RICH Temperature Monitor
Jianchun Wang
(Created: 12/20/00)
Key Word: thermistor, crate, slot,
chain, sector, quarter, chip carrier
Introduction
Precise
monitoring of temperature is quite important to CLEO RICH detector. The CaF2
windows are fragile. A large temperature gradient might cause CaF2 windows
to crack. Also the front-end electronics is sensitive to the temperature.
Temperature fluctuation will change the working point and might cause a
malfunction. This note serves to describe the temperature readout system and
thermistor calibration.
Thermistor and Readout Circuit
We use a
surface mounted negative temperature coefficient (NTC) thermistor produced by
MMC Electronics (http://www.mmea.com/). The
thermistor (TH20-3H103FB) has nominal resistance Ro of 10 KW (± 1%) at
temperature To of 25°C. The value of the coefficient b is 3370
(tolerance: ± 3%). The resistance (R) as function of temperature
(T) is expressed in Eq 1, where the temperature T is expressed in Kelvin.
Each
chain has one and only one thermistor mounted to monitor the temperature. The
location of the thermistor is shown in Fig 1. The thermistors are mounted on
either the first chip carrier of outer chains or last chip carrier of inner
chains to monitor the maximum temperature difference (for nomenclature of chain
and chip carrier, please refer to page channel ID).
There is also a thermistor on each data board to monitor the board temperature.
In total, we have 360 thermistors to measure the chain temperature and 120
thermistors for board temperature measurement.
Fig 1. Location of thermistors on one RICH sector
The
readout circuit diagram is illustrated in Fig 2. The thermistor is serial connected with a 10k
resistor. A 2.5 V reference voltage is applied to them. The voltage across the
thermistor is measured via a 10-bit ADC on data board. The resistance of the
thermistor is used to calculate the temperature according to equation 1.
Fig 2. The temperature readout circuit
diagram
With ADC
readout value adc, the formula to calculate the temperature is
Thermistor Calibration
Even
though we chose high precision thermistors, there are still few thermistors
delivering values very different from expectations. Fig 3 shows the thermistor
value distribution before calibration as the function of the location in the
RICH detector. The system power had just been turned on after a long enough
time. We could reasonably assume that the system had reached thermal
equilibrium. We expected that all the thermistors would have similar values.
Obviously there are a significant number of thermistors read higher than
typical values. A calibration is necessary.
Fig 3. The thermistor value before calibration
It is
difficult to perform a full calibration on all thermistors, especially after
the detector installation. However, the powerful cooling system makes it
possible for us to make an one-point calibration. Around 25oC, the
largest temperature error is contributed from the value of Ro, the
nominal resistance at 25oC, which is the one we decide to calibrate.
To make things easy, we will determine the temperature To, at which
the thermistor resistance is 10 KW.
Shown in
Fig 4 are the thermistor value distributions for the inner and the outer
chains. The center values are slightly different for the inner and the outer
chains due to limitations of the cooling system. We use the fitted center values as the
expected temperature of the thermistors. The value of the calibrated To
is listed in the file (therm_calib.txt). We will use these values in an online
monitoring program.
Fig 4. Thermistor value distributions before calibration
Performance
After calibration, the entries that have weird thermistor values disappeared. Shown in Fig 5 is the thermistor value distribution with the correction. We see that for inner chains the average temperature is about 1.5oC higher than outer chains. This is due to the configuration of the cooling system, and is anticipated. We also noticed that there are some entries, mostly in the middle chains that have higher values than others. This may reflect real temperature differences corresponding to differences in current consumption between the readout chains. It is also possible that this is due to b value errors. As the inner chains have larger temperature difference from 25oC, the effect of b uncertainty becomes significant. Not until we have at least two-point calibration, will we know the final answer.
Fig 5. Thermistor value distributions after calibration and after 1 day stable run