Single-Board Universal Module

Also, the via stitching between the top and bottom layer ground planes should be ~0.25 inches in the RF section of the board. Generally a larger diameter via is preferable where possible to reduce inductance.

victzh wrote: Another thought - do you think it make sense to provide an extension header for the radios which are not there? There is at least one major player - Open LRS - which uses Si4432. It does not need PA and it works in another band - 433MHz. May be provide SPI + couple of CSs for extension?

The main issue is we are just out of room on this size board. I already had to remove headers I wanted due purely to board area. Adding another 6-pin header is probably infeasible. There is one option I see...hook up CSN pins to the unused pins on the 10-pin Devo header. You could then use a long '2 sided' header to allow stacking a 2nd board on top. I'm not sure it will be possible to get any wires to those pins, but it is probably not a bad idea if we could.
The main downside is that it only supplies 5V, so youd need to grab 3.3V from elsewhere on the board

Edit: We'd need to verify that the DevoTx doesn't ground those pins internally...

SadSack wrote: And if where asking secound PPM in/out for none Deviation Tx....

What do you need a 2nd PPM for? You already have one. There aren't many pins available that can act as an accurate PPM (using an interrupt vs sampling).

My expectation is to have the 'PPM' pin be able to be used in different ways. In a 'dumb' mode, it would act purely as PPM and you'd use bluetooth to configure the module. In a 'smart' mode, it would act a as a high-speed I2C pin creating a i2c->SPI bridge. This would give all the capablities of the Devo interface but with a PPM socket (i.e. support via Taranis with only custom software

The boards need to fit in the enclosure (either JR module or Devo Tx) so there isn't room to make them any bigger than they are right now. We're actually already pushing it as it will just barely fit in both cases right now.

If you can find a place for a pin on the existing design (but only in the MCU section) and can get it wired to a free pin on the MCU we can try to make it work. But none of the existing connectors can be moved to do this.
Actually, I have no issue pinning out as many of the unused pins on the MCU as possible. I just don't see a lot of room to do so.

Edit: I forgot, we already exposed the SWDIO and SWCLK pins on a header, so those are already available for various functions if need-be. But everything I said above still stands

Unless someone knows of a low-volume supplier of the BK2423, I think I'll stay with the NRF24L01+. It is easy to find in low volume, and while it costs up to $1 more, not having to worry about the supply chain makes it worthwhile for me.

After more discussion, we've decided to redesign the board on a 4-layer board. It will increase costs a little, but should provide a much more robust design. Of course it'll take some more time, and means it is no longer possible to use the free version of Eagle.

We finalized the board design, and placed an order from Oshpark.
SadSack is going to try to hand-assemble the prototypes (I can't even imagine doing such a board myself).
We'll see how it goes from there.

The final board can be found here:
oshpark.com/shared_projects/if6JSkqO

I've also tagged them with the v0.8 tag in bitbucket:
bitbucket.org/SadSack/universaltx-4-layer