http://www.projectmechatronics.com/2008/03/02/e-drives-conference/ The synergistic application of interdisciplinary engineering fields. Thu, 03 Jul 2008 23:06:18 +0000 http://wordpress.org/?v=2.5.1 http://www.projectmechatronics.com/2008/03/02/e-drives-conference/#comment-9 Erlendur Kristjansson Thu, 06 Mar 2008 21:19:47 +0000 http://blog.projectmechatronics.com/2008/03/02/e-drives-conference/#comment-9 The sad thing about the DSP craze in the motion control industry is that 90% of the things that are handled by a motion controller have nothing to do with signal processing. The only reason for such high usage of DSPs is that until in that last few years we have not had any options. DPS’ have been the only show in town and companies like TI have cashed in on it. Actually, most of the control tasks that happen in a motion controller are better handled by a microcontroller. It is possible to say that DSPs are good for contouring calculations, but again there the reason is that they are available with fixed point or floating point capabilities at price points that are reasonable. There still no true signal processing going on. DSPs are optimized to handled big blocks of data or data streams, and do complex calculations on it. This is not what motion control is about. Fortunately, as Steve Meyer has pointed out, there are options out there. Luminary was the first to introduce a microcontroller based on the ARM CortexM3 core, which is a very powerful core optimized for microcontrollers. Last year STMicroelectronics also introduced a product family based on this core, with even higher performance. Both have made their products very attractive to motor control. Also, compared to a lot of DSPs out there this micros can run at their maximum clock speed from flash memory and not RAM, which is the norm for most DSPs. Luminary and ST run at 50MHz and 72MHz, respectively, giving us 62 MIPS and 90 MIPS performance. More than enough for the most demanding motion control tasks at a very attractive cost (at least compared to the DSP ;) The sad thing about the DSP craze in the motion control industry is that 90% of the things that are handled by a motion controller have nothing to do with signal processing. The only reason for such high usage of DSPs is that until in that last few years we have not had any options. DPS’ have been the only show in town and companies like TI have cashed in on it. Actually, most of the control tasks that happen in a motion controller are better handled by a microcontroller.
It is possible to say that DSPs are good for contouring calculations, but again there the reason is that they are available with fixed point or floating point capabilities at price points that are reasonable. There still no true signal processing going on.
DSPs are optimized to handled big blocks of data or data streams, and do complex calculations on it. This is not what motion control is about. Fortunately, as Steve Meyer has pointed out, there are options out there. Luminary was the first to introduce a microcontroller based on the ARM CortexM3 core, which is a very powerful core optimized for microcontrollers. Last year STMicroelectronics also introduced a product family based on this core, with even higher performance. Both have made their products very attractive to motor control. Also, compared to a lot of DSPs out there this micros can run at their maximum clock speed from flash memory and not RAM, which is the norm for most DSPs. Luminary and ST run at 50MHz and 72MHz, respectively, giving us 62 MIPS and 90 MIPS performance. More than enough for the most demanding motion control tasks at a very attractive cost (at least compared to the DSP

]]>