There is a doubt about the S12ZVM_FOC_Sensorless model from NXP

[edenli]

Shown as below,how to deduce the id_frac_est = (ud_frac*K1_n + w_n*iq_frac*K2_n + ed_frac*K3_n + id_frac_est*K4_n)? 

I don't know how to deduce it,and the id_frac_est  is deduced form  EMF?

Original Attachment has been moved to: S12ZVM_FOC_Sensorless.slx.zip

[Daniel_Popa]

Hi Eden Li, 

Please allow me to begin by saying that i understand your doubts about this. I also had some hard time to understand it, just to realize that sometimes the comments and notations within a model may be misleading you. I think in this case that is the root cause.

Let us start from the observer equation highlighted below. This equation is written in a synchronous reference system gamma/delta.

 

The Simulink model is used Id_est/Id_obs and Iq_est/Iq_obs but if you go over this inconvenient and replace with gamma/delta the d/q quantities you will reach to the conclusion that the blocks id_frac_est and iq_frac_est implements the equation in the red box above.

The maths behind is a bit more complex - i do not pretend to be an expert - but for an easy explanation you can start from the the non-salient motor model - where you can express as:

For the salient use case where Ld!=Lq the model becomes:

Now, if you take this equation and solve it in current observing frame you will end up with a similar model as the one implemented in the Simulink example.

The K1,K2,K3,K4 are just the coefficients of i = f(u,i,w,e) of the derived equations that are basically f(Lq,Lq,Rs,K). Unfortunately, i do not have the original values used to derived the K1---4 coefficients used in the model but all that i can say those are strongly dependent on the motor you are going to use.

If you search on the web about salience observers you will find a lot of literature that will cover this subject.

Hope this will clarify some of your doubts.

Best regards,
Daniel

[edenli]

Hi Daniel,

Thank you for your reply,i wanna confrim some parameters such as ud_frac,id_frac_est,ed_frac,

My conclusion:

ualpha=ud_frac,we(Ld-Lq)=W_n*K2_n,ibelta=iq_frac,Ealpha=ed_frac,ialpha=id_frac_est,

I do not know if is right?

Best Regards!

Eden Li

[Daniel_Popa]

Hi edenli‌,

Sorry i did not had a chance to compute it by myself but your math looks about right.

Anyhow, keep in mind that the exact coefficient might need to be adjusted/tuned for the final model since it's pretty hard to guess all those motor parameters exactly. You also have measurements error that you need to consider.

From my past experience - and i think that is the reason why those K1-4 factors are hardcoded in the model - you need to adjust those parameters manually via Freemaster in real time until you reach a proper stable system.

It's quite difficult to compute those based on the motor manufacturer datasheet - just think about the inverter transfer function that is not accounted for the math model.

Best regards,
Daniel

[edenli]

Hi daniel,

Thank you for your reply!

Best Regard!

Eden Li