Single-Board Universal Module

Just to get a feel for it, I priced out the BOM for:
1 part: ~$40
10 parts: ~$34
100 parts: $28
1000 parts: $23
There are parts I can't get from the same supplier, so those prices could shift around by about $5. Also, I used Digikey as the supplier because it was easy to price that way. They are certainly not the cheapest supplier.

Lastly for the bluetooth version (PWM/JR module) you'd need to add another $10 to the cost

I've been told FCC compliance would run ~$5000 (though I don't know how reliable that number is)

I'm guessing here, but you should assume that anyone who would want to sell these would charge ~2x the BOM cost to make any profit at all.

Right. That is why the PWM capability is there. It provides the ability to use Deviation's protocols with any radio. We know there is an interest in the Anylink/Orange/Magic Cube, so I assume there is an interest in something that would do what all 3 of those do in one package.

Whether there is enough interest, I don't know. As I mentioned in my 1st post, this may never come to fruition.

As far as costs go, prices can vary wildly based on supplier. Once you get past ~10 units, the main costs come from the 7 primary chips (below 10 units you pay a huge premium on the passives)
These chips don't get huge price-breaks on the volume were talking about. For the 10 pieces kit from Digikey, I estimated these at about $22
When priced from China, the price appears closer to $10
There are other pieces that could additionally reduce costs. For instance the crystals in the board are around $4/board. From china those would be $1/board instead. Unfortunately for many passives, you take a risk in tolerances as you go cheap, which could result in the board not performing well. This is one of the reasons I need help from RF experts who have experience choosing appropriate parts to balance price and performance.

I did some work on placement. The outline here shows the size of a Devo module, which was my target size. It doesn't look like that will be possible though. I was able to squeeze all the components in, but I don't think the ground plane will be robust enough, and I don't think I have enough room to actually route the design (the bottom layer needs to be a ground plane under the RF portion to perform properly). I'm currently trying to figure out how big I can make the module and still fit it into all Devo radios.

I wasn't able to fit the headers onto the board, so those are likely to be casualties (only preserving the Bluetooth and Devo ports)


If I can add 5mm in width and 2 in height, I think everything will fit.

Honestly, I'm not even considering doing it myself. My hands just aren't steady enough, and therse parts are really small. Ideally I'll team up with someone to get the boards produced. Worst-case, eventually I'll pay to have a set of prototypes run through an APCB shop.

I got my F072 discovery board in the mail today. I'll need to hack up one of my nrf modules and see if the interrupt is fast enough soon. Probably not this weekend though. Other stuff going on.

I haven't routed it yet except for half of the A7105 module (which I did just enough of to get an idea that it was feasible). And yes I expect it'll take me many hours to complete. As it turns out the RF portion is all based on the reference designs (except the CYRF module) so generally placement can be done on a single layer, which makes things a lot easier.

I want to finalize the board dimensions before I actually spend the effort to route it. Seeed does PCBA, but only using their stock parts which are larger than what I need. I got an online quote from smart-prototyping, which was ~$1400 to do 5 boards (or $1200 for 1) as I recall. I'll likely shop around more if it comes to that.

I think I was looking at another company actually.
I just recalculated with smart prototyping.com:
I ended up with a price of $55 for one board or $24 each for 5.
That is using their manual service, and the prices seem incredibly low to me. It'll definitely be worth contacting them through once I'm further along.

I'm now at the hard part. I've got the RF fully routed (though it may need some tweaking/component adjustment) and just need to complete the digital section. Trying to squeeze the routing into the available space is proving difficult and I'm not having great success. I could go to a 4 layer board and solve the issue, but I don't think it should be necessary. I know there are some folks here who do layout, so maybe someone will take a crack at it. Otherwise I'll keep plodding along on it.

Here are the eagle files for the circuit as things are. The only placement critical stuff (not in the RF section) is the 1x5header on the left and the 2x5 header in the bottom middle. Anything else is fair game. Also, most pins on the STM32 can be rearranged/reordered to improve congestion. The power pins need to be able to handle 500mA, so wide metal will be needed out of the LD1117. the LD1117 should really remain on the top of the board if possible. Design is being done to OshPark rules, though I'm sticking to 8mil traces where possible (dru file is included).

Anyhow, I'm taking a break for a little while. Feel free to have at it if you have the inclination, time, and skill.

BTW, I am working with some folks who have RF design experience who can possibly bring this to market (and will likely end up doing their own layout of the board). However, I have no commitment that will come to pass, so doing the design myself is a hedge against that progress.

Edit: Removed old version. See below

Well, I couldn't leave it at that. So here is the fully routed board:





There is still a lot of optimization to do (as well as the RF matching). I also need to re-verify the schematic since I moved everything around. Still, I'm glad I was able to get it all done in 2 layers.

Edit: See below for updated eagle files with the brd file