As of January 2019, we have a perfectly working GSM+GPRS modem solution, currently realized in the physical form factor of a development board. Our current FCDEV3B modem works perfectly, running firmware that implements 100% complete functionality of a standard GSM modem, but with full source code freely published, and the only shortcoming of our current solution is its physical form factor: it is a development board, good for demonstration and evaluation, but not really meant for end use applications.
This situation makes it quite clear what our next step should be: the logical next step is to repackage our proven-good FreeCalypso modem solution into a different physical form factor that is more appropriate for end use applications. The Mother's preferred approach is to repackage our modem into the form factor of a compact SMT module that would directly compete against the mainstream proprietary ones from the likes of SimCom, Telit, Gemalto, Quectel and so forth — this approach would allow us to present our modem in its most pure and most native form.
As it turns out, one historical commercial manufacturer has already once produced a packaged GSM+GPRS modem module in the form factor we are thinking of, containing a TI Calypso chipset inside! The historical modem module in question is BenQ M32 (datasheet, pictures), featuring a slightly earlier version of the Calypso chip than the version used in our FreeCalypso solution, a triband radio front end and a Silabs Aero+ RF transceiver. The world's remaining supply of these historical BenQ-made modem modules is extremely scarce (we have about 40 pieces, and what we have is the world's last remaining supply), hence an effort to port our source-enabled FreeCalypso firmware to this historical BenQ-made hw would be rather pointless — instead the idea is to design and build our own newly-made FreeCalypso modem module inspired by BenQ's historical design, and then we could produce at least a hundred thousand pieces of our modem module before we exhaust the supply of chips and have to take the bigger cost step of getting the chips themselves cloned for us.
However, the specific proposal presented here is not to make a verbatim clone of BenQ's historical modem module, but rather to make a semi-clone: keep the same physical form factor (33.8x36.8 mm SMT module), keep most of the interface signals and most of the internals unchanged, but make a few deliberate changes compared to BenQ's historical version:
BenQ's modem brings out only an analog voice interface, no digital. Our version most definitely needs to bring out both analog and digital voice interfaces: the analog voice interface is already built into the core chipset and we are not removing it (artificially restricting functionality is not what we do in FreeCalypso), but we need to add a digital voice interface option. We are going to accomplish this feat by way of Calypso's MCSI: it is an auxiliary interface that can be used as a digital voice channel; all we need to do is to bring it out to module pins, which BenQ neglected to do in their version.
BenQ's modem is triband and so is our current FreeCalypso development board (FCDEV3B), but if someone funds the embedded modem module idea presented here, there is a very high likelihood that we will be able to make our new modem version fully quadband instead, by using the same quadband RF FEM component that was used in TI's Leonardo and E-Sample reference designs.
We will use the same version of the Calypso chip as used on our current FCDEV3B product, which is newer than the earlier version used by BenQ. It is only a matter of populating a different part onto the same PCB footprint, but
ourversion of the Calypso silicon has a newer DSP code in its built-in mask ROM than BenQ's version, and this newer DSP code features many important fixes and improvements from TI.
BenQ's historical design uses the Aero+ RF transceiver from Silabs, which is a 3-chip solution. The Mother's current plan is to use one of the newer single-chip GSM RF transceivers instead, either TI Rita or Silabs Aero II. Using TI Rita would involve much less work, time and cost (any of the Silabs Aero options would involve significant additional firmware work at very significant cost), hence the Mother's current development plan is to try to squeeze a Rita RF block into the available PCB space as our first-choice option, but if we are not able to fit it in, then fall back to Aero, either the original Aero+ from BenQ's version or Aero II.
Make it happen
Now comes the part which most people probably won't be too happy to hear: I, Mother Mychaela, am not in any position to build this proposed product with my own money, hence the proposed idea has any chance of becoming a reality only if someone other than me pays for it. If no one steps forth to fund this idea, then it will never become a reality — unlike my other planned project of building a handset UI development board, I will probably never be in a position to do this project with my own money.
How much money are we talking about? My estimate for the total cost to completion is around 30 kUSD, but that full amount would not be required upfront; instead the project will need money in incremental installments to pay for each step as the development proceeds. To get the development process started, we would need to receive 6 kUSD as the first installment, and we can deliver the finished product in 6 months from the moment that first 6 kUSD payment arrives, provided that subsequent additional payments to cover the cost of subsequent work steps arrive promptly when needed, without delay.
If you are interested and able to help this project, please email Mother Mychaela.