Freescale makes a 32-bit processor for the 8-bit crowd.
Well, it's not exactly how I spent my summer vacation but the week before my vacation, nonetheless. I attended Freescale's Technology Forum and learned a few interesting facts. Interesting, because at first glance, one announcement in particular seemed a bit off kilter.
The Flexis microcontroller family apparently offers 8- and 32-bit variants that are pin-for-pin compatible and work with a common set of on-chip peripherals and development tools. The 8-bit MCU is dubbed the MC9S08QE128, and it's based on the S08 core, while the 32-bit device, the MCF51QE128, is the first device based on the ColdFire V1 core.
Right away, my antennae went up. How can you run 8-bit code on a 32-bit processor and vice versa? Well, it obviously has to be recompiled. Not an automatic, no-brainer process, but definitely doable.
Next was the notion that the two parts can use the same peripherals. In some cases that makes sense. But in others, the 32-bit peripherals are designed to take advantage of the available 32 bits. The explanation I received was that the performance would be reduced but by a negligible amount.
Third, I asked how you could make the two parts pin-compatible when the 32-bit part obviously (or so I thought) needs more pins. The answer there is that some of the features were either disabled or multiplexed.
When I dug deeper into the story, it turns out that I was looking at it from the wrong angle. The answer is that if you're looking for a blazing, ultimate-performing 32-bit microcontroller, the Flexis family is not for you. The Freescale folks were quick to point out that there are many other Coldfire devices that easily fit that bill.
But if you are currently an 8-bit user, and you're looking for a higher level of performance (either now or at some point down the road), this may be just what you're looking for. The migration path is definitely clear and as painless as can be.
Or, another scenario is that if you're planning on building two versions of a product, one for the low-end and one with higher performance, you can pop the 8-bit MCU on one board, the 32-bit part on the other, and off you go. The example I was given is the security camera–the one built for the consumer with a $99 price tag can be populated with the 8-bit part, while the industrial grade camera, which can carry a higher price, can use the 32-bit MCU.
Once explained, it makes sense. I'm glad I had a chance to explain it to you.
Richard Nass is editor in chief of Embedded Systems Design magazine. He can be reached at .
I've just gotta say, I'm very impressed that Freescale (formerly Motorola Semiconductor) finally got a clue. So many other semi cpu suppliers have been doing this for some time (Hitachi / Renasas for example), it was bound to catch on. I'm not knocking them, but I've had a lot of experience in the HC05, HC11, HC08, & HC12 device development that just made me want to pull out my hair if I had to go between them, as none of peripherals were the same. They used to tell me that the different families were designed by different groups in different locations – and I just couldn't figure out why an 8-bit timer had to be so different from one device to the next. Thanks a lot, Freescale – for coming to the party just a little bit too late for me (I've moved on to other devices and architectures…