Developing a universal module

we've talked about it before. Assuming you drive it at a high frequency, it probably isn't much slower than our current solution. Though it would need to be tested. I'm not sure how you could de-assert CSN with this module. When you assert !OE, all registers go into a Z-state. You likely need to add pullup resistors to all pins to reset them, but it is possible not all pins should be pulled up so you need to figure out which pins need a pullup and which need a pulldown (for instance, I have no idea what to do with the nrf CE pin in this case)

Assuming it does work somehow that I'm not seeing, it has the benefit of being faster to change the non CSN values like the CE pin. You could also choose to chain 2 of them together to get 16bits, and then you wouldn't need to play around with the chip GPIOs (though I could get the same effect by using a larger AVR)

But I've actually tested my board, and I know it works fine, and it is a tiny bit easier to build (no diode, no resistor, smaller IC). You don't need a programmer for it because I developed a dfu to program it, and most importantly I already have it.

So your solution may work if you have something up your sleeve to de-assert CSN, but I don't see any tangible benefits over mine, and I am not going to support multiple ways of doing the same task.

The problem I'm having with the board is making it small enough to fit in all the transmitter modules I want it to and to support as many variations of the modules as possible. The circuitry is simple, and the avr itself works fine for its task.

Edit: Also note that the number of (non transceiver) components in my design is 1. if you want to hand-wire the AVR, it is no harder than hand-wiring the HC595. The circuit board is only there to make it easier.

I still see no way to guarantee that CSN behaves as expected when using the shift register. it is either in the z-state or the programmed state, but I see no way to go from one programmed state to the next without transitioning through the z-state (which would require pull-up resistors) to guarantee that csn doesn't spontaneously transition.

show me a set of steps to send:
CSN=0
byte = 0xa5
CSN=1
without any other bytes being consumed by the module and without falsely triggering any other of the shift-register bytes low.

Good news. the version 1.1 of the board is fully functional.
I have tested:
A7105 with and without PA
NRF24L01
CC2500
CYRF6936 (Devo and AWA24S)

All working as intended.
I created a version 1.2 of the board with the only change being to slightly widen the drill holes to make it easier to install the 2mm pin headers

I have enough confidence if some adventurous folks want to try it:
oshpark.com/shared_projects/nNRSCkjI

Schematic:


Gerber files:


Kicad files:


The board is built with specific components in mind:
ATTiny24A/ATTiny44A/ATTiny84A (note the 'A' it is important)

NRF24L01 (others will likely work):
www.ebay.com/itm/400611170500

A7105:
www.ebay.com/itm/181091377178
or (no PA):
www.ebay.com/itm/A7105-2-4G-Wireless-Mod...ver-AA-/171154521980

CC2500:
www.aliexpress.com/item/CC2500-PA-LNA-wi...-2-4G/359941652.html
or this one (needs antenna soldered)
www.skyartec.com/ProductShow.asp?ArticleID=1977

CYRF6936:
www.digikey.com/product-detail/en/AWA24S/748-1003-ND/1687150
or
www.ehirobo.com/walkera-wk-devo-s-mod-de...-upgrade-module.html

The modules are designed to be soldered directly to the board. it is possible to do this without pin header, but most users will find it much easier to purchase 2mm pin headers like these (only needed for a7105 and cc2500) (note these are metric headers, and it is very unlikely you'll find them at a local hobby shop):
r.ebay.com/K8zPRJ

If using the skyartec module, you need 0.05" headers:
r.ebay.com/GlelB8

Note that the board design was done to minimize space so it could fit in a Devo7, Devo8/10/12, or X9d. This means that some of the modules sit on top of each other.
specifically:
The A7105 must be installed before the nrf24L01
the ATTiny and connecting wires must be installed before installing the cyrf6936
It is recommended to shorten the pins of the AVR chip so they don't stick through the board as far.

I will try to put together detailed instructions with pics in the future, but at the moment. The soldering is pretty easy but must be done in a specific order.

I am going to try to design either a 2nd board, or add onto this board to support the 'XL' series A7105 and CC2500 modules. They use standard 0.1" headers, which are easier to deal with, but at a cost of a larger board.

I was able to squeeze the XL2500 onto the board without growing it any. I think that is it for this size board though. I haven't posted the v1.3 design yet as I haven't actually tested the changes.

I placed an order from Seed for the v1.3 board. Their service is nowhere near as polished as Oshpark, but I was able to order 20 boards for $18.

If there is no issue with the final design, I'll likely offer the boards to the forum at cost. I went with 0.8mm boards from seed, since it'll be easiest to de-panelize them that way, so in the end they'll be nowhere near as nice as the Oshpark boards. If you are interested, and need more than one, Oshpark will likely still be the better deal. For $9, you can get 3 very high quality boards anywhere in the world.

I started work on some documentation of building the module. Mostly it is just a place for images I took while putting one together. I'll fill in the text in the future as well as include instructions on how to connect it to the transmitter.
bitbucket.org/PhracturedBlue/deviation/w...20the%20Multi-Module

The module will not fit in a &3 (or probably any Devo Tx) with the Cyrf board attached. So no you don't need it. If you did want to replace the low-power on-board module in the 7e with a high-power one (i.e. not do the Diode mod), you'd likely need to use jumper wires so you could mount the moudule side-by-side.

The pics I show are of the 2nd board I assembled. The 1st board is all socketed as MassiveOverkill described. This is a lot more convenient but means it won't fit inside any Tx. Because the 0.1" female headers are taller than 2mm female headers, the stacking of the A7105 and nrf24l01 still works.

Note that if you want to socket the cyrf6936, you must buy 2-row 2-mm headers. with the 0.1" headers, you can mount 2 1-row headers adjacent to each other to make a 2-row, but this doesn't work with 2mnm headers.

I'll take some pics of mine. I use a male header on the board, and a 10-wire ribbon with female header (only 8 pins used) to connect to the Tx. I'm in the process of wiring all my Tx this way so I can swap the modules around for testing.

The AVR chip is an ATTiny24A/44A/84A (any of these will work fine) in a 14-pin DIP package:
www.ebay.com/itm/Atmel-8-bit-Microcontro...&hash=item565a2593fb
If you buy them in packs of 10, they are ~$1 each.
I would probably recommend the ATTiny44A or 84A instead of the 24A. We only use half the space the 24A has, but who knows what firmware upgrades we may have in the future.

Your Tx will be able to program the chip, so there is no need to buy a programmer, though for someone like MassiveOverkill, I'd recommend buying one as it is easier and more reliable (you may need to try a couple times before the tx programs the chip successfully)
I use this one:
www.ebay.com/itm/USBasp-USBISP-AVR-10-Pi...&hash=item258405ff05

Assuming the boards I get from china are good enough, I will likely include an AVR chip, since I have a bunch of spares. I'm just not sure about how adding a DIP to the package will affect shipping.

I recommend reading this, I found it very interesting regarding FCC regulations:
www.sparkfun.com/tutorials/398

As I am not a lawyer, I will not make a statement on my interpretation of the details.