A custom chip containing a tiny piece of Klipper

General Discussion
1

Eight months ago I submitted a silicon design that contains a tiny piece of Klipper. Last week, just in time for Christmas, I received that chip!

testboard
testboard1920Ă—1440 308 KB

In the center of those pink boards is the custom chip - it is the prominent black square in the center of the raised pcb.

It took eight months to obtain the chip, but it was a busy eight months. Lithographic masks were manufactured, the silicon was dyed, exposed, etched, ionized, diced, and more. The silicon manufacturing was just the first step. It next made a journey around the world, at least twice - from wafer creation, to chip packaging, then on to pcb soldering, and finally to delivery.

I took on this effort as a technology demonstration - to “kick the tires” on a new open source chip manufacturing process - and as a kind of tribute to the history of Klipper. A tribute to all the contributors, volunteers, and passionate users that has made Klipper what it is today. The chip contains a tiny piece of Klipper’s motion control logic - one of the first ideas during the genesis of Klipper. Send simple movement commands containing “count”, “interval”, and “add” attributes, and the lower level system can generate thousands of well timed step pulses from that. A small series of these commands can generate precisely timed acceleration, cruise, and deceleration movements. It is a core piece of logic that enables Klipper to split processing between micro-controller and host.

But, does the chip work? Can it actually move a motor? Designing silicon chips without bugs is no easy task. In contrast, software systems are much easier to build. Software changes can often be tested in minutes, and if there is an issue a fix can often be deployed in minutes as well. With a silicon chip it’s not so simple. If it takes eight months to get the chip, well then, it’s eight more months if a change is needed.

To find out it it works I built a test setup:

testsetup
testsetup1920Ă—1440 211 KB

The test contains the custom chip and its test board (upper left) wired to an a4988 stepstick (lower left) wired to a stepper motor (upper right) and powered by 12V (wire at lower right).

It works! See it in action at: Tiny Tapeout Stepper Movement

I hope everyone enjoyed the holidays and I want to wish everyone a Happy New Year!!

-Kevin

14 Likes
2

…speechless… Happy new year!

edit
Did you consider about an FPGA approach?

edit
May I ask how much you spend? Which process (nm) was used for the die? Tiny Tapeout, never heard of them, pretty cool.

3

Congratulations!

What comes next?

4

Happy new year to all here and the proud father @koconnor!
That looks great and very interesting!

5

I didn’t even know you could get a custom chip made! That looks amazing!

Happy New Year!

6

Thanks everyone!

I’ve developed on FPGAs before. The goal here was to “kick the tires” on a new silicon based development process.

You can find that info and more at: https://tinytapeout.com/

I don’t have any hardware design goals right now. Reprap 3d printers were in their infancy 15 years ago - I don’t think we’ll know what open source silicon can really do unless we try out experiments like this.

Happy New Year!
-Kevin

5 Likes
7

Happy new year to everyone here. Let us hope it becomes more peaceful again.

Forgive me my ignorance, because I am but a mere machine design engineer. :wink:

What are the advantages of putting “some” of klipper on a chip?
The way I see it, probably wrongly, is that a chip is hardware so it is a fixed piece of code. Klipper is flexible and can be adjusted to our needs. Sometimes this is a bain, but most of the time it is a good thing.
If the code is fixed on a chip, what parts of klipper do you think would be useful to have as a fixed code on a chip?

With my nearly non existing insight into electronics, I wouldn’t even know where to begin to look for advantages. So, please elaborate a little bit, even if only to humor this 59 year old mechanical dinosaur.

In the meantime I think I should congratulate you on your accomplishments, since it does sound a truly dificult thing to do. And as the video showed us, the motor did move so it was successful.
Take care and hope to hear soon.
Kees

8

There are indeed a lot of advantages to software. The goal of this testing is to test out some interesting new technologies in silicon development, and as a kind of tribute to the history of Klipper. I don’t have any plans to convert Klipper to hardware. I think it is interesting to run tests like this to get a feel for new technologies - it’s also a good way to find out how and where it could be useful in the future. After all, reprap style 3d printing itself was a completely new technology just a few years ago.

Happy New Year!
-Kevin

1 Like
9

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.