Skip to content

separate hardware read and write #12

@shadowmanos

Description

@shadowmanos

There is a function void EthercatHardware::update(bool reset, bool halt) that does both read and write in Ethercat. The real-time loop should be:

while(true) {
read from sensors
update controllers
write to actuators
wait until 1 msec has passed
}

But what happens instead is

while(true) {
read from sensors
write to actuators
update controllers
wait until 1 msec has passed
}

That means when a discrete-time PID controller is tuned it must assume a sample-period of 1msec and a latency of 1msec. That's because the controller action calculated for current sensor readings will actually be applied in the next iteration. That means the tuning has to be less aggressive (less responsive control) compared to almost zero latency. Split the EthercatHardware::update to read and write function.

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions