Rockbox Development > Feature Ideas

Raspberry Pi for a new generation player

(1/1)

jimbo70:
I'm sure it has been already discussed before, but what do you think about engineering a new generation player based on Raspberry Pi Zero or Pico?
Wouldn't these models, or similar others by other brands, be ideal for developing a new player from scratch with open, documented features?
There are also cheap displays fit for these mini boards, even e-paper type for low power draining, so the only difficult thing would be sourcing a proper case; would a Kickstarter funding, or the Rockbox site selling them be a good starting point?

speachy:
The RPi family is poorly suited for something ike Rockbox.  The Pico is simply too limited to run Rockbox (_far_ too little RAM).  As for the others, even the original RPi has more than enough raw performance/resources for Rockox to run well, where they fall down hard is how much effort would be needed to create an integrated device.  Just off the top of my head, the RPis either completely lack audio functionality (zero series) or it's complete garbage (everything else); there's no meaningful power management (ie operating from a battery, charging control, plus meaningful suspend/sleep modes); bootup times are pretty awful (at least if we want to stick with Linux); performance of typical SPI-attached displays is pretty atrocious and can't take advantage of the RPi's GPU; and the resulting form factor would likely be about the size of an early-gen (ie thick) ipod if you wanted it to have reasonable battery life.

By the time you add up the necessary bill of materials and the work it would take to assembe/integrate (custom PCB, lots of fly wires, etc), it's more cost-effective to just design and manufacture an integrated PCB to begin with, and that would give you far more options with respect to the final form factor.  But that leads to the real problem -- the enclosure.  The only way for it to be remotely cost-competitive would require mass producing a case, which in turn requires very expensive tooling.  Literally everything else can be done incrementally on a shoestring budget, but a case is looking at a six-digit price tag, up front. 3D printing can be used for low-volume prototyping but again, cost-prohibitive for anything we'd expect the general public to purchase.

Remember, this hypothetical player would be competing with hardware that can be purchased new for $70-120.  Not to mention second-hand hardware (eg hdd-based ipods) that can be purchased and modded in the same general price range.

The sanest path forward would be to find the folks who built an existing case (or better yet, an entire player with the hw functionality we need) and get them to quote doing some small to medium production runs for us (=~1000-10000 units).  10K units is the magic point where volume discounts begin to kick in, but even with an existing case, that's still a lot of up-front capital to be tied up in inventory -- perhaps a kickstarter would help there, but before a KS pitch would be viable,  the design (and initial prototypes) would have to be essentially complete.

(Yes, a lot of thought has gone into this whole thing.  And all roads lead to needing a pile of $$$$ up front)

Frankenpod:
Don't suppose there's any possibility something could be engineered to fit into an already available casing?

Ipod classic/video parts are very widely available from the usual Chinese manufacturers (I notice they've started producing cases and clickwheels in a variety of new colours now).  The limiting factor is the supply of second-hand ipods with still-working logic-boards.

Maybe there are all sorts of legal/IP issues there, though?

speachy:
Hmm, a quick ebay and aliexpress perusal shows replacement case + clickwheel assembles can be had in single-unit quantities starting at about $20, depending on the color.  And if we placed a sufficiently large order, I'm sure we can have them put whatever logo/writing on the back that we want.

Still, even without the apple logo/etc on the case, given how litigious Apple tends to be, I'd be _very_ wary of trying to sell a new mp3 player that is externally indistinguishable from one of their own (even long abandoned) product lines.

__builtin:

--- Quote from: speachy on December 18, 2021, 10:24:03 PM ---By the time you add up the necessary bill of materials and the work it would take to assembe/integrate (custom PCB, lots of fly wires, etc), it's more cost-effective to just design and manufacture an integrated PCB to begin with, and that would give you far more options with respect to the final form factor.  But that leads to the real problem -- the enclosure.  The only way for it to be remotely cost-competitive would require mass producing a case, which in turn requires very expensive tooling.  Literally everything else can be done incrementally on a shoestring budget, but a case is looking at a six-digit price tag, up front. 3D printing can be used for low-volume prototyping but again, cost-prohibitive for anything we'd expect the general public to purchase.

--- End quote ---

I've given some thought to this problem as well -- the electronics/PCB design work is within the realm of possibility for an open-source project such as ours. There are plenty of PCB houses out there that will spin and assemble fairly complex boards for reasonable-ish prices ($100 per board in low quantities, falling rapidly as you scale), on rather fast short lead times as well.

Mass-production of the mechanicals is, as speachy says, a much thornier issue. The first problem we run into is the design software. Unlike PCB design, where there exists a very nice open-source package on parity with the professional options (KiCAD), for mechanical CAD, you're stuck with either OpenSCAD or FreeCAD as open-source options, neither of which is very desirable for anything beyond a single low-complexity part (in fact, it's basically impossible to have a OpenSCAD part sent out for injection molding or machining - it's a purely polygonal modeler). Once you're talking complex assemblies, as we'd undoubtedly end up with, the proprietary options (SolidWorks, Autodesk, CATIA) are the only game in town. Fusion 360 does have a free "hobby" license for personal use, but that's a gray area for us. On this point, the best bet would probably be to hope that FreeCAD matures to the point of viability. (I haven't even touched on the issue of human capital - who would do the mechanical design?)

Assuming we've got a viable design, manufacturing is the next issue. Initial quantities of 10-100 can be done by a couple of FDM 3D printers running for a few weeks. The unit costs here aren't too bad (I'd estimate $2/case, assuming $20/kg filament, and a 100g case), but it doesn't scale. Injection molding is the logical choice for qty >100. I've heard that Protolabs will do low-volume injection molding runs for $1500-$10000, depending on mold complexity. (For something like an MP3 player, we'd probably be in the medium complexity range, so maybe $5k for a mold). There's also a new market for "DIY injection molding" machines, but I haven't looked into those too much. In any case, you still need to make the mold, and that doesn't come cheap - you need a skilled machinist and a CNC milling machine.

If we can come up with the funding, it could be done, but I do wonder how much of a market there still is for this kind of thing...

Navigation

[0] Message Index