Another opportunity for P2
Seairth
Posts: 2,474
http://www.altera.com/technology/system-design/articles/2014/motor-control-computer-architects.html
The article is primarily providing an overview of how control systems for PMSMs (permanent magnet synchronous motors) are evolving beyond the capabilities of simple MCUs. The premise is that SoCs are increasingly needed to control PMSMs. Apart from the fact that it was an interesting read, I saw this as an ideal opportunity for the P2.
The article is primarily providing an overview of how control systems for PMSMs (permanent magnet synchronous motors) are evolving beyond the capabilities of simple MCUs. The premise is that SoCs are increasingly needed to control PMSMs. Apart from the fact that it was an interesting read, I saw this as an ideal opportunity for the P2.
Comments
http://www.xmos.com/news/14-dec-2011/xmos-platform-simplifies-high-performance-multi-axis-motor-control-applications
There are all sorts of existing options that are suited for PMSM control, I'm sure. That doesn't preclude the P2 from being a contender (once it's released, admittedly).
As they say there, the devil is in the details.
One detail missing from their Block Diagram is fault sensing and dead bands in the Bridge drive. Easy to craft into a FPGA, but the P2 timers are going to need some help for serious Motor control.
Maybe Chip's smart pins willbe smart enough to allow one COG to control 6+ Bridge Pins, with hardware fault sense ?
The change from pure analogue to mixed digital processing occurred around 20 years ago. There would have been some rare digital cases earlier I guess.
Yup, another way to look at the servo control needs, is in the time-domain.
Most quadcopters are a great example of responsive real time servo, and they seem to manage fine with 2~2.5ms control loops.
Altera mention 16Khz trending to 100Khz, but that is the Bridge Switching frequency.
It is natural to try to then work at the Bridge Switching Frequency, but there is scope here for a 3rd time domain.
Full maths is not needed every PWM cycle, but single core controllers just find it easier to develop that way.
( and Altera, of course, will pitch a solution that uses many DSP cores, as that is what they sell )
A prop could work quite well in 3 time domains - an outer servo loop at the QuadCopter like ~2ms, an innermost PWM one, at the upper 100KHz, doing sampling and protection, and a mid-range one, maybe 100us~300us that manages the motor physics and angle extractions. 3/16 of a Prop 2
I think they are talking about the Bridge Frequency there, not the motor field currents.
The 16KHz -> 100KHz trend makes sense, as it allows smaller L's and C's, and IGBTs and MOSFETs are improving all the time. Likewise for Solar inverters.
A P2 could also make a nice merging Solar Micro-inverter, one where a number of panels feed a single switching controller. Uses multiple inverters to match each panel, without a package price adder on every single panel.
With multiple inverters you can also do clever things with the phase, to slash ripple currents.
["To allow seamless system expansion using off-chip memory, the STM32F446 includes a flexible memory controller (FMC) and dual Quad-SPI (QSPI) interface. The 90MHz FMC eases control of external SRAM, SDRAM, Flash NOR/NAND, or LCDs with embedded controllers, and supports memory-remap mode for enhanced performance. The dual QSPI can work independently from the FMC, and allows direct or memory-mapped connection to up to two external SPI NOR-Flash memories at single or dual data rate"]
This FMC hardware block also supports SDRAM.