Jump to content


Photo

Instrumenting a Dob with Temperature Probes

  • Please log in to reply
40 replies to this topic

#26 don clement

don clement

    Vendor (Clement Focuser)

  • -----
  • Vendors
  • Posts: 812
  • Joined: 02 Feb 2011
  • Loc: Running Springs, California

Posted 28 June 2013 - 08:52 AM

I am sure there are five figure lab units that also make good voltmeters.


Yes Kiethley instruments come to mind http://www.keithley.com/

Ah the late lamented HeathKit! The website assets "the news of my death is greatly exaggerated" but that is so far the only sign of life.


I am sure there are many others that believe that the vacuum tube is dead. There are also the few of us still channeling those dead vacuum tubes for the clairvoyant sounds of long deceased vinyl.

Don Clement

#27 Pinbout

Pinbout

    Cosmos

  • *****
  • Posts: 8032
  • Joined: 22 Feb 2010
  • Loc: nj

Posted 28 June 2013 - 09:03 AM

I am sure there are many others that believe that the vacuum tube is dead. There are also the few of us still channeling those dead vacuum tubes for the clairvoyant sounds of long deceased vinyl.

Don Clement



but the day has long been dead that you drive to the 7-11 to check out if the tubes are blown.

I still have my gt101 accoustic amp

#28 careysub

careysub

    Surveyor 1

  • *****
  • Posts: 1906
  • Joined: 18 Feb 2011
  • Loc: Rancho Cucamonga, CA

Posted 28 June 2013 - 09:39 AM

why not use the one-wire sensors? Super-small in size and cheap as well. If you would not like to put the necessary sensors to a bus-cable (actually, a simple cable like the Cat5 or so and all sensors connected as a bus to this cable) then you even could use the one-wire buttons called iButton. Sensor with battery in a small housing, like a coin. So you could equipp your scope with this "buttons" wherever you like. In the end, you pick them out and put them into a little reader unit and read the data. Buttons can be configured to take a measurement all n-minutes or so.


Interesting. I did not know about the one-wire technology.

This sort of low cost, low performance digital microlan looks promising. Here is a good introductory page from a hobby component provider:
https://www.hobby-bo...es.php?pageid=4

I found this site interesting:
http://www.digitemp.com/
Temperature monitoring software for Linux, I have two EEPC netbooks running EEBuntu that might be just the thing for this. This would solve the controller/monitor problem with looked like the most costly/time consuming part of any digital solution.

The iButtons look a lot more expensive on a per sensor basis.

Are there any books out there providing a good grounding in 1-wire network design and set-up; e.g. how to use these little digital switches and other network components I see? Something along the lines of the Arduino Cookbook for example?

This looks like a cheap and convenient and versatile enough technology, with a low enough cost of entry, to merit closer examination.

#29 freestar8n

freestar8n

    Soyuz

  • *****
  • Posts: 3987
  • Joined: 12 Oct 2007

Posted 28 June 2013 - 10:00 AM

The ds1820's that I used in my sct cooling study mentioned earlier also operate in one-wire mode, allowing multiple sensors to be connected on the same input and individually addressed. The newer version, ds18b20, has 12-bit precision, or 0.06C. This sensor family worked well for the specific application of measuring temperature at different points in a telescope, with no analog signals floating around. But I was comfortable writing custom PIC code for the whole thing.

Frank

#30 Pinbout

Pinbout

    Cosmos

  • *****
  • Posts: 8032
  • Joined: 22 Feb 2010
  • Loc: nj

Posted 28 June 2013 - 11:35 AM

shouldn't be a problem to wire into an arduino or even a basic stamp?

#31 Jon Isaacs

Jon Isaacs

    ISS

  • *****
  • Posts: 44369
  • Joined: 16 Jun 2004
  • Loc: San Diego and Boulevard, CA

Posted 28 June 2013 - 02:52 PM

My thinking

Repeatability between probes is very important. So... When it comes to repeatibility as well as accuracy, it it hard to beat the old fashioned thermocouple. Thermocouple wire is relatively cheap so you can make your own probes. The voltage output is quite low so an amplifier of some sort is needed. I have an old Fluke multimeter with a seperate module that fits any multimeter and outputs 1 mv per degree either F of C. The Fluke has a millivolt scale that reads to 0.1 mv so this is 0.1 degree.

With other sorts of sensors, the repeatability is questionable between sensors.

As far as using an IR device to measure the temperature, the amount of IR emitted by a body depends on both its temperature and its emissivity so one cannot make accurate measurements of different surfaces without calibration.

Jon

#32 don clement

don clement

    Vendor (Clement Focuser)

  • -----
  • Vendors
  • Posts: 812
  • Joined: 02 Feb 2011
  • Loc: Running Springs, California

Posted 28 June 2013 - 04:41 PM


Thermocouples are non-linear. One must provide a lookup table to linearize thermocouples. Some of the old Fluke DVM did do that for K thermocouples IIR. There is a simple way to match the output on LM35 linear temperature sensors using 10K trim pots.
See: http://www.investiga...tura/AN-460.pdf

Don Clement

#33 freestar8n

freestar8n

    Soyuz

  • *****
  • Posts: 3987
  • Joined: 12 Oct 2007

Posted 28 June 2013 - 04:41 PM

shouldn't be a problem to wire into an arduino or even a basic stamp?



Yes - many chips and environments would work for development - but my main point is that custom work is helped by being able to write code specific to the needs of the project.

As for accuracy, when it comes to air measurement I think a key limitation is a good shroud for the sensor so it isn't heavily affected by radiation from nearby surfaces and the sky - and is a true measure of the local air temperature rather than the radiative properties of the sensor in its environment.

Frank

#34 Jon Isaacs

Jon Isaacs

    ISS

  • *****
  • Posts: 44369
  • Joined: 16 Jun 2004
  • Loc: San Diego and Boulevard, CA

Posted 28 June 2013 - 05:44 PM

Thermocouples are non-linear. One must provide a lookup table to linearize thermocouples. Some of the old Fluke DVM did do that for K thermocouples IIR. There is a simple way to match the output on LM35 linear temperature sensors using 10K trim pots.
See: http://www.investiga...tura/AN-460.pdf

Don Clement


It is true that thermocouples are nonlinear over their full range but the full range of a K type is up to 1370C. Ktypes are reasonably linear to a 1000C. In the range of interest here they are linear and in any event these measurements are relative rather than absolute.

Jon

#35 don clement

don clement

    Vendor (Clement Focuser)

  • -----
  • Vendors
  • Posts: 812
  • Joined: 02 Feb 2011
  • Loc: Running Springs, California

Posted 28 June 2013 - 06:31 PM

Omega is a good place to get thermocouples,connectors and other types of temperature probes. http://www.omega.com...rature/tsc.html

Don Clement

#36 Jon Isaacs

Jon Isaacs

    ISS

  • *****
  • Posts: 44369
  • Joined: 16 Jun 2004
  • Loc: San Diego and Boulevard, CA

Posted 28 June 2013 - 06:40 PM

Omega is a good place to get thermocouples,connectors and other types of temperature probes. http://www.omega.com...rature/tsc.html

Don Clement


That's where I get our stuff for the lab. We have a wonderful old digital thermocouple reader, it's 12 channels, switchable, reads to 0.1 degrees. But I think a data logger would be slick for this app. There are some inexpensive USB data loggers out there that would do the job.

Jon

#37 careysub

careysub

    Surveyor 1

  • *****
  • Posts: 1906
  • Joined: 18 Feb 2011
  • Loc: Rancho Cucamonga, CA

Posted 29 June 2013 - 02:03 AM


Thermocouples are non-linear. One must provide a lookup table to linearize thermocouples. Some of the old Fluke DVM did do that for K thermocouples IIR. There is a simple way to match the output on LM35 linear temperature sensors using 10K trim pots.
See: http://www.investiga...tura/AN-460.pdf

Don Clement


It is true that thermocouples are nonlinear over their full range but the full range of a K type is up to 1370C. Ktypes are reasonably linear to a 1000C. In the range of interest here they are linear and in any event these measurements are relative rather than absolute.

Jon


Putting the Omega type K specs into a spreadsheet I see that average mV/C over the range of -10 to 50 C is 0.0403; over the more restricted range of interest 5 to 35 C it is exactly the same.
The maximum degree differential range shown (-10 to 50 C) is 0.039 to 0.042.
The average mV/C per temp decade:
-10 : 0.039
0 : 0.040
10: 0.040
20: 0.041
30: 0.041
40: 0.041

And it appears that over this entire range, if you used the average value of 0.040 mV/C the largest deviation would get anywhere would be 0.05 C, and the largest degree-to-degree variation 0.01 C.

From this I conclude that the output is effectively linear in this range, below the error of a typical voltmeter. Of course the voltage needs to be boosted about 100 fold for most milli-range voltmeters.

#38 careysub

careysub

    Surveyor 1

  • *****
  • Posts: 1906
  • Joined: 18 Feb 2011
  • Loc: Rancho Cucamonga, CA

Posted 29 June 2013 - 02:57 AM

OTOH - significantly more work seems necessary to incorporate a thermocouple into device that actually reads temperature from multiple probes.

Looking on-line I see lots of sources descrbing how to make a thermocouple thermomenter, that focus on making the thermocouple probe itself, which can be bought for a few dollars, then describes using equipment costing thousands of dollars on a lab bench to actually use it. Rather topsy-turvy in emphasis.

The 1-bit technology at least has ready made components, wiring, and free Linux software designed to monitor tempertures.

The LM-35 IC package puts out enough voltage to monitor directly with a half-decenbt multimeter.

To use the thermocouples effectively it looks like getting into Arduino building and programming (I an an assembly programmer from way back so this hardly fazes me, but it is more work).

BTW - isn't compensating for the wire resistance of the wire from the thermocouple to the amp a problem?

#39 tjmck

tjmck

    Vostok 1

  • -----
  • Posts: 177
  • Joined: 06 Oct 2007
  • Loc: Winnipeg, Manitoba, Canada

Posted 29 June 2013 - 07:03 AM

Look at Adafruit Industries. They have all the solutions you are looking for. If you want to go with K type thermocouples - $10 per, DS18B20 - $4 per, TMP36 - $2 per. Most popular controllers including Arduino and Rasperry Pi, Thermocouple Amplifier MAX31855 breakout board. They also have a data logger sheild for Arduino that incorporates a real time clock and SD card for data storage. Most importantly, they have tutorials on how to setup all of the items and a support community to solve issues.

As for measurement precision, a Class 1 K-Type thermocouple is only rated ±1.5°C Accuracy from -40°C to +375°C. For the DS18B20 ±0.5°C Accuracy from -10°C to +85°C.

There a lots of ways to do what you want. I recently assembled a lab freezer temperature mapping system using an Arduino, Data Logger Shield and 16x DS18B20s that was programmed to take measurements off all 16 sensors every 6 seconds. The data was stored as a CSV file and easily read into a spreadsheet. You could also add a LCD and keyboard or touch screen and even run the whole thing from batteries if you wanted.

Tom

#40 Jon Isaacs

Jon Isaacs

    ISS

  • *****
  • Posts: 44369
  • Joined: 16 Jun 2004
  • Loc: San Diego and Boulevard, CA

Posted 29 June 2013 - 10:42 AM

BTW - isn't compensating for the wire resistance of the wire from the thermocouple to the amp a problem



In general the amp will have a high input impedance so current flow in the wire should not be a problem. I buy my thermocouple wire in rolls and make my own...

How much accuracy are you hoping for?

Jon

#41 careysub

careysub

    Surveyor 1

  • *****
  • Posts: 1906
  • Joined: 18 Feb 2011
  • Loc: Rancho Cucamonga, CA

Posted 29 June 2013 - 03:03 PM

Look at Adafruit Industries. They have all the solutions you are looking for. If you want to go with K type thermocouples - $10 per, DS18B20 - $4 per, TMP36 - $2 per. Most popular controllers including Arduino and Rasperry Pi, Thermocouple Amplifier MAX31855 breakout board. They also have a data logger sheild for Arduino that incorporates a real time clock and SD card for data storage. Most importantly, they have tutorials on how to setup all of the items and a support community to solve issues.

As for measurement precision, a Class 1 K-Type thermocouple is only rated ±1.5°C Accuracy from -40°C to +375°C. For the DS18B20 ±0.5°C Accuracy from -10°C to +85°C.

There a lots of ways to do what you want. I recently assembled a lab freezer temperature mapping system using an Arduino, Data Logger Shield and 16x DS18B20s that was programmed to take measurements off all 16 sensors every 6 seconds. The data was stored as a CSV file and easily read into a spreadsheet. You could also add a LCD and keyboard or touch screen and even run the whole thing from batteries if you wanted.

Tom

]

The accuracy figure quoted for the Class 1 Type K might be the absolute accuracy over the entire range, I am only interested in relative precision in 5C to 35C (41F to 95F) roughly (OC to 40C absolute limit - 32F to 104F).

Thanks, Omega had K-type thermocouples for only $3.60 per (pack of 5) and the voltage output datasheet in the range of interest suggested precision of 0.05C or better - but of course that is only one component in overall system precision.


Precision of 0.5 C or better between sensors (not overall accuracy, which could be less) would be what I am looking for. What is critical is the precision of the differential between sensors at a particular time. Neither absolute accuracy nor stability of individual sensor reading over long periods is important, they just all have to agree closely at the same time.

It sounds like I should get into Arduino at some point, just because. Not sure how soon I want to tackle it though.






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics