Drag Your Office Aircon Into The 21st Century With Wi-Fi Control

We’ll all have worked in offices that have air-conditioning, but a little too much of it. It’s wonderful on a baking-hot day to walk into the blessèd cool of an air-conditioned office, but after an hour or two of the icy blast you’re shivering away in your summer clothing and you skin has dried out to a crisp. Meanwhile on the other side of the building [Ted] from Marketing has cranked up the whole system to its extreme because he’s got a high metabolism and an office in the full force of the midday sun.

Wouldn’t it be nice if individual air-conditioning units could be easily controlled. To that end, [Maya Posch] has made a rather nicely designed board that takes a NodeMCU board with its ESP8266 processor, and uses four of its outputs as PWM to produce 0-10 volt analogue outputs via filters and op-amps to control individual units. In addition there is an onboard CO2 sensor and a temperature sensor, with provision for an external temperature sensor. The whole fits very neatly into a standard electrical outlet enclosure.

Software wise, the system uses the Sming framework providing an MQTT  communication with a backend server that allows the users to control their aircon experience. This is very much a work in progress, so the software has yet to be put up. (Hint, [Maya], hint!) The whole project though is an extremely tidy build, in fact a thing of beauty to a standard you’d expect from a high-quality commercial product. It’s this that tipped the balance into our featuring it before the software is released, it’s one to keep an eye on, because quality like this doesn’t come every day.

This isn’t the first aircon control we’ve brought you, take a look at this one controlled through Slack.

32 thoughts on “Drag Your Office Aircon Into The 21st Century With Wi-Fi Control

  1. Poor building design and cost cutting lease holder “improvements” aside, I’ve pondered the thermostat war dilemma. With IoT devices being super cheap, why not outfit every cube/desk with a simple device that users can operate to indicate if they want it cooler or warmer. They don’t get to pick the temperature. The system uses democracy of the devices to nudge the set point up or down.

    The device would have 3 LEDs – “I want it warmer”, “I’m fine” and “I want it cooler”. A single momentary dual throw toggle switch that lets the user choose warmer or cooler. And a timer that after an hour, returns the device to “I’m fine”.

    Feel cold? Look a the control, an if the “I want it warmer” LED is not lit, toggle up to light it up (state will change from either make it colder or I’m fine to I want it warmer).

    If the majority of devices in a control zone want a temperature change, start a slow PID type algorithm to start nudging the temperature. Add some time of day, day of week, plus holiday understand to anticipate based on history. Maybe even a bit of “outside” influence if you want to get clever enough.

    Buttons for everyone!

    1. It alrwady exists. Not necessarily at the individual level, but at the zone level. VAV boxes can be programmed on a voting system to determine need for heating or cooling. Typical system give each zone 1 vote if they are within 1 degree or 2 votes if they are 2 or more degrees away.

    2. The timer would lead people to believe it was not working and they would ignore it. On top of that, comfort and pain are in balance here. Trying to make most people comfortable may put some people in physical pain, and that could be a huge liability. Say you want to make it cooler so you and your dudes are not sweating so much, but now the arthritic old lady now can’t move her fingers. Could be a big problem. Maybe someone should just invent a working environment with internal doors between thermally isolated (well, loosely thermally coupled) rooms?

        1. The idea of controlling hvac for comfort at work is silly. Companies want efficiency so they set it to provide the best temperature to work at and cost the least. Companies have room based thermostats that damper each room based on zones. But its not simple to heat one room and cool it with the same hvac system, it can’t do both so something has to give. Usually that means the controls are fixed and usually 2 degree variance. Employees will still complain since that one person had control and its not right.

          Even motion thermostats are out, meeting rooms that go into a programmed temp when occupied. yet still get complains its not right since one person shows up 15mins early and its not the ideal temp. Yet after the meeting never hear from them and it probably got to the desired temp in a few mins.

          Again this design would be for programmed set temps and not personal control. Work from home if that is an issue, or wear more clothing. Much harder to wear less clothing for most, and HR would disapprove of that.

          1. Depending on the type of system, it is entirely possible to heat one room and cool another. A VAV system (what you seem to be describing with dampers) with reheat coils in the terminal units could do exactly that. A VRF/VRV Fan Coil Unit system could do it too. Whether it’s a good idea is an entirely different question!

            I think the easiest control reigime for most workplaces is going to be a set of sensors for the space depending on it’s size, and remote temperature controllers (in a locked room) – mainly so that people aren’t constantly changing the setpoints.

            Once the temperature is in the recommended comfort band (ASHRAE/CIBSE) i completely agree that it should be down to the occupants themselves to adjust via clothing, or getting up for a brief bit of exercise, warm drink etc.

      1. agreed. buildings are generally far too warm these days. If you are too cold, you can always put another jumper on (or wear a hat or some thermal underlayers), but if you are too hot there’s a natural limit on how little you can wear.

    3. I’d call this new installation a definite improvement, to be honest. The current setup sees the sensors/controller units mounted on the wall, with most of the AC units mounted at the opposite side of the room near the windows. As a result it’s always a number of degrees colder near the windows compared to the wall-side.

      As for adding democracy to the mixture, I fear that there the problem is that one seeks to find a technological solution to what is essentially a social issue and to some extent a biological one. ‘Comfort’ is a really hard thing to quantify.

      The main goal for this project is therefore essentially to undo some of the worst design failures, and to get to a point where we can at least contemplate the concept of having control over the temperature in the office. After that we would like to explore the idea of having different temperature zones in the open office we’re dealing with.

      Consider it a big experiment :)

      1. Hi Maya, as an engineer who designs the layout of air conditioning systems for a living, I can tell you there are several good reasons for having the air conditioning grilles near the windows and the controllers not so near the windows.

        Windows are good at letting heat out in the winter, sun in the summer. Much more so than a good old opaque wall – have a look at U and R values, even of high performance glazing vs a even a timber frame wall with a bit of insulation.
        The effect of this is to cause localised heating caused by the sun, and draughts caused by cold air outside (depending on respective day, season, weather etc). To offset these factors, we design for grilles as close to the windows as possible, especially with floor to ceiling glazing. When selecting grilles themselves, a good engineer will work out the throw distance of air the grille will produce based on a fairly standard set of design parameters, the grille parameters, and the data calculated for the air conditioning unit connected. Essentially, we don’t want to hose whoever is underneath the grille with cold air as much as the person who will be sitting there wouldn’t want to be.

        The controllers are further away for the exact same reason. If they were right next to the window they would be reading the localised climate. Getting hosed by the sun they would be put the units on constant cooling mode. I’ve actually got a wireless temperature sensor in the window at home (just happens to be propping something up, and it’s hilarious to read the temperature and RH from it on a sunny day.

        Thermal comfort actually already has appropriate, specific, and measurable metrics – though measurement in a actual office of all the necessary inputs could be expensive and tricky in reality, which is why air air temperature is usually the only thing assessed for the control of the AC.
        Predicted Mean Vote ( fangar ) and Percentage People Dissatisfied are based on Air temperature, Mean Radiant Temperature, Mean air velocity, occupant clothing (in units of clo which can be calculated) and metabolic rate. As fangar pointed out from asking people, you can never please all the people all the time!
        The BREEAM ratings scheme has a whole section dedicated to thermal comfort based around these metrics.

        I hope you find my comments enlightening. I actually love what you’re doing, and would love to see the data you collect!
        Honestly I don’t get the opportunity to get the feedback I’d like from occupants to feedback into the design process.
        I hope my knowledge will be of value to you.

        I’d suggest you look at adding Radiant temperature sensors to your sensor lineup. Sontay do black bulb globe sensors, or one can be made with a ping pong painted a particular shade of grey with a thermistor installed at the centre. How you integrate that input will depend on how well it syncs to the air temperature – as the air temp will change far more rapidly than the MRT/’surface’ temperature of the room, but it would give a improved feedback input for AC control.
        In actual fact the CIBSE Guide A for the design temperature for offices and other spaces is actually based on the Operative temperature, which in pseudo code is:
        OperativeTemperature = mean(AirTemp, MRTemp).

        Of course, saying all this if your Air conditioning system was not designed by a decent engineer, was designed by an engineer in a hurry, was installed on the cheap, or the office design has changed and the AC hasn’t been reconfigured for the new layout needless to say someone is even more likely to get too cold or too hot.

        1. Also, it should be Fanger (with an e). Google Fanger and PMV and you’ll come up with a shedload of info on thermal comfort. (much more fun to experiment i know! ;) )
          Your Co2 sensor looks like a MH-Z19? I have the Z14! Make doubly sure the ‘thermostat’ unit is sensors away from direct sunlight as these NDIR sensors use Infrared and will give erroneous readings in sunlight. I’ve taken someone else’s suggestion and housed mine in a scooped out smoke alarm.

          Are you going to hook up to the vent systems too using the Co2 sensor as a IAQ measure?

          1. Yeah, we’ll be ceiling-mounting these controllers, comfortably away from direct sunlight and direct airflow :)

            It’s an MH-Z19 CO2 sensor, yes, as described on my HaD.io project page as well. On my desk I currently have an MH-Z14 as well, but I haven’t really noticed a difference between the Z14 and Z19 other than that the former is hugely more expensive than the other, not to mention much bulkier. I like the Z19 for being compact and easy to integrate :)

            We’re currently looking at how far we can (non-invasively) integrate into the system. Sadly it’s a centrally distributed system (coil fan units, with either hot or cold water supplied and air mixing done centrally as well), so we are limited in what we can do. Having dedicated (stand-alone) AC units would make our life easier, I imagine.

            The CO2 readings at this point are mostly to keep an eye on CO2 levels throughout the office, since CO2 levels have such an impact on one’s sensation of well-being and ability to concentrate. One study I read suggested that a rise from 400 to 800 ppm dropped work performance (focus) by as much as 21%. The ~2,000 ppm we often see in the meeting rooms here is at a level where it will make people feel lethargic and distracted.

            Not much we can do about that just yet (building manager decides how much fresh air gets mixed into the air we get), but we can at least crack open a window when it gets bad :)

        2. A lot of places are seemingly designed as general purpose so the air handling is – dump at one end and suck back at the other. Leading to polar conditions near the vents and sweaty people complaining at the returns. I expect that to do otherwise requires multiple fan stages and servo driven panels to redirect the appropriate flow rate as required.

          One place I worked had the office manager and guy in charge of the thermostat, a particularly obese man**, in a sunny side office. The rest of management also had sunny-window offices. Summer or winter, the separated engineering area fed from the same system was 55F.

          His F350 leaned when he got in. He was 5’6″

          1. Hmm, i suspect you’re talking about as general purpose air handling is a ‘central air’ system, or what we call a constant volume system in the UK. What you desribed as an improvement is a Variable Air volume system, which usually consist of a central air handling unit (say on the roof) ducting to individual terminal units (VAV units) which at minimum have a motorised damper that regulated the opening of the ductwork to a grille/outlet. Usually the fan can be ramped up and down and this is seen as a benefit of the system as it can reduce the energy usage. Modern fan motors use Variable Frequency Drives to achieve this.

            Was 55F (12C) really the setpoint of the AC? Usually we aim for supply temperatures near that to minimise drafts and discomfort.

            Oh i also didn’t mention in the above, but there’s a newer guide, CISBE TM52 that says that temperatures can be allowed to rise somewhat above usual prescribed comfort ranges if air velocity is allowed to increase, and taking into account that people wear less clothes in summer conditions.

          2. wow, 55F – ie 13c for the rest of the world – is a bit cool… I have my office (I work by myself) at 24C (75F) summer or winter, though at summer I do sometimes let it warm a up a bit so I can take my jumper off ie 27C (80F) – but then again I’m in Aus and it gets to 45C (113F) outside and I quite like it.. :-)

          3. Dddave you could in theory push and pull air throughout the ceiling void by using a string of fans in series though you’d have to deal with fire dampers through fire curtains (in the UK).


            Void. ×fan×. Void. ×fan×.
            Ceiling———–. .————–. .——-

            Shunt the air like in an underground car park smoke vent system. I sized a boost fan like this for the British Museum’s 6th floor restaurant. Works okay AFAIK. Was a bit concerned about the pressure further down the system. Maybe should have fitted some Non return dampers… Hmm ????

        3. Thanks for your comment, Gazzat, it’s very useful :)

          Your assumption is that thought was put into the design layout of this particular room, but sadly there was no. This particular office building is part smaller rooms (used for meetings mostly) and one big open office plan. The latter was originally designed to also be a range of small rooms with one AC unit each, but after the AC units were already installed this got changed.

          All of the installed AC units were then hooked up in two strings to two room controllers (next to one of the two doors leading into the room). As a result there is an excess capacity near the windows, and a dearth of capacity elsewhere. Since there’s no way to regulate individual AC units this way, the room controllers will just kick the AC units into high gear a lot of the time until after a long delay the cooler air finally reaches the sensors. Since there are a lot of warm bodies between said air and those sensors, this can take a long time.

          As a result the people at the windows are basically freezing and those near the sensors complaining about high temperatures. The building owner has made it clear that nothing can be changed about this, also saying that the next company to rent the office will likely want to have the many smaller rooms, in which case the AC layout starts making sense again. We have already tried all the non-technological solutions so far, and thus we have arrived at this compromise solution to at least be able to tweak the output of individual AC units, allowing us to (theoretically) make things suck less than they do currently :)

          As for the meeting rooms we have, the problem we have with the AC units there is that they are constantly adjusted by occupants, do not automatically regulate back to a set temperature when the room isn’t occupied, and tend to get really loud during meetings (causing people to change the temperature settings to shut them up). By using ceiling-mounted temperature sensors, we also avoid the issue of having direct sunlight hitting the currently wall-mounted room controllers (put on a dark wall, even), and kicking the AC unit into high gear because it believes the air is 30C when it’s actually 24C. The motion sensor we’ll be adding to our controller should also make the unit behave more politely during meetings.

          1. Luckily I’m often employed to design layouts. It’s odd though – thats the opposite way round to how we do thiings. We usually design for open plan for base build, then the Tennant decides what rooms they want to install, they send a proposal to the landlord, we check it and are sometime employed to redesign the AC to suit the new layout.

            Also, the BCO design for offices says that there should be at minimum a zone adjacent the windows – which the AC system should reflect in terms of control. “perimeter control zones should typically be no more than 6m wide and 4.5m deep”. You could certainly use that strategy when you rewire your units.

            For the meeting rooms, sounds like you need a controller where you can set some lock-outs. Often you’d use controls able to lock down the available temperature range that users can set. Often times the controller would reset itself after X hours or on a schedule. Depending on the controller installed, this could be logic/configurable in the controller or in a Building Management System which is just a computer that controls the buildings systems. The benefit of a BMS is also being able to set schedules for the entire system – setpoints to acheive at certain times of the day. Thus heating/cooling could come on before occupants are in the room and shut down at the end of the day/weekends/holidays. If you were feeling particularly clever you could integrate the meeting room cooling with the meeting room booking system so that it cools to a setpoint in anticipation of occupancy.

  2. My experience with the BME280 is that it reads high unless there’s some airflow over it. self heating I think. I’ve seen reports on forums of the same from other people. something worth keeping in mind.

    1. The BME280 seems to show a litttle self heating but usually its placed near other hot IC’s with no airflow. Place it in an isolated box with a lead or external mount it and it seems dead accurate.

    2. The self-heating thing with the BME280 from what I have been able to determine is due to breakout boards with a built-in Vreg to allow for 5VDC supply voltage. Usually an AMS1117 3.3 linear regulator. These dump the 1.7V as heat, which would explain the higher readings. It’s why I went for pure 3.3V breakout boards (GY-BME280), to avoid this issue. So far there do not appear to be any self-heating issues, reading pretty close to a (calibrated) stand-alone thermometer put next to it.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.