Posted by vaughn on November 3, 2011, 6:52 pm
Morris, (On the huge assumption that I understand what you are trying to do).
Any time you heat a mass with a point temerature source, you end up with a
temperature gradient. The heater temperature will be significantly hotter than
the material you are heating. The short interval approach you mention will give
you the temperature of the heater, not the temperature of the mass you are
heating. Is that what you really want? To accurately measure the temperature
of the mass using the heater as a sensor, you will need to have a longish time
delay so that your sensor can cool down and finally match the temperature of the
Posted by Morris Dovey on November 4, 2011, 7:15 am
On 11/3/11 1:52 PM, vaughn wrote:
You're right - but there'll be a fair amount of testing with an unfueled
reactor to get a handle on things like the heat transfer rate you
My thinking is that I can use the heat transfer information along with
predictive logic to project what temperatures /should/ result from the
heating so that significant variances from the predicted values
(signalling ignition) can be identified fairly quickly. Knowing the heat
transfer rate should allow me to shorten the delay between powering down
the heater and capturing the heater resistance, and to use that value in
context to accurately /infer/ the temperature in the reaction chamber.
Keep in mind that during heating I only care about detecting ignition,
that with ignition electric heating stops, and that I only care about
accurate temperatures /after/ that point.
Aside from the electronics and logic issues, I still have to be able to
build this thing. The reactor has to be hydrogen-tight at a couple
hundred PSI and every "feature" adds difficulty to the build (and I'm
neither skilled machinist nor expert welder).
Posted by Jim Wilkins on November 4, 2011, 12:23 pm
If I were doing this I would scare up a surplus PID temperature controller,
Omron is a good name, a thermowell for the temperature sensor and a vacuum
oven feedthru for the heater power leads. The rest can be fabricated,
including the thermowell if you have a decent thread-cutting lathe (almost
essential for physics experiments). The feedthru on my vacuum oven was a
pipe plug with Glyptal sealing the wires. It leaked and thus I bought the
oven cheap, but making a new one fixed it fine.
If you attempt to measure temperature from heater wire resistance you will
have to wait many seconds for it to cool down exponentially to the internal
temperature. You could sacrifice one to find the input power that damages it
and stay below that with the second one.
Usually we keep the heater and sensor separate, with the heater upstream and
the sensor in a location that better averages the temperature of the
contents. Applications that measure and control the heater resistance
generally keep its temperature constant and read out the necessary power,
for instance the hot-wire detector in a mass spectrometer or the mass air
flow sensor in a car engine. Otherwise PID control which avoids temperature
overshoot is much better.
The two accessible controls on your A/D converter are gain and offset. Use
the offset to get your zero and the gain, the reference voltage, for full
Posted by Morris Dovey on November 4, 2011, 5:05 am
On 11/3/11 8:13 PM, Bob F wrote:
Sorry, I should have described this better: 1 sample/sec should be
enough during initial heating. If it takes 1 msec to capture the
reading, then heating time can be as much as 999 msec (less relay
make/break latencies) out of each second.
If/when there is ignition, heating will stop altogether and the rate
will be increased to 10 samples/sec during the self-sustain testing.
All the while, beta- and gamma-radiation counts will be accumulated into
100 msec counter "buckets".
I'm hoping none of these modes will strain any components.
Posted by CWatters on November 4, 2011, 4:20 pm
On 04/11/2011 05:05, Morris Dovey wrote:
Rossi suggests that perhaps kW are needed so losses fropm the cell must
be quite high?
I think you might find the temperature of the heating element falls
rapidly when the power to the heater is turned off until it reaches the
temperature of the bulk material around it. If you sample the
temperature immediatly you turn off the heater you might end up
measuring the temperature of the heater not what the bulk material would
be with the heater off. Isn't it the latter that matters?
Perhaps experiment to work out what that sampling delay should be, eg
heater off then sample the resistance every half second for say 10-20
seconds, plot the graph and see what it looks like. If the temperature
stops falling with the heater off then some kind of reaction is going on.