Ian Hattwick

The best laid plans for DIY sensors: laid to waste

Ian Hattwick — Wed, 04 Aug 2010 21:39:40 +0000

Well, the whole mbira is put together now- and it went basically exactly according to plan. Except for one major caveat: the foam that I used for the pressure sensors basically just doesn’t work for this application. On a tabletop, it works great- going from 2M to 2K resistance by pushing down on it. However, in the mbira, the wrapping between the bridge and the tine keeps the foam at between 80-200k. The real problem though, is that foam is very slow to react, and stiff- so if you push down on it it takes a lot of force to get a big change in resistance – so in a voltage divider circuit with a 120k pull-up resistor, there is about a 10% change in voltage. (Which is exacerbated by the lever mechanics of the design.) And when you let go of a tine, the tine rebounds to a much higher value than its original value, and then slowly returns to the middle.

At this point, I am going to stick with this design for a little while and see if the foam ‘breaks in’ and gets less stiff and more responsive. However, I’m not expecting much. I think I will also search for some other conductive fabric and foam and see if I can’t find something more flexible. As a last resort, I could get small FSR’s from sparkfun, but I’d like to stick to this design as a proof of concept.

The mbira looks great, though- and the CdS photocells on the tips work great – I get around 90% range on most of them. The interface with max/msp is also coming along- I’ve got max reading all 30 tine inputs. I’ll talk more about the max/ firmata interface later.

Here are the wires connecting the pressure sensor to the PCB. I soldered small pieces of copper tape to the wires- the copper tape is then stuck to the foam, and the whole assemblage placed in the heatshrink with the tine.

The finished mbira! I haven’t shrunk the heatshrink since the sensors aren’t working so well. I don’t have too much hope that sealing the heatshrink will improve the performance of the foam so I’m holding off on it until it breaks in or I get better conductive foam or fabric.

The photocells are the top 15 graphs, laid out in the same order as the tines. The pressure sensors are on the bottom. You can see how the pressure sensors are a lot less sensitive than the photocells.

Sensor details

Ian Hattwick — Sat, 31 Jul 2010 07:37:00 +0000

I had hoped to have the whole mbira put together today- and I am close- but I ran into a problem I hadn’t solved yet. Let me explain how the sensors on the mbira work.

Each tine has two sensors- a cds photocell on the tip and a pressure sensitive sensor by the bridge. The photocell is fairly straightforward, although the wiring to connect it to the circuitboard is messy, but the pressure sensitive sensor was tricky. I was using a standard pressure sensor from sparkfun, but at $10 a pop 15 sensors was getting pricey, and I had read about an alternative solution using conductive foam (Adrian Freed from cinmat at UC Berkeley also made a kalimba using pressure sensitive fabric). Basically, you take a piece of conductive foam, like the kind IC’s are shipped with, and sandwich it between two metal conductors. The resistance of the foam changes according to how much the foam is squeezed. It’s actually quite sensitive- on a tabletop you can go from 2 megaohm to 2 kohm just by squeezing.

So the mbira has 15 small pieces of foam, one for each tine, sandwiched between conductive copper tape on the tine and a wire leading to the pcb. The metal tape on the tine leads to a strip of metal tape connected to ground, and the wire is connected to a voltage divider.

The problem is finding a way to mount 15 of these systems on a bridge 6 inches wide in a way that will make a good electrical connection, not let there be any crosstalk between tines, and will be dependable. I’d kind of been hoping the pressure of the tine would keep the foam in place, but I had a feeling I needed a better solution. Sure enough when I started putting the tines on, it became a nightmare keeping things aligned. As much as I hated to do it I was forced to use electrical tape to keep each pressure sensor intact.

Electrical tape, by the way, sucks in ways I am sure everyone understands. Also, after spending all of this time making beautiful laminated wooden tines I don’t want ugly electrical tape glaring at me everytime I pick up the mbira.

Thankfully, after putting together the first five tines I came up with the perfect solution- 1/2″ heatshrink. It will still go over the top of the tine but will look a million times better and also be more stable and dependable.

So, still have to finish the tines but at least have a gameplan for how to approach this nagging problem.Here is a picture of the tines with tape. Notice the black and white wires sticking up on top of the aluminum bridge- those are what connect the conductive foam to the circuit board. It looks like wider spacing than it actually is because the photo is so close- it’s actually about 3/8″ from wire to wire.

digital mbira electronics

Ian Hattwick — Fri, 30 Jul 2010 07:29:16 +0000

I finished making the circuit board for my digital mbira today – man building circuit boards is incredibly detail-oriented. I’ve come to dread it a little bit, especially since you always try to get the most electronics in the minimum space, which ends up being so fiddly to work with. Nonetheless, this board is done! Take a look first and then I’ll describe what’s on it.

The board is built around an arduino nano which is the big blue chip on the left. The IC’s at the top are 4051 analog multiplexers, there’s one to the right of the arduino as well, which turn 5 of the arduino’s analog ins into 34. At the bottom of the board is a 10LED unit, controlled by a shift register (which isn’t put into its socket in this photo, and the shift register can only control 8 LEDs, so 2 LEDs are directly wired to the arduino). There also two pushbuttons on the top for thumb control, and two on the bottom that you can’t see for control with your middle finger.

Almost forgot the accelerometer (the red board on the right) which goes into one of the 4051′s.

So a pretty simple board, but still a lot of work to cram everything on! Tomorrow I am hopefully going to finish the phyical construction – I’ve put copper tape on the tines so I just have to wire them up. Here are the tines so far:

mbira~ version 2 – in progress report

Ian Hattwick — Thu, 29 Jul 2010 08:45:50 +0000

While at CCRMA earlier this summer, I prototyped a digital musical instrument (DMI) based on an african mbira- I affectionately named it the mbira~ (pardon the in-joke). Basically, mbira~ was four tines with different kinds of sensors to measure the pressure between the tine and the mbiras bridge. This information is sent to max/msp where it is interpreted and used to trigger a physical model of a flute. There is a video of the demo I gave at CCRMA – I’ll post it when I get a chance. Anyways, here is a picture of mbira~:

I enjoyed the way mbira~ turned out – it was satisfying to play. So far, I have stuck with traditional instruments and avoided DMI’s (I think midi ruined them for me), but as I step more into the world of electro-acoustic music it seems like having a DMI to experiment with would be useful, so when I got home I began planning for a full-scale digital mbira (I haven’t come up with a name for the new instrument yet).

Unfortunately, with all of the cool things I did this past year, I didn’t have time to document the creation of anything. (There are links to documentation of the final products on my website ianhattwick.com though.) So as I build the new mbira (please, help me come up with a better name than digibira!) I am going to document it as thoroughly as I can stomach, and get in the habit so I continue to document my projects in the future.

For now, it is getting kind of late, so I’ll post a few pics of the work so far and later will give more details.

Basic frame (it’s a picture frame!)

With wooden tines:

Tines with cds photocells in the tips:

There’s fifteen of those suckers!

Looking pretty good stained:

And that’s as far as I’ve gotten so far. I’ve had to wait until the physical construction is complete before doing the circuit board, and haven’t programmed the arduino firmware yet. That will have to wait til a later post!