA C7, CN700 North
bridge and VT8237R south bridge.   It has the super i/o Fintek F71805F.

I'm trying to get basic fan control working .  Right now I've got it to
the point that when i type 'sensors' I can see all the voltages and fan
speeds.

This post is a similar request:
http://lists.lm-sensors.org/pipermail/lm-sensors/2007-March/019278.html
that Jean Delvare was able to address.

Is there some magic register I can poke w/ isaset like the one in that
email in order to slow down fan#2? 

> isaset 0x295 0x296 0x90 0x00

Or can I possibly get the full fancontrol program working?    I have zero
experience in this realm and have already burned 5 hours on it (no pun
intended).

My basic problem is that i'm building a mythTv box and the case fan is on
at 8,000 rpm and it sounds like a hair dryer.  I think I could run it at
about one quarter that duty cycle and still achieve adequate cooling.

Thanks!
-m

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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Max Baker]

Update.

I found the data sheet for the fintek f71805f and started PEEKing and
POKEing in its register bank.

: FAN1 OPERATING Control Register -- Index 60h
: FAN1 OPERATING Control Register -- Index 60h
: FANPWM2 STOP DUTY-CYCLE . Index 72h
: FANPWM2 Output Frequency Control . Index 73h
: FANPWM2 STEP Control Register -- Index 74h
: FAN2 PWM_DUTY -- Index 7Bh

I am now relatively sure that this board does not have any hardware
control of the fan (PWM or DC mode).  The registers were mostly at their
default values from the datasheet and would change accordingly, so I think
i was in the right place.  I tried setting the fan to manual mode and
turning it down.  I also tried any number of other settings in the temp
monitoring registers.

All that said, is there anything else you experts could recommend I try
before packing up this board and returning it?   I don't think I want to
go with the 'roll-your-own' solution -- I'ld rather it work out of the
box.

Can anyone recommend an Mini-ITX motherboard with CN700 or CM700x chipsets
that has working hardware fan control built in?  It's critical to my
application (big drive+tv tuner+small case = heat).

Thanks all!
-m


On Sat, Jun 09, 2007 at 12:12:29PM -0700, Max Baker wrote:
> Hi,
> 
> Newbie question here.  I have a Jetway J7F4K1G5D, w/ a VIA C7, CN700 North
> bridge and VT8237R south bridge.   It has the super i/o Fintek F71805F.
> 
> I'm trying to get basic fan control working .  Right now I've got it to
> the point that when i type 'sensors' I can see all the voltages and fan
> speeds.
> 
> This post is a similar request:
> http://lists.lm-sensors.org/pipermail/lm-sensors/2007-March/019278.html
> that Jean Delvare was able to address.
> 
> Is there some magic register I can poke w/ isaset like the one in that
> email in order to slow down fan#2? 
> 
> > isaset 0x295 0x296 0x90 0x00
> 
> Or can I possibly get the full fancontrol program working?    I have zero
> experience in this realm and have already burned 5 hours on it (no pun
> intended).
> 
> My basic problem is that i'm building a mythTv box and the case fan is on
> at 8,000 rpm and it sounds like a hair dryer.  I think I could run it at
> about one quarter that duty cycle and still achieve adequate cooling.
> 
> Thanks!
> -m
> 
> _______________________________________________
> lm-sensors mailing list
> lm-sensors@lm-sensors.org
> http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Phil Endecott]

Max wrote:
> Newbie question here.  I have a Jetway J7F4K1G5D, w/ a VIA C7, CN700 North
> bridge and VT8237R south bridge.   It has the super i/o Fintek F71805F.
>
> I'm trying to get basic fan control working.

Hi Max,

I guess that by now you've found my thread about fan control on the 
J7F2, which is a quite similar board.  If you haven't, search for my 
name on this list over the last couple of weeks.

My 1.2GHz board is fanless, i.e. it has a passive CPU heatsink, and all 
three of the fan connectors are non-controllable.  However the board 
has unpopulated positions for control components for the CPU fan, and 
my guess was that they are fitted on the 1.5GHz+ versions (which ship 
with a CPU fan, as I presume yours does).  So you should be able to 
control the speed of your CPU fan.

I would be most interested to know whether you get reliable fan speed 
readings when the CPU fan is not set to full speed, so if you could 
possibly check this before you return your board I would be very grateful.

One possibility would be to connect your case fan in parallel with the 
CPU fan, allowing both to be controlled together.

You could also use a quieter, lower-volumne fan, or reduce the voltage 
to your existing one; one common idea is to wire it between +5 and +12 
so that it runs at 7v.  You can even buy inline cables for voltage reduction.

But are you sure that you need a case fan?  I have built a number of 
things with mini-ITX boards and none have had any fans at all, until my 
present 1U project.  What sort of power supply are you using?

Regards,

Phil.





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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Max Baker]

On Sun, Jun 10, 2007 at 11:35:28AM +0100, Phil Endecott wrote:
> My 1.2GHz board is fanless, i.e. it has a passive CPU heatsink, and all 
> three of the fan connectors are non-controllable.  However the board 
> has unpopulated positions for control components for the CPU fan, and 
> my guess was that they are fitted on the 1.5GHz+ versions (which ship 
> with a CPU fan, as I presume yours does).  So you should be able to 
> control the speed of your CPU fan.

No Dice.   I set 0x60 to 0x22 (Fan #1 - Manual Mode) and 0x6b to 0x01
(lowest PWM setting) and did not see any change in the fan speed.


> I would be most interested to know whether you get reliable fan speed 
> readings when the CPU fan is not set to full speed, so if you could 
> possibly check this before you return your board I would be very grateful.

Since I can't affect any change, I can't test this one for you excpet by
dragging my finger on the fan :-)  Any other tests I'll be happy to try in
the next day or so before I return the board.

> One possibility would be to connect your case fan in parallel with the 
> CPU fan, allowing both to be controlled together.

Thought about this if fan control would work on #1 and if the motherboard
could source enough current.

> You could also use a quieter, lower-volumne fan, or reduce the voltage 
> to your existing one; one common idea is to wire it between +5 and +12 
> so that it runs at 7v.  You can even buy inline cables for voltage 
> reduction.

I'm not fond of using the +5 as a ground on the +12 leg.  I'ld rather have
the full control.

> But are you sure that you need a case fan?  I have built a number of 
> things with mini-ITX boards and none have had any fans at all, until my 
> present 1U project.  What sort of power supply are you using?

I have a 500G 3.5" SATA drive and a Win-TV (happauge) PVR-350 card in
there... both of which make a lot of heat.  Inside the mini-itx case it
just gets too hot to not have a fan -- I'm worried about the long-term
reliability of both items if they sit at 50-60C all the time.  the case
was physically hot to the touch w/out the case fan.

Does anyone have any success stories for another Mini-ITX motherboard
about having working fan control?  Epia?  MSI Fuzzy?

Cheers,
-m

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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Phil Endecott]

Max Baker wrote:
> On Sun, Jun 10, 2007 at 11:35:28AM +0100, Phil Endecott wrote:
>> My 1.2GHz board is fanless, i.e. it has a passive CPU heatsink, and all 
>> three of the fan connectors are non-controllable.  However the board 
>> has unpopulated positions for control components for the CPU fan, and 
>> my guess was that they are fitted on the 1.5GHz+ versions (which ship 
>> with a CPU fan, as I presume yours does).  So you should be able to 
>> control the speed of your CPU fan.
>
> No Dice.   I set 0x60 to 0x22 (Fan #1 - Manual Mode) and 0x6b to 0x01
> (lowest PWM setting) and did not see any change in the fan speed.

Max,

I've put a couple of photos of my J7F2 board here:
http://chezphil.org/tmp/j7f2_cpufan_top.jpeg
http://chezphil.org/tmp/j7f2_cpufan_bottom.jpeg

In the top photo, you can see an unpopulated position for a transistor 
("Q1") next to the CPU fan connector.  In the bottom photo, you can see 
a "zero-ohm resistor" and some more unpopulated positions (the three 
pins on the left of this picture are the bottom of the fan connector).

If your board looks similar, you're right that you can't control the 
fan speed.  If you have a transistor and other components near the fan 
connector, then keep trying!


Phil.





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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Jean Delvare]

Hi Phil,

On Mon, 11 Jun 2007 12:55:06 +0100, Phil Endecott wrote:
> I've put a couple of photos of my J7F2 board here:
> http://chezphil.org/tmp/j7f2_cpufan_top.jpeg
> http://chezphil.org/tmp/j7f2_cpufan_bottom.jpeg
> 
> In the top photo, you can see an unpopulated position for a transistor 
> ("Q1") next to the CPU fan connector.  In the bottom photo, you can see 
> a "zero-ohm resistor" and some more unpopulated positions (the three 
> pins on the left of this picture are the bottom of the fan connector).
> 
> If your board looks similar, you're right that you can't control the 
> fan speed.  If you have a transistor and other components near the fan 
> connector, then keep trying!

Very interesting. This trick seems to work for other motherboard brands
too. I have an Asus board here where I can't control the CPU fan and
bingo! there's unpopulated room for a 3-leg component labelled Q12 next
to the CPU fan header. So it seems that they designed the board with fan
control in mind, and finally didn't solder the required components.
Maybe because they share the same design between different models.

-- 
Jean Delvare

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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Max Baker]

Phil,

I will "see" your motherboard pictures and "raise" you a schematic :)

http://warped.org/linux/j7f4r30fan.pdf

The fine, fine folks at Jetway were nice enough to send me an actual
schematic for the J7F4 board -- well only the page w/ the fans on it.
They get an A+ for customer/Linux support, buy their products.

I have some follow-up questions with them to see if this is the same
circuit for their newer boards or if they have added fan control on the
other plugs.

Here's my review of that schematic :
    * CPUFAN has PWM control, implemented in just about the same way as
      the fintek data sheet sample 3-wire fan circuit.
    * SYSTEMFAN and CASEFAN do not have control, only monitoring
    * CPUFAN PWM control is optional : Either 
        A) R3 (0-ohms) is populated and there is no fan control, or 
        B) R3 is left open and the rest of the devices connected to CPUFAN
           Pin 2 (VCC) are populated.
      I will check my board when I get home.

You will see that there is more than one transistor that needs populating.
Without the schematic on other boards, I would be hesitant to go soldering
things onto the board.  There are two transistors, two resistors, and one
resistor removal in this case.

Fan control circuit analysis:
    Q2) P-channel BJT that is the 'on-off' switch for the fan
    Q3) N-channel FET that is used to isolate Q2 from the Fintek chip
        and provide logical inversion of active-high Fintek to Q2 (pnp).
        Probably a level-change going on here too.
    R3) "option" zero-ohm resistor that selects between fan control or 
        hardwired on. essentially a jumper
    R2) Limits current into the base of Q2 -- needed to bias the BJT
    R6) Pull-up on the base of Q2 for when Q3 is off and the base would
        be floating (not good)
    C3) Bypass cap for the power supply node of the fan?
   C10) Bypass cap, noise filter?

If I won the lottery and I have Q2,Q3,etc then I will probably swap my CPU
and SYSTEM fan connections and be happy.  If instead I have R3, then I
will not be pulling out the soldering iron and scope, but will be
returning the board and getting an EPIA board.

On Wed, Jun 13, 2007 at 01:15:32PM +0200, Jean Delvare wrote:
> Very interesting. This trick seems to work for other motherboard brands
> too. I have an Asus board here where I can't control the CPU fan and
> bingo! there's unpopulated room for a 3-leg component labelled Q12 next
> to the CPU fan header. So it seems that they designed the board with fan
> control in mind, and finally didn't solder the required components.
> Maybe because they share the same design between different models.

It's sad, but they probably didn't stuff this part of the board just to
save $0.50 on the bill of materials, but left it on the board just in case
there was a huge customer demand or special order.

Cheers,
-m

On Wed, Jun 13, 2007 at 01:15:32PM +0200, Jean Delvare wrote:
> Hi Phil,
> 
> On Mon, 11 Jun 2007 12:55:06 +0100, Phil Endecott wrote:
> > I've put a couple of photos of my J7F2 board here:
> > http://chezphil.org/tmp/j7f2_cpufan_top.jpeg
> > http://chezphil.org/tmp/j7f2_cpufan_bottom.jpeg
> > 
> > In the top photo, you can see an unpopulated position for a transistor 
> > ("Q1") next to the CPU fan connector.  In the bottom photo, you can see 
> > a "zero-ohm resistor" and some more unpopulated positions (the three 
> > pins on the left of this picture are the bottom of the fan connector).
> > 
> > If your board looks similar, you're right that you can't control the 
> > fan speed.  If you have a transistor and other components near the fan 
> > connector, then keep trying!
> 
> Very interesting. This trick seems to work for other motherboard brands
> too. I have an Asus board here where I can't control the CPU fan and
> bingo! there's unpopulated room for a 3-leg component labelled Q12 next
> to the CPU fan header. So it seems that they designed the board with fan
> control in mind, and finally didn't solder the required components.
> Maybe because they share the same design between different models.
> 
> -- 
> Jean Delvare

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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Phil Endecott]

Max Baker wrote:
> Phil,
>
> I will "see" your motherboard pictures and "raise" you a schematic :)
>
> http://warped.org/linux/j7f4r30fan.pdf
>
> The fine, fine folks at Jetway were nice enough to send me an actual
> schematic for the J7F4 board -- well only the page w/ the fans on it.
> They get an A+ for customer/Linux support, buy their products.

Absolutely!

I agree with your analysis, but note that the control components are 
all marked "NC", which I imagine means "not connected", while the 0-ohm 
link is not marked "NC".

>     C3) Bypass cap for the power supply node of the fan?

This is odd; the Fintek datasheet shows a capacitor in this position, 
but doesn't give a value.  Putting a largish capacitor here (this is 10 
uF I think) will rather screw up the PWM control in exactly the way 
that Jean and I have been discussing, i.e. even if the "on" period is 
quite short it will be enough to charge the capacitor for the whole 
cycle.  Having experimented I have not added a capacitor in that 
position on my hack.

They may be imagining that it has the effect of "smoothing" the PWM 
signal (e.g. 50% 12v smoothed to 6v).  To make this work you would need 
an inductor and a diode and you'd have built yourself a simple 
switch-mode supply.

> If I have R3, then I
> will not be pulling out the soldering iron and scope, but will be
> returning the board and getting an EPIA board.

Do you have any firm indication that the EPIA C7 board does have fan control?


Phil.





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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Max Baker]

On Wed, Jun 13, 2007 at 11:17:34PM +0100, Phil Endecott wrote:
> I agree with your analysis, but note that the control components are 
> all marked "NC", which I imagine means "not connected", while the 0-ohm 
> link is not marked "NC".

Ahh, very good... err bad.

> >    C3) Bypass cap for the power supply node of the fan?
> 
> This is odd; the Fintek datasheet shows a capacitor in this position, 
> but doesn't give a value.  Putting a largish capacitor here (this is 10 
> uF I think) will rather screw up the PWM control in exactly the way 
> that Jean and I have been discussing, i.e. even if the "on" period is 
> quite short it will be enough to charge the capacitor for the whole 
> cycle.  Having experimented I have not added a capacitor in that 
> position on my hack.
> 
> They may be imagining that it has the effect of "smoothing" the PWM 
> signal (e.g. 50% 12v smoothed to 6v).  To make this work you would need 
> an inductor and a diode and you'd have built yourself a simple 
> switch-mode supply.

I was confused on this too, since you would now have an exponential
discharge at every pulse... so you get that weird ramp / triangle-wave
behavior you were seeing.   Is there something about the fans that makes
it better to always have some sort of DC available instead of switching to
ground?  The only other thing I could think of would be for noise / EM
reasons -- so that you are not switching all the way to ground?

> Do you have any firm indication that the EPIA C7 board does have fan 
> control?

Only annectodal evidence of people getting lm-sensors/fancontrol working
on some EPIA boards.  I have a request open with the local Via FAE to get
an official answer.  I'm hoping that the local electronics store might
have some in stock and I can go eyeball em.

-m

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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Max Baker]

On Wed, Jun 13, 2007 at 11:17:34PM +0100, Phil Endecott wrote:
> I agree with your analysis, but note that the control components are 
> all marked "NC", which I imagine means "not connected", while the 0-ohm 
> link is not marked "NC".

You were correct -- all the fan control circuit devices are not on the
board, just the zero-ohm resistor.

Cheers,

-m

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

Re: [lm-sensors] Fintek F71805F on Jetway J7F4K1G5D - fancontrol?

[Jean Delvare]

On Wed, 13 Jun 2007 17:43:15 -0700, Max Baker wrote:
> On Wed, Jun 13, 2007 at 11:17:34PM +0100, Phil Endecott wrote:
> > This is odd; the Fintek datasheet shows a capacitor in this position, 
> > but doesn't give a value.  Putting a largish capacitor here (this is 10 
> > uF I think) will rather screw up the PWM control in exactly the way 
> > that Jean and I have been discussing, i.e. even if the "on" period is 
> > quite short it will be enough to charge the capacitor for the whole 
> > cycle.  Having experimented I have not added a capacitor in that 
> > position on my hack.
> > 
> > They may be imagining that it has the effect of "smoothing" the PWM 
> > signal (e.g. 50% 12v smoothed to 6v).  To make this work you would need 
> > an inductor and a diode and you'd have built yourself a simple 
> > switch-mode supply.
> 
> I was confused on this too, since you would now have an exponential
> discharge at every pulse... so you get that weird ramp / triangle-wave
> behavior you were seeing.   Is there something about the fans that makes
> it better to always have some sort of DC available instead of switching to
> ground?  The only other thing I could think of would be for noise / EM
> reasons -- so that you are not switching all the way to ground?

Maybe this was an attempt to guarantee valid fan speed readings even
when PWM is in use? I don't know much about electronics, unlike both of
you, but Phil said he observed the phenomenon only on high frequencies.
If the F71805F is capable to sample the fan speed at times it knows the
PWM signal is up, it may require that the sampling periods are not too
short. Just a random idea...

OTOH you wonder why Fintek had the chip generate these high frequencies
if it can't cope with them.

> > Do you have any firm indication that the EPIA C7 board does have fan 
> > control?
> 
> Only annectodal evidence of people getting lm-sensors/fancontrol working
> on some EPIA boards.  I have a request open with the local Via FAE to get
> an official answer.  I'm hoping that the local electronics store might
> have some in stock and I can go eyeball em.

Every board is different. I have a Jetway motherboard here where it
works, when Phil and you have Jetway boards where it doesn't.

-- 
Jean Delvare

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