|
Message
From: "Jim Dempsey" <tapedisk@a...>
Date: Wed, 22 Oct 2003 12:09:05 -0500
Subject: Re: [oc] Basic FPGA question (OT)
I believe you have to be careful to describe the
situation as either cause or affect (with an a).
1) Running a device at a higher frequency causes an
increases the temperature.
or
2) An increase in temperature affects the maximum frequency
at which you can drive a device. An increase in temperature
decreases the maximum frequency at which you can drive
a device.
Therefore I believe Renaux's statement should read
"The temperature is proportional to the frequency" and
not the other way around.
You could use frequency for a forced speed control for cooling
if you at some time you measure the temperature at various
frequencies (calibration). Then when the device is put in
use you could guess the temperature at a particular frequency
and thus guess the cooling requirements.
Guessing in this manner would not
produce the minimal power consumption for cooling.
If you are interested in minimal power consumption for
cooling then you would likely have to factor in something
that is affected by the actual temperature. A thermistor
could be used (optionaly in combination with one other
resistor) to construct a one-bit A/D converter which
could be read off an I/O pin or used directly to turn on
or off a power transistor for the fan. You can expand
on this to use 2 resistors to drive two input pins (or
two power transistors) giving the fan Off/Low/High.
Another technique you could employ is to eliminate
the fan and have the thermistor included in the
clock circuit such that the hotter the device gets
the slower the clock ticks. Or have the clock halve
frequency if the temperature exceeds a preset value
etc...
Jim Dempsey
----- Original Message -----
From: "Kirill 'Big K' Katsnelson" <kkm@d...>
To: <cores@o...>
Sent: Wednesday, October 22, 2003 4:10 AM
Subject: RE: [oc] Basic FPGA question (OT)
> Some time ago, Renaux Jacky wrote...
> >that counter
> >frequency will be proportionnal to the
> >silicon temperature ( silicon timing is
> >quite close
> >to 10% every 10 degres ----) . for a
> >given device your counter will be a silicon
> >thermometer
>
> Is this "thermometer" practically usable for fan speed control in a
> forced-cooling system? It reports the temperature right on the spot.
>
> Nice idea anyhow! :)
>
> -kkm
>
>
>
|
Browse
