In my day job, I have the privilege of being the head of electronic design for Astrium Products Payloads in the UK. Primarily, we develop telecommunication processors for telecommunication satellites. Not surprisingly, these satellites require the very highest levels of reliability and use very high-grade components. We also implement complex redundancy architectures and perform extreme amounts of design analysis.
The components we use can be very expensive -- many tens of thousands of dollars for a one-time programmable (OTP) FPGA, for example. It is therefore crucial that we have a staged development schedule that retires risk as it progresses, ultimately resulting in a space-qualified product.
In fact, the way we do this is pretty much standard across most space engineering companies. The development is split into numerous phases, which we refer to as the "model philosophy." The most commonly used flow is as follows:
- Engineering Bread Board (EBB): This is the initial prototype and is only functionally representative of the final module. This allows the more complicated packaging design to be taken into account in the next phase. This EBB will use commercial components and will hopefully be the least expensive of the models. This model will employ, for example, commercial reprogrammable FPGAs as opposed to OTP space-grade FPGAs. Not surprisingly, this is the model that may well experience a number of revisions until the function is finally as required.
- Engineering Model (EM): This model is form, fit, and functionally the same as the final flight model; however, it will use a mixture of flight-grade components (usually spare or out-of-life in stores) and commercial components.
- Engineering Qualification Model (EQM): This is the model that is subjected to the qualification campaign. This usually includes design updates from the EM and uses identical components to flight; however, these components do not need to be screened to the flight standard.
- Proto Flight Model: This is a flight model that is subjected to the same qualification levels as the EQM. Sometimes, for cost issues, this is also flown.
- Flight Model: The models actually flying; these are tested to acceptance-level testing.
The ability to prototype in the lab -- be it testing a custom-designed PCB specifically for a task (EBB in the above philosophy) or a "lash-up" consisting of vero-boards and an engineer's attempt at soldering -- allows one to quickly determine whether a concept is viable... or not.
This is especially true when working with complex systems. Although modelling tools do provide a good indication of how the product should perform, there really is no substitute for getting hardware in the lab early and then learning on the hardware. This allows engineers to demonstrate to the customer and internal management that the required performance can be achieved. At the same time, with careful planning, it allows some risks to be retired early.
Of course, the prototype is just the start of the journey. You will have a number of design analyses and -- I am reasonably confident -- a re-spin or two of the board as you progress towards the final qualification of your product. This will have all started on your bench in the lab, which may seem a distant memory by the time you get to deliver the system.
How do you prototype products in your field? Do you even prototype, or do you use a version of the "suck it and see" approach?