Statistics: November 2012
werner at almesberger.net
Tue Dec 4 17:17:51 EST 2012
Paul Boddie wrote:
> Well, I think there's an obvious direct correspondence between enthusiasm for
> concrete future developments and the list traffic such enthusiasm generates.
Yes, we have basically the following project "threads":
- Ben NanoNote software: the system/distribution is still being
actively maintained, mainly by Xiangfu, but there's little
visible activity otherwise.
Perhaps that's also a sign that the platform has matured. E.g.,
I use my Bens quite a lot to help with developing DIY circuits
(combined with UBB, they make great in-circuit programmers or
in-circuit testers), but all these uses are project-specific.
- NanoNote successor: we talked a lot about making a possible
successor but nothing concrete has happened in this direction.
One of the main problems is of course the lack of money to fund
Past Qi-Hardware developments were financed by participants,
mainly Wolfgang, but these pockets are pretty empty these days.
If someone feels up to setting up a major crowdfunding project
that finances R&D and production, that may be an option. A few
- it's probably hard to convince people to pay for R&D since
a) the outcome is uncertain and b) it'll take a while until
there will be tangible results.
- in addition to volunteers, you'll need 3-5 people who can
work full-time (with plenty of overtime) on this, including
occasional travel to external suppliers and fabs.
- you'll need to raise a substantial amount of money. If
things go very smoothly, such a project would need probably
at least 3 months for design (with partial prototypes) and
sourcing, then another 3 months spent on making and testing
prototypes and preparing for production.
Each prototype run will have external costs in the order of
at least USD 5-10 k(PCB and SMT cost, components, shipping,
- it may be hard to agree on a feature set. Perhaps projects
like PengPod (a Linux-friendly tablet, but AFAIK not Open
Hardware) are helping us there, because they already satisfy
some of the need for tablets, narrowing the range of potential
- Milkymist One R3: while there seems little enough left to do on
the hardware side (all that went into Milkymist One R4, see
below), and development of the "VJ station" software has come to
a halt, there are two follow-on activities that are still alive:
- the MMU, headed by Yann: this is advancing slowly but steadily,
with the expectation that we'll have a solid basis for Linux at
some point. Practical value of this will be limited, given the
overall state of the platform, but that won't diminish the
value of the learning and perhaps teaching experience.
- improved SoC architecture, including a much faster memory
interface, headed by Sebastien: this seems to be in a working
state but includes the burning of some bridges. I.e., it may be
a good basis for the Linux port (which should in general be
agnostic to how the SoC is implemented), but since not all the
functional blocks used for the VJ application have been ported
to this new Migen-based system, there is no trivial migration
path for existing uses.
- Milkymist One R4: this would be the next revision of the
Milkymist hardware, with fixes from R3, and several
improvementes. This is more or less at a point where we could
consider a prototype run, but since there is very little prospect
of being able to commercialize future Milkymist hardware, making
boards would be nothing but a very expensive extravaganza.
- the Milkymist experience has spawned two projects that focus on
specific details. One is Sebastien's Migen, which is sort of a
build system that generates "smart glue" to interconnect pieces
of Verilog, and the other is Wolfgang's fpgatools, which aim to
end the Age of Darkness surrounding FPGA synthesis.
Both projects are quite active but have their main audience ouside
the Qi-Hardware community (in the case of Migen) or are still
pretty much a "lone wolf" activity (in the case of the fpgatools).
Neither aims to lead directly to the production of large
quantities of hardware at the moment.
- I have a few small projects that experiment with LEDs, among them
"antorcha", a wand with 32 LEDs that can display text or images
when waved back and forth rapidly, and "tornado" (under
development), a similar but more advanced line of 64 LEDs that
should also display things when whirled at the end of a chain.
I made these specifically for protest marches.
There's also the "LED Toy", a smaller and more streamlined device
with only 24 LEDs that's intended for left-right symmetric images.
Some of these projects could be turned into products, but again,
they'd need some R&D and production setup budget to get there.
- there's a bunch of ancillary projects (eda-tools, cae-tools,
kicad-libs, ben-wpan, etc.) that move in little leaps from time to
time, e.g., when new features are needed, when bugs are found, or
to track changes in their environment.
Did I forget anything ?
More information about the discussion