Temperature Measurement System Design

[Davidson]

Hi, everyone. I have a project concerned with a temperature
measurement system design. My proposal is to do a body thermometer
that display the result on a PC(therefore I set the temperature
measuring range to about 25 ~ 45). One project requirement is to
design a "measurement system" - we can't use an off-shelf sensor
package (e.g. LM35)- we must use a temperature transducer together
with linearization and signal conditioning circuit. The result must be
outputed to a computer.

My problem is, are there any good (simple, accuracy around 0.5~1
centigrade is fine) circuit design that I can use? I thought I would
use a thermistor, but somebody said that a thermistor is not used
unless I want to design a boiler?! Could I use a thermistor to
construct a simple circuit? Or are there any other better transducer I
can use?

My project is only paper-design.

Thanks in advance.
Davidson

 

[Robert Baer]

Davidson wrote:

Hi, everyone. I have a project concerned with a temperature
measurement system design. My proposal is to do a body thermometer
that display the result on a PC(therefore I set the temperature
measuring range to about 25 ~ 45). One project requirement is to
design a "measurement system" - we can't use an off-shelf sensor
package (e.g. LM35)- we must use a temperature transducer together
with linearization and signal conditioning circuit. The result must be
outputed to a computer.

My problem is, are there any good (simple, accuracy around 0.5~1
centigrade is fine) circuit design that I can use? I thought I would
use a thermistor, but somebody said that a thermistor is not used
unless I want to design a boiler?! Could I use a thermistor to
construct a simple circuit? Or are there any other better transducer I
can use?

My project is only paper-design.

Thanks in advance.
Davidson

Thermistors are fine, and would seem to "fit" the requirement of
"necessary" linearization.
Most thermistors have specs at room temp anyway.

 

[Rene Tschaggelar]

Davidson wrote:

Hi, everyone. I have a project concerned with a temperature
measurement system design. My proposal is to do a body thermometer
that display the result on a PC(therefore I set the temperature
measuring range to about 25 ~ 45). One project requirement is to
design a "measurement system" - we can't use an off-shelf sensor
package (e.g. LM35)- we must use a temperature transducer together
with linearization and signal conditioning circuit. The result must be
outputed to a computer.

My problem is, are there any good (simple, accuracy around 0.5~1
centigrade is fine) circuit design that I can use? I thought I would
use a thermistor, but somebody said that a thermistor is not used
unless I want to design a boiler?! Could I use a thermistor to
construct a simple circuit? Or are there any other better transducer I
can use?

My project is only paper-design.

I'm selling temperature controllers that have a stability
approaching 0.01 degrees Celsius. They are based on NTC
sensors and need calibration for absolute values.


Rene
--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net

 

[legg]

On 27 Apr 2004 00:17:26 -0700, DavidsonChen833@hotmail.com (Davidson)
wrote:

Hi, everyone. I have a project concerned with a temperature
measurement system design. My proposal is to do a body thermometer
that display the result on a PC(therefore I set the temperature
measuring range to about 25 ~ 45). One project requirement is to
design a "measurement system" - we can't use an off-shelf sensor
package (e.g. LM35)- we must use a temperature transducer together
with linearization and signal conditioning circuit. The result must be
outputed to a computer.

A thermometer is a 'patient applied part' in medical product safety
classification.

You might get browney points by taking this into account, in your
design, and allowing for suitable isolation in interfacing to the
'PC'.

RL

 

[Fred Bloggs]

Davidson wrote:

Hi, everyone. I have a project concerned with a temperature
measurement system design. My proposal is to do a body thermometer
that display the result on a PC(therefore I set the temperature
measuring range to about 25 ~ 45). One project requirement is to
design a "measurement system" - we can't use an off-shelf sensor
package (e.g. LM35)- we must use a temperature transducer together
with linearization and signal conditioning circuit. The result must be
outputed to a computer.

My problem is, are there any good (simple, accuracy around 0.5~1
centigrade is fine) circuit design that I can use? I thought I would
use a thermistor, but somebody said that a thermistor is not used
unless I want to design a boiler?! Could I use a thermistor to
construct a simple circuit? Or are there any other better transducer I
can use?

My project is only paper-design.

Thanks in advance.
Davidson

Where on/in the body is the measurement to be made? Skin, tongue, anus?

 

[Bill Sloman]

Rene Tschaggelar <none@none.net> wrote in message news:<408e29d6$0$720$5402220f@news.sunrise.ch>...

Davidson wrote:
Hi, everyone. I have a project concerned with a temperature
measurement system design. My proposal is to do a body thermometer
that display the result on a PC(therefore I set the temperature
measuring range to about 25 ~ 45). One project requirement is to
design a "measurement system" - we can't use an off-shelf sensor
package (e.g. LM35)- we must use a temperature transducer together
with linearization and signal conditioning circuit. The result must be
outputed to a computer.

My problem is, are there any good (simple, accuracy around 0.5~1
centigrade is fine) circuit design that I can use? I thought I would
use a thermistor, but somebody said that a thermistor is not used
unless I want to design a boiler?! Could I use a thermistor to
construct a simple circuit? Or are there any other better transducer I
can use?

My project is only paper-design.

I'm selling temperature controllers that have a stability
approaching 0.01 degrees Celsius. They are based on NTC
sensors and need calibration for absolute values.

You can buy glass encapsulated "interchangeable" thermistors that have
stabilities of the order of 0.001C per year below 45C, and initial
accuracies of +/-0.05C

http://www.ysi.com/extranet/TEMPKL.nsf/447554deba0f52f2852569f500696b21/34e470aaf6efd3d885256ac400478e95!OpenDocument

Newark stock some of the range, which means that you can buy small
quantities via Farnell.

Similar parts have been used in "microdegree" temperature controllers
stable to down to +/-3.5uC. E-mail me - my e-mail address is real - if
you want more details.

--------
Bill Sloman, Nijmegen

 

[Davidson]

Firstly, thanks for all the replies above.

This project is meant to test my ability to design a measurement
system (i am a 4th year university student), therefore we are not
allowed to use any prebuild, linearized sensor package (e.g. LM35 is
not allowed). Also, this is a paper design project, we don't need to
build the instrument, but we need to do all the other tasks (e.g.
ananlysis, simulations, etc.)

Davidson

 

In article <7c584d27.0404270643.2d42d463@posting.google.com>, Bill
Sloman <bill.sloman@ieee.org> writes

You can buy glass encapsulated "interchangeable" thermistors that have
stabilities of the order of 0.001C per year below 45C, and initial
accuracies of +/-0.05C

http://www.ysi.com/extranet/TEMPKL.nsf/447554deba0f52f2852569f500696b21/34e470aa
f6efd3d885256ac400478e95!OpenDocument

Newark stock some of the range, which means that you can buy small
quantities via Farnell.

Similar parts have been used in "microdegree" temperature controllers
stable to down to +/-3.5uC. E-mail me - my e-mail address is real - if
you want more details.
You may be able to answer a long standing uncertainty:

Platinum resistance, NTC thermistor of the best type.
Quartz crystal resonator specially cut for temperature measurement.
Which yields the best absolute accuracy initially if cabrated ie thermal
signal to noise.
Best long term drift.
Ive generally used NTC thermistors for crystal oven temperature control
the absolute temperature is fairly unimportant as the temperature is set
for each crystal.
Hammond of HP designed a crystal with a linear tc and 1Hz? / deg K
that was used in a meter that indicated udeg, A linear correspondence of
freq/temp is no longer valuable and the crystal is not made anymore (nor
is the instrument)
I also remember that the mechanical trimming process to take out
thermistor manufacturing tolerance degrades the aging performance of
some thermistors.

Note dont reply to demon emailmail

Doug.
--
ddwyer

 

[Bill Sloman]

dougfgd@hotmail.com wrote in message news:<04N9GIAugVkAFwPf@ddwyer.demon.co.uk>...

In article <7c584d27.0404270643.2d42d463@posting.google.com>, Bill
Sloman <bill.sloman@ieee.org> writes

You can buy glass encapsulated "interchangeable" thermistors that have
stabilities of the order of 0.001C per year below 45C, and initial
accuracies of +/-0.05C

http://www.ysi.com/extranet/TEMPKL.nsf/447554deba0f52f2852569f500696b21/34e470aa
f6efd3d885256ac400478e95!OpenDocument

Newark stock some of the range, which means that you can buy small
quantities via Farnell.

Similar parts have been used in "microdegree" temperature controllers
stable to down to +/-3.5uC. E-mail me - my e-mail address is real - if
you want more details.
You may be able to answer a long standing uncertainty:
Platinum resistance, NTC thermistor of the best type.
Quartz crystal resonator specially cut for temperature measurement.
Which yields the best absolute accuracy initially if calibrated ie thermal
signal to noise.
Best long term drift.

Standards laboratories use platinum resistors as temperature sensors.
Some thermistors are pretty good, if you don't dissipate too much
power in them, nor use them at too high a temperature - one of my
ex-colleagues found that his went slightly but definitely unstable
above 55C.

Quartz crystal resonators for temperature measurement appear to be a
minority taste.

Ive generally used NTC thermistors for crystal oven temperature control
the absolute temperature is fairly unimportant as the temperature is set
for each crystal.

The main advantage of themistors is that their resistance varies ten
times as fast with temperature (-4%/C) as does the reistance of
platinum (closer to +0.4% per degree C).

Hammond of HP designed a crystal with a linear tc and 1Hz? / deg K
that was used in a meter that indicated udeg, A linear correspondence of
freq/temp is no longer valuable and the crystal is not made anymore (nor
is the instrument)
I also remember that the mechanical trimming process to take out
thermistor manufacturing tolerance degrades the aging performance of
some thermistors.

-------
Bill Sloman, Nijmegen