Hi (I use the STM32F107 and the Keil PLUS license)
I got the following challenge: My own developed HW card/application (let us just call it PCB) shall be operating in parallel with 10 other PCBs in the following manner (roughly): - The PCBs each have an Ethernet plug (runs UDP/IP) through which they communicate with a common PC (through a router) - The/each PCBs outputs shall be directly connected (230V DC) to each other (running in parallel operation) - An UDP broadcast message from the PC to the PCBs shall start activation of a xxx Volt DC output pulse (simultaneously generated from each PCB) - The starting point of the PCBs output pulses must not vary/deviate (all together) more that 100us (between the PCBs)
I have some questions regarding this: Q-1: Will this (strong(?) timing requirement) be possible to implement under the Keil-RTX OS when using the IPv4 Network MDK Middleware? Q-2: Will we have clock-ticks/clock-interrupts and task-switching which may destroy/disrupt this tough(?) timing requirement? Q-3: Is it possible to implement the IPv4 Network stack SW without the Keil-RTX OS - to ensure a tighter/faster timing performance (less timing deviations between output pulse generation for the different PCBs)?
I really appreciate answers to these questions. Best Regards Terje Bohler
Board level synchronization via IEEE1588 or a 1PPS?
There is no board level synchronization. Terje
My basic question is actually as follows: Q-1: How can I be guarantied a max time (delay) of (ca) 100us for (starting) processing an incoming UDP/Broadcast message? (the time from the UDP/Broadcast message arrives at the MCU-port, until my application starts executing the massage)
I then also have the follow-up Q's: Q-2: Can I use the KEIL Middelware (UPD/IP/Ethernet CMSIS functions) without the OS (RTX)?
I really appreciate comments to these important Q's. best regards Terje Bøhler