[time-nuts] Capacitive temperature sensing
XDE-L2G3 at myamail.com
Sat Aug 23 11:43:25 EDT 2008
Bruce Griffiths <bruce.griffiths at xtra.co.nz> wrote:
> Magnus Danielson wrote:
>> Hej Bruce,
>> Indeed. However, I was trying to hint on a alternative approach
>> or temporary approach which would not need much design to achieve
>> equal performance if the electrodes could not be cleaned or
>> replaced easilly.
>> I trust the capacitive measurement would be sufficiently good
>> even at moderate design efforts.
> The biggest problem is that a stainless steel thermoregulator and
> not a mercury in glass one may have been used.
> Retrofitting temporary capacitive sensing would be difficult to
> impossible in this case.
> Metals other than platinum or iron are precluded from use as they
> tend to form amalgams with the mercury.
> However one could probably temporarily substitute a mercury in
> glass thermometer with capacitive position sensing.
> The only potential problem being that there are moves afoot to ban
> mercury thermometers.
> A capacitance of around 1pF or so may be expected for a 1cm long
> cylindical capacitive sensor electrode and an instability in the
> capacitance measurement of 100ppm or so would be required to
> achieve millidegree stability (provided the mechanical instability
> of the glass allows this).
The problem is very simple. If the sensor lead that contacts the
mercury has a diameter of 0.010 inch, with a flat end, and is 100
microinches away from the mercury, the capacitance is 0.35 pf. An
Analog Devices AD7747 24-bit capacitance-to-digital converter could
be used to measure this. It sells for $4.60 in quantity.
It has a resolution of 20 aF (1e-18F), and an accuracy of 10 fF.
This would allow it to detect a change in distance of 0.00566
microinch or 0.144 nanometers. This is on par with commercial
capacitance sensors such as Physik Instrumente. But since it is
completely integrated, the drift specifications are about an order
of magnitude better.
If the mercury contacts the sensor wire and shorts it to ground, a
7.5pF cap could be placed in series. This would limit the maximum
capacitance to within the range of the AD7747.
The AD7747 can accept up to 17pF of stray capacitance. The existing
sensor wire should be much less than this, especially if the leads
were as short as possible.
The problem would be a bit more complicated if the sensor lead had a
v-shaped end caused by cutting with pliers instead of the shearing
action of scissors. But the resolution of the AD7747 should allow it
to measure the capacitance just before making contact.
It also has an on-board temperature sensor with an accuracy of 2
degrees C. This would allow the system to do coarse temperature
measurements to confirm it was working.
More information about the time-nuts