Interactive lighting elements for the mobility impaired. Designing a power supply for architectural scale interactive EL installations

Similar projects worth following
A modular and re-configurable interactive lighting system that is easy to install and can be used to make almost any space safer at night. Powered by an open source (arduino compatible) circuit board!

We document the development of a ~5 watt, quiet, wearable EL power supply (that outputs ~100v at ~1000Hz under load).

Integrating Electro-Luminscent (EL) materials into interior lighting makes it possible to design a friendlier lighting system for the mobility impaired.  

One useful feature of EL materials is that it is possible to detect touch anywhere along their surface (using this arduino compatible circuit board).

We propose using this feature in combination with motion sensors to create environments where motion activates EL strips that aid in guiding an individual to the kitchen (for instance) in the middle of the night.  These strips can be configured to initially turn on in a low brightness mode (to minimize the disturbance from accidental triggering).

If an individual in this environment needs more light in order to navigate, they can get it by simply touching the "rails" anywhere to increase their brightness, or to activate other (possibly LED based) lighting.  This improves safety by making lighting environments that more immediately and seamlessly respond to the needs of their (mobility impaired) humans.

For our hack, we document the development process of an external power supply compatible with the touch sensing functionality of the Whoa Board. 

Here is a statement of our main technical challenge!

We then:

Surveyed some existing design resources

Proposed a "simple" solution

Thought about component selection

Gave an example of an experiment that need not be repeated

Documented a working example!

In our earlier logs, we document the context and design process of this project:

EL background and Assistive Tech brainstorm

Exploring applications and interactions

Project History/Context

User Research

A Physical Therapy application

User Story!

Survey of existing solutions and desired technical features

  • 2 × Whoa Board A circuit board capable of performing capacitative sensing on EL materials.
  • 4 × Electro-Luminescent Strips Strips of "paper" that can glow in response to touch
  • 2 × nrf24L01 radio To wirelessly pass messages between boards.
  • 2 × SM02B-SRSS-TB(LF)(SN) Connectors and Accessories / Board-to-Board and Card Edge Connectors
  • 1 × MIC4833YML-TR Interface and IO ICs / Display Interface

View all 7 components

  • High voltage DC?! Seems safe...

    fool09/04/2017 at 00:19 0 comments

    The next step of this project was to move beyond using an existing "self-oscillating" supply and instead try to design one of our own (that didn't spike without a load, and that didn't make an annoying 1-2KHz buzzing noise).  

    We decided that it would be nice if this supply ran off of 5V, so the first step was attempting to design a DC-DC boost converter that transformed 5V into ~150.

    To do this we first tried using the LT8304 for the control circuit, and then tried the AZ34063UMTR.

    Both of these had problems with efficiency though, and ultimately, we decided to return to the problem of designing a better DC-DC converter at a later time.  For testing, we took a standard wall wart, and rewound the transformer so that it would output 170V DC. 

    This is not a particularly safe thing to do (and would almost certainly violate some safety guideline regardless of where you attempted to distribute such a thing), but for testing it's been really handy :).

  • A concrete simple instance!

    fool09/02/2017 at 20:18 0 comments

    Here is a better, and brighter way to power things with an external supply:

    And here it is in action, powering a large panel and a surface painted with electro-luminescent paint (developed by a company called LumiLor).


    fool09/02/2017 at 18:58 0 comments

    Here we're going  to document our first experiment with connecting an external supply.  

    It worked, but it is a BAD idea (for a few reasons that we will detail here). Please learn from our foolish explorations - don't electrocute anyone!

    Ok, so what are we looking at here.  We are looking at a regular american power plug, that fits into a supply that provides ~120V (RMS) at 60Hz.  We attached a resistor (and didn't include a TVS diode because there is one on the board and because the voltage in the wall shouldn't exceed 170V as long as there isn't a spike).  We flip the supply switches on the board, and connect this as an external supply. 

    Okaaaay, so why is this a BAD idea?

    The main issue is that because we are driving the panel directly from the wall, if you touch the panel then there is a potential to create a ground loop through your body, because there is no galvanic isolation between the supply and your feet. 


    This is particularly worrisome because the board is exposed, and so it's easy to touch an exposed lead, and because the panel in question isn't sealed/laminated, and so can shock you when you touch the edges of it (as has happened to us, particularly when touched with sweaty/moist hands).  

    If you want to draw power from the wall, it is much better practice to pass the signal through an isolating transformer.

    That said, EL materials really glow much more effectively when driven at a higher frequency than 60Hz, and so it is better to use a more specialized supply, as we will describe in later posts! 


    There are, however, some redeeming things about this experiment, as we illustrate in the image below:

  • Choosing a TVS diode!

    fool09/02/2017 at 18:54 0 comments

    Fill fill out more information here later!

  • A simple solution!

    fool08/31/2017 at 03:48 0 comments

    As mentioned previously, the limitation of the whoa board is that we are not allowed to put more than 3A and 150V (300 peak to peak) through it. 

    Accordingly, a simple solution which can make it compatible with any power supply is simple to limit the current and voltage going into the board.  


    This means that they have to be chosen appropriately to withstand the situation where the whoa board doesn't provide a load (as happens for part of the time during every touch sensing cycle).  

    This is important because the self oscillating power supply designs (like the ones in the videos in the previous post, and generally most commercially available EL supplies) increase in frequency and voltage as the load decreases.  In fact, some supplies cannot be turned on without load because they will fry themselves.

    Annnyway, in the interest of documentation, let's work out the math of how to appropriately choose these resistors and diodes here.  

    Let's suppose at it's max, the external supply outputs 400V peak to peak, and to be conservative, let's limit the max current to 2A. 

    In order to limit this current, we need 100 Oms of resistance.  The path that the current flows through is already ~50 Om, so we need another ~50 externally.  For the sake of symmetry (and to distribute the power dissipation) it's probably better to instead include two ~25 Om resistors, as drawn in the above schematic.  

    Because EL loads are capacitive, they have a much higher peak current than average current, which makes it difficult to compute a power rating for the resistors you choose.  For most applications, a 1W resistor will probably be fine, but if you aren't sure, there's no beating experiment.  If it starts getting uncomfortably hot, increase the power rating :).  

    In the next log, we'll talk about power dissipation in the above TVS diode! 

  • EL Power Supply Resources

    fool08/29/2017 at 16:16 0 comments

    Online resources for designing a high frequency, high voltage power supply that is happy driving a capacitative load are not particularly easy to find. 

    The closest project to ours which we could find was this one:,  (and for those that haven't seen it, the rest of the Kobakant site is also worth exploring!)

    Jeri Elsworth also made a few videos about EL materials (above is her exploration of a self-oscillating EL supply). 

    And here is a video that tears down another (self-oscillating) commodity EL supply:

  • A technical challenge!

    fool08/27/2017 at 04:30 0 comments

    So far we've defined an assistive technology product for integrating interactive lighting into spaces for the mobility impaired. Now we can start hacking to figure out what needs to be done in order to get this implemented!

    The main challenge here is that the on-board power supply on the Whoa Board is based on the MIC4833.

    The way that it works is by first generating a 110 volt DC signal (from 5v, using a high frequency boost circuit), and then generating an alternating current out of that by passing the DC signal through two alternating half bridges (which we have set to oscillate around 700Hz).

    We chose because it was quiet (ours is one of the only EL drivers that doesn't make a 1-2KHz buzzing noise), but is limited in it's power output, and cannot power EL elements on an architectural scale.  Fortunately, we manufactured the whoa board with the ability to accept an external power supply!  

    The way it works is as follows:  At the core of the Whoa Board sits a fast high voltage switch (that we control using a bit-banged serial protocol). In order for the switch to switch, it needs a "min" and "max" voltage reference as an input to it's level shifters.  

    We obtain this reference for an external supply by disconnecting the onboard supply (to protect it) and then by passing it through the coupling circuit at the top of this post.

    Note: If you forget to flip the dip switches and connect and external supply, you will probably fry the onboard EL drive circuit (and maybe something else as well).  

    The ground marked optional is something that we've found helps protect the board, but is not currently integrated.  We are happy to provide instructions on how to add this for anyone interested.  

    Finally, note that the touch sensing doesn't work on "first edition" boards with the external supply without a bit of rework.  Again, if you have a board and want more insight, happy to answer questions. 

    The onboard switch also has a few additional constraints which end up presenting challenges.  First off, it is rated for a peak current of 3A, which is worth bearing in mind because EL loads are capacitative.  Secondly, it cannot safely bear a peak reference voltage of more than 150V.  

    Now that we have this circuit, we are ready to define our hack!

    Namely, we set out to create a power supply capable of driving a capacitative load at ~100V and ~1KHz, that can safely output ~5 watts, and is also compatible with the above constraints!

  • Y Du Dis?

    fool08/26/2017 at 21:38 0 comments

    Existing Solutions

    There are a number of existing solutions for adding night time lighting to living environments.  There are of course light switches and lamps on your bedside table. There are the ever present night lights.  There is the infamous clapper (which is sure to startle your partner).  More recently there has been an influx of motion sensitive LED lights, as well as time sensitive lights that only activate at night.  

    There are also a number of app based wireless lighting controllers that are emerging.  

    Why Moon-Lites?

    We have spent the past two years developing a technology for prototyping applications for making electro-luminscent materials touch activated.

    This board makes it possible for MOON-LITEs to offer several novel features over what is otherwise currently available.  

    • Touch them anywhere to turn up the brightness

    The Whoa Board at the core of each MOON-LITE is the new technology which makes this possible!

    • A low brightness motion activated mode

    MOON-LITEs are wirelessly networked and are configured to turn on in a very low brightness state when they detect motion.  This makes them easy to find, while making it less likely that they'll wake someone up if they roll over in bed, or if their cat comes into the room in the middle of the night. 

    • They can be turned off.

    The Light-Lite integrates into any light switch cover, making it easy to tell your MOON-LITEs that you are done using them and you are ready for them to turn off, something not currently possible with more traditional motion activated systems.

    • Easy to integrate.  

    EL materials come in paint, panels (like sheet of paper or strips of tape), or wires.  You don't need to call a contractor to install a complicated lighting system only to find that it doesn't work for you.  Just stick some tape on your wall.  If it doesn't make sense, move it around.  There is a low barrier for experimentation to help people arrive at the best solution for them.  Stick some tape on the edge of your furniture to make it easier to avoid at night.  

    • Seamless operation.

    MOON-LITEs are simply evolutions of existing household objects.  There is no new interface to learn, you don't need to find a phone.  Simply continue moving around your space as you did before, but with more confidence!

  • A user story!

    fool08/26/2017 at 19:57 3 comments

    After doing these interviews, we came up with an archetypal human that we'd like to help with the technology we have been developing.  In the following post, we detail a scenario where we imagine interactive lighting could help a mobility impaired person live a fuller life.

    For Rose, we created three modules which work in tandem to make her life safer in seamless way, by augmenting her existing interactions.

    Consider how these three MOON-LITEs interact in the following scenario!

    Rose wakes up hungry in the middle of the night.

    The MOON-LITEs detect her motion and enter a low brightness mode, highlighting obstacles in the room, and making it easy for her to get a bit of extra guidance if she needs it.

    When she grabs on to any MOON-LITE, all other lites (as well as other optional LED lighting) enter a high brightness mode to give Rose a bit of extra assistance as she moves around.

    On her way back to bed, she can tap on the outside of the light-switch to deactivate the MOON-LITEs, something not usually possible with a motion activated system or turn on the room lights if she needs to find a book on her shelf.  

    If she forgets to tap the Light-Lite, the MOON-LITEs will all turn off automatically if they don't detect motion for a few minutes.

  • A physical trail marker!

    fool08/15/2017 at 00:09 0 comments

    One suggestion that received lots of enthusiasm was working with physical therapists to create a rehabilitation game.  

    We designed a very crude visualization, however there is lots of room to flush out and improve on our idea! 

    The sense that we got though is that it would be helpful to create an interactive "mat" that physical therapists could easily program to create exercises for people recovering from strokes!

View all 15 project logs

Enjoy this project?



fool wrote 10/02/2017 at 03:50 point

Hi @Wassim, getting to the point of popping this off the stack for a moment.  So what I think I'm interested in is experimenting with self-registration.  I.e. install several wireless elements, and they all discover each other, and then all attempt to mirror each other's state (i.e. no dedicated master/slave). 

The use case I'm imagining is a "synchronized light switch" where you touch one element, and it wirelessly triggers all the other elements in the house, in a "natural" way while minimizing the amount of state you need to maintain in configuration.  Maybe you tell the devices a "range" of ports, so that it's possible to isolate devices in one room from devices in the next.

I am also curious to experiment with bluetooth connectivity.  According to this video, it does actually seem that it is possible to both send and receive bluetooth instructions using these radios: 

Any clue about the voodoo?

Also here is the wireless course I'm following:

  Are you sure? yes | no

Wassim wrote 10/02/2017 at 14:00 point

Hi @fool ,

What you are trying to acheive is a good challenge, because unlike the MQTT, you don't want a server. You will have to dig into network topologies to set an appropriate one for you, so if the nodes keep rounting tables or not. Service discovery is also common in other fields but at a higher layer, once the network is abstracted. Keep in mind that mirroring states might involve an update rate with an update latency.

If you want to experiement with bluetooth, and if it is more than theoretical inspiration, then do not waste your time with the nRF24, a good idea is always to start with a reliable devkit with working examples, now they come cheap with debuggers included, I do have the nRF dongle as example.

Good luck with your activities, I'm currently working with the nRF51 series, so kind of close.

  Are you sure? yes | no

Wassim wrote 09/10/2017 at 15:13 point

Hi @fool, this project looks great, if I get it right you'd like to enhance it with full connections capabilities (user, mesh, IoT, device to device...). Then you need a gateway and probably some bridges. Using the nRF24 you can have a look at my Flood mesh protocol it is working, but have limitations depending on your use case. I developped it as ultra low weight protocol with 8KB STM8 not even half full. Then I moved to STM32-M3 with nRF24 still but dropped the low power constraints, My latest move is finally towards the nRF51822 which was my favourite choice from the beginning but the cost became reasonable only now same for reference designs. With that you have the M0 embedded and optionallly your custom RF protocol (nRF24L01 compatible) or stick with standards such as BLE. Reasons why I dropped Bluetooth in my project is complextiy, power hungry and no mesh support. The mesh just got adopted in July 2017. Advantage of Bluetooth is that you have direct smartphone connection and drops the need of additional gateways.

  Are you sure? yes | no

fool wrote 09/13/2017 at 15:50 point

Hi @Wassim , this looks interesting.  I'm taking a wireless networking class this semester, so will be implementing something, this looks like a good starting point!  Do you have a reference implementation?  The whoa board uses a atmega32u4, and there is potentially some trickiness relating to timing issues. 

Have you seen that it is possible to use these radios to broadcast bluetooth (for a hack-y phone connection?

The nRF51822 looks like something that you'd use in place of a micro?  Probably not a good fit for our project.  In the latest revision, I believe we're using every gpio pin on the 32u4, as well as some weirder features like  If we do a complete redesign, we'd switch over to something that can run circuit python, but until that happens probably sticking to this.

  Are you sure? yes | no

Wassim wrote 09/13/2017 at 19:21 point

Hi @fool , yes, reference implementation of the simple mesh is on github free for any type of use, partial copy there you find the sensor rf node, the dongle with receiver and repeater code, the web 3djs gui and some other stuff.

As you're having this as a serious course, then a reference for you is the Nordic document  nWP-005 Introduction to wireless networks. Simple and efficient.

The nRF51 is not worth as second chip as it is supposed to be used in very agressive cost reduction with single SoC for both RF and CPU. First generations started with 8 bit 8051 embedded, now Nordic alligned like everyone else using ARM 32 bits, here an M0 for the 51 series.

Of course your SoC selection depend on a lot of parameters that makes your selection the best for your usecase.

Implementing your own protocol is the best way to learn, Bluetooth on with nrf24 is only for broadcast with limitations.

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates