I have a successful application that has been deployed many times. It is based on an AVR ATMEGA16, and right now I am working in adding remote control and telemetry by using a wireless module. I have concluded in using the Q2687 which is much more powerful that the AVR and in addition will offer the wireless functionality I am seeking for. My current approach and plan is to completely eliminate the AVR and use just the Q2687 to do what the AVR currently does plus the wireless part. So, my questions and discussion topic is this:
In these cases, is it better to use a microcontroller (uC) as a master and use a wireless module just for the communication part, or using just a wireless module can be a perfectly good solution. Sierra Wireless is advocating the the latter, but I have seen several people taking the former approach. Of course, using just the module eliminates parts, some cost, etc. But, I am asking for advantages and disadvantages of one approach vs. the other, and any other point somebody might feel worth while sharing here.
To get the discussion started: one disadvantage of using just the module is that (for the US at least) if I develop all my application in C on the Q2687 then I cannot replace the module with the AirPrime Q26 Elite CDMA in case I want to use Verizon’s network (my understanding is that the Q26 Elite can be controlled just with AT commands, and it cannot be programmed in C). In addition, if I decide to migrate from Sierra’s hardware completely and I have all my application developed in C on the Q26, then using another module from another vendor would mean major rewriting of the software. Otherwise, if I a use a uC to send AT commands to the wireless module, migrating to a different vendor might be easier since the uC code would be maintained, a big part of the AT command scripts could be transferable.