1. Torque oscillations at frequency LF
Torque oscillations at frequency LF*s (Milovan’s motor, J reduced to 20)
Unrelated oscillation
after reach full speed...
this one dependent
upon inertia etc
see LC circuit
in other PPT
0-speed
1*LF osc
half-speed
LF*s = 0.5*LF osc

2. q-d currents show same oscillation pattern at frequency LF. s
q-d currents show same oscillation pattern at frequency LF*s. (M’s motor, J reduced to 20)

3. a/b/c currents have dc offset
a/b/c currents have dc offset. The offset has initial transient decay (red) followed by oscillation(blue) that varies at frequency of (1-s)*LF (M’s motor, J=104)

4. Zoom-in on oscillating dc component 0-0. 8 sec
Zoom-in on oscillating dc component sec. This is the frequency of the rotor currents when viewed in the rotor reference frame.

5. Rotor current in rotor ref frame (corrected)

6. Locked rotor start - current shows no oscillation. as expected s
Locked rotor start - current shows no oscillation ...as expected s*LF remains 0)

7. Locked rotor start - torque oscillations remain constant frequency of LF (as expected s*LF remains LF). Also interesting that the oscillations die down much slower during LR start.

8. Krause’s equivalent circuit for transient analysis of SCIM
Krause’s equivalent circuit for transient analysis of SCIM. I believe that current will circulate in the loop shown at frequency (w-wr) = LF*s
w = ref frame radian speed *
wr  (Poles/2)* 2*pi* Rotor Speed **
Te=1.5*P/2*Lm*(Iqs*Idr-Ids*Iqr)
* Most convenient reference frame is synchronous
i.e w = we = 2*pi*LF. In that case Vqs and Vds are dc. (With proper choice of supply phase Vds = 0)
** note wr is NOT 2*pi*RotorSpeed, instead it is adjusted by number of pole pairs so that it matches we when
rotor speed is zero
Lamba_qs = Lls*Iqs+Lm*(Iqs+Iqr)
Lamba_ds = Lls*Ids+Lm*(Ids+Idr)
Lamba_qr = Llr*Iqr+Lm*(Iqs+Iqr)
Lamba_dr = Llr*Idr+Lm*(Ids+Idr)

9. Rough proof of claimed resonant frequency from previous slide
Assume the rotor currents in sync frame oscillates at some unknown frequency wn
Neglect resistive losses.
Linear system -> neglect other currents flowing in magnetizing branch (response is sum of responses)
We have loop currents Iqr*cos(wn*t+theta) and Iqd*cos(wn*t+phi)
Let s be derivative operator. Let Ltot = Llr + Lm
KVL around q axis rotor loop: Ltot * s*Iqr = - (we-wr)*LAdr = - (we-wr)*Ltot * Idr
Cancel Ltot and solve for Iqr
s*Iqr = - (we-wr)* Idr [1]
KVL around d axis rotor loop: Ltot * s*Idr = (we-wr)*LAqr = (we-wr)*Ltot*Iqr
Cancel Ltot and solve for Idr
Idr = (we-wr)*Iqr/s [2]
Plug Ird from [2] into [1]
s*Iqr = - (we-wr)* (we-wr)*Iqr/s
s^2*Iqr = -(we-wr)^2*Iqr
d^2/dt^2(iqr(t) = -(we-wr)^2*iqr(t)
This differential equation is satisfied by sinusoid of frequency (we-wr)
we - wr = s*LF is a resonant frequency associated with circulating current in the path shown in the previous slide
From equation [2] we observe Iqr =s*Idr/(we-wr) so we expect Iqr leads Idr by 90 degrees... that is exactly what we see in simulation (next slide).
Stator currents also oscillate at the same frequency because they share the magnetizing branch so their flux linkage is affected. They go along for the ride.

10. Iqr leads Idr