Can I use any sensor via an analog input?

[beelzerob]

I'm looking for a good solution for measuring water temp in a copper pipe, without having to open the pipe. This is the best thing I can find.

I was hoping I could use my secu16 analog inputs to measure whatever voltage this thing produces and convert that to a temp value. I just don't know how interchangeable these sensors and I/O things are.

The sensor link doesn't indicate what the voltage range is for that. I also realize that the secu16 only has an 8-bit A/D, so that's 255 discrete values...that should be enough. The sensors range is -10 C to 85 C, so that's an LSB of about a third C.

The sensor writeup however indicates that it's a digital sensor with a 12 bits resolution. Sooo...does this thing NOT produce a simple voltage from 0 to 4 or 5 like my other temp sensors I own do?

 

[ericvic]

I'm looking for a good solution for measuring water temp in a copper pipe, without having to open the pipe. This is the best thing I can find.

I was hoping I could use my secu16 analog inputs to measure whatever voltage this thing produces and convert that to a temp value. I just don't know how interchangeable these sensors and I/O things are.

The sensor link doesn't indicate what the voltage range is for that. I also realize that the secu16 only has an 8-bit A/D, so that's 255 discrete values...that should be enough. The sensors range is -10 C to 85 C, so that's an LSB of about a third C.

The sensor writeup however indicates that it's a digital sensor with a 12 bits resolution. Sooo...does this thing NOT produce a simple voltage from 0 to 4 or 5 like my other temp sensors I own do?

The probe you mention uses the 1-Wire protocol which is a digital protocol so you won't be able to read it from an analog input.

Eric

 

[beelzerob]

Well nuts. That's the only decent, simple surface mount water temp sensor I can find.

So I guess I'm either at putting water sensors in the pipe itself, or using a 1-wire solution that works with premise. The former I'm familiar with....what about the latter?

 

[123]

To my knowledge, no one has ever published a 1-wire driver for Premise. The options would be to write a driver for the Maxim DS9097U, Midon Designs Temp08, Embedded Data Systems HA7NET, or DataNAB BarioNet 1-Wire Sensor Gateway (listed in order of increasing cost from about $28 to north of $300).

Probably not the challenge you want at this stage in your project.

The HA7NET is around $150, has a LAN port, and is controllable via URLs so it seems to represent the best value. The DS9097U is cheap, is available in serial or USB format, but I don't know much about its API nor how difficult it's be to interface to Premise.

 

[BraveSirRobbin]

What is your budget (I'm guessing low since you are using Premise)? I know we used to use (if I'm remembering correctly) Relco 1018P temperature probes wrapped with black tape against copper pipes to measure temperature of water for a heat exchanger/cooling system, but this was for a work application and the sensors are pretty expensive. We would then wrap insulation around them to prevent ambient air influence. This prevented us from having to intrude on the piping interior.

They had an incredible LINEAR analog output though and came with calibration curves so you can custom fit an equation for that particular probe.

The difficulty in using other less expensive alternatives is the lack of a linear response and a packaging that does not lend itself well to exterior water temperature measurements inside a pipe.

I can find out more detailed info if you like.

 

[damage]

depending on how much accuracy you need, what about an lm34? you end up with 2 degree accuracy. http://www.national.com/mpf/LM/LM34.html

 

[beelzerob]

Well, budget is definitely an issue, and Premise seems to have the basic core functionality I'm looking for....but I *really* don't want to have to get into driver writing for THAT software too. So I'm hoping to only use devices that are supported already.

That pretty much means using a secu16, which I'm fine with, as it's versatile enough to do most everything I need I think. I'm looking to measure maybe 4 temps, 2 or so switches, and control a few relays.

I appreciate any help for anything you can find, BSR. But what damage linked to looks awful interesting. I had stumbled across that site before in my many google attempts, but I couldn't figure out how to use a circuit diagram on a pipe!! Now I finally discovered the "packages"....those things all look like they belong on circuit boards. If I get one of those with 3 pins on them, how will I connect that to, say, a secu16 to read values from it?

 

[jwilson56]

If you search the ADI Ocelot forum you will find some instructions for using the LM34 with the SECU. I have some new ones laying around here somewhere I was selling for a couple bucks each. I don't need them since I have a 1-wire masterhub and sensors in place.

 

[beelzerob]

If you search the ADI Ocelot forum you will find some instructions for using the LM34 with the SECU. I have some new ones laying around here somewhere I was selling for a couple bucks each. I don't need them since I have a 1-wire masterhub and sensors in place.

Excellent! That takes care of reading water temps then. Looks like this might be a good working solution after all. I didn't think to look in the ocelot forum....

 

[BraveSirRobbin]

The reason I did not recommend an LM34 is that it has a very small millivolt output and it's high dynamic range will not lead to very good accuracy with an eight bit A-D converter.

Also, I don't see how you can efficiently mount that to a pipe and expect it to accurately measure the water temp inside.

JMHO though, your mileage may vary...

 

[beelzerob]

Oh, well...rest assured I haven't looked that far ahead either!

I'm not *too* worried about accuracy....but are we talking within a degree or so? Two?

I figured the flat side of the thing was the temp measuring area, and if I can just press it against the side of the pipe, maybe with some thermal paste....well, I guess we'll see.

I'm still open to other suggestions if you have them, BSR.

Do you think I might do better if I get this analog voltage thermistor from datanab, and just tape it on the pipe?

 

[BraveSirRobbin]

The problem with that "resistive type" sensor is it is not linear. JeffVolp explains how you can try to get this to a linear output in THIS post from ADI's forums, but it can get messy as you have to select precision resistors based on the type of sensor (output curve) you are using.

The LM34 is linear, but it has a very low output which is not good for running the signal over any length of wire. The resolution you would obtain from an 8-bit A-D converter would be around 2 volts (deg F). (More details HERE).

The output is 10 millivolts per degree F, so if you were measuring say room temperature at 70 degrees F this sensor would output 0.7 volts.

If this were me and I HAD to use the SECU16 as an analog to digital converter, I would probably look more into trying to linearize a resistive (thermistor) type temperature sensor as JeffVolp did in that ADI post.

I wish someone made an inexpensive true linear output temperature sensor such as the LM34 with about ten times the voltage output.

Again, I emphasize, this is all JMHO!

 

[roussell]

You could use the 1-wire sensor you originally found if:

You purchase an HA5-ASCII. It converts the 1-wire protocol into simple ASCII text and interfaces through a serial port, you could then use Premise or most any other HA app to read the messages from the serial port - should be minimal coding, 123 could most likely help there.

OR

There is also this PHAnderson assembled kit which interfaces to the serial port and is for multiple DS18S20 1-wire temp sensors.

OR

You can run OWFS under Windows using cygwin and a DS9490 adapter. OWFS exposes the 1-wire network as a folder/file structure, so to see the reading of a particular device, you navigate to the "directory folder" of that device and read the data in the "file" contained within. Again, minimal programming would be needed to read the file content in the directory structure. OWFS runs in a Linux environment so you'd need cygwin to emulate that environment in Windows. As an alternative, you could create a virtual Linux box in Sun's VirtualBox (free, supports USB), share the OWFS through samba and read the sensor data that way (that's how I do it).

OR

You could also run Linux natively on the box and use something like Misterhouse (with OWFS) instead of Premise to read the sensor data.

OR

Lastly, you also get something like this that plus into a USB port and exposes the temp but it might be hard to get it to read from the pipe surface but I think you're on track with the thermal paste and aluminum tape. Again, I think some programming will be required to get it into Premise.

From the same site you could instead get this and this for better surface temp measurement.

EDIT: If you're feeling especially cheap, yet like to hack (as I do) the you could buy this for $8.76 (free shipping) and try to extend the sensor out of the casing an inch or two. The device is registered as a serial port and apparently just sends ASCII data. It includes an app with full source and an SDK of sorts.

I just found this one on the same site and for $14.97 it includes a remote temp sensor and has a broader range. No hacking required. I might order one of these myself to see how well it works...

It looks like you may have to choose between developing hardware, or software - I guess it depends on the lesser of the two evils from your perspective...

Terry

 

[beelzerob]

*sigh* Why is nothing EVER easy?

Thanks for all the suggestions Terry. If anything, I've now got too many ideas of what to do...it's somewhat paralyzing.

I'd really like to utilize the secu16 for this project, as it's pretty inexpensive, it does digital and analog inputs and has relays, and is already supported by Premise. I'd also like to keep this to a single hardware device if possible, rather than adding others specifically for measuring just the pipe temps.

Does no one in these forums use a secu16 to measure temps, water or otherwise?

Hmmmph. I think I need to come up with a definitive list of what I'd like to measure and make sure the secu16 is going to be able to handle it as is.

Edit: Looking at some more of terry's suggestions, it might be worthwhile to get one of those devices that allows a lot of 1-wires to be attached and provides simple ascii data...that *should* be within my programming capabilities in Premise.

 

[BraveSirRobbin]

Sorry, I wasn't trying to make life difficult for you, just giving you the pros and cons of the methodology (SECU16 use) you were trying to use.

Why don't you send JeffVolp (he's also a Cocooner) a PM and ask him to look at this thread and get his opinion directly on best methods of using an SECU16 for your application?