If you're posting a request for installation help, please provide as much info about your system as possible. For example, for help regarding connecting a smart thermostat, please include details about your system including whether you have gas/electric, heat pump, number of heat stages, A/C condenser, etc.

Ecobee 3 Temperature Accuracy



  • ericandiericandi OrlandoMember

    I installed two Ecobee3's in my house for my two AC/heater systems. I have two remote sensors connected to one of my Ecobees and no sensors hooked to the other. My Ecobees seemed to work Ok for a while but now they are both showing huge temperature fluctuations upwards of 7 degrees plus or minus in the course of less than an hour. For example my upstairs Ecobee (the one with no remote sensors) was showing a room temp of 72 degrees and my comfort settings were 74 cool to 68 heat. All of a sudden over the course of 30 minutes the temp reading on that Ecobee went up to 78 degrees. It was evening and the outside temp is 60 degrees. My Ecobee is mounted on a wall with no vents in the area and no heat source nearby. When I saw the temp reading has risen so much I changed my comfort setting for cool to 80 so my AC wouldn't stay on. All of a sudden 30 minutes later this Ecobeee read 72 degrees and this happened without the AC on. The same thing happens with my other Ecobee where the temp reading fluctuates up and down huge swings when there's no cool or heat coming from my system and no external cause. I am positive that the actual room temperate is not actually changing this much and there is a software or hardware issue causing false temp readings. These are brand new Ecobees with the latest firmware. Has anyone else experienced this issue and if so how can I fix it?

  • KRUZNBYKRUZNBY Raleigh, NCMember

    Plug the hole in the wall behind the you thermostat with plumber's putty. There may be a draft in the wall.

  • BPMBPM AlbertaMember

    Kruznby, you are not the first one to say this is the solution and I would think people would be more suggestive to conquering this fault. As I have stated, this is with calibrated thermometers etc. not just saying that it turns off and on and looks good. There is some constant and accuracy involved here. Although that may be the fix at times, what is your suggestion when the hole is completely plugged with plumber's putty?

  • manoficemanofice CharlotteMember

    If you guys have plugged the hole behind the stat and are running the latest firmware... And it sounds like you guys know your stuff about the device.... I don't think there is anything else anyone can offer?? Best bet is to communicate with support so they can fully document your issue and maybe correct it in a future firmware release or issue you a new device under warranty... I don't remeber if you did that already.

  • briansusernamebriansusername UtahMember
    edited March 2015

    I think @manofice is right. I think no one is offering any more insight or suggestions because we haven't seen this issue ourselves. Personally, my ecobee was right in line with my old thermostat and another digital thermometer placed nearby when I tested. Maybe you do have a faulty unit? I don't know... I do hope you get it figured out!!

  • EcocechEcocech MontrealMember
    edited March 2015

    BPM said:
    Kruznby, you are not the first one to say this is the solution and I would think people would be more suggestive to conquering this fault. As I have stated, this is with calibrated thermometers etc. not just saying that it turns off and on and looks good. There is some constant and accuracy involved here. Although that may be the fix at times, what is your suggestion when the hole is completely plugged with plumber's putty?

    For myself, I definitely noticed a difference after plugging up the hole. But I like you, still noticed a difference in room temperature readings vs the Ecobee reading. By simply placing my old WR thermostat on a table in the room, the difference is roughly 1.5 degree celsius (higher on the E3). Now, there may be calibration handicaps between the two devices, but I did use an old analog thermometer that came closer to the WR reading than the E3.

    What I did this week, was I actually placed the WR digital thermostat right on top of the E3 (sitting on it) and the temperature slowly rose and came close to 0.5 degrees off. What this means is that the exact placement of your calibrated thermometers is important when comparing the readings to the E3. I know I haven't tested temperature fluctuations for you to compare help resolve your issue, but this may be something that can help you or others troubleshoot similar issues.

  • I think some ecobees are clearly defective, like mine. I noticed my thermostat reading was several degrees lower than the remote sensors. With continuous fan running all three remote sensors were within one degree of each other and they are pretty much spot on with my NIST calibrated thermapen. I ended up taking one of the remote sensors and taping it to the wall next to the ecobee. Sure enough, the remote sensor there reports the correct temp and the ecobee itself consistently reports 4 degrees low. I could use the temp correction, but I don't trust this ecobee at all because the humidity reading is also clearly too high. It is reporting 10% higher than the Nest it replaced.

    I discovered that if you remove power to the HVAC system, when you bring it back online the ecobee will do the calibration again. Unfortunately, it doesn't matter. My ecobee still settles on 4 degrees low and too high humidity.

    And yes, the hole behind the ecobee is plugged.

  • Ok, the temperature being off appears to be solved. It occurred to me that the leftover screw holes from two previous thermostats were still there. I taped over them since plumber’s putty doesn’t work for them. I have been running for about 15 minutes now and the temp from the ecobee is exactly the same as the remote sensor.

    However, I also notice I now have a newer firmware version. I was on and am now on I was working with support for this issue so I'm assuming they updated it. Thus, I do not know if taping the screw holes fixed it or the firmware fixed it.

    Unfortunately, the indoor humidity is still way too high to be trustworthy. It is showing 40%. The outside humidity is 15%. Before the fixes above it was displaying 44%. Just before taking off the nest, the nest was reporting 36% indoor humidity when the outside humidity was 27%.

  • FlyespressoFlyespresso SeattleMember

    ToBeFrank: I'm having the same issue, I sealed up the holes from the previous honeywell stat and I too have insane humidity. (~60+% when another reader a few feet away reads a dry 31, and it's well, dry in the place!)

    I am also having the temp issue, it's on average 8-10* off when cold and not so much off when warm. This variance makes me think it's just not calibrated to begin with? I've contacted support and yet to hear anything back-I'm on .3957

  • larrywalarrywa OntarioMember

    I am currently having the same problem with a Venstar ColorTouch thermostat. I know this isn't a Venstar stat group but the problem is driving me to the ecobee3 as a Smart Stat alternative. Now I am reading that people that actually understand calibration and deadband setting function are having the same problems.

    I also had this problem on an older electronic Honeywell programmable from 20 year back. Not as pronounced but after the heat cycle was done the stat displayed temperature, and operating logic temperature would increase by a degree or two.

    This problem effectively increases the deadband of the stat on my Venstar T7900 to 2.0 to 2.5 degrees C despite being set to 0.5C.

    The temperature display rise takes between ten to fifteen minutes after the call for heat shuts off. Multiple highly accurate temperature probes do not reflect this artificial temperature rise on the display. This causes it to report an incorrect temperature reading to Home Automation devices also. This artificial temperature rise also seem to be length of cycle dependant. I have a slow rising house temperature when setback comes off in the morning and it takes hours to complete the first heat cycle raising the house temp from 21.0C to 23.0C. This may take 4-6 hours depending on the outside temperature.

    Now after the heat call stops and the temperature is satisfied and the heat stops, the stat display rises to 1.5 to 2.0 degrees C higher. After the temperature falls to the setpoint - deadband = 22.5 degrees the stat call for heat again but in reality the room temperature is 21.0 degrees as measured on half a dozen other probes and thermometers.

    While the call for heat is in progress the stat drops the display temp back to normal, calibrated temperatures so the heat runs correctly fulfilling the setpoint again.

    The result is the stat deadband is the set deadband (0.5C) plus the display creep (1.5-2.0C)

    My joist cavity behind this stat was filled with newspapers and spray foam sealed right to the attic 2x6 plates.

    Next I may try to hang the stat into the room air and insert my probes behind the stat into the cavity. This may prove some arguments here. Otherwise I don't know which stat to turn to get some decent steady temperatures in my house.

    I am sensitive to these cold waves every day. Some people just aren't and 21C heat setpoint and 26C A/C setpoint is just fine for them. I would be sweating in summer and frost bite in the winter with those setpoints :) I can't tolerate these temperature swings.

  • sstevetxsstevetx DallasMember

    I know this is an old thread, but wanted to offer my thanks to all because I had the same issue with my thermostat reading 2 degrees low making the A/C run more than it needed to. I entered the correction factor in settings and that worked for the most part. After reading this, I used some clear silicone and filled up the wall hole and now the temperature reading is more stable and doesn't require any correction factor.

  • LarrylLixLarrylLix Northern Southwestern OntarioMember

    Now with an ecobee3 I find it has more problems, not with joist cavity air but with attempts at using the display register for the overshoot predictor. This does cause a good cycling feature keeping the room temperature constant by avoiding temperature overshoot of the HVAC system, but it means this:


    It may be totally inaccurate. Use a second accurate temperature sensing device mounted close but not bove the electronics in the stat under question.

Sign In or Register to comment.