As some of you may already know, the Bela team has been working on a new family of sensors for creating beautiful touch interaction.

We just launched our Trill sensors on Kickstarter today.

Trill is a family of touch sensors that make it easy to develop high-resolution, intuitive touch interfaces. They’re designed for Bela but are also compatible with any system that uses I2C, such as Arduino. There are three types of Trill:

  • Trill BAR, a multi-touch slider that can detect up to 5 touchpoints;
  • Trill SQUARE, an x/y sensor with two axes of sensing that can also detect touchpoint size;
  • Trill CRAFT, a 30-channel breakout board so you can make your own touch interfaces out of any conductive material

Trill sensors use I2C, meaning they are compatible not only with Bela but with a variety of popular maker boards. Trill comes with everything you need to get started, including a library for both Bela and Arduino, and examples for visualising and making sound with the sensor signals. Trill is a versatile sensor that can be used in a huge range of applications and products. Trill is open hardware, and the firmware is under a CC-BY-NC-SA license.

Feel free to ask anything else you want to know about the Trill sensors or the Kickstarter campaign!.

congratulations, i think i might have been backer #1 πŸ™‚

can i get a couple of points of clarification....

at superbooth, i think you mentioned technically the BAR could also do 3 axis (slide, Y, pressure) as its all the same tech (e.g. as SQUARE) ... is this still the case?
also, is the SQUARE multi touch? one of the video seems to show it is, but your description implies not... again seems like it should be given its the same tech
i don't mind going 'lower level' to get these features , if its been simplified for the UI which i could understand
but id really like a multi touch x/y/z capability

EDIT: ok, i can see the FAQ says square can do up to 3 touches, but cannot differentiate x/y for each.
... though, I'm a bit confused what that means... what does it report for multi touches?

also.... are the sensors active right to the edges, ie. if you mount multiple next to each other will there be any gaps in sensing?

hmm, wondering if i might explore the CRAFT a bit more for larger surfaces....
id be interested in some tech examples of how craft is setup and works.

    Ordered the 5 sensors, I guess you get in touch to ask us what ones we want?

      thetechnobear

      The BAR cannot do 3 axis I am afraid, as the sensors are arranged differently to the SQUARE. However, you can get up to 5 different touches on the same axis.

      The SQUARE on the other hand cannot do multi-touch in the same way the BAR does. You could detect 3 different touches but would not be able to differentiate between their XY positions.

      All this is due to how the pads are arranged on the sensors themselves.

      thetechnobear also.... are the sensors active right to the edges, ie. if you mount multiple next to each other will there be any gaps in sensing?

      Yes, the sensors are active right to the edges. We have done several experiments with sensors connected edge to edge and there weren't any gaps that we could tell.

      AndyCap Ordered the 5 sensors, I guess you get in touch to ask us what ones we want?

      Yes, we will send a survey once the Kickstarter has finished.

      wow, you are approaching you goal very fast πŸ™‚

      with reference to CRAFT,
      does anyone have a link on how to build say a slider or x/y using a cap sensor chip like craft?
      also possibly as 'inspiration' how the the TRILL BAR/SQUARE are laid out in terms of channels to get their functionality?

      here is my assumption on how it might work... is this vaguely correct?
      so I believe a strip of conductive material with a sensor, will tell you how far along that strip you have touched. so this gives you 'x' with one channel on the 'craft' (or similar) - this is how a ribbon pot works.

      to get X/Y you form a grid of channels. and then you interpolate from this.
      if you make the grid small enough, you can then 'infer' the touch size (since you assume close points are not actually individual points but the same touch)

      so this gives us x/y/z for a single touch

      but this will work for actually work for multi touch too. because you can different touches using the grid with certain limitations i.e. as long as you can differentiate each touch on one axis.

      (iirc there is a 3 touch in a 'box' configuration limitation... but my memory is a bit sketchy in this area)

      so with crafts 30 channel , the limitation is forming that grid, the closer the spacing , the higher the resolution (particularly important for touch size) ... but of course the smaller the surface area.

      BUT you can chain the crafts over I2C, to get more channels ...
      I guess then you become mostly limited by the data rate of I2C? (give i2c allows for 127 slaves)
      (hmm, and I guess there might be a limited i2c address count on CRAFT)

      is this roughly correct?

        Separately can I make a suggestion πŸ™‚

        TRILL CRAFT allows you to turn anything that conducts electricity into a sensor: you can make sliders or buttons out of copper tape, conductive paint, e-textiles and many other materials such as water, metals and even fruit.

        Ive seen this kind of statement before on other projects, about easily getting 'required' materials to build things (notably the Koma Field kit)..
        usually what happens is backers when they get the kit get pretty frustrated trying to hunt down the stuff... partly because they dont know the suppliers, or what is suitable because of 'lack of experience' - and you see loads of posts saying why don't you sell this stuff?

        my suggestion is:
        Bela could stock some things like copper tape, conductive material, conductive paint - that could be added onto the kickstarter order at the end. (or even just a 'starter kit')

        (Roli recently used this on their LUMI kickstarter when they allowed you to add all sorts of accessories to your pledge - i guess its part of the 'backer kit' software?)

        the advantage of this, is when backers CRAFT turns up, they will immediately have what they need to build a project rather than having to go hunting around for other materials.

          thetechnobear They way sensing with the Trill BAR and SQUARE sensors works is actually different to that approach. In this case, one axis is given by interpolated readings between adjacent sensors.

          You can read a little bit more about this sensing technique here: https://www.nime.org/proceedings/2012/nime2012_195.pdf

          However, we plan on having a tutorial for building different sensors using the Trill CRAFT and conductive material.

          thetechnobear I guess then you become mostly limited by the data rate of I2C? (give i2c allows for 127 slaves)
          (hmm, and I guess there might be a limited i2c address count on CRAFT)

          In its current configuration, you can program up to 8 different addresses per sensor (on hardware, via some external pads) and each family of sensors (CRAFT, BAR & SQUARE) will have a different address range, meaning you will be able to use 8 of each on the same I2C bus.

            thetechnobear Bela could stock some things like copper tape, conductive material, conductive paint - that could be added onto the kickstarter order at the end. (or even just a 'starter kit')

            That's a great suggestion! Some of these materials would be a hassle to stock & ship for us (conductive paint, for example), but some others we may be able to provide with the kits.

            Stretch goal, maybe? πŸ™‚

            adanlbenito

            ah, thanks for that paper ... all clear now...

            so each channel is capacitance on a gives surface, then you interpolate that to get position - neat.
            the X/Y is particularly a neat implementation, having the longer connected strips allows you to reduce the number sensors (over having a simple grid of sensors) - very cool.

            I guess, creating that grid by hand is pretty tricky, as you need to find a way to get a wire to each block to take back to the sensor.i could imagine some kind of (wood) back plane with holes (for wires)... but creating a good contact with the copper layer is tricky ... hmmm, have to think about that.

            yeah, I think copper tape is probably the thing id most like.... also the plastic template tool you show for cutting could be useful too.

            the thing Ive noticed with DIY, is whilst I can track down most of this stuff - you end up buying from multiple online suppliers, which not only takes quite a while to track down - but also means shipping starts to really add up. (as they all tend to have a minimum shipping or order size)

            also generally, I think backers really like 'complete' kits , a bit like lego or eurorack kits, where you order it, and you know you can just start assembling... then of course, with that experience, you can then start venturing off and modding/hacking πŸ™‚

            How sensitive are the sensors?

            I'm just wondering if they could be used with pots/encoders to tell when someone is touching them?

            You can actually sense a tiny bit of capacitance through a small plastic knob, so in principle, assuming the pot's case is NOT grounded, then you could give this a try. I tested one just now, but it will probably depend heavily on the actual layout.

            Hi Guys,

            I got the email about the Hex and Ring.

            How many sensors are on the hex? Do you have some design docs?

              AndyCap The Trill Hex will behave in the same way the Square does and will use the same number of sensors (30 in total).

              So the hex just a different shaped Square?

              My maths isn't great but I could understand 19 sensors if the edges were not straight, or 19 full + 12 partial sensors if the edges were straight.

              I'm trying to work out how you could combine 8 of these to make an isomorphic controller?

                AndyCap The Hex will use the same sensor tech as the Square, but most likely with a finer spacing of the pads. Like the Square it will have two sliders of 15 elements each, at 90 degree angles to one another (but truncated to the overall outline of a hexagon). Since we expect the Hex will be smaller overall than the Square, the resolution can be somewhat finer.

                You could cut down a Square to get a Hex, but this variation is more convenient in that you get consistent, repeatable results (which not everyone will have the tools to do), and as above, slightly finer resolution.

                When we say you can use 8 sensors at a time, this is the number of unique I2C addresses that are available on each type of sensor. There's no limitation on using more sensors, if either (a) you use multiple I2C busses, or (b) you put custom firmware on the device with a different range of I2C addresses. Ultimately the main limitations to how many devices you can put on a single bus are the rise time of the bus and its total bandwidth. I would expect 20-30 devices should run relatively comfortably provided they had unique addresses.

                Hi @andrew

                Thanks for the info, I'm still a little confused though!

                The email mentions isomorphic controllers which I took to mean something like:

                Is each trill hex multi touch so would contain multiple smaller hex keys as in the image?

                Or is each trill Hex single touch like an individual hex key in the image?

                  with CRAFT, i assume it reports all 30 sensors data (continuously)

                  for SQUARE/BAR/HEX etc is this still the same... i.e. they report all 30 sensors data, and then the 'interpolation' to touch is done on Bela (or whatever) , or is this done on the hardware? if the later can you get it instead to report the underlying sensor data?

                  AndyCap i have one of those :-) and i added "aftertouch" to it (with an axoloti and some fsrs). and as you i was wondering what exactly they meant with that email...