Making the Linnstrument

very nice and concise video by Roger, covering how ee created the Linnstrument.
does from prototype to release, showing prototypes and how he evolved it, problems he faced and how he overcame them, including a complete with teardown of the (released) linnstrument.

as well as being clear, what I like about this video is… it really shows how anyone can prototype ideas with freely available parts etc… inspirational.
and yet, doesn’t pull its punches, on how its very different having a prototype/idea and something thats actually playable and can be released :slight_smile:

9 Likes

Well he glosses over several points in order to keep the video length short and accessible to non-technical viewers. You can read more about the LinnStrument’s design in the patent that I linked on the wiki.

That’s what make it interesting…
I don’t think many watching this want to build a linnstrument - it’s more interesting to hear about the approach he took, and I like his enthusiasm about giving it a go.

At the end of the day if want to build something like this yourself , it’s about rolling up your sleeves and giving it a go.
But also recognize it takes quite a few different skills to build something like this… it’s mechanical, engineering, electrical and software…
and Roger worked with others to achieve the end result.

Anyway, overall I think many of us just enjoy the story behind his journey :slight_smile:

2 Likes

Roger also chose to keep the components easily replaceable too. There is not much complexity with the folded aluminum chassis, silicone playing surface, steel top panel, and wooden sides. Some of them may be expensive to manufacture, but they are not custom-fabricated parts that takes years of reverse-engineering to painstakingly reproduce.

1 Like

very insightful and inspiring!

1 Like

Finally arrived here, too.

After the Loom kickstarter died I thought I’d have a look in this direction, see if there is a way to design a DIY-friendly, MPE-capable multitouch PCB for a reasonable cost.

Off to “how the sensor works”!

2 Likes

Not directly Linnstrument related, but the maker of the Soundplane has written an article about a 8x8 DIY version with BOM, Max patches etc.: https://madronalabs.com/DIY

1 Like

That’s interesting, who’d have thought it’s possible to throw audio with different frequencies at it like some sort of sonar.

I am starting to believe that a well-made resistive sensor strip can outperform a capacitive touch panel type IC as far as response and expressivity are concerned?

1 Like

Not directly Linnstrument related, but I came across a PDF with inside photos of the Joue Play.
https://device.report/manual/10378377
(Find “Internal Photos”)

This looks like what I originally had in mind: a PCB designed to have capacitive fields which are picked up by a PsoC touch screen sensor of some sort (actually two in the Play), then translated to useful data by a MCU.

Not sure what sorcery created this site, if you search for Linnstrument you can actually find some brochures and various other documents that mention it:
https://device.report/linnstrument/

2 Likes

The difference between a force-sensitive resistor and capacitance is that the FSR actually has a usable variable signal based on force, so you get pressure-sensitivity. I’ve never seen a capacitive sensor that gives you that. I’ve read that the ipad sensor used to provide radius of the contact, which some software used as a measure of pressure, but I’ve read that it didn’t work out to be accurate enough to be useful (for either drawing or music).

(ed: with replies, now I’ve seen a few! love this forum).

2 Likes

That is still the case, e.g. Musix Pro or GeoShred support this. It’s nowhere near Linnstrument/Morph etc. level, but usable (Musix uses this for velocity mainly. Not sure whether they just look for contact radius or also e.g. for vibrations(?)).

2 Likes

So in short, there still ain’t nothing out there quite like the Linnstrument!

So in a musical instrument context, is there no way (that we know of) around a FSR or FSR in combination with capacitance?

Wondering how other recent pad-based MPE devices (Seaboard, Sensel Morph, Joue Play, defunct Loom, Embodme Erae) do their sensing?

1 Like

Sensel Morph and ERAE Touch use force-sensing multi-touch trackpads, which make tradeoffs regarding sensor density/sensitivity against latency. Roger has an annual workshop at Stanford University specifically targeted towards creating musical interfaces into a commercial product:

This page explains the course contents along with a brief overview of each sensor’s benefits and drawbacks.

2 Likes

Re: capacitance and pressure, isn’t the idea to put squishy keys on it and measure distance…? I thought that was how the Exquis works. They did a great job of extracting velocity and pressure from the keys in this MPE device.

If I’m not mistaken the PCB shows a two-zone interlaced cap touch design which also measures vertical movement. (The centre component is an RGB LED.)

You can see the same principle more clearly in the simpler design of the Snyderphonics Manta, which guesstimates pressure/velocity from coverage IIRC.

image

Artiphon made a good job of programming the firmware for the Orba’s cap touch response.

5 Likes

I went down a bit of a Loom rabbit-hole after Ksoloti’s post. The patent is here.

https://worldwide.espacenet.com/patent/search/family/088315878/publication/WO2024088720A1?q=aodyo

As mentioned in the Loom thread, it’s apparently a combination of capacitive touch with piezeo. It has some interesting ideas, such as using an array of piezo sensors, and associating these with touch positions. There’s a certain amount of dicussion of proximity sensing; making calculations about the velocity of an impending strike, and even its angle. The basic idea is to combine the technologies to get the best of both worlds; position and impact.

The cap touch part is apparently based on a standard diamond-type array of the kind found in trackpads.

I’ve realised the value of cap touch controllers in interpreting the data to make it more digestible for the microcontroller. By rummaging through various MIDI controllers I’ve broken open I can see that the Orba uses the CYC4045AZI-S413…based on previous hacking attempts, I’m pretty sure it’s loaded with custom firmware. And the touchscreen on the Misa Tri-Bass uses the SSD2533QT6.

Given how well different devices track position/pressure/velocity, I’m interested in the idea of a cap touch design that can do all of them. It sounds like the Arturia MicroFreak handles velocity, though I’ve never tried one. It seems to have a Manta-like approach, using coverage based on an unusual pattern.

The tricky part would be the programming. There’s an open-source trackpad design called the Peacock on Github which can track touch (and other gestures like pinch-zoom), but it would be completely beyond me to try and add pressure/velocity based on calculations about, say, how quickly elements in close proximity were activated, to read how fast you’re pressing it.

Although it’s a trackpad, this project uses a Maxtouch IC designed for touchscreens. (“I have chosen to use a touchscreen controller as they tend to be designed for much larger sensor areas than touchpad controllers.”)

This uses the same diamond layout for the cap touch elements. These are wired up in rows and columns…the basic idea is explained in docs like these.

But I can’t find any open-source code for velocity estimations with this type of design, or cap touch controllers that have it built into the firmware.

The only available velocity code I’ve come across is for the simpler KontinuumLAB cap touch keyboard project.

So…interesting idea, but recreating the Loom is out of my depth!

3 Likes

Some good results, detective!

Coincidentally I was looking at another Cypress touch controller series, CY8C20, simply because Bela Trill uses it. But i also found Maxtouch (whatever is currently active and recommended for new, larger designs? ATMXT2912?) because of the larger screen size they support.

However this still leaves questions regarding musical multitouch. Apparently the new maxtouch controllers support “true multitouch” but the CY8C types do not.

They do give you two lists, x and y, of all currently pressed areas but x and y “cannot be reliably matched one to another”.

So I’m wondering if newer controllers like Maxtouch give us better options there.

As to the Peacock, I’ll look into that, thanks for the pointer. Though this may also be limited multitouch (if you only need to check for “one/two/three fingers” and “pinch/zoom” or “rotate” etc., and the actual IC might have a function that recognizes them. Or I imagine you can detect them in code more easily than “there are five points being held, the first is being moved left and right, and the second and fourth are being moved up”)

I actually ordered some FSLP and will breadboard something resembling the first Linnstrument design idea. Force resistance still seems much more usable (also I have more confidence in resistance, from a hardware design and coding point of view, and never understood enough about capacitance)

If I can get anywhere with this design and can see this becoming some sort of kit, I will contact Roger himself to make sure all is respected.

1 Like

Love this discussion – @thetechnobear, perhaps we could use a separate Category for Sensor Technology. Then I can throw out my idea of building a bow based on high-speed gaming mouse sensors ;-).

3 Likes

Cool, yes. I didn’t really follow how the Linnstrument works its magic, and it’s not going to be cheap to get started. I contacted Tangio…

…to ask about the practicalities of numbers and costs, but they didn’t reply. (edit They have now.)

Also got lost in discussion about something called “Quantum Tunnelling Composite”, and the other day I received a plastic bag containing five tiny squares of the stuff for a tenner…but that’s probably a musical dead-end. (Some kind of FSR substitute.)

Peratech look quite interesting too. They didn’t reply either.

1 Like

(I’m obsessed with MIDI bows. Long overdue.)

3 Likes

Interlink seems to be a big player (they may even have been the original design) and another big one is Ohmite. They both offer PDF guides for force+position sensor design not unlike the Linnstrument. I’ve come across Ceradex and asked to buy a sample of their XYZ pad but as with many such companies their mode of communication is a bit “ineffective”, no offense meant

2 Likes