[Huub3]

Dear all,

yesterday I managed to have my AC motor running for the first time. Although still on the grid, and just with standard parameters of my SJ300 Hitachi VFD, this was quite a nice moment.

However, I noticed that at low speeds (up to 2 Hz, but dominantly below 1 Hz), the motor was shaking quite noticable. This vibration has nothing to do with any mechanical unbalance, as it was not related to the rotational speed.

The motor (ABB 160 size frame, ca. 120 kg weight) was standing just on the concrete floor. No mounting, but also no elasticity, so the forces must be quite substantial.

Is this effect known or observed by someone? Can this be some internal movements of the coils?? What could be a resolution to counteract this? I am a bit concerned about this, as I want to avoid that I am internally destroying my motor.

At higher speeds (between 10-25 Hz), I noticed a higher pitch tone, seems also not related to the rpm. I assume this is the 6 kHz switching of the VFD, not that concerned about that right now.

On a positive note, I opened the motor (pretty easy with a AC motor), pictures will follow. I had purchased this motor, with Dahlander wiring, with the intention to get the internal start connection brought outside, and use this to go to a delta parallel switching, allowing higher base frequency.

I was very happy to discover that this internal star point did indeed exist -)), and that it was reasonably simple to reach. After some hacking on the resin soaked binding ribbons (please do wear safety goggles, that stuff is knife sharp!), I was able to free the connection, so now I am up to make a good connection to the outside, and start experimenting with the different switching opportunities.

Happy to hear your remarks and suggestions on the vibrations, regards,


Huub

 

[Coulomb]

Dear all,

yesterday I managed to have my AC motor running for the first time. Although still on the grid, and just with standard parameters of my SJ300 Hitachi VFD, this was quite a nice moment.

Yes, I've been there recently; it's great!

However, I noticed that at low speeds (up to 2 Hz, but dominantly below 1 Hz), the motor was shaking quite noticeable.

I believe that low frequencies is where AC controllers really need position feedback to prevent vibration. Are you running open loop (no position encoder)?

 

[Coulomb]

... and just with standard parameters of my SJ300 Hitachi VFD

That could also be a big part of the problem. In other words, if you don't have an encoder, maybe you can still get away without one.

I think it's time for an auto tune, so the controller learns the characteristics of your motor.

 

[Coulomb]

See for example this post:

http://www.diyelectriccar.com/forums/showpost.php?p=226800&postcount=70

etischer tuned his VFD (ok, it took him a year) to work very well sensor-less.

 

[Huub3]

Coulomb,

thanks for your quick reply!

Would you think that these vibrations do come from some superposed rotational ripple on the 1 Hz rpm? So that it still is mechanical vibrations of the rotor, that just are not visible.

I want to avoid that I am shaking the coils into pieces, by some effect that I do not yet know.

Indeed the motor was running open loop, without encoder, and without auto-tune.

After I opened up the motor, and uncovered the star point, I became aware that I will not be able to run the motor until I securely bind the coils and wiring together again. A bit too enthusiastic, and a bit stupid, as indeed I noticed that I should have done an autotune before the modification, to learn the parameters and write them down.

Well, one learns on the go so to speak.

Any idea where to get this binding ribbon, and how to use it.

Furthermore, are you guys using an encoder? I might use an encoder (with some additional interface board on the VFD), but was unsure what type and operation to use.

Regards, and thanks again!


Huub

 

[Coulomb]

Would you think that these vibrations do come from some superposed rotational ripple on the 1 Hz rpm?

Yes, surges of current and hence torque.

A bit too enthusiastic, and a bit stupid, as indeed I noticed that I should have done an autotune before the modification, to learn the parameters and write them down.

Well, the auto tune needs to be done on the final configuration(s) of the motor. For star/delta, you will need to change from one motor configuration to another (different V/Hz, for example), on the fly. So you'll need to do a "flying start" (different controller manuals have different names for this). I was reading yesterday about some experienced people that said they never got that to work.

We were going to do star/delta switching, before we decided to retain the gearbox. It sounds like you might have windings to switch from series to parallel or something. How many contactors will you need for that?

Furthermore, are you guys using an encoder?

Yes, we are. We decided that for the cost that they are (surprisingly high for such a small device, but cheap compared to the motor+controller), it would be worth the expense and hassle of wiring and mounting. Also, while an encoder was optional for our original industrial controller, it is mandatory for our current controller (a Tritium WaveSculptor 200).

 

[etischer]

If you are running in sensorless vector (or with encoder feedback) and the speed is unstable at low rpm, you might try running in the simpler "volts/hz" mode. V/Hz mode will run the motor without any speed or torque corrections, and will probably be smoother. You can't really drive an EV in V/Hz mode, but it is useful for troubleshooting.

You might check for noise on the encoder line with a scope. I assume the encoder feedback direction matches the motors direction?

 

[Huub3]

Eric, Coulomb,

thanks for your replies and good tips.

Related to the wiring of the motor, I was not planning to do switching. I want to go direct drive (on a BMW E36 (most likely) or Mercedes C-class) on the rear-wheels.

For this I need to increase my base speed at full voltage (or with other words, reduce the voltage V/Hz, for the 50Hz).

That is the reason I stepped into buying this Dahlander motor. If you look at the high-speed circuit:

you will notice that there is an internal star point (look at the lower right picture) in the high speed parallel star wiring.

By opening up this point and bringing the three phases outside separately, I can wire the motor in parallel delta circuit.

This would allow to get to 50 Hz. with ca. 230 Vac, meaning that base speed at 460 Vac would go to ca. 100 Hz. This way I would have max. torque till ca. 2900 rpm, instead of till 1450 rpm, making it feasible to go direct drive (I hope).

This is similar to the rewiring experiments done by a.o. 4x4kiwi and acmotors at the AUS forum. The dahlander circuit has the benefit that you do not need to rewire individual coils, but rather only have to look for the internal star point.

I will look into an encoder, but indeed those critters are awfully expensive compared to their size.

Regards,


Huub

 

[etischer]

Most industrial motors have windings wired out to the junction box allowing 230 or 460 volt operation. Is your motor originally 460v and you are converting it to 230? Could you do two deltas instead of two stars to reduce the voltage to 133?

 

[JRoque]

Hello Huub and congrats on getting your motor started!! What was your SJ300 HP/kW rating again? Are you running off standard 220V mains or 3 phase input?

etischer, are you implying that a 9-lead, Wye/Delta, 460/230 motor can be configured to further cut it's voltage to 133V? I'm all ears because it would certainly solve a lot of issues in my planning.

JR

 

[Huub3]

Hi Eric,

that is an interesting thought. Indeed I am converting from 460 V to 230 V. I am not aware of having more windings outside (the motor in standard dahlander form has only 6 leads, I am going to add another three, but not sure this is the same as the 460/230 wiring). Could you explain the double delta winding a bit more, I seem to miss an important point here.

A picture of the 460/230 Wye/delta 9 lead wiring would be very helpful I think, I am going to look for such a schematic.

On the other hand, this is an EU motor, and I somehow vaguely recall this 460/230 windings thing to be a US system, to cope with both 1 and 3 phase?

JRogue: my VFD is the SJ300-550 HFE, so nominal 55 kW, and max 75 kW. Should be sufficient for a midsize car. I am running now from 3 phase mains, this VFD is not suited to run on 1 phase unfortunately.

Looking at my current VFD, going to a lower voltage would most likely not give much additional value, as I now am just before hitting the current limit (by sheer luck, I would lie when saying I designed it like that). But, for the future, a 133 V would open up using a real EV-VFD like the wavesculptor.

Regards,


Huub

 

[etischer]

I don't know if your motor will run wired as a delta, but if it can, you could possibly lower your voltage by another 57% or 1/sqrt(3)

 

[etischer]

A picture of the 460/230 Wye/delta 9 lead wiring would be very helpful I think, I am going to look for such a schematic.

http://www.joliettech.com/3ph_motors_single_speed.htm

Pics from link above

 

[Coulomb]

Most industrial motors have windings wired out to the junction box allowing 230 or 460 volt operation. Is your motor originally 460v and you are converting it to 230? Could you do two deltas instead of two stars to reduce the voltage to 133?

I think that's what he's doing now.

It's a 460 V motor, with all winding in series star. By putting it in parallel star, it becomes a 230 V motor. By getting at the internal star point and converting to parallel delta, it becomes a 230/√3 = 133 V motor.

[ Edit: This is wrong! I was confusing Dahlander connections, which change the motor from 4-pole to 2-pole, with series/parallel connections. The Dahlander parallel star configuration (and parallel delta also) are both inverse parallel. In other words, the windings are in parallel, but with the opposite polarity to standard, so that the windings oppose each other. I don't pretend to understand this, but I know enough to know that the above is wrong. Dahlander configurations are always for a fixed voltage, so the "high speed" connection is still 460 V. ]

So if it's a 4 pole with nominal 1450 RPM @ 460 V in series star, it becomes 1450 RPM @ 230 V or 2900 RPM @ 460 V in parallel star. In parallel delta, it becomes 2500 RPM @ 230 V or 5000 RPM @ 460 V in parallel delta.

Related to the wiring of the motor, I was not planning to do switching. I want to go direct drive

Alas, direct drive is where switched windings (series/parallel or star/delta) becomes useful. It gives a kind of "electrical gearbox" for a vehicle that no longer has a mechanical gearbox. Hence, we were planning on a star/delta switch for our MX-5 when we were thinking of losing the gearbox.

This becomes more important if your motor or controller are not very powerful. 75 kW peak might be marginal for direct drive. Remember that for direct drive, you need the torque and current at low speeds for acceleration, hill climbing, and gutter climbing, and sheer speed at the top end. By making the motor only get to full power at 5000 RPM, as this rewind will do, you'll get really great performance at the top end (full torque to 5/6 of top speed, assuming you balance the motor for 6000 RPM), but taking off/hill climbing/gutter climbing will be like doing thise things in top gear in a V8. It will do it, kind of, but how many V8 cars don't have a gearbox?

You will be able to get to all the connections, so you will be able to do series/parallel switching (equivalent to about 2nd and 4th on a 5 speed manual transmission) with 9 wires and a box full of contactors.

See for example here: http://www.aeva.asn.au/forums/forum_posts.asp?TID=1574&PID=19215#19215. In fact, TJ seems to be implying that you didn't need to open the motor to get to the internal star point! I'd have to think about how that is done. I'm pretty sure it is on the AEVA forums somewhere.

Series parallel switching is more important for the Civic, as it has an ironless radial flux motor; these can't be field- or flux-weakened.

For direct drive, I think it would be better (if you don't do series/parallel switching, or even if you do) to have the speed where you run out of voltage (not really "nominal speed" any more; I suppose it's "new base speed") about the middle of the motor's speed range. Then each winding gets the maximum current your controller can supply, so the motor has twice the torque up to half speed. After that, you get approximately constant power from new base speed to maximum speed, effectively giving you a gearbox. So then the average torque and hence average power is higher when accelerating from stop to maximum motor speed. This tips the balance towards low speed driving, where you need the most torque, and you spend most of your time. You still get nearly full power available on the freeway.

Sorry to contradict your assumptions, but I think that a bit of a rethink may be in order.

 

[Coulomb]

I am running now from 3 phase mains, this VFD is not suited to run on 1 phase unfortunately.

All the larger industrial controllers are designed for 3-phase input, but they just convert to DC internally. We modified our industrial controller (415 V 3-phase nominal) to work on single phase 240 V converting the input to a half wave voltage doubler:

http://www.aeva.asn.au/forums/forum_posts.asp?TID=1237&PID=14128#14128

Since it's only half phase, you can't use it for serious power, and the external surge protection is a pain, but it's convenient for testing if you don't happen to have 3-phase mains available.

 

[etischer]

It's a 460 V motor, with all winding in series star. By putting it in parallel star, it becomes a 230 V motor. By getting at the internal star point and converting to parallel delta, it becomes a 230/√3 = 133 V motor.

I think that's what you meant to say above?

Switching windings on the fly is going to be pretty tough, I wouldn't count on this working. There may be some reason to do it as a soft start when running a motor across the line, but a VFD controlling torque??? I won't say it's impossible, but I don't think it is necessary.

 

[Coulomb]

I think that's what you meant to say above?

Yes, sorry, and thanks for pointing that out. Edited now.

Switching windings on the fly is going to be pretty tough, I wouldn't count on this working. There may be some reason to do it as a soft start when running a motor across the line, but a VFD controlling torque??? I won't say it's impossible, but I don't think it is necessary.

Well, if you want to "change gears electrically", you have to do it. When we were planning on doing star/delta switching, we fortunately had the ear of an expert to ask many questions about, and we paid extra for the fancy controller software option. We would not switch under load, of course; power would be ramped to zero for the tens of milliseconds of the switch. It could be disconcerting if you have the pedal to the metal at the switch point.

I believe that the Civic with the axial flux motor does this (or at least plans to do it soon).

 

[Coulomb]

See for example here: http://www.aeva.asn.au/forums/forum_posts.asp?TID=1574&PID=19215#19215. In fact, TJ seems to be implying that you didn't need to open the motor to get to the internal star point! I'd have to think about how that is done. I'm pretty sure it is on the AEVA forums somewhere.

It is here:

http://www.aeva.asn.au/forums/forum_posts.asp?TID=585&PID=13333#13333
From that post:

Six changeover contactors required. Of course, with no change from star to delta, access to the star point is not required.

However, the above diagram requires 6 connections, not 9, though none of these are the internal star join point.

The problem is that a Dahlander configuration (needed to get the number of terminals down to 6) has windings opposing each other in one of the configurations. This does not allow the full potential of the motor to be made available, if my understanding is correct.

 

[Coulomb]

http://www.joliettech.com/3ph_motors_single_speed.htm

Pics from link above

The above is of course a 12-terminal star/delta change. I don't think it is possible to do star/delta with only 9 terminals.

 

[Coulomb]

Huub, My understanding is that you want about a 2x over-frequency motor (2x over-voltage). What you want therefore is a parallel to serial change, but not a star to delta change. So what you want is the high speed configuration as per your diagram, with one important change. That change is you want to invert the polarity (exchange the wires at each end) of one set of windings. So instead of inverse parallel, you will have normal parallel windings.

[ Edit: I very likely have this wrong. It seems that it is the series star mode where the polarity is inverted. So the "centre tap" isn't really a tap, but a place where opposite ends of two windings join. ]

I agree that a rewiring to 133 V is too much for you, since you will be current limited by your controller. What you were proposing with the parallel delta winding will give you either a 266 V 2-pole motor or a 133 V 4-pole motor, depending on the polarity of the windings compared to each other (I hope I got that correct). A 230 V 4-pole motor is what you want, and you get that with normal (non-inverse) parallel star wiring.

If when you were testing you were in a Dahlander configuration (e.g. the low-speed [ Edit: was high-speed ] configuration in the diagram), then it will have had less than ideal (not very sinusoidal) back EMF, which at low speed could cause torque surges. So that could also have been part of your "vibration" problem. By connecting in a more standard pattern (e.g. straight 4-pole), hopefully the back-emf will be more sinusoidal, and this may improve the vibration problem at low speed.

It may seem strange to talk about the "polarity" of an AC winding, but in fact the relationship of the two windings is important. Think of a brushed "universal" (AC/DC) motor. It will only turn in one direction, whether is is fed DC, opposite polarity DC, or AC. However, if you reverse the connections of the field but not the armature (or the armature and not the field, but not both), the motor will now turn in the opposite direction (again with either polarity of DC, or AC). In a similar way, how the two windings are connected makes a big difference to the Dahlander motor. The magnetic fields of the two windings can either oppose or assist each other. Further detail than that I don't know, except that in one configuration, you end up with a 4 pole motor with 2 wound and 2 consequent poles, or a 2-pole motor with all poles wound.