search

worlickwerx / pi-parport

retro parallel port for raspberry pi
GNU General Public License v2.0
64 stars 11 forks source link

explore possible kit offering #48

Open garlick opened 3 years ago

garlick commented 3 years ago

There are some barriers for people getting this project off the ground:

garlick commented 3 years ago

A test fixture might be nice, to check out a board that has only the SMT parts soldered.

It could

Or am I getting off in the weeds for a demand that may add up to 5 boards? :-)

BTW, PCBWAY estimates $60 for 40 boards including international shipping.

quorten commented 3 years ago

I think a test fixture is getting into the weeds for SMT-only soldered boards. For other small-volume projects, it seems sane to ship out factory-soldered boards as-is and leave testing up to the user who populates on the through-hole components to complete it.

quorten commented 3 years ago

Oh yeah... pi-parport is much cheaper in volume than RaSCSI because it doesn't need relatively rare and expensive transceiver chips. It could just as well make sense to only offer fully assembled boards.

garlick commented 3 years ago

Maybe we should put the BOM into digikey and see where the price breaks are? (Probably dominated by the price break on the buffer chip which is at qty 10, 25, and 100)

At minimum it seems like a no brainer to put together a parts kit that could be provided to people at cost + shipping, since it will be so much cheaper for someone that wants to assemble one or two boards.

quorten commented 3 years ago

The 26-pin header, 40-pin header, and buffer chip together have the least discounting at larger volume and are about $3-$4 total.

For the BOM quantities as stated (i.e. not including spares for small passives), quantity 10-50 is largely similar price breaks. So, price breaks for 5 boards, 10 boards, and 100 boards, these are the totals for the BOM of one board.

garlick commented 3 years ago

Thanks for doing that!

To add to the data, here is the cost per board from PCBWAY (shipped) with a stencil thrown in:

Putting together a kit with the SMT parts soldered on as suggested by @Geato in #38 might not be so bad.

quorten commented 3 years ago

Including the DIP jumper is probably a good idea for the kit offering, I'll update the BOMs to include the Digi-Key PN of the DIP jumper I've been using.

Edit: See issues #51 and #47.

quorten commented 3 years ago

It's probably a good idea to perhaps set something up to gauge interest from folks who might be interested in pre-ordering a kit. Could be as simple as a note in the top-level README.md for adding a comment to this GitHub issue.

But yeah, my hunch is that an initial batch of 5 boards could last at least a year, 10 boards twice as long. But I'm basing this off of my (possibly false) assumption that the pi-parport market has already been relatively "saturated" by those who assembled their own boards.

quorten commented 3 weeks ago

I've been thinking about this for a long while and will jot down my thoughts here.

To avoid sitting on too much inventory, we can think of things this way: we are providing kits first as a convenient source of spares for current users, and second as an easy way for new users to get started.

Starting out by making a batch of 10 kits seems to be a reasonable ask. That way, the total up-front cost for one batch at a time is somewhere between $120 and $300. I don't know the current prices, but it seems like a manageable ballpark.

garlick commented 3 weeks ago

Sure. I could maybe send out a small batch and have the board house source the parts and do the PCBA?

I'll try to make a pass and update to the latest kicad, run through the issues and see what improvements can be made, etc. Would you have time to review any changes I propose? I probably won't have time until the weekend at the earliest to do anything.

Meanwhile, I'll open a couple of issues to float some TODOs I had in mind.

quorten commented 3 weeks ago

Sure, I have time to review some changes.