170227-1326 EST
To clarify my point 3 in the previous post.
When you look at wyerman's photos of the plots and look at a single common time period of 10 seconds, look at the current for each of the phases for this time period, then it appears the three transients are all about the same magnitude.
Look at different 10 second time periods and the amplitude of the transient varies from one time period to another, and appears to bo nonexistent in other periods. Is this logger instrumentation error or a real world variation.
I did some experiments with simple test equipment I have using a single phase motor as a load. This motor has no external mechanical load. Thus, its internal inertia and available starting torque determines how long starting current lasts. Basically the shortest time for this motor to start. This waveform is very consistent starting from 0 RPM for every cycle based on scope measurements. This was done with a Fluke Hall device probe and a Rigol scope. 100 mS to where the centrifugal switch opens, and another 20 to 30 mS to reach full speed and a little overshoot.
Waveform peak measured current was 40 A (4 V), this lasts for 6 cycles to where the centrifugal switch opens, then about 3 to 4 cycles to reach steady current with a peak value of about 4 A.
With an old, 60 years, Amprobe analog clampon meter set to the 40 A range I consistently read a peak value of 15 A. In this time range it is working like a ballistic galvanometer. On the 15 A range there was no way to see if there was overshoot. On the 15 A range the steady state reading was about 4 A.
In addition to the scope I added a 1N4148 diode and 1 ufd capacitor to provide a limited peak hold circuit across the Fluke Y8100 current probe. Two different shunt resistive loads were placed across the capacitor. The first resistance was 1 megohm scope input in parallel with a 10 megohm Fluke. In the second case the scope input was changed to a 10 megohm probe in parallel with Fluke thus equalling 5 megohms. This determines the discharge time constant.
I used both a Fluke 27 and 87 for measurements. The 27 peak hold seemed too slow in responxe, inconsistent results.. The 87 is labeled as 100 mS.
Neither the 27 or 87 did a good measurement in AC MIN-MAX mode on this duration of signal. The 87 did well on the diode-RC peak hold with 5 megohms as the discharge shunt. Note that using a simple diode and capacitor as a peak hold results in an error from diode drop. For several trials DC readings of 2.32, 3.24, 3.24, and 3.24 were read. At 100 mS the red curve is about 3.3 V, and 3.6 V on the longer time constant. AC line voltage could be the source of some variation. It appears the Fluke current probe saturates at about 40 A, and we are right at that point.
Following are the plots:
.
.
.
Blue is AC current, red is the peak hold current. Fluke scaling is 20 A = 2 V.
With the Fluke 27 on AC fixed range and MIN-MAX mode using the same Fluke Hall current probe as used with the scope I had more like 30% variation in the peak current. It was reading the equivalent of 5 to 8 A. Other meters may perform quite differently.
To try to sort out instrumentation errors from data one needs to know how their instruments function.
.
️can you help for a couple hours on Tuesday? Have need of your experience & intelligence thx !