Improved stability of sensorless control under difficult operating conditions such as rapid acceleration or step load change.
Reduced the default PWM frequency from 47 kHz to 32 kHz while at the same time reducing the state estimation errors thanks to new DSP architecture. This change reduces the switching losses and thus lowers the heat dissipation in the inverter.
Implement feed-forward injection in the velocity PID controller. The new architecture is as follows:
target velocity --> [ramp] -+---> [PID] ----> current ---> (+) --> current
| ^
+-- [Feedforward] --> current --^
Address the spurious motor temperature warning problem introduced in v0.4.
When migrating configuration parameters from an earlier revision, do not migrate the PWM frequency setting (vsi.pwm_freq); keep the default value of 32 kHz. Higher PWM frequencies are not supported starting with v0.5.
The new PID+FF tuning helper is implemented in this spreadsheet (copy to edit):
Parameters for some of the commonly used propellers are provided in the below spreadsheet:
The motor parameters are likewise available here:
The spreadsheets are currently sparsely populated, but we are going to be extending them over time. You can ship your hardware to our lab for characterization; our address is:
Zubax Robotics
Akadeemia road 21/1, Room 356, Tallinn 12618, Estonia
+3726891919
Great news!
I’m upgrading from v0.3. How do I fill in the new parameters?
Do I have to run the motor identification again to get those filled in?
In @pavel-kirienkopost he mentions how critical the accurate measurement of the motors parameters is to the stability of the operation under high loads.
We have experienced issues with instabilities under high load with our motors.
I’m not sure if this post implies that the standard motor identification process is not accurate enough and the motor needs to be accurately measured with an LCR in the lab.
Do you have plans to get the T-Motor MN501S 300KV tested for PMSM parameters?
The new parameters can be computed with the help of the linked spreadsheet. The other ones can be simply exported into a file before the firmware update, and then imported from the file afterwards.
This shouldn’t be necessary, as there were no significant changes to the motor characterization algorithms in this release.
The motor identification methods implemented in Telega are accurate for most cases, but they occasionally produce incorrect results in some marginal scenarios (hub motors of micromobility vehicles are one such case). If you experienced stability issues, this might indicate that the motor parameters require adjustment, so an LCR meter would help.
We have an extensive collection of various motors, but the one you listed is, unfortunately, not in it. If you shipped one to our lab (preferably with a propeller) we would run the system identification on them and update the spreadsheets.
I was wondering if this update included the ability to set active braking when the setpoint/CAN command/PWM is set to 0? It was also discussed in this post: Electric brake for VTOL. Is there a parameter which could be set in Kucher with this new update by any chance?
This is really useful for eVTOL and forward flight vehicles and we’re currently using the Myxa ESCs in a VTOL configuration.
Hi Zen, thanks for the info! I was wondering if a version of Telega v0 may have the electric brake feature sometime in the future by any chance? We’re currently using DroneCAN since that’s what our components are programmed with, so I think we would need Telega v0 with the braking feature if that might be available. Does Telega v1 only work with Cyphal?
