How would MacGyver build a remote freezer temperature alarm?
September 25, 2013 10:27 AM   Subscribe

I work in a research lab and one of our most critical pieces of equipment is a large chest freezer that normally runs around minus-80-degrees Celsius. It is an older unit with audible alarms built in, but no in-built remote monitoring capability. We'd like to install a simple means of detecting out-of-range temperature conditions and sending out a text message, etc., if the unit starts getting too warm. What's the best way to go about this?

Specifically, the thing we want must be able to:

(1) Detect and accurately sense temperatures in a range down to -80 degrees Celsius.

(2) "Sample" the temperature at least once every 5-10 minutes and check it against a user-determined limit value (-50 degrees C or thereabouts).

(3) Flag the condition if the limit value is reached and send a text message to at least one cellular phone number.

I have done a fair bit of searching for pre-fab temperature monitoring products, and there are certainly a bunch out there, but very few of them actually go down to -80 Celsius. Many stop at -50 C, which would be useless for this application. A simple power-monitoring device would also be insufficient, given that one possible failure mode of this freezer is a shutdown of one of the two compressors (which doesn't cause the unit to shut off entirely, but which would result in an unacceptable temperature rise).

Probably the closest thing to what I'm looking for is this thing (used in conjunction with the available ultra low temperature sensor.)

...but since we don't actually need multiple inputs or the means to connect multiple types of sensor, that product is a tad over-featured.

Getting to the point, it seems like this is the sort of thing one ought to be able to do with a thermistor and an Arduino-type board (I have an OSEPP Uno here)*...is there any good reason why not?

And if not, can anyone recommend a specific thermistor and accessory board (for sending SMS alerts)?


* I'm a little embarrassed to admit that despite being an EE I've never actually used an Arduino board and I have no past experiences incorporating thermistors into designs, just because it's never come up until now. I'll recommend a pre-fab product if that turns out to be the best option for my employers, but I'd really prefer to take this as an opportunity to save them some money AND enhance my skillset.
posted by aecorwin to Technology (11 answers total) 3 users marked this as a favorite
 
There are absolutely off the shelf systems for this. I've worked in labs where any problem with the -80 automatically contacts a list of people usually including the facility/lab manager.This has a low temperature probe rated to -80 and sends some kind of cellular message. Here's another offering. You might try the freezer manufacturer, especially if they have an applications group, or your lab equipment distributor.
posted by JonahBlack at 10:36 AM on September 25, 2013


You've checked that the built-in alarm doesn't have external circuit contacts, right?
posted by zamboni at 10:42 AM on September 25, 2013


Response by poster: zamboni: There may be a way to "tap into" the built-in alarm but no, there are no readily-accessible output ports (that I'm aware of, I'll check again to see if I missed something obvious, but this freezer model definitely pre-dates, say, the USB interface.). Most likely I would have to intercept whatever signal makes the temp alarm LED light up, which would entail directly accessing the control circuit board.
posted by aecorwin at 10:57 AM on September 25, 2013


Instead of tapping into the sensor and temp alarm LED, why not use a light sensor placed on top of it? So when the LED comes on, the light sensor is activated, and then you can go from there. No sense recreating the thermometer and alarm circuitry.
posted by Jairus at 11:02 AM on September 25, 2013 [1 favorite]


Best answer: If you can find a temperature sensor for the right range that talks to a serial port, you may simply build a system yourself with a raspberry pi. I've done something like that for a pH sensor, so if you go this route, memail me.
posted by DreamerFi at 11:09 AM on September 25, 2013


Response by poster: Update: there is what appears to be a DB9 connector on the back of the unit. Here's a photo of it: http://www.flickr.com/photos/aecorwin/9937823726/

Not sure what its actual purpose is...in the process of hunting down a manual that might give a clue.
posted by aecorwin at 11:41 AM on September 25, 2013


Best answer: We use these things from Omega, as an off the shelf solution. A standard, cheap K-type thermocouple would work fine. I really like these because I can just run an ethernet cable to them and talk to the units as if querying a remote webserver. They're very dependable, and we've been using a dozen or so without a single major problem for years now. These things log too, so even if the PC getting the data goes down, tracability isn't lost.

You can go much cheaper though if you are willing to code the application yourself. These provide a thermocouple to PC bridge, with a very simple software interface. You'ld need to write something to do the alerts yourself, but that's not a huge problem.

Note in both cases you need a dedicated PC to do the alarm function.
posted by bonehead at 12:37 PM on September 25, 2013 [1 favorite]


Best answer: That DB9 could well be an RS232 port that you could query for current temperature and status info, but do you really want to rely on the unit itself to tell you when it's not working properly? It seems better to me to have an entirely independent sensor and monitoring system, so that if the freezer loses power or goes insane you can still see what the actual current temperature is and have a log of how bad things got before proper operation was restored.
posted by contraption at 1:00 PM on September 25, 2013 [2 favorites]


By the by, don't waste money on wireless solutions. In an electronically noisy environment like a typical lab, with industrial concrete and metal , wireless transmitters just aren't dependible. We tried a couple of manufacturers.
posted by bonehead at 1:30 PM on September 25, 2013 [2 favorites]


Best answer: I designed and manufactured controllers for cryo freezers for one of the major vendors that were -90c and did work on earlier two compressor versions. Depending on the model, and if it's one of them, I can tell you how to access telemetry. you can decode it and get temps. serial rs232. no response or no change in telem means problems.

If you want an independent sensor to give you alarms above a deeply cold level, you may have to resort to something better than an arduino and thermistor. alarms usually don't have to be too accurate and if you are going cheap, you can go redundant. if you want precise, however, it takes some work.

my design used a TSC500 dual slope a2d and a chopper-stabilized op amp with isothermal ratiometric gain setting networks for zero drift conversions. slow, but 200+ counts per degree in the configuration i employed. micro was a 80c552 (8052 variant) and programmed in assembly code. 90's era. the dual slope converter can be traded off for speed versus accuracy, but slow is good. nice design. worked like a champ. still does. i think there are 20,000-30,000 or more of them out there in various ranges. not all went as low as -90. sensor was a 100 ohm RTD with a test resistor built in for heating it up a little to test the integral alarms. used a howland current pump for linearization of the RTD over the interval of interest, which was -whatever to about +5C. 1 mA excitation current, IIRC.

the dual compressor types were military, though, and used a solid state sequencing relay (which I also designed) and a more primitive controller that was not micro-based.

of course, if it's a different manufacturer, i can't help. memail some model numbers and we'll chat.
posted by FauxScot at 9:04 PM on September 25, 2013 [1 favorite]


I use/used Xbee Pro 100mW 802.11.14 modems and they worked pretty well. Open air they have range of nearly a mile. I had two nodes in a separate building sending data to a collector computer. One node sent readings once per minute. The other node sent readings at a much lower rate, but irregularly. The only time I lost data was when the packets collided. A third display node watched all the traffic and grabbed a temperature reading out of it for display. Works pretty well. Did this in two separate factories with lots of wifi computers running here and there. They are about $30-35 a node. Way more flexibility than I exploited. Might be one of my fave little subassemblies. The 1 mW model... good for same room line of sight. Cheaper. Digikey item. In this mode, they are multi-drop, so you can use as many as you want, you just have to design a protocol to handle whatever the heck you want to ship.

FYI, one was connected to a two channel Omega HH303 thermocouple thermometer with a polled serial interface. I'd send a read command and get back thermal data. The other one was connected to the serial output of a scale. When a weight appeared on the scale and stabilized, an autoprint function emitted a packet via serial I/F and it went into the radio. The display node ignored the packets from the scale, but when it saw a thermal packet, it would snag it, remove and format the data and display it on a large LED display. (I was weighing pizzas and correlating weight to oven temp in a wood fired oven scenario. )

In your scenario, likewise, perfect reliability isn't that important. It's low data rate and a slowly changing value. Simple reasonableness checks and timeouts will let you accomplish what you are trying to do.... get early warning that something is wrong. If x minutes go by and you don't get a reading, that's an alarm.

The controller in that freezer requires that you send it a "T" to get back an 8-byte data package formatted as ASCII. There is some status, alarm condition stuff, set point data and freezer temperature. I've got several of the controllers here and will verify all this if/when you decide to use it. I looked briefly at the source code and am embarrassed at my coding of 20 years back. Ick.
posted by FauxScot at 5:58 PM on September 26, 2013


« Older Songs with Whistling   |   sew lost Newer »
This thread is closed to new comments.