how is direction determined on AC traction motors?

[daveklepper]

The modern AC motors used in railway service are non-sychronous "induction" motors.  They do not have slip rings or commutators and do operate on the induction principle by use of the hystoresis effect.  In a synchronous induction motor, the rotating aluminum or copper bars are parallel to the motor shaft.  In railway motors, the bars are slanted.  There is always slip between the rotating field and the rotating bars, with the slip greater with increasing load.  The rotating field is slow at start-up, so the frequency of the AC current is low.

[gregc]

rdamon

For those of us who have lost the attention span for reading :)

thanks.    that certainly cleared up my understanding.

i found the following, which explained how multiphase sinusoidal currents are re-generated.

however, still curious about startup.   i think in particular if wheel slip can be prevented based on the frequency or if some other mechanism is needed at ~0 speed.   (there must be a youtube video that describes this)

[daveklepper]

Don't call it wheel slip.  It's magnetic-field-rotatitonal slip.  And although the staring frquency, cycles-per-second or Herz, is low, it is not as low as you might think.  Because in a three-phase non-synchronous, hystorises-induction motor, there are more than just three field coils acound the motor, as many as twelve or twenty-four.  So one cycle takes the magnetic field only one-fourth one-eighth of a complete rotation.

 

[gregc]

Dave

i do mean wheel slip.

AC Traction vs DC Traction describes an advantage of AC motors is that if the wheels slip the magnetic field slip (the angle between the stator and rotor fields) diminishes along with torque (thanks erik).

i think it's become clear to me that this can't work at 0 rpm, hence there must be some other mechanism (after all there is a computer)

daveklepper

Because in a three-phase non-synchronous, hystorises-induction motor, there are more than just three field coils acound the motor, as many as twelve or twenty-four.  So one cycle takes the magnetic field only one-fourth one-eighth of a complete rotation.

the fact that there may be more than 3 poles has been confusing me.   this is why i've been thinking of an AC motor as a stepper motor.    with such a motor, does the rotor also develop multiple magnetic fields?

[Overmod]

daveklepper

Don't call it wheel slip.  It's magnetic-field-rotational slip.

No, what he's talking about when he says that is wheelslip, in the sense of preventing it at starting.  It would be better to distinguish the railroad importance of prompt slip detection and remediation within a small effective degree of wheel rotation from the magnetic forces creating torque in the traction-motor rotor.

When I was younger, I assumed that a slip system would involve some sort of high-resolution means of detecting both angular rotation and absolute position (e.g. relative to fixed elements like poles in the motor structure).  There turn out to be ways to implement 'creep control' that don't depend on feedback from this (I believe the original Super Series 'creep control' is one such) and in fact something like a combination of monitoring true ground speed vs. individual motor (or inverter) power consumption could be used to determine the onset of wheelslip and compensate quickly for it.

What I think is being implicitly asked is how an induction traction motor using a rotating magnetic field inducing bars in the rotor handles developing wheelslip.  This might be done by quickly slowing the 'speed of rotation' of the field (which as noted could actually provide braking torque on a spinning motor and the wheel it is geared to) or by reducing the developed torque by reducing the peak stator current inducing the field.  In something like IGBT architecture these could be done very quickly and positively.  I would be interested to learn the actual practical methods used in locomotive practice, and what experience may have guided and informed that development.

[daveklepper]

There is some inherent wheel-slip control in the modern ac traction motor, in that the wheel through the gear-power-transfer system cannot cause the motor to rotate faster than the fields are rotating without, as has been described in a previous post, the motor acting as a breaking force.  Whether this is sufficieint wheel-slip control for all sitiuations is doubtful.

And yes, a modern AC railway (or bus)  motor has multiple magnetic fields.

[Overmod]

daveklepper

Whether this is sufficient wheel-slip control for all situations is doubtful.

It is not, and for essentially one of the reasons gregc was invoking.

If an inverter-synthesized drive is right at its adhesion limit and then slips under conditions that constitute a relatively persistent decrease in physical adhesion, something interesting happens.  The wheels will not wildly spin faster and faster -- but they will continue to rotate with considerable torque at the speed corresponding to 'field rotation' less slip.  As even under ideal traction conditions the coefficient of 'sliding' friction is lower than engaged, the wheel will happily keep turning until sufficient power is cut or field 'rotation' is slowed to where induced current in the bars drops or disappears.

I look at this as analogous at least to a vehicle like a '70s Eldorado with its automatic transmission.  The engine has a required idle speed, and with the transmission in even low gear there is a corresponding final-drive speed (absent torque multiplication this would determine how fast the car would move on the level with brakes released).  Now if you were to think that at idle and at rest the drive wheels won't turn -- put them on ice and they happily turn, not very fast but enough to utterly destroy steering integrity.  Switching to equivalent of neutral will quickly, and dramatically restore steering integrity...

[daveklepper]

Overmod, thanks for you excellent explanaition and analogy.

And practical DC motors (and AC-comutator motors, now obsolete) also have multiple magentic fields.  Count the number of field-coils and the identical number of armature coils on a typical EMD or GE locomotive motor.

Don't be confused by there being only two brushes contacting the commutator.  The armature coils are connected in parallel or series-parallel to the same pair.

 Direction of rotation reversal of a DC motor requires reversal of armature polarity with respect to field polarity.  With an electrically-produced field, reversal of the whole motor's polarity will not reverse rotation direction.  Thus, scale-model trains usually use DC motors with permanent-magnet fields.

[Erik_Mag]

daveklepper

 Direction of rotation of a DC motor requires reversal of armature polarity with respect to field polarity.  With an electrically-produced field, reversal of the whole motor's polarity will not reverse rotation direction.  Thus, scale-model trains susually use DC motors with permanent-magnet fields.

Dave's comment explains why "universal" motors work on both AC and DC. Lionel trains used used a motor with field windings instead of a permanent magnet, thus requiring a reversing circuit. Placing a bridge rectifier on either the field or armature (but not both) would allow polarity to control direction.

Note wrt variable frequency drive:

The use of variable frequency drive allows for getting good starting torque with a high efficiency (low slip) rotor. The low slip means that for a given drive frequency, the motor torque will drop rapidly with any increase in wheel speed. The experience with three phase traction motors on the GN, N&W and VGN showed that it was possible to get significantly higher levels of adhesion than what was possible with steam locomotives.

[daveklepper]

How such a rectifier works is illustrated.  The arrows represent diodes, formerly a kind of vacuum tube, today a solid-state device, passing current in only one direction.

Using such a device for the field and not the armature or the converse csn permit polarity revesal for a "universal" motor, if the source is DC.  If it is AC. the motor will try to reverse itself 120 times a second with 60 Hz house current.  Lionel understood that.

[daveklepper]

In short, applying rectification to either the armature or the field alone converts the uniuversal motor to a DC-only moior.

[Al DiCenso]

Absolutely not. It's the "direction" of the a.c. cycles against the rotor; in engineering technology, the control system "sends" the three phase a.c. wave form in either a-b-c or c-b-a direction, depending on whether forward or reverse is selected. Simple basic a.c. engineering stuff.

[daveklepper]

You are not discussiung the same subject that Overmod and I diverted to discussing: The universal comutator motors used in toy trains running on DC or single-phase house current.  He and I had already covered the control of multi-phase non-synchronous hysteresis-effect induction motors used in both diesel-electric and straight-electric railway and transit (including buses) equipment.

To finish up the toy-train reversal matter, with universal comutator motors that run on AC  and DC. my memory says that both Lionel and AC Gilbert American Flyer used sequence reveral relays, changing the polarity of the commutor with respect to the field.  Activate your "train-set," and the train goes forward.  Stop it, and when you start it the second time, it goes backwards.  So every time you make a station stop with your Lionel "train set." you have to "take slack!"

Is the control more realistic and sophisticated with toy trains today?

[daveklepper]

Oh, ususlly the sewquence-reversal relay was in the "coal car." (Most toy train buyers did not use the word "tender.")

Any complaints that this belongs on the Toy Trains Forum can be answered that Overmod and I are simply trying to be as complete as possible in answering the questions posed by this thread's originator.

[daveklepper]

And Al, the 3-phase AC motor that you refer to, NOT the AC-DC universal motor that Overmod correctly pointed could be made current-supply potarity reversel directional reversal (but for DC operation only), any application of any rectifier between the inverters under computer control and the field coils woild obviously make the motor inoperable, defeating the purpose of the inverters to begin with.  I suspect you simiply read the post previous to yours, without reading the whole thread, and assumed it was the 3-phase motor I was discussing.  I think there may be a lesson there. And yes, I have made that kind of mistake on a Kalmbach forum, possibly more than once.

[daveklepper]

I note that the video explained very well how the non-synchronous induction motor worked but did not explain the synchronous.  The explanaition is quite similar except that it is simply the lag between the fields produced in the rotating bars with respect to those in the stator that produces the magnetic fields and torque. Instead of slipping and falling behind in speed, the rotary never catches up to be exactly in step, but lags behind at the same rotating speed.  The video showed the slanted bars of the non-synchronous motor.  But Tesler's original invention (if memory is correct) was for the synchronous type with bars parallel to the shaft.

While a Senior at MIT, I had a part-time job as a transformer designer at Mystic Transformers in Winchester, at the same time as I was testing load-regulator-control performance on B&M GP-7s 1567 and 1568 for my SB thesis.  So commuting to Winchester wasn't a problem, with my B&M engine pass, and my Raleigh bike on the front platform of the B&M wood open-platorm commuter train.  I got to know the properties of various grades of iron used in transformers.  Obviously, a permanent magnet uses different iron aditives than an audio transformer.  Too much hysterises effect, and the audio trainsformer won't pass high-frequency signals, first distorting them, and then being useless as the frequency is raised.  At the same time, a high-frequency power-transfer transformer requires much less iron and fewer windings than a low-frequency power transformer.  Iron with a greater hysterises effect is not useful at high-frequencies.  So, I suspect that the iron used in the variable-frequency-driven motors now in use has to have a more limited and controlled hysterises effect than comutator motors, used for single-phase AC, 60, 50. 25, and 16-2/3 Hz and DC.  (And locomotives and rail cars that must draw power from 25 Hz as well as 60 Hz require transformers far heavier and larger than those running on 60 Hz only.)

When I had to design transformers for very-high-frequency applications (100,000 Hz and higher), iron available was not really up to the job, and hysterises introduced distortion.  I went back to my teacher at MIT, and he suggested introducing an air-gap in the magnetic structure, putting up with about a 20% loss in efficiency while drastically reducing distortion from really unmeasurable, most harmonics outside the range of my test instruments, to less than 2%.  The purchaser decided the solution was satisfactory for the particular application.

[Overmod]

daveklepper

... I suspect that the iron used in the variable-frequency-driven motors now in use has to have a more limited and controlled hysteresis [note sp.] effect than commutator motors...

This is an interesting point which I had not carefully considered in design detail. Of course the armature bars are intentionally designed to have zero remanent magnetism, and permag armatures designed to resist transient changes in magnetism right up to the effective Curie point with any transient hysteresis effects accounted for empirically, so the question relates to the stator architecture.  A relatively easy way to check is to see if the stators use some transformer-grade alloy (for example including silicon); another would be to see if they use a multiplicity of thin laminations.

I have a funny story about how the 'better' alloys for large transformers led to my predicting the 1987 market crash, in writing, to a former head of Morgan Stanley.  It would have made a good episode of the show 'Connections'... 

 

[Erik_Mag]

The stator on any rotating field machine needs to be made of laminations to prevent eddy current loss. Possible exceeption would be using ferrite or some other powdered high frequency magnetic material. The rotor on high speed synchronous machines is often made from a solid cylinder of steel with slots milled  in for the armature windings.

One advantage of a synchronous traction motor is higher efficiency and lower rotor heating at vey low speeds. Disadvantages are requiring an angular position sensor for the rotor and a dedicated inverter per motor.

[Former Car Maintainer]

This thread fails to mention several weak points of the AC motor as it applies to railcars operated by VVVF inverters. The VVVF inverters essentially create sine waves from square waves. The fast on switch-off switch of the square waves pound the stator insulation requiring continuous hipot testing for breakdown. Additionally the eddy currents between the rotor shaft and the rotor bearing casing result in bearing asperity, pitting and early bearing wear.

[Erik_Mag]

Correct, the high frequency currents from the switching frequency (and harmonics from the rapid rise/fall times), typically dictate the use of inverter rated wire. OTOH, the higher switching frequencies possible with SiC and GaN FET's should allow for some filtering, which would greatly reduce the high frequency currents.

[daveklepper]

Armature windings?    Synchronous and non-synchronous INDUCTION motors do not have armature windings.  But there are other types of AC motors that do have windings.  Which type are you discussing?

[Overmod]

Former Car Maintainer

This thread fails to mention several weak points of the AC motor as it applies to railcars operated by VVVF inverters. The VVVF inverters essentially create sine waves from square waves.

One point about synthesis is that the inverters can generate a pretty good shaped waveform when switched at 'modern' frequencies (even at power currents, this can be high frequency corresponding to good waveform in the 'rotating field').  Conventional methods of smoothing will also work, albeit adding expense and packaging complications to the final drive.

The fast on switch-off switch of the square waves pound the stator insulation requiring continuous hipot testing for breakdown.

Smoothing this would be almost childishly simple at each winding if it is an observed concern.  On the other hand, since this isn't an application where the square waveform has to be imposed on the power (as, amusingly, it is in NMRA/Lenz DCC model railroading!) using higher frequency combined with the natural inductance of the stator windings pretty well solves any risetime problems the modern insulation I'm familiar with would have difficulty with long-term.

Additionally the eddy currents between the rotor shaft and the rotor bearing casing result in bearing asperity, pitting and early bearing wear.

This is as childishly simple as proper third-brush installation -- I can say with great pride that I solved this problem for the Precor treadmill company in the early Nineties when they started experiencing disintegrating bearings in their early large AC-drive treadmill offerings.    A simple low-impedance connection around the bearings solves the issue definitively, as do certain other methods of neutralizing current across intalled bearing structure.

[Overmod]

Former Car Maintainer

This thread fails to mention several weak points of the AC motor as it applies to railcars operated by VVVF inverters. The VVVF inverters essentially create sine waves from square waves.

One point about synthesis is that the inverters can generate a pretty good shaped waveform when switched at 'modern' frequencies (even at power currents, this can be high frequency corresponding to good waveform in the 'rotating field').  Conventional methods of smoothing will also work, albeit adding expense and packaging complications to the final drive.

The fast on switch-off switch of the square waves pound the stator insulation requiring continuous hipot testing for breakdown.

Smoothing this would be almost childishly simple at each winding if it is an observed concern.  On the other hand, since this isn't an application where the square waveform has to be imposed on the power (as, amusingly, it is in NMRA/Lenz DCC model railroading!) using higher frequency combined with the natural inductance of the stator windings pretty well solves any risetime problems the modern insulation I'm familiar with would have difficulty with long-term.

Additionally the eddy currents between the rotor shaft and the rotor bearing casing result in bearing asperity, pitting and early bearing wear.

This is as childishly simple as proper third-brush installation -- I can say with great pride that I solved this problem for the Precor treadmill company in the early Nineties when they started experiencing disintegrating bearings in their early large AC-drive treadmill offerings.    A simple low-impedance connection around the bearings solves the issue definitively, as do certain other methods of neutralizing current across installed bearing structure.

[Former Car Maintainer]

My railcar maintenance experience in both primary and secondary repair, showed the DC motor brush replacement, flash overs, and commutator cutting was soon replaced by AC motor stator replacement, rotor bar separation and bearing asperity. An attempt to replace the bearings with ceramic coated races lessened the asperity but not completely. Stator breakdowns were diffcult to detect with the standard hipot tester. The leading edges of the VVVF inverter square waves could overshoot with 5kv spikes appearing across the stator. A purchase of a stator signature analysis tool...some kind of a magnetic sniffer...helped pinpoint shady stators.....a stator removal process requiring dry ice.....just sayin...

[Overmod]

If you were getting that level of spiking something a bit more agile might be needed to avoid 'bearing asperity'; I admit I'm surprised good ceramic race coatings wouldn't increase dielectric strength to where "achievable" current flow across the bearings would not produce sufficient plasma for microarcing. 

I take it there was insufficient adhesion or excessive loading to use actual ceramic rolling elements.  In my experience it was asperities on those, more than on the races, that led to the greater problems.

The problem with the spikes is inherent in your description.  That's a synthesis artifact, and a truly nasty one to be of that magnitude.  Sometimes very fast current risetime capability is NOT a good thing!

Likely you had someone build the 'megger' equivalent of a logic probe to be able to read those spikes at what might have been microsecond or even nanosecond duration... but with voltage high enough, and current following good enough... there would be your sparks across the bearing contacts...

Be glad you didn't have to use something like liquid nitrogen on the shaft... and preheating the stator assembly.  

[Former Car Maintainer]

The AC motor stator tester I described was only used after the rotor was removed while in the traction motor repair shop. It was a non destructive test method. There was never a hipot test placed across the bearings...but all in all the MTBF of AC motor railcars were 8 times or more than  that of DC motor rail cars...

[Overmod]

You have accurately put your finger on the likely issue, though, which indicates you knew more about both AC powertrain design and likely fault analysis than some of the 'engineers' from wherever those cars came from.

The ideal thing to do this testing today might even be one of the portable recording oscilloscope replacements, with an accurate time base reference so you could identify spikes even in the nanosecond range and associate them with various external conditions or control actions.  Be fun to identify what was causing the actual spiking and even more fun seeing what the builder would do 'bout it...

[Former Car Maintainer]

Ah yes...and many hours spent cutting wheel flats...

 

[Overmod]

Former Car Maintainer

Ah yes...and many hours spent cutting wheel flats...

You have my admiration.  Thankless out of all proportion to its necessity...

[Erik_Mag]

Former Car Maintainer

The leading edges of the VVVF inverter square waves could overshoot with 5kv spikes appearing across the stator. A purchase of a stator signature analysis tool...some kind of a magnetic sniffer...helped pinpoint shady stators.....a stator removal process requiring dry ice.....just sayin...

The overshoot sound like the product of a high frequency resonance in the stator windings being excited by the high dV/dt of the inverter. These resonances are due to the distributed winding capacitance interacting with the various ways inductance expresses itself in the windings. Having built a number of inductors and then measuring impedance as a function of frequency, it doesn't surprise that large motors would have these issues. Other fun part of designing/building/measuring inductors was running into proximity effect, which is one reason why special wire is needed for inverter driven motors.

Putting a filter between the inverter and motor can help, though you probably want at least a decade between the highest VVVF frequency and the filter cut-off frequency as well as another decade between the filter cut-off frequency and the pwm switching frequency. Large IGBT's tend to like witching at relatively low frequencies, which make filters impractical. The other issues is designing a high power high efficiency filter is non-trivial.