Fan speed control

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.

The easiest way to tell that, is to see which Super I/O chip
with Hardware Monitor is being used. Some of the chips only
have one mode or the other. Some chips are PWM, some are
linear, some support both. If your chip is one of the first
two types, then you can tell what to expect from looking
at the datasheet for the chip. If you'd mentioned the motherboard,
that would have made the answer to your question more precise.

Also, the noise characteristics of PWM versus linear control
are different. At low speed, a PWM controlled fan will make
"gear noises" or "grumbling", which is aliasing between the
PWM energy and the rotation rate. A linear control method
should make less noise at low speed.

The pass transistor for linear control should run warmer
than if a PWM method is being used. If the big transistor
near the fan header runs hot, when the fan is set to a
low speed, then chances are the control method is linear.

Paul

>The easiest way to tell that, is to see which Super I/O chip

Thanks, you've confirmed pretty much everything I've said in this
thread:

"Strange fan behavior":
http://groups.google.com/group/comp.sys.ibm.pc.hardware.chips/browse_frm/thread/300a63d5a2adb445/a731e34f62adfb57?lnk=st&q=&rnum=1&hl=en#a731e34f62adfb57
http://tinyurl.com/yleslb (short)

I'm not the owner of the motherboard (Asus A8N-E), but I posted a
general question in this group because the owner does not appear to
have the wherewithal or the inclination to pursue his own question.

My understanding of the issue is that his motherboard uses the IT8712F
multi-IO chip which produces a PWM output with a minimum frequency of
about 6kHz. If this chip is not supported by an external linear
circuit, then I cannot see how such high frequency PWM can provide any
intelligent control over a brushless DC fan rotating at a frequency of
the order of 20-50Hz (the supply to the fan's electronics would be
chopped at ~100 - 300 times per rev). The OP in that thread appears to
be using a thermally controlled fan which further complicates the
issue. AFAICS, such a fan must not be run at any voltage other than
its rated one.

The only issue I would have with your statements is your contention
that some super I/O chips are linear. This would mean that a 5V chip
would have to have a 12V output and a built-in DAC. Are there any such
super I/O chips?

On a prior Asus board (A7N8X) with ITE IT8708 Super I/O,
just before the fan header there's a 100uF cap, then an
HJ772 (PNP transistor) that looks driven by an LM358.

This chip has both PWM and linear as output options. PDF page 133
shows a two transistor circuit for level shifting between 5V
logic and 12V fan (PWM circuit - saturating). The 10uF cap
is probably pretty important. The next page has a linear level
shifter (gain block), it consists of a charge pumped opamp driving
a MOSFET for power, with closed loop feedback. The charge pump ensures
sufficient gate voltage to control the MOSFET. The 83391 is the charge
pumped opamp.

http://www.winbond-usa.com/products/winbond_products/pdfs/PCIC/W83627EHF_EHG.pdf
http://www.winbond-usa.com/products/winbond_products/pdfs/PCIC/W83391TS_TG_0%2053_Ga.pdf

I cannot find any tech data on the AMD site for AMD fans.
This is the only mention of the fans being thermally controlled.
Usually, this is via a thermistor mounted in the hub of the fan.
The effect of modulating the voltage, and the thermistor, work
at the same time. If you had a cold room, and only 7V on the fan,
it might just stop turning, for example. Intel does give a spec
for their fans, but being cagey guys, they don't spec the RPMs
as such. They spec the temp at which to expect min RPMs and
max RPMs. But I cannot find similar info from AMD.

http://www.amd.com/us-en/assets/content_type/DownloadableAssets/Cool_N_Quiet_Installation_Guide3.pdf

As for the thread you quote, a fan speeding up as the control
goes down, doesn't make a lot of sense. I can't think of
a reason for it to happen.

Paul

>On Thu, 26 Oct 2006 06:09:06 +1000, Franc Zabkar

Thanks. That definitely looks like a linear circuit. I bet it follows
AMD's own recommended design. There is an example in Fig
15, section 8.4, page 69 of this document:

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/32200.pdf

It looks like the OP's motherboard in the other thread probably does
produce a DC voltage. Despite this, he is witnessing a *doubling* of
the fan speed as he *reduces* the supply. I suspect that his absurd
results are probably due to the fan's thermal sensor misbehaving at
the lower voltages.

Thanks for the correction. I notice that the application circuit has
three options covered, namely DC, low frequency PWM with a 3-wire fan,
and high frequency PWM with a 4-wire fan.

Yes, I had already found this document. It states that "all AMD Athlon
64 Processor-In-A-Box packages include thermally controlled fans".

Also, see page 30 of the "AMD Athlon 64 and AMD Opteron Processors
Thermal Design Guide":

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/26633_5649.pdf

"Systems with high levels of acoustic noise can generate customer
complaints and can influence the acceptance of a product. Therefore,
AMD recommends using fans that have speed control implemented on the
fan.

Usually, a thermistor is used to sense the airflow temperature at the
fan to determine the fan speed."

The following explanation is one I proposed in the other thread.

If the fan's internal temperature sensing utilises a potential divider
such as the following, then a fall in the supply voltage could
conceivably be sensed as an increase in temperature:

|---> transistor ---> fan motor
|
+12V o--- R --|-- NTC ---o ground

As the NTC warms up, its resistance decreases and the voltage from the
divider also decreases. A reduction in the supply voltage produces the
same result.

Franc Zabkar wrote:

<<snip>>

I'd prefer to see an actual schematic of the circuit used,
but so far I cannot find one. I tried the SanyoDenki site
(as they make the Intel thermistor cooler), but no luck there.

Paul

>Franc Zabkar wrote:

>I'd prefer to see an actual schematic of the circuit used,

I found this device:

ZXBM2001/ZXBM2002/ZXBM2003 Variable Speed 2-phase Fan Motor Controller
http://www.zetex.com/3.0/pdf/ZXBM2001.pdf

Notice that the block diagram on page 4 shows an SPD input at the
centre of a potential divider powered from an internal 2V reference. A
thermistor connects between SPD and ground. Such an arrangement would
accurately sense the temperature, irrespective of supply variations.

OTOH, an alternative device, the ZXBM2004 Variable Speed 2-phase BLDC
Motor Controller, has a similar sensing arrangement to the one I
proposed (see Fig 1, page 8).
http://www.zetex.com/3.0/pdf/ZXBM2004.pdf

AFAICT, the circuit in fig 1 would be sensitive to large supply
variations. However, because the speed control is open loop, I'm not
convinced that it would produce the symptoms I have described. Still,
there may be other fan control ICs that behave slightly differently.