Scarcity and Precarity in Small Manufacturing

I remember reading that the different core options in current CPUs is a result of how many actually fail during the manufacturing process - ex. all chips are built to have the maximum number of cores, but only a percentage of them is successful in all areas and most have at least one failure. So they simply sell it as cheaper version with fewer cores. And sometimes they will intentionally disable cores to make a chip less powerful to balance their stock. I can’t find the source right now but if I can I’ll add it.


cm3+ restock listed as feb 2024


Yeah, the example I recall was Apple was making decisions on how many cores their GPUs would have based on manufacturing yields rather than straight product positioning.

Another example, particularly relevant here, is that Raspberry Pi boards have always allowed user-controlled overclocking, with the degree of success a mix of custom setup (providing extra cooling, &c.) and manufacturing luck (how far does that particular board exceeds specification).


these are good questions.

it’s good to keep in mind that monome has been around for 16 years as i write this. compared to the number of “breaking changes” from the tech industry on a whole, i’d say we have a pretty good record of maintaining compatibility and respecting the community. for example, the serial protocol has been stable for about a decade.

most reasons for change have been in pursuit of “improvement” on various levels— mostly manufacturing optimization (ie, make things easier to build, more consistent, less expensive)— but also improving on performance and (rarely) features.

keypad press detection required constant attention and revision in the early years. we only changed the keypad size once— from the 40h to the “series” in accommodation of a dimension of 16 keys. circuitry changes were attempted for better key detection, as were conductive silicone compounds. the most recent keypad design (last year) is fully new, and in the end also had a lot of design and production issues despite my (now very extensive) experience. generally if you have keypad trouble (especially within GS versions) it’s likely worth replacing the keypads (email

variable brightness began as a literal hack to a new LED driver technique. we decided to release it. the next year a much better LED driver chip came on the market, and we improved the circuit. last years’ redesign used a more modern chip to provide the same capability at lower power consumption.

one major issue we had with apple is their ever-changing FTDI driver which caused perpetual frustration over the course of about a decade. and last years’ redesign now uses a new STM chip with built-in USB capability… but there are now operating system level issues with class-compliant serial transfer if using un-optimized bulk transfer— ie, sending tons of single LED updates at high rates, versus using /led/map packed messages (btw, this is how you fix the issue, until we publish some sort of hack workaround).

regarding scarcity, we’ve been hit hardest with this new revision (the STM chip) and i’ll have to do a revision with a different microcontroller (not a small task). truly, scarcity has never been a real problem since we started. (i mean, sometimes you might have to wait a month for something in high quantity).

so in conclusion— yes i’d love to time-travel back to the beginning and give my past self some advice. but the chips we use today and have access to weren’t available back then (ie, USB C, or tiny cheap white LEDs, etc).

fun fact: the first circuit board i ever produced was the 40h logic board. all my grid prototypes were hand-wired. these were early days— arduino and the phrase “physical computing” didn’t exist yet.


As it says in the title: “rumored”, but some more info nonetheless.


Thank you so much for your answer @tehn ! As I wrote in the other thread it is so interesting to get a peak behind the curtain! Really appreciate it!

One thing I was thinking about is that since most of these things are open source maybe a system to share the work load other than just whoever happens to see it on Lines or get contacted for you may be something to consider? Not totally sure how to do it, but maybe some of these things (like nerding out harder to get the new 2021 Grid to work on the Organelle) is something someone would be interested in contributing on if it was some clear list of things people could contribute to?

Or maybe that would just flood you with emails of sollutions to test, I am not sure. Anyways Monome + its devices + its community is still the best in this space :slight_smile: .

And just to be clear in this thread we sorta concluded that it is just double press filtering in software since pressing different buttons is super fast, and in this thread I got the sollution to the single message / overload thing.

1 Like

Some time ago we made the decision not to post monome-related stock announcements or any sort of commerce hype here on the forum. I’m instead posting this information prominently on the monome site.

Consider this post something of the opposite of a stock announcement, then.

It will be very difficult to continue producing norns in the short-term: estimates for restock of the necessary CM3+ module at February 2024. I’m reaching out to the Raspberry Pi Foundation hoping for some help acquiring these sooner, but I’d be grateful if anyone here might have industry connections that could assist securing what we need. I know it’s unusual, but we may as well disavow the idea of “normal.” (For example, Adafruit has a video series they put out where they basically beg large companies for relatively small quantities of semiconductors.)

Basically, I’m not when we’ll have norns in stock again. I’ll keep the site updated with news.

I’m in the process of redesigning the circuitry of the grid to use a different microcontroller, so that we can have more ready by the end of the year. This is a substantial burden, but STM continues to be an absolute mystery. (This same chip is holding up any more crow production).

Lastly, we’ve discontinued the norns shield. I’ll be preparing the files for the newer versions and add them to the github repository, so that anyone can DIY a shield using two different audio codec options.

Thank you for your continued support.


If monome ever started a patreon I would gladly sign up. I feel very fortunate being a part of such a special community.


Yangtze river dries up in drought, causing 50% drop in hydropower output in China’s heavily populated Sichuan province, impacting surrounding cities like Hubei. Aside from disruptions to shipping, thousands of factories incl. Foxconn (Apple devices), Tesla, and Toyota have suspended operations until supply resumes



It’s really starting to feel like scarcity-and-precarity-in-small-planet-existence…

Is there anything we can do as a community to help? Even something pretty far outside the box?


Do you mean something related to this other thread ?


Yes that, but specifically in light of potential impact on availability of materials for monome…

I know that my trust level doesn’t grant me access to things like the for sale/trade thread and that’s fine with me, but is it the same case with the climate-collapse thread? I get an “Oops! That page doesn’t exist or is private.” when clicking the link.

1 Like

Yes, that thread is in the open category which is also limited by trust level.


Thank you for the clarification!

As seen a few posts above, @tehn has expressed interest in potential individual help.

You’re probably right @eblomquist, I suppose that a more collective help could also be welcome.

I’m curious to see what ideas about this can be presented here.


I’m sure that something like this has been considered, but I’m just throwing this out there. In the average house, in the average person’s inventory there is actually a huge surplus of computing power.

Have you ever considered building a norns that utilizes the CPU of a desktop/laptop computer? If norns is running in docker, it feels like it could be possible to host the CPU tasks on a standard computer while putting the dedicated audio IO, serial IO, display and controls in a norns-lite container tehtered to another computer with a single cord. Or, what about a way to make norns work with a commodity ARM based micro compute platform? A fire stick or a rooted chromebook or something like that.

What gave me the idea was my high schooler’s ingenuity during covid in rooting their chromebook so that they could appear to be in class while playing Game Boy Advance games. She and her cohort played through dozens of GBA games during covid this way… the platform was emulated, but it was still beautiful and fully functional.

I know this would break the very cool concept of standardization of the platform… but in our day I think that this paradigm could be extended as an abstraction. The platform is standardized conceptually and in implementation while the hardware is virtualized.

And I continue to have a dream of a client/server virtual architecture that would allow virtual hosted norns instances to work as vsts within a daw.

Just thoughts. WIshes. Etc. Silly me.


Hmm norns (or audio software more generally) running, in batch mode, on BOINC or A fascinating idea actually…

You are right, many of us have potent computers just laying around (peers around the room; yup guilty!).

I can imagine there are thousands and thousands of unused RasPi 3B+s too in people’s cupboards, once purchased and then abandoned after initial curiosity.


I love this so much. The future might not be so dark after all.


norns seem to have been designed as tightly coupled hardware platform.

maybe are you looking for something else?

yes. the server/client architecture is cool for lots of applications (chess programs).
but i personally don’t “get” the added value here vs having everything run on a single computer.

maintaining support for CM3, 3B & 4 looks already quite daunting, and the amount of effort it requires when adding more platforms would certainly not scale linearly…

virtualization & standardization are not totally incompatible!
using a qemu-based emulation of a raspberry pi (w/ a docker wrapper for convenience such as this or this) it may be possible to load an “official” firmware image on most computers.

above comments may sound negative but they are in no way dismissive!

we need more people like you challenging of how things could be. out of many ideas that may not work in practice, a world-changing one sometimes pop. keep on dreaming :kissing:


I think the point here is that if you can’t get hold of that single computer (for the reasons outlined in this thread), why not sell a box with the display and ins and outs and offload the actual processing to the other computers that everyone has around them at all times?