One wire temp sensor Accuracy and Calibration Questions for my GARN sensor network.

[Gooserider]

Is there a way to have a data logger adjust the values your sensors are tracking? Say for example, the sensors are off 7 degrees in 180 water, netting 173. Or 150 degree water its off 5 degrees. The sensor data sends the data back to the computer, the program references to a % error chart and records the correct temperature. Not an exact science but plus or minus a degree or two seems "good enough"? Just some food for thought. I would think Microsoft Excel/Access could handle the conversion/tracking easily.

Data can be manipulated and massaged quite readily at almost any stage of the process from initial collection to final display, just takes the right software tools, which are easily found... The problem with massaging data in that way though is that you don't have any solid way of knowing how well what you get post massage is a reflection of reality... As long as the massaging matches the real world it is fine, but the computer is totally dependent on the programmer to make the massage accurate.

Thus it is quite proper to work with the sensor placement and installation to get increased reading accuracy, and it's reasonably safe to adjust with a known valid calibration curve to deal with a verified sensor error, but it is vital to ensure that you are correcting a "real" error, not engaging in wishful thinking of "it should be X, so I'll adjust the sensor to match..."

Gooserider

 

[sgschwend]

I had taped my wires to the heated device to provide good stability so they are heated just like the probe. After I used the goop the boiler probe can up to match the thermistor.

After watching the boiler/storage probes for an evening I can see the storage probes are matching the boiler probes.

I was surprised on how long it took for them to synchronize. The boiler hasn't been run for three days so the storage temps were low.

If one knows the temperature of the hot water leaving the boiler then one should know the temp of the water entering the storage. At some point the two temperatures should match. Then the lower probes temperature rising to eventually match the upper prob. It took it awhile but after an hour or two I could see all the probes were measuring correctly.

You can quote me, "goop is good", "not just for thermistors anymore".

 

[Gooserider]

If one knows the temperature of the hot water leaving the boiler then one should know the temp of the water entering the storage. At some point the two temperatures should match. Then the lower probes temperature rising to eventually match the upper prob. It took it awhile but after an hour or two I could see all the probes were measuring correctly.

Well you do need to allow for heat losses from the piping, and stratification... Beyond that difference between the theoretical ideal and the real world, you are right...

Gooserider

 

[heaterman]

If you are looking for absolute accuracy the best is a directly immersed sensor. A surface mounted sensor will always be a few degrees off unless you use thermal grease under it and insulate very well over the top.

 

[dirttracker]

You can deal with the error in the data logger software, but I suspect it's more complicated than just a percent. You will probably find that different locations require different calibration factors and the calibration factors will be dependent on the difference in temperature between the local ambient temperature and the temperature being measured by the sensor. I agree with not making this into a science project to get out the last .1 degree, you will have to decide how accurate you want the results to be.

The following items may or may not be important depending on the accuracy and temperature range you require:
Wire gage, type of insulation, and construction
Location of the wire in the neighborhood of the sensor (close to other objects, above or below items that will block natural convection airflow)
Amount of airflow around the wire/cable
Type of pipe where the sensor is attached (usually greater thermal resistance here will result in more error)
Attachment method (close contact, thermal grease, insulation, etc)
Length of wire inside the insulation with the sensor

Basically you are trying to account for a complex heat transfer system that is setup between the water and the local ambient temperature. The heat transfer system depends to a greater or lesser degree on a number of factors. Keep in mind, you are usually working with better than a 100 F difference between the sensor location and ambient, this tends to make heat flow in places you may not expect.

Eric

 

[rkusek]

I had taped my wires to the heated device to provide good stability so they are heated just like the probe. After I used the goop the boiler probe can up to match the thermistor.

After watching the boiler/storage probes for an evening I can see the storage probes are matching the boiler probes.

I was surprised on how long it took for them to synchronize. The boiler hasn't been run for three days so the storage temps were low.

If one knows the temperature of the hot water leaving the boiler then one should know the temp of the water entering the storage. At some point the two temperatures should match. Then the lower probes temperature rising to eventually match the upper prob. It took it awhile but after an hour or two I could see all the probes were measuring correctly.

You can quote me, "goop is good", "not just for thermistors anymore".

Did you just get some Radio Shak goop or something off the internet? I have some old quicksilver stuff from about 5 yrs ago I used on a computer so I will try that.

 

[sgschwend]

Mine was older than that, you know how you have that junk box of stuff from work, some of my stuff comes from the 80's.

It is kinda cool, you can see it work by watching the temperature rise. The bigger the blob the better the job!

 

[heaterman]

Any heating supply house should be able to hook you up with the real deal. Grainger or any Johnstone Supply would have it also. Ask for heat transfer paste. I have used copper based anti-seize compound in a pinch and it works well also.

 

[deerhntr]

As promised, I said I would post my results from my "Science Project" once I had something. So here goes...

First as others have already said, the thermal grease makes a difference. For me, that difference was about a 4 °F improvement when compared to the temp measured in the thermowell. I also improved the thermowell measurement when compared to the Face plate thermometer by about 2 °F when I applied the thermal grease to the sensor in the well. Plot 1 shows the result during an overnight cycle. For those interested I got the grease frombroken link removed to http://cgi.ebay.com/New-X-15-CPU-THERMAL-COMPOUND-paste-grease-XBOX-360_W0QQitemZ220546549924QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item33599948a4)

I included a couple of pictures that show the with and with out grease sensors I used. Basically, I tried a couple of ideas pointed out here. I am going to go with the lead wrapped around the pipe, thermal grease under the sensor, then aluminum tape taping the whole thing to the pipe. Then cover in insulation. None of this done on PEX, only Cu.

Plot2 shows another burn cycle where I now have a sensor on the supply pipe at the HX in the house. You can see that I am measuring virtually no difference between the GARN barn and the house. The difference is +/- 0.5 °F on average. That is a 110 foot underground run between the house and the barn. That is why I wanted to try to eliminate measurement error, so I could make a solid determination of system performance.

Finally, from this whole exercise, I think I now understand the GARN burn cycle much better. If you take a look at the Plot 2, you can see very soon after I start a burn, the temp of the H2O in the supply moves up to match exactly the temp in the thermowell. Then once the burn has completed, the H2O temp in the supply pipe begins to drop below the thermowell temp. I believe when the temps match, the water is moving and mixing in the tank during the burn and BTU tranfer, therefor they match.. Once the burn has completed, the H2O in the tank begins to temperature stratify. The thermowell is at a different height in the tank then the source of the supply H2O, hence a different temp in the supply pipe. I also have a sensor measuring the ambient temp of the boiler room. The temp rises 10 °F during the burn cycle, but the change does not follow the supply sensor change, so I don't think the boiler room temp is causing the difference in the thermowell and supply sensor over the burn cycle. I think it is the GARN mixing, then stratifying.

Thanks to everyone for the help, input, and insight. You guys are good.

 

[deerhntr]

Just one more plot. This shows the difference between the Garn Barn sensor, and the House sensor. This really make me feel good that the Pex with the spray foam insulation is really working well. THANKS JIM K

 

[sgschwend]

You just got to love this stuff. Great work, thanks for the follow-up.

I have found with my stuff I can choose the set points that control the circulator and find the best value for both wood burn periods and also shut down heat recovery period. Very handy. I wouldn't be able to do it without the data.

 

[deerhntr]

You just got to love this stuff. Great work, thanks for the follow-up.

I have found with my stuff I can choose the set points that control the circulator and find the best value for both wood burn periods and also shut down heat recovery period. Very handy. I wouldn't be able to do it without the data.

Steve,
No doubt! I love it. You are right, the data really gives insight in to the subtle operation of the system Good stuff.

BTW, your portable saw mill is pretty cool too! Good video.

 

[dirttracker]

I wish I had the time to set up a data logging/control system like that. I've looked at the 1 wire sensors before, how easy were these to set up? I have not done much programming since I did some assembly language programing for a DSP almost 15 years ago. Most of the lingo being used now is foreign to me. What did you use for your logger/controller? I've seen some of the other controllers folks have worked up here, I'd like to play a bit myself when I get the time.



Eric

 

[Jim K in PA]

Just one more plot. This shows the difference between the Garn Barn sensor, and the House sensor. This really make me feel good that the Pex with the spray foam insulation is really working well. THANKS JIM K

LOL - you're welcome Russ! It really is a good solution and a cheaper alternative to the Micro-flex. Luckily we both had ready access to the service . . .

 

[deerhntr]

I wish I had the time to set up a data logging/control system like that. I've looked at the 1 wire sensors before, how easy were these to set up? I have not done much programming since I did some assembly language programing for a DSP almost 15 years ago. Most of the lingo being used now is foreign to me. What did you use for your logger/controller? I've seen some of the other controllers folks have worked up here, I'd like to play a bit myself when I get the time.



Eric

Eric,

I used the arduino usb board from Sparkfun Ardunio. There is also a entry on the wiki page here at hearth.com that gives some backgroundHearth Wiki Arduino. There is a very large user community for this microcontroller that has developed a bunch of code. Not just for HVAC stuff. So you don't have to be an expert to get it up and running. There are a bunch of examples, and that is where I started. I do have to admit that I have some C-programming experience that comes in handy, but the vast amount of documentation and examples out there really help speed the learning curve. The arduino code is "C/C++ like" which makes things very easy, and understandable.

You should be able to get a simple system up and running for $50 and a little bit of time. Hey, this is a neat winter time/cabin fever project that will yield a little benefit.

 

[dirttracker]

Russ,

Thanks for the info. I've looked at the Arduino in the past, but only with the thought of using it with thermocouples. The 1 wire sensors would be a great addition, you don't need as many channels of analog IO to make a system work. I need to run a new underground line to my shed this summer, I'll include some Cat 5 cable to network in a future boiler controller.

Thanks

Eric

 

[deerhntr]

Russ,
The 1 wire sensors would be a great addition, you don't need as many channels of analog IO to make a system work. I need to run a new underground line to my shed this summer, I'll include some Cat 5 cable to network in a future boiler controller.

Eric,

As far as one-wire sensors, you don't need any Analog I/O pins. The one wire network consists of 3 wires:VDD(3V -5V), GND, and DQ(Tri-state Data). That is it. Then back on the arduino, you consume 1 digital pin for read/write. Each sensor has a unique 64-bit address, so you simply wire all the sensors in parallel on the bus, and you are up and running. You can use the analog pins on the arduino for your thermistors, or sensing relays, or something like that. I found something on the web that would allow you to sense a current with a one wire sensor. There are a great deal of possibilities.

 

[dirttracker]

Russ,

Yeah, I was thinking of the analog inputs I would need to use with thermocouples. The 1 wire sensors would take care of most of the temperature measurements, there are still a few measurements to do with TCs, but 2 or 3 lines of analog IO is not a big deal. If I had to do all of the temperatures with TCs I would have to use a switch to get all of the analog inputs into the controller plus all of the circuits I would need to build for the different TC interfaces. This stuff is so cheap, I just need to get off my backside and order some hardware to play around with. I agree the possibilities are almost limitless. I really want to make a variable speed controller for my main circ pump so I can use the pump speed to control the return temperature from the house into the tank.

 

[deerhntr]

Russ,

Yeah, I was thinking of the analog inputs I would need to use with thermocouples. The 1 wire sensors would take care of most of the temperature measurements, there are still a few measurements to do with TCs, but 2 or 3 lines of analog IO is not a big deal. If I had to do all of the temperatures with TCs I would have to use a switch to get all of the analog inputs into the controller plus all of the circuits I would need to build for the different TC interfaces. This stuff is so cheap, I just need to get off my backside and order some hardware to play around with. I agree the possibilities are almost limitless. I really want to make a variable speed controller for my main circ pump so I can use the pump speed to control the return temperature from the house into the tank.

Eric,

The the analog inputs a very powerful. Take a look at the Data logger Thread. Yesterday someone posted a link of a one wire A/D thermocouple circuit for flue temp measurement. Really neat, and sovles the problem of needing the aruduino close to the thermocouple. You just build up the one wire board, remote it close to the thermocouple, but not in the flue, and then power/read//write over you one wire bus. I like it.