No announcement yet.

Temperature Monitoring

This topic is closed.
  • Filter
  • Time
  • Show
Clear All
new posts

    Temperature Monitoring

    Hi Folks,

    I've had my Ocelot and a bunch of SECU-16I's working well monitoring all the doors and windows in my home but I want to add temperature monitoring to the system.

    I've run 4c22 telephone wire to each room for future motions sensors so I'm pre-wired for a number of options:

    1) Buy a 1-wire interface, sensors and a plug-in to build a new network.
    Star network (each sensor on own leg, ~50' max/leg)
    Cost will be a few $100's of dollars if I need hub's.
    2) I already have a small bag of Microchip MCP9700 analog temperature
    sensors in a TO-92 package so I could wire them up and connect them
    as analog inputs to my Ocelot and calculate the voltage from the reading.
    I'm not sure it will be as accurate/reliable as I would like. The sensors
    output 10mV/degree C and the SECU-16I inputs are sampled at 8-bit
    resolution from 0-5V (~20mV resolution). It sounds like this approach
    is a non-starter without amplification since the SECU-16I sampling
    resolution is 1/2 the sensor resolution but I thought I'd ask the community
    what they thought.

    Has anyone used the Ocelot in this manner?

    If I have to add amplification to get a usable signal then I might as well go with 1-wire.


    Yes I do montior temperature with my Ocelet. I use the LM35 using an LM317 to produce 4-20 amps to feed to the Ocelost. I will attach the circuit and circuit board layout to this reply.

    Attached Files


      Thanks Ken,

      To use your circuit I would have to skip the board and instead build this up by hand into a pencil like assembly (potted in epoxy) that can be inserted into holes in my ceiling... cumbersome but doable.

      I'll consider this further but my best option might just be to to go 1-wire.



        Have you considered the temperature bobcat? I'm not sure how many sensors you factored into your budget quoted above but if you translate severeal hundred into $200 that could get you 4 bobcat-Ts if you shop aorund. It's a plug and play process after that with all the engineering you mentioned done for you.

        I realize some of us enjoy the engineering part, but when it came to time spent and ROI I finally opted for the bobcat-T route.


          I've looked at that but it isn't suitable. I have 10+ locations where a 4c22 wire is brought out to a hole in the corner of the ceiling in each room. My goal is to connect the wires to a narrow sensor device and insert this back in the hole with the sensor protruding enough to get a good reading.

          I can get 1-wire DS18B20 temp sensors for <$2 on eBay so my cost per point goes way down with each sensor I add to a 1-wire network. I like that there are plug-ins to use as well so I'm leaning towards 1-wire.

          Time to buy some more hardware!


            Sure, that makes sense. Not being a 1-wire user, what are the fixed start up costs with such a setup? Plugin + Hardware before adding the cheap sensors? Does that include the amplifier circuits you mentioned earlier to address the sensor resolution limitiations?

            Once you've invested in that, how many sensors can it scale to?


              I just paid for an EDS HA7E 1-wire interface for about $70 delivered. I also bought 15 of the DS18B20 temp sensors for $18.

              When these come in I'll see what I can do with just this equipment. I've read that I can put a 100 ohm resistor at the head of each pre-wired data line (star network) to reduce reflections so if that does the trick I won't have any more cash outlay. If that doesn't cut it then I'll look at a hub.

              Looks like this isn't an Ocelot discussion anymore!


                I think you will find that the HA7E does not support the DS2409 (hub switch). There is a degree of black magic to setting up a 1-wire network and the more branches you have from the main trunk the greater potential that exists for waveforms that tend to cancel the primary digital bit stream. When you do get a configuration that works using with branches then be cautious to change it with new sensors etc as a single modification can bring down the entire network. That is what is nice about the hub switch. It isolates each branch electrically for greater probabilty of initial success and then later for fault isolation.


                  1 wire is pretty cheap and stable once you get it up and ruinning. I have had over 10 sensors going in the two houses I have installed for years with no issues. I don't have hubs, I just used a telco punch down panel for tieing all of the branches together. With Star topology I needed resistors in some of the shorter runs to clean up the network, I would say that a bus topology would probably be better if it is possible but for me that was not really feasible.