create a custom arm board for $5 @ 1mi units

July 3, 2016 at 15:34:18
Specs: Linux
If I had a budget of $5M and had a year to ship 1 million ARM units in a tiny form factor, what steps could I use to design them?

Since the Raspberry Pi Zero doesn't actually exist (isn't available in unlimited quantities) / is a marketing stunt, I have lowered the required specs to:

1) Size of raspberry pi zero

2) No video output or camera input, just SDCard and 1 single OTG USB and 1 single power connector

3) 256 MB of RAM

4) 700 Mhz

5) Not necessarily compatible with Raspberry Pi Zero, but must support a generic Linux

However, unfortunately I am adding a requirement:

6) Has a microcontroller chip which is interfaced to the Linux side and and can shut it down. (For power reasons.)

This means it's not a simple clone but would have an actual design process.

In practice, if I haven't shipped software but had the money, how would this process look? Who would I hire? Where? How would the chip get designed?

Thanks for your help and thoughts. Please assume I would have the money for it. This is just a blue-sky question on how design process even works at this scale.

Do not budget for any profit for me.

See More: create a custom arm board for $5 @ 1mi units

Reply ↓  Report •

July 3, 2016 at 16:56:30
700MHz, 256MB RAM, no graphics output? What would it be used for?

Reply ↓  Report •

July 3, 2016 at 17:14:45
Hi riider, thanks for your comment. I'm just trying to understand the whole process. I suppose it would be used for any embedded application that requires higher processing than a microcontroller can provide. though I don't specify any outputs except usb - which can run a video camera, or a variety of peripherals, or connect to the network (over a usb wifi) and do whatever. there is a lot that raspberry pi's can be used for, even headless (meaning without a screen), and the first ones had just 256 MB ram.

here is the product I'm talking about -

if I wanted to manufacture a million similar ones, but with a custom design... (as described above) what is the whole proess? Assume I have the money, bot only about $5 per board, and want a quantity of a million in a year. doable? what are the steps? Also accounting for the custom design that I mention in 6. above.

message edited by curious4

Reply ↓  Report •

July 4, 2016 at 00:22:36
$5, including all development costs, is just unrealistic. Not doable.

Reply ↓  Report •

Related Solutions

July 4, 2016 at 02:37:36

Could you share your detailed calculations? You can post an excel spreadsheet, for example Google "easy free small filesharing upload" or any similar terms (I just tried it, was a result and seems okay but it's not my specific recommendation or anything.)

I would like to know how you conclude that $5 (or a dollar or two more) is not reasonable. At $5Million it means that three engineers could work for a year for $120,000 and spend an additional $100,000 on tooling and equipment and still leave $4.54 million for the BOM and manufacturing costs for the board. I also said if absolutely necessary a dollar or even a bit more could be added to the BOM. So you could still spend $5.5 per board (BOM and manufacturing) after having three engineers @ $120k work for a year on it and also have a budget of $100k for whatever generic things they needed.

So I am not sure I can see your reasoning that the development costs are excessive. It's just some ARM boards and some routing, right? Please share your detailed thoughts. Thanks!

I realize it would be unreasonable at a much lower quantity.

message edited by curious4

Reply ↓  Report •

July 4, 2016 at 03:28:36
There are 65 people in the Raspberry Pi team. It took seven years to develop the computer. I'm not going to do your financials for you.

Reply ↓  Report •

July 4, 2016 at 04:29:20
ijack, that is pretty ridiculous. How many people are on the Orange Pi team and how long did it take them to make a $10 compatible board (orange pi one)? except with on-board ethernet, additional USB host (1 OTG + 1 host as well), and quad-core CPU. Which happens to also support raspbian -

I very much doubt they have sold a million, either. The raspberry pi foundation does a ton, and for your information I think a single person laid out the PCB for the raspberry pi zero (since there was a video about it.) Granted they pared down existing designs, but still.

If you don't want to answer my question or offer your opinion, that's absolutely fine. But this is for having these types of questions answered. Someone else could step up.

or you can give your impression. How much do *you* think it would take in development cost? (how much did the orange pi team spend)? How about development time?

It's fine that you don't think it's possible.. heavier than air flight was considered impossible too. All we're talking about is laying some PCB around any one of a large number of chips from a variety of ARM manufacturers. (samsung, sony, broadcom, allwinner, etc).

it might be difficult, but that's a question of budget and time. I will take your suggestion of "65 people * 7 years" ( * $100k loaded cost per employee = $45,500,000) under consideration. Since that's a very large multiple of their actual budget, you'll excuse me if I consider it a wildly high estimate.

Reply ↓  Report •

July 4, 2016 at 04:42:04
Let us know how you get on with this project.

Reply ↓  Report •

July 4, 2016 at 06:12:17

>Let us know how you get on with this project.

Sure. I'm rather new to - will you get a notification if I post in this thread later? Since the board I have in mind includes a microcontroller side (which I listed under 6) I was also on the arduino forums - there, they suggested that since it's not pure Arduino but half ARM / half arduino, I should consider making my own forum for the project. I may do so and would invite you to follow along with the project (I consider community input to be very important.) The only question is how I would let you know, how to invite you. I didn't notice PM functionality in, though I may have missed it. I dont know where that forum would live, yet. Possibly reddit.

Let me know how to keep you in the loop if you'd like to follow along. Your input is also always valuable. :)

message edited by curious4

Reply ↓  Report •

July 4, 2016 at 06:38:49
In your initial post you asked many basic questions. It seems to me that anyone with this many questions would not be in a position to make meaningful cost estimates. If this project is real you are very much in need of professional advice that is beyond what any forum could provide.

But your cost estimates do seem more than a little optimistic.

Reply ↓  Report •

July 4, 2016 at 07:22:28
Thanks. All this is true. :) I've found that there is a PM center, I'll PM you when my forum on this project is up.

There's some background that I haven't written here - originally I wasn't designing my own board but wanted to just move raspberry pi zero units., in part for my own very small-scale embedded project.

message edited by curious4

Reply ↓  Report •

Ask Question