[PMSM Motor Control] [MCAT] Why do I get different power stage characterization curves depending on Vdc level?

[lucabarbiero]

Hello,

I am using the FreeMASTER based MCAT application for PMSM motor parameters identification. It was not available for the microcontroller I use (which is the MPC5777C, with motor control functions allocated to the eTPU), therefore I ported the target side software from the most recent reference project I found, to my project. I am using a custom HW board and inverter, therefore I need to do the power stage characterization in order to successfully identify the stator resistance Rs.

I do not understand why I get different characterization curves for different Vdc voltage levels applied to the inverter. In the picture below, I plotted the voltage error relative to the Vdc. As you can see it is different for 24 Vdc, compared to 48Vdc. Why is this the case? I also checked if the absolute error in volts was maybe the same (instead of the relative error) but no. Is this expected?

Thanks in advance for any help!

Luca

[petervlna]

Hi,

I have forwarded your questions to application team, if they can comment on your issue.

As it is application specific.

regards,

Peter

[lucabarbiero]

Hi petervlna,

Thank you for doing so. Is the application team going to review and reply to this thread?

After discussing this with my team we found out that it is probably related to the impact that PWM dead time has. For higher Vdc, the same inverter output voltage means a lower duty cycle, and for smaller duty cycles the dead time is more relevant, leading to a larger voltage error introduced by the inverter. In the screenshot I posted, voltage error Uerr is relative to Vdc, but the absolute Uerr increases for higher Vdc.

If this is the reason, it means that the power stage characterisation curve is only meaningful for a particular Vdc level. In our application, Vdc cannot be assumed constant, because it is provided by batteries. In general the Vdc available when we identify Rs will not be the same as when we derived the power stage characterisation curve (this is supposed to be done only once for a given power stage, by connecting a motor with known Rs according to the MCAT concept).

So does this mean that I need to derive different curves for different Vdc levels and then interpolate, to get the actual voltage error I need to identify Rs?

Thanks for supporting,

Luca

[petervlna · Accepted solution]

Hello Luca,

According to application team you have already answered it by:

After discussing this with my team we found out that it is probably related to the impact that PWM dead time has. For higher Vdc, the same inverter output voltage means a lower duty cycle, and for smaller duty cycles the dead time is more relevant, leading to a larger voltage error introduced by the inverter. In the screenshot I posted, voltage error Uerr is relative to Vdc, but the absolute Uerr increases for higher Vdc”

Regarding the MCAT version 1.1 which is actually available on NXP web the Rs is used for regulation parameters calculation, BEMF, observer, etc...

It is assumed that user already know the parameters (eventually he measured them)

For the resistance measurement it is possible to use MCAT control structure (Control Struct->voltage FOC) and monitor the current response on the voltage.

For the parameters measurement we can recommend you the application note:

https://www.nxp.com/docs/en/application-note/AN4680.pdf

If you want to use online resistance estimation and you want to tune the regulation parameters online, then you need to count also with converter parameters.

We do not offer algorithms for online parameters estimation.

regards,

Peter