temp measurement with TC

Archiver

Hello,
I'm trying to measure the temperature of water flow (5-40 deg C)
through a 1/2" pipe using the BS2. I have little experience with the
stamp, nor with temp measurements.
My best idea is to use a type T thermocouple probe in a 1/16"
compression fitting. Then I would use an IC temp sensor (maybe LM35)
to measure the temp of an a isothermal block. Voltage outputs of the
IC sensor could be digitized by an ADC (Maybe one that's MUX capable
from Analog Devices - or could use multiple LTC1298's). Then, I will
tackle the software issues to compute the temp of the thermocouple.
Is there a cheaper way to do this? Do I need to amplify the signal
from the thermocouple before trying to measure with ADC? Does my
choice of components look reasonable?
Thanks for any feedback,
Eric

Archiver

Hi Eric,

I did a similar thing for a hot tub. I did not use a thermocouple
because of its very small output. Instead I constructed a probe using
an LM34 (°F) (you can use a LM35 for °C) along the lines the probe
described in http://www.dalsemi.com/news/resource/rsc5_1820.html. I
brought the 3 wires out thru a well shielded cable and ran it 20 feet
or so to the BS2. I got my stainless steel probe from Omega as
suggested. I grounded the probe but did not ground the sensor to the
probed.

For an ADC I selected the Maxim MAX187 for its simple, time
independent, interface and for its 1 count per millivolt of input.
A snippet of my code is:

' Maxim MAX187 Analog/Digital Converter (conversion time <8.5us)
' 12 bits, full scale is 4.096 volts
' ADC reads in millivolts = decidegrees
adcclk con 13 ' Output pin for ADC clock line
adccs con 14 ' Output pin for ADC chip select
adcdl con 15 ' Input pin for ADC serial data line
adcctr var byte ' counter for ADC clocks

low adccs ' enable adc chip select
tempmeas = 0 ' init result
for adcctr = 1 to 12 ' data read time about 20ms total
tempmeas = tempmeas << 1 ' shift left
pulsout adcclk,1 ' 2 us clock pulse
tempmeas = tempmeas | ins.lowbit(adcdl)
next
pulsout adcclk,1 ' thirteenth clock
high adccs ' disable chip select
' we're done with adc

This will give you a resolution of 1/10 degree.

Finally, to display the temperature I used the LED AppMod from
Parallax. The code for it is a little tougher.

I hope this helps.

--- In basicstamps@y..., capnmeier@y... wrote:
> Hello,
> I'm trying to measure the temperature of water flow (5-40 deg C)
> through a 1/2" pipe using the BS2. I have little experience with
the
> stamp, nor with temp measurements.
> My best idea is to use a type T thermocouple probe in a 1/16"
> compression fitting. Then I would use an IC temp sensor (maybe
LM35)
> to measure the temp of an a isothermal block. Voltage outputs of
the
> IC sensor could be digitized by an ADC (Maybe one that's MUX capable
> from Analog Devices - or could use multiple LTC1298's). Then, I
will
> tackle the software issues to compute the temp of the thermocouple.
> Is there a cheaper way to do this? Do I need to amplify the
signal
> from the thermocouple before trying to measure with ADC? Does my
> choice of components look reasonable?
> Thanks for any feedback,
> Eric

Archiver

I now see that my link to Dallas Semi has expired. Roughly what it
described was a probe made from an Omega SS-14 or similar. Attach
wires to the sensor. Put heat sink compound in the end of the probe (I
used a straw and puffed it in place). Slide the sensor into the tube
and seal the open end of the tube with a moisture resistant epoxy...

Sorry, I should have tested the link before I quoted it.


--- In basicstamps@y..., fdavidson@m... wrote:
> Hi Eric,
>
> I did a similar thing for a hot tub. I did not use a thermocouple
> because of its very small output. Instead I constructed a probe
using
> an LM34 (°F) (you can use a LM35 for °C) along the lines the probe
> described in http://www.dalsemi.com/news/resource/rsc5_1820.html. I
> brought the 3 wires out thru a well shielded cable and ran it 20
feet
> or so to the BS2. I got my stainless steel probe from Omega as
> suggested. I grounded the probe but did not ground the sensor to the
> probed.
>
> For an ADC I selected the Maxim MAX187 for its simple, time
> independent, interface and for its 1 count per millivolt of input.
> A snippet of my code is:
>
> ' Maxim MAX187 Analog/Digital Converter (conversion time <8.5us)
> ' 12 bits, full scale is 4.096 volts
> ' ADC reads in millivolts = decidegrees
> adcclk con 13 ' Output pin for ADC clock line
> adccs con 14 ' Output pin for ADC chip select
> adcdl con 15 ' Input pin for ADC serial data line
> adcctr var byte ' counter for ADC clocks
>
> low adccs ' enable adc chip select
> tempmeas = 0 ' init result
> for adcctr = 1 to 12 ' data read time about 20ms total
> tempmeas = tempmeas << 1 ' shift left
> pulsout adcclk,1 ' 2 us clock pulse
> tempmeas = tempmeas | ins.lowbit(adcdl)
> next
> pulsout adcclk,1 ' thirteenth clock
> high adccs ' disable chip select
> ' we're done with adc
>
> This will give you a resolution of 1/10 degree.
>
> Finally, to display the temperature I used the LED AppMod from
> Parallax. The code for it is a little tougher.
>
> I hope this helps.
>
> --- In basicstamps@y..., capnmeier@y... wrote:
> > Hello,
> > I'm trying to measure the temperature of water flow (5-40 deg C)
> > through a 1/2" pipe using the BS2. I have little experience with
> the
> > stamp, nor with temp measurements.
> > My best idea is to use a type T thermocouple probe in a 1/16"
> > compression fitting. Then I would use an IC temp sensor (maybe
> LM35)
> > to measure the temp of an a isothermal block. Voltage outputs of
> the
> > IC sensor could be digitized by an ADC (Maybe one that's MUX
capable
> > from Analog Devices - or could use multiple LTC1298's). Then, I
> will
> > tackle the software issues to compute the temp of the
thermocouple.
> > Is there a cheaper way to do this? Do I need to amplify the
> signal
> > from the thermocouple before trying to measure with ADC? Does my
> > choice of components look reasonable?
> > Thanks for any feedback,
> > Eric