Up/down switches for roller shutter

Does anyone had found a solution to command up/down a roller shutter ?
ideally small enough to fit in the wall box behind the mechanical switches.

Sonoff ? (sonoff basic (1 IP) with a second relay on pin 14 with Tasmota firmware, or is that too complicated), thank for your help …

1 Like

I have thought about using these. They have remote control that perhaps can be hacked. One controller can also be programmed to control multiple windows.

I am trying to stick with Google Home. I don’t know that tasmota directly will support GHome.

I am having difficulty determining what roller shades to use with these. They need hollow metal tubes and everything so far that I found have solid cardboard tubes. I am also not sure what size tube to buy for these.

I have also read that Ikea is coming out with smart shades that are GHome compatible.

thank’s @donkjr, but the roller shutters are already in place with mechanical switches (2 push buttons, not On or Off). I would love to replace those with a remote one (from bed), and still be able to have a mechanic one next to the shutter. (I just find out some chinese products (Xiaomi Aqara) are too big to fit the EU standard wall box, another parameter to take account of).

my shutters are the type with slots in for daylight, if not completely down they let some light to go trough. so I would like to “say/ask” put the shutter X% or full up full down day/night/siesta (based on full operation time/cycle).

as per standard (xbee, etc …) I would prefer to stay just IP/Wifi (like sonoff) an be free to any proprietary softwares or big data (Snips voice control) in case I lose internet (which happen in my country side).

As per home server my research lead me to Jeedom or Hass.io at the moment, still digging.

but I have to say, it’s big bit to swallow, a lot to read and learn … more I dig in more I realise my goals are very basic in terms of devices (few on/off unit, media center, temp control) , but the scenarios (scenes) on server side.

MQQT, NodeRED, home server, Snips, … plus huge amount of other things to read & understand before discarding them. If only I could find a good article to summarize all these techno with +/-.
We are not here yet, but I dream of home server’s graphic UI (OS dedicated to home use make a lot of sense to me in terms of devices/unit, users&group, events , security), where you pick and set your components, link them with arrow (function), then create your scenes yourself with no coding. (dream vs nightmare)

fit in without enclosure, could fit in with enclosure with a bit of persuasion.

EU box is 90x70mm.

I had a similar problem in that I wanted to turn on my bed heater remote so that I could have it toasty when I arrived for bed. I did not want any chance a wifi hack could turn on a heater so I used a 433mhz controller with a fob that I keep next to my TV chair.

I ended up hacking the control panel on the bed heater and inserting a relay(s) where the up down buttons were.
I found a RF remote control and controller with a fob (car remote like) to turn the beds on. You can have four functions.
Perhaps there is some combination of this setup albiet manual-remote not voice that would work for you.

However while getting a link for the above from amazon I found this up down remote motor controller that could be wired directly to your shades motor? I notices quite a few versions of this type of motor controller.

@donkjr Don’t think 433MHz is safer than WLAN. It’s the oposite. Most cheap 433MHz remotes just use a simple ASK modulation without any encryption. They just send one code for on and another for off. Anybody in the receiption area could capture your on/off code and send it out when they wish.

I agree but I am not much worried about my blinds or bed heater being hacked :)!

OK I found a solution witch also allow me to keep the push buttons, the sonoff 4ch describe there :

How do you plan to control the motors with this?
Configure relays in an polarity reversing format?

simple connection as per following design.
the 4ch pro is compatible with tasmota, Mqtt will be the way.
my problem is the size of 4ch pro it don’t fit the wall box behind the push button.
if only it was a 2CH pro …that would have been fantastic.

Look like the Shelly 2.5 is a far better option as the built in server offer the possibility of setting a time for motor up/down. Sonoff 4ch don’t and have a max value that could be too short for your shutter.

It is far smaller fit in wall box (Sonoff need to be in the shutter roller box and waste 2ch).

Hi, I am planning to add remote control to an existing installation with 7 roller shutters. I thought of using a Loratap module. They come either in the newer version with both 433MHz and wifi capabilities (about 21€) or the previous version, wifi only, at about 12€.
On the other hand, there is the Shelly 2.5 that is slightly more expensive than the latest Loratap.
I am concerned with the possibility to control the shutters from my system, sending wifi frames, rather than using the provided Android application. Thus I would like to know if anyone here has an experience with any of these modules, and if they know of the availability of the technical information to implement this custom control.

Hi @Phono,
I just received my Shelly, to early to say, I can’t make my mind between the two following road :
Keep Shelly firmware or flash them with tasmota. With Shelly that make install into home assistant a bit more difficult as you need to go for hacs first.
Tasmota would be more easy (like any sonof), but you may have to recreate some features like % opening.
I haven’t decide yet, still digging…

Interestingly, I just started a project to automate cheap roller shades.

My plan is to modify the cheap shades from big box stores to interface with a gear motor and brackets much like this inspiring video shows: https://youtu.be/2bWeKEtdMFA

My home is using Google Assistant so I wanted to stick with the app I use, eWelink.

My plan is to use a Sonoff compatible relay in the “inching” mode, i.e. it outputs a pulse.
That relay will connect to a Nano that is connected to a motor controller then the gear motor which is attached to a modified shade.

Attached are some pictures including my circuit scribbling and the prototype electronics. Also, you can see the start of my mod to the shades.

I am going to start by simply turning the shade motor on and off for a period of time in the direction opposite of its position which is stored in non-vol memory. Not closed loop but in my application if the shade does not stop in exactly the same position that will be ok.

If that does not work I am going to try optical end-stops that read punched holes in the shades.

The shades can be controlled from Google Assistant, a 433mhz remote or local at the controller.

This design probably has one more u-processor than it needs but I did not want to write my own interface & app to Google Assistant. The sonoff compatible device I am using is 2 channel and I only need one. The amazon add suggested you could control a shade motor with this two-channel device. You can command a motor forward or reverse but there is no STOP state …huh!

All these parts are pretty cheap and on amazon. The one controller is expected to drive 4 shade motors. Depending on the torque required I may have to add 1 more driver. More info is available if it’s useful.


I am afraid you will not do without the end of stroke switches. Counting time is ok for crude positioning, but you will need an accurate control at least at the top and maybe also at the bottom of the motion.
You may omit the bottom sensor, at the expense of unaccurate lower end of stroke using time counting.
Optical sensors might not be the best choice near a window because of ambient light. If you do not want to use a microswitch, a good choice would be a hall effect sensor and a small magnet attached to the bottom of the curtain. An example of such a sensor is found here.
A good placement of the magnet is necessary for a reliable operation, but it is not difficult to achieve.
Yet another solution would be a reed switch and a magnet. Both are very cheap.


Thanks, all good things to consider:

The optical sensors are IR and up to now other than extremely strong and directed light I have not had trouble with ambient. That said the hall effect switch is likely a better idea.

My shades are installed under a valance and can afford to stop with an accuracy of 1-3 inches at the top and on the down side they only have to open to a position below the horizon of the couch so that error can be large.
Both up and down positions have a lot of tolerance and will still be out of sight.

Due to the high gear ratio motor I am expecting it to stop pretty quickly and the gear train to provide a lot of holding friction. In addition the end mounts are being replaced with bearing to help with smooth and more repeatable operation.

As I said in the first post, open loop is not my first choice but I am going to try it anyway since the benefit of not having lots of sensor wiring was the only goal given from the WIFE.
Worse case I add sensors when she is not looking!

I will report results.

1 Like

Update on the shade project:

Control schema:
Time-based control worked … kinda… not!
Things I found out:

Although I still think it is possible, it is not practical, two many variables made this complicated to control and multiple pots would be needed to make it useable.
My statistical model told me I could be off as much as 11", although I doubt this would ever happen.
The case where the power goes off in the middle of the shades movement dictated a “Home” sensor. See details on using EEPROM**

Practical Control variables:

  • Roll diameter changes with roll position… doh!
  • The motor speed varies from spec. 10RPM motor turns on average about 9.5 rpm
  • One revolution of the motor is approx 4.5" so that is the potential error in positioning if the movement starts just after the shaft interrupt.
  • **Saving the roll position on power cycles. I planned to store the position count in EEPROM. To ensure the position is saved accurately you have to save it in the ISR. This means that you will exceed the 100,000 write cycles in 8 years (estimate) and have to replace the electronics. It gets worse if I increase the shaft magnets.
    This was the deciding factor causing me to add a home sensor. Now I have the ability to reset the position each time the shade goes up making 1/2 of the control problem easy.

What’s changed from the original design concept:
I have added 2 sensors (yah we all knew it would come to this):

  • Shaft sensor that counts revolutions
    – A small magnet inserted in the shade tube adapter and a hall effect sensor.
  • A home sensor that detects the up (home) position of the shade.
    – A magnet on the backside of the shade positioned so that the proper amount of shade is left hanging when in the up position. *** Thanks @Phono
  • Added a 2 channel Google Assistant wifi control that also supports a 433mhz FOB.
    – One channel commands up and the other down.
    – The wifi connected app will not allow me to change the name of the channel so saying " Hey Google, Shade Socket 1 ON" for going up … is kinda cheezy.
    – The FOB next to the TV chair may be the main means of controlling the shades!

Meeting the boss’s main objection visible wiring: the new sensors have not exposed the wiring much as its all still hidden up under the Valance.

Remaining challenges:

  • Final bench testing, install the proto and run for 1 month.
  • The mounting/sensor bracket has gotten complicated and there is not much room above and to the left-right of each window to mount this. Will probably need a carefully designed integrated bracket, sounds like multiple 3D printed iterations :(.
  • Add more magnets to the shade tube adapter to increase the resolution and decrease max error.
    – Magnet placement accuracy might be a challenge, use my lathes indexing.
  • An OLED panel… some bling!
  • Controlling multiple shades in sync will require a master-slave kind of setup. The original design accounts for this. Build modular motor drivers that mount on bracket?
  • The wife wants the controller to be on the other end … reverses everything … sigh!
  • $ is higher than budget … shocking! Still much lower than store-bought remote control!