Francisco's Electric Pajero

[francisco.shi]

Finally got the last phase in.

Top view all phases. (Inverter side)

20200617_175535.jpg (1.91 MiB) Viewed 1491 times

Bottom view. (Shaft side)

20200617_180031.jpg (1.69 MiB) Viewed 1491 times

Now the wires line up with the IGBT module output tabs.

Output terminals and IGBT module.

20200617_175820.jpg (1.73 MiB) Viewed 1491 times

All I need to do now is join the neutrals, tie up all the coils with string, apply varnish and crimp the output terminals.

[brendon_m]

Looking good.
I'm very impressed, not to mention interested to see how it turns out

[francisco.shi]

I am confident it will work as expected.
I have already tested the power stage with a single IGBT module to 400A 380v with no problem at all. For this winding i will have two IGBT modules so the load on each module will be the same on each module as already tested. The current sharing will be pretty close to half because each module is driving its own winding.
The output torque will be the same as before. The only difference is it will spin to twice the speed as before before it starts dropping in torque. So it should roughly give me double the power.
As for the top speed I will still run within the motor specs.

[francisco.shi]

I finally got the motor almost finished.

Inserting wedges.

20200628_104416.jpg (1.66 MiB) Viewed 1317 times

Wedges all in.

20200628_115959.jpg (1.7 MiB) Viewed 1317 times

Windings tied.

20200703_001031.jpg (1.8 MiB) Viewed 1317 times

Other side windings tied up.

20200703_001018.jpg (1.6 MiB) Viewed 1317 times

Will post more pictures later of the assembled motor.
Now just need to put the terminals to the transistor pack.

[jonescg]

Looks legit! I always wanted to try winding a motor for the learning process.

[francisco.shi]

The more I work on it he easier it gets.
Here is a video from Indian university explaining winding of electric motors.

Reminds me of a lecturer I had at uni.

[francisco.shi]

I did a bit more work on the motor today. I put the terminals and installed the power stage from the inverter.

Motor with inverter output stage installed.

20200704_191804.jpg (1.36 MiB) Viewed 1282 times

Another view.

20200704_192049.jpg (1.33 MiB) Viewed 1282 times

The next step was to measure the back EMF and compare it to the old motor.

Old winding response at 1110rpm

20200704_183844.jpg (1.33 MiB) Viewed 1282 times

And here is the new winding.

New winding response at 1095rpm

20200704_184210.jpg (1.35 MiB) Viewed 1282 times

The old winding was short pitch wound and the new is full pitch wound.
The difference is with short pitch the 3rd and 5th harmonics are almost eliminated. You can see that the old winding produces a perfect sine wave. The full pitch winding has the disadvantage of having more harmonics but has the advantage of using the full lux from the magnets which will result in slightly more torque capability. You can see the output wave is not as cleaned. The peak of the wave has a small lump.
And the last thing to confirm I got the two parallel windings wound with the correct phase I connected two terminals on the same phase together and measured the voltage between the two terminals on another phase and spun the motor. If I got it right there should be no voltage between them. And here is the result
(See next post. I couldn't add so many files)
I have a Hyper9 motor I am making an adaptor for so I though I should do the same test and see what it is like. I was curious as to the back EMF constant (volts/rpm) and here is the result.

Hyper9 response at 1117rpm

20200704_193526.jpg (1.36 MiB) Viewed 1282 times

The Hyper9 is a 4 pole motor so the frequency is lower.
It is difficult to say anything good about this motor without having to lie.
The harmonics on this thing are horrendous. No wonder they overheat.
The other problem will be trying to measure the angle of the current when doing the Parke transforms. I can't imagine how hard it must be to get the inverter to work correctly.
Here is a side by side view of the two motors.

Hyper9 vs my motor comparison.

20200704_193954.jpg (1.56 MiB) Viewed 1282 times

[francisco.shi]

Here is the voltage between the two parallel windings.

Voltage difference between two parallel windings.

20200704_184457.jpg (1.29 MiB) Viewed 1283 times

As you can see it is negligible.
I am quite happy with the results so far.
Now I need to finish the logic board and test it and compare it to the original winding.
I am pretty confident the results will be as expected.

[coulomb]

And the last thing to confirm I got the two parallel windings wound with the correct phase I connected two terminals on the same phase together and measured the voltage between the two terminals on another phase and spun the motor. If I got it right there should be no voltage between them.

Neat. Was this before closing up the motor and crimping the final terminals?

Really interesting work, thanks for sharing.

[francisco.shi]

I had to assemble the whole motor including crimping the neutrals which are inside so I could spin the rotor inside the stator to get a voltage waveform.
The inverter has two IGBT modules in parallel in order to get the current capacity. But to make things more effecient I run one winding per IGBT module. This way the winding resistance acts as a sharing resistor. This allows the maximum current rating of the IGBT to be used.
IGBTs have negative temperature coefficient which means I can not just connect them in parallel or one will start taking more current than the other and eventually blow up. To get around this problem you can go several ways. One is derate the current capacity of the pair to less than double of each device. Another is to get match devices (unlikely to happen unless I get them direct form the manufacturer) and the last one is to add a resistor in series with each device. The resistor will generally help to keep the current thru each device shared more evenly. The problem with the last method is you increase losses. What I have done us use the winding resistance as sharing resistor.
The tricky part was that I had to wind one winding back to front so that the terminals ended in the same spot. If I had wound them the conventional way I would have ended up with the terminals of each of the parallel windings on opposite sides of the motor (which would have been very inconvenient for packaging)

[coulomb]

But to make things more efficient I run one winding per IGBT module.

Duh! 6 wire bundles coming out; I somehow missed that. And of course you told us about the separate windings earlier. Thanks.

[francisco.shi]

I have been doing some work on the inverter logic board.
I am using an SAME70 Xplained development board.
The processor seems to be perfect for the job. It has all the things needed in the one chip. Unfortunately documentation isn't great. The datasheet is not detailed enough to explain how things work. It tells you it has the features but it doesn't tell you how to use them clearly. There are libraries and examples on how to use the different periferals but they have been written with the idea of hiding the complexity but this really defeats the purpose of an example. The libraries have no documentation at all. I have to search thru several layers of code in the examples just to get an idea of what it is doing. Things like the example uses a pin to flash an LED but instead of writing what pin the LED is on it has 3 levels of #define so the pin name is LED 0 instead of the port number that the LED is connected to.
Another problem is the examples have so much stuff added to them with no comments that give you useful information that it is very difficult to follow what is going on and what all the registers do.
Anyway, here is a video of the motor running.

It is only controlling the magnitude of the current but it is mainly to test that I have all the basic routines working.
So far I can read the resolver angle, current magnitude and angle (using 2 current sensors) have implemented the output waveforms to drive the 6 PWM signals that go to the gate drive including dead band and watchdog timer that turns off the gate drive in case the processor gets stuck.
The motor in the video runs by using the angle from the resolver to set the angle of the 3ph sine waves. The angle of the resolver is 90° to the magnets on the rotor. The 3Ph voltage is in phase with the resolver. Then the amplitude of the 3Ph currents is adjusted to keep a constant current. It is not the correct way to drive the motor because as the motor is loaded the phase angle between the current and the voltage changes which causes the motor to not run effecintly but it is a way to check I have the input and output functions working correctly.

[JD Solar and EV]

Here is a video of Francisco speaking about his Pajero project from Wednesday night's meeting. Enjoy.

https://youtu.be/Hdt5l9f_Iis

[francisco.shi]

I was doing a bit more work on the inverter today. I few days ago I finished the brackets to hold the bus capacitor and decided to install them. I had to disassemble the whole inverter to do that so I figured I would take some pictures and document how it fits together.
This is the rewound motor with 6 terminals.

Motor without inverter 1

20200727_195421.jpg (1.52 MiB) Viewed 1039 times

Motor without inverter 2

20200727_195439.jpg (1.67 MiB) Viewed 1039 times

Then the bracket that holds the inverter is bolted to the back of the motor. The screws that hold the black plastic blocks are screwed from the back.

Bracket for holding inverter.

20200727_203409.jpg (1.56 MiB) Viewed 1039 times

Then the bracket is bolted to the motor.

Motor with bracket installed.

20200727_203940.jpg (1.56 MiB) Viewed 1039 times

[francisco.shi]

Next I need to bolt the IGBT blocks to the cooling plate.

Thermal grease on cooling plate.

20200727_201053.jpg (1.99 MiB) Viewed 1039 times

I need to make sure the IGBT block is facing in the correct direction. The coolant ports need to go on the side that connects to the motor.
Then I need to turn it around add grease to the other side and bolt the second IGBT block making sure the output of both IGBT blocks are on the same side.
This time I have to support the cooling plate so the gate drive board doesn't get damaged.

Thermal grease on second side.

20200727_201612.jpg (1.64 MiB) Viewed 1039 times

Assembled IGBT blocks.

20200727_201819.jpg (1.94 MiB) Viewed 1039 times

Next I need to bolt the IGBT blocks with the cooling plate to the busbars.

Busbars

20200727_201903.jpg (1.7 MiB) Viewed 1039 times

Making sure that they are the correct way around so the coolant ports end up on the correct side (Next to the coolant ports for the motor)

IGBT blocks in correct orientation.

20200727_202058.jpg (1.55 MiB) Viewed 1039 times

Then bolt the busbars to the IGBTs and make sure the busbars holes line up with the capacitor terminals.
Tighten the input bolts because once the bus capacitor is bolted down there is no access to the bottom IGBT input bolts.

IGBTs on top of Bus capacitor.

20200727_202610.jpg (1.73 MiB) Viewed 1039 times

[francisco.shi]

Then bolt the capacitor and the bracket that holds the capacitor to the plastic blocks.
Then slide the whole assembly into the black plastic blocks and bolt the motor terminals (with great difficulty) to the IGBT terminals. I need to think of a good way to do this.

Assembled IGBT blocks.

20200727_210652.jpg (1.61 MiB) Viewed 1044 times

I have made a prototype board with some signal conditioning. This board is mainly to test the interface for the temperature sensors (3), the two current sensors, bus voltage, the resolver exciter circuit and filters for the resolver phases and the plugs for the wires going to the gate drive boards. The gate drive boards are flipped so the logic signals also need to be flipped so the pwm signals end up on the matching output of the IGBT modules.
The empty space on the right hand side of the board is for placing the processor and power supply.

Prototype board for inverter front end.

20200727_211715.jpg (1.58 MiB) Viewed 1044 times

All the signals go to a connector that I am currently wiring to an evaluation board.
The processor I am using is an Atmel SAM E70.

Wiring to evaluation board.

20200727_211749.jpg (1.67 MiB) Viewed 1044 times

Unfortunately the documentation for this chip is not the best so I have decided to get everything wired to the evaluation board and once I am happy that everything works I will make another board with the processor and isolated power supply.
I have got the software reading the current sensors, the bus voltage, the resolver and the temperature sensors and generating the signals for the gate drive.
I also got the Clarke and Parke transforms done and the PID loops for controlling the quadrature currents.
I have attached a pot to one of the analogue inputs and I can control the output torque in both directions and regen is also working.
Now I just need to tidy up the code and add more safety features and an easy way to set up the parameters. The torque input to the inverter will be via CAN bus. So there will only be 4 wires for control (2 for CAN bus and two for power) and two thick wires for power.
It is still early stages but I am hoping to have a running motor to show on the next meeting.

[jonescg]

I'm well impressed. Winding your own motor and building your own inverter is hardcore

[esyjam]

This is Impressive Francisco, the workmanship is of next level Level . The only Thing to worry about is Insulation as the circuit and control will be beneth the car and also condensation maybe vacuum seal the circuit area .

My v75 project is still on the infancy, still getting some items . Have attached the motor to the transfer of v5a51 for trials first. Using leaf motor with open inverter control system planning to see if I can push it to 140Kw safely down the line.

[francisco.shi]

Thank you for the compliments.
I do this kind of project as a job. So I have experience in mechanical and electronic design.
As for the electronics I will need to put some conformal coating on all the electronics and coat the busbars in the rubbery goo used on switchboards. I will also have to remake the busbars because I still have not figured out a way to get the power connections and coolant lines out neatly. Originally I was going to take the power terminals out the back of the capacitor but that makes the whole assembly longer and still doesn't solve the problem of how to get the cooling in and out.
I am trying to keep the diameter of the inverter case the same as that of the motor.
If I had made the casing of the motor then I could have put the coolant connections internally and feed them thru the front of the motor so the coolant does the motor and electronics. That would have been really neat but I guess I can't make everything.
The plan so far (not that happy about it) is to get the coolant and power thru the side. This will make the case for the inverter a bit bigger and will have a funny shape so we will see how it goes.
As for sealing the inverter, the whole thing will be airtight including the motor. I will have an air hose connected to it and plan to presurise it slightly to make sure nothing gets in. It needs to be able to be submerged in water without any problems. The slight pressurisation will also help to check the integrity of the seals. I am planning to do the same for the battery.
I am hoping the inside will always be warm enough that there won't be any condensation. I am planning to feed chilled water to the inverter so I can get more power out. This will include the motor.
The motor is rated at 100kw continuous (after the rewind) with a coolant temperature of 70°C and the IGBTs 400A at a case temp of 85°C (500A @25°C) if I run water at under 25°C I should be able to have a higher continuous power but not really sure if it will be useful unless I am racing it on a track (unlikely to happen on a tall SUV).
With two motors able to put out 100kw each continuously I doubt there will be any situation where it will overheat. Even towing a heavy load up a steep hill will probably run the battery flat before it overheats.

[francisco.shi]

Just a quick note on resolver.
Here are some waveforms from the resolver.

The yellow and purple traces are the output of the resolver. The output is sine and cos of the angle of the resolver. The blue trace is the exciter coil (input) the exciter coil needs a sine wave so the out puts are also sinusoidal. The reason the input is not sinusoidal is because it is being driven by a pwm output. The input coil is setup as a tunned circuit which resonates at about 20khz. The pwm runs at 20khz so the sine wave in the input is synchronised with the pwm. Then to read the outputs the ADCs are triggered from the pwm output so they read the peak of the output waveforms.

Resolver outputs and decoded angle.

20200730_215158.jpg (1.74 MiB) Viewed 931 times

Here is a picture showing the outputs and the decoded angle while the motor is spinning. Assuming that the speed of the motor is constant the decoded output should be a a ramp. The ramp should be perfectly straight.

[francisco.shi]

Some videos of motor running.
This is slow acceleration which shows current meter reading a negative current for a very short time and power supply showing zero current and voltage increasing slightly. The motor was running from a 12v lead acid battery so I could run more current than the power supply can produce and also to be able to absorb the energy during regen. If the battery is not there the power supply will overvoltage.


This video is with higher torque setting.

[francisco.shi]

Working on the front now. I decided to make new half shafts and get rid of the spacers between the diff and the cv joints.

Front diff with extended half shafts.

20200811_205746.jpg (1.86 MiB) Viewed 656 times

Now I have to fix the back part of the diff and finish the other gearbox and should be able to install the front motor.

[francisco.shi]

Back of the diff is attached.

Diff back attachment.

20200816_130040.jpg (1.88 MiB) Viewed 616 times

I will bring the rewound motor with inverter to the meeting on Wednesday.

[T1 Terry]

Wow, that is a very high oil filler/level plug, is that standard? I would have thought that much oil being churned up by the crown wheel and pinion would cause issues with seals ......
The diff in my big bus project has been rolled over and because it is a hypoid set up, pinion is below the centre line of the crown wheel, it is now high and dry so I need to develop an oil feed to pump oil between the teeth to protect them from wear etc. If your diff is designed to fill to that level then maybe I can get away with just increasing the oil level.

T1 Terry

[T1 Terry]

Wow, that is a very high oil filler/level plug, is that standard? I would have thought that much oil being churned up by the crown wheel and pinion would cause issues with seals ......
The diff in my big bus project has been rolled over and because it is a hypoid set up, pinion is below the centre line of the crown wheel, it is now high and dry so I need to develop an oil feed to pump oil between the teeth to protect them from wear etc. If your diff is designed to fill to that level then maybe I can get away with just increasing the oil level.

T1 Terry

Closer look, the level plug is on the side at around the centre line ..... back to the oil feed plans, or flip the diff back over and mount an electric drive motor and converter the drive engine into a generator .....

T1 Terry