Francisco's Electric Pajero

Post up a thread for your EV. Progress pics, description and assorted alliteration
Post Reply
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

About the quietness the inverter is running at 7khz and surprisingly there is no 7khz noise coming from the motor. I was very surprised. However when I lift the wheels and run it there is a lot more noise from the diff gears than I was expecting. I have to wait and see how noisy it is once it is driving.
At the moment I am only running it off 48v so it can't go very fast.
User avatar
JD Solar and EV
Noobie
Posts: 25
Joined: Thu, 21 Mar 2019, 18:20
Real Name: Jon

Re: Francisco's Electric Pajero

Post by JD Solar and EV »

Good Job Francisco! Can't wait to hear more about it at this months meeting.
Current AEVA QLD Chairman

Check out my EV videos at JD Solar and EV YouTube Channel
https://www.youtube.com/channel/UCNMxXc ... bUTfpHU5LQ
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

I have been working on the BMS today. I got 4 modules I have been using for testing to make sure there are no mistakes in the boards. I was assembling and calibrating the modules in sets of 5.
Here are some pictures of the boards in a subpanel.
15849701403783566395928291262880.jpg
15849701403783566395928291262880.jpg (1.47 MiB) Viewed 4190 times
15849701078307081356832782851835.jpg
15849701078307081356832782851835.jpg (2.3 MiB) Viewed 4190 times
I used different colours for the connectors because it is too easy to plug it in the wrong plug. The first time I made that mistake the com's wire burst into flames. The transmit has an optocoupler but the receive is connected to gnd. If I plug two receives together it shorts the battery thru the thin com's cable. With the different colours I just need to plug into sockets of the same colour and all is fine.
Here is a picture of the test pack.
Test module
Test module
15849710330875985588895374606538.jpg (1.73 MiB) Viewed 4190 times
While I was doing the assembly I decided to balance the battery I have been using for testing. The cells were 88mv out of balance.
Before balancing at 6:40pm
Before balancing at 6:40pm
20200323_184007.jpg (1.16 MiB) Viewed 4190 times
I set the cutoff to 15mv difference.
After balancing 11:18pm
After balancing 11:18pm
20200323_231811.jpg (1.73 MiB) Viewed 4190 times
It took 4 and a half hrs to balance the 189Ah cells for a 55mv change. The energy recovery system seems to work well. The low cells charged up and the high cells discharged. The average battery voltage only went down by 10mv.
T1 Terry
Senior Member
Posts: 1952
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Re: Francisco's Electric Pajero

Post by T1 Terry »

What voltage is considered 100% capacity? In other words, at what voltage does the cell voltage rise rapidly yet only a small current going into the cell. LFP and LYP cells will voltage balance quickly below fully charged but that has nothing to do with the state of charge being balanced. The only points that balancing really counts is fully discharged or fully charged, and you can't have both, it's one or the other.
Top balancing for house batteries and bottom balancing for EV batteries seems to be the way to go, you never completely discharge a house battery intentionally, yet it is the norm for an EV.

T1 Terry
Green but want to learn
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

4.2v is supposed to be full.
The bypass circuit dumps the energy at the pack's terminal. So I can balance at any time.
The way the balancing works is the master works out the average voltage of all the cells and it sets the target voltage to that value. Then if the difference between the max cell voltage and min cell voltage exceeds a certain value it issues the bypass command to all nodes.
If the cell voltage is higher than the previously set target voltage the node will bypass the cell. The energy goes to the pack which acts as a charger. So all the cells that are not being bypassed will start to charge (and increase in voltage) the ones that are being bypass will decrease in voltage.
So the balancing is always active regardless of the cell voltage or wether the pack is charging or discharging. The end result during charging is the cells with lower voltage will have slightly more charge current and the ones with higher voltage will have less charge current.
During discharge the cells with higher voltage will supply more current and the ones with lower voltage will supply less current.
So ultimately the stronger cells will supply/absorb more energy than the weaker cells so all the cells will reach top and bottom at the same time.
User avatar
4Springs
Site Admin
Posts: 926
Joined: Thu, 23 Dec 2010, 01:14
Real Name: Christopher Walkden
Location: Selbourne, TAS

Re: Francisco's Electric Pajero

Post by 4Springs »

T1 Terry wrote: Wed, 25 Mar 2020, 11:04 What voltage is considered 100% capacity? In other words, at what voltage does the cell voltage rise rapidly yet only a small current going into the cell. LFP and LYP cells will voltage balance quickly below fully charged but that has nothing to do with the state of charge being balanced. The only points that balancing really counts is fully discharged or fully charged, and you can't have both, it's one or the other.
Interestingly, these NMC cells don't behave the same as the LFP that we're used to using. Within reason, you can actually use the voltage to figure out the state of charge. Here is the table from the spec sheet:
E63_table.jpg
E63_table.jpg (36.57 KiB) Viewed 4152 times
The difference between 5% and 15% is 0.07V or so.
The difference between 90% and 100% is 0.1V or so.
So at a higher state of charge the voltage does move more rapidly, but nothing like LFP.

This means that the cells can be balanced over a very large voltage range. Which is great! Because holding at a high state of charge is somewhat detrimental to cell life. We can balance at pretty much any state of charge, and can do it while we're charging, discharging or resting. High bypass currents are not required, we can use low currents over long times.
User avatar
brendon_m
Senior Member
Posts: 1315
Joined: Sat, 28 Oct 2017, 11:00
Real Name: Brendon McCarrol
Location: Perth, Western Australia

Re: Francisco's Electric Pajero

Post by brendon_m »

4Springs wrote: Wed, 25 Mar 2020, 16:12 Within reason, you can actually use the voltage to figure out the state of charge. Here is the table from the spec sheet:
E63_table.jpg
I tried to find a similar chart for my ex imiev cells and couldn't really find anything with actual values. Lots of small graphs that if you squinted you could get a voltage vs soc. I got a couple of hard numbers but nothing for the whole range of charge levels.
I ended up getting my own from readings from my outlander.
100% 4.1v
90% 4.05v
80% 4.01v
70% 3.96v
60% 3.93v
50% 3.9v
40% 3.86v
30% 3.83v

These figures were taken after driving without letting the pack rest so they're not exact.
Interestingly the manual states 3.75v is 30% and from the graphs I've seen I get
20% 3.5v
10% 3.25v
0% 3.0v
Dead flat danger zone% 2.75v
You just destroyed 1000's of $$% 2.5v

So the official Yuasa numbers match with your chart but my numbers from the outlander seem too high even with an unrested pack. I wonder if 30% (the minimum soc you can go in an outlander) is not 30% soc but 30% of allowable usage.
T1 Terry
Senior Member
Posts: 1952
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Re: Francisco's Electric Pajero

Post by T1 Terry »

4Springs wrote: Wed, 25 Mar 2020, 16:12
T1 Terry wrote: Wed, 25 Mar 2020, 11:04 What voltage is considered 100% capacity? In other words, at what voltage does the cell voltage rise rapidly yet only a small current going into the cell. LFP and LYP cells will voltage balance quickly below fully charged but that has nothing to do with the state of charge being balanced. The only points that balancing really counts is fully discharged or fully charged, and you can't have both, it's one or the other.
Interestingly, these NMC cells don't behave the same as the LFP that we're used to using. Within reason, you can actually use the voltage to figure out the state of charge. Here is the table from the spec sheet:
E63_table.jpg

The difference between 5% and 15% is 0.07V or so.
The difference between 90% and 100% is 0.1V or so.
So at a higher state of charge the voltage does move more rapidly, but nothing like LFP.

This means that the cells can be balanced over a very large voltage range. Which is great! Because holding at a high state of charge is somewhat detrimental to cell life. We can balance at pretty much any state of charge, and can do it while we're charging, discharging or resting. High bypass currents are not required, we can use low currents over long times.
Thank you for pointing that out, I hadn't realised the NMC cells had a sort of linear voltage that could represent the SOC. That would mean the 3.818v that the Pajero pack balanced at was actually some where between 65% and 70% SOC?

T1 Terry
Green but want to learn
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

That is correct. The batteries were not fully charged. The way the BMS is setup it will keep the cells balanced at any voltage no matter if the pack is charging or discharging. Because the bypass energy is recovered I can balance at any time without losing range if anything I will gain range because I can get the maximum energy capacity of each cell even if the cells are not all exactly the same capacity. Remember that the same capacity cells are only within 3 to 4% of each other. So if I used the conventional method of top or bottom balancing and one cell was 3% lower than the rest then I would get 3% less than the true potential of the pack.
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

I have been thinking about the power steering and I didn't want to have a pump running all the time. After talking to a friend he suggested I use an electric power steering module instead. He had one that he used for testing before. It came out of a Toyota Corolla.
Power steering module.
Power steering module.
20200413_065245.jpg (1.91 MiB) Viewed 4026 times
The current draw when sitting idle is about 100mA.
Peak current when hitting the end of travel is 25A.
Current when turning the steering wheel fast is about 20A.
It is a bit more involved than an electric pump but it is a much neater and efficient solution and it fits under the dash leaving more space for the frunk.
User avatar
brendon_m
Senior Member
Posts: 1315
Joined: Sat, 28 Oct 2017, 11:00
Real Name: Brendon McCarrol
Location: Perth, Western Australia

Re: Francisco's Electric Pajero

Post by brendon_m »

Power steering columns are a much neater and efficient idea.
What are the engineering repercussions of modifying the steering? Just a looky loo and a "yeah she's chipper", or something more involved?
antiscab
Senior Member
Posts: 3183
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Re: Francisco's Electric Pajero

Post by antiscab »

could you pretend the power steering is original and hope no one notices?
Matt
2023 BYD Atto 3 - 21k km
2017 Renault zoe - 147'000km
2012 Leaf - 101'000km - soon to be trialing a booster battery
2007 Vectrix - 197'000km (retired)
2007 Vectrix - 50k km
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

I am not going to say anything and see what happens.
The shaft on the power steering module is straight thru. If the module fails the car will still steer.
User avatar
blwnhr
Groupie
Posts: 31
Joined: Tue, 02 Apr 2019, 05:10
Real Name: Adam
Location: Launceston, TAS
Contact:

Re: Francisco's Electric Pajero

Post by blwnhr »

What year/model Corolla is that out of?
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

I need to ask my friend but I think it was a 2010.
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

I have been doing some further testing on the power stage of the inverter. So far I have been running only one transistor pack so current is limited to 400A. I have been testing at 280A RMS which is about 400A on the peak of the sine wave.
I am quite happy with the performance of the gate drive and bus bars.
Here are some pictures of the waveforms.
Timebase 40us/div
Yellow bus voltage.
Bus voltage at full load (400A peak)
Bus voltage at full load (400A peak)
20200420_212151.jpg (1.99 MiB) Viewed 3837 times
The next picture shows gate signals.
Timebase 2us/div
Yellow voltage across gate of high side transistor
Blue voltage on output of half bridge.
Green digital input to gate drive circuit for high side transistor.
Purple digital input to gate drive circuit for low side transistor.
All the ringing is noise picked up by the probes. it is not really there.
Gate signals.
Gate signals.
20200421_220405.jpg (1.51 MiB) Viewed 3837 times
The next picture shows something worrying that turned out not to be unavoidable.
Timebase 0.4us/div
Yellow output of half bridge.
Blue bus voltage.
Green high side transistor input to gate drive.
Purple low side transistor input to gate drive.
Worrying spike
Worrying spike
20200421_221014.jpg (1.94 MiB) Viewed 3837 times
You can see the bus voltage drops to almost 1/3 for 232ns. This had me worried because it means there is a significant amount of current being drawn from the bus.
I though both transistors were turning on at the same time. But the gate signals were off already so all measurements of the gate signals say that the transistor should be well off by this time. All gate signals were perfect and all swithching was within spec. The dead time was more than enough. Something interesting was that the pulse got wider as the current increased.
It turns out there is nothing wrong and there is an unavoidable reason for the behavior.
Points for anyone who can explain it.
Will post the answer in a few days. :lol:
User avatar
coulomb
Site Admin
Posts: 6357
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Re: Francisco's Electric Pajero

Post by coulomb »

francisco.shi wrote: Tue, 21 Apr 2020, 21:20 It turns out there is nothing wrong and there is an unavoidable reason for the behavior.
Points for anyone who can explain it.
OK, I'll have a go. I'll consider the case where the output current is positive (conventional current is flowing away from the IGBT pair towards the motor). At this point, the lower transistor had been conducting, but stopped conducting (earlier in time, so to the left of the DSO image). There is always inductance when driving a motor, so something has to keep the current flowing. At this time, it's the lower transistor's free-wheel diode. It's conducting because the inductor suddenly made the output a little negative (it was a volt or two positive with respect to bus negative when the transistor was conducting, now it's negative a volt or so to keep the output current flowing).

When the upper transistor starts conducting, it has to wrestle the current away from the lower transistor's free-wheel diode. It takes a while for hundreds of amps worth of carriers in that diode to sweep out of the junction and cause the diode to turn off. While this is happening, there is a high current path across the bus, through the upper transistor just starting to conduct, and the lower transistor's free-wheel diode, just starting to ramp down its current. So the bus has to provide the load current, plus this high current path. This continue for the hundreds of nanoseconds it takes to turn the diode off. That's one of the reasons that you need a very stout snubber capacitor across the bus, physically as close as possible to the transistor pair, and it needs fat, low inductance leads. The capacitor has to be specially built for low impedance and high current spikes. Often they are designed to bolt onto the IGBT module terminals.

I'm sure I've gotten some of the details wrong, but I suspect that this is the broad outline.
MG ZS EV 2021 April 2021. Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
Patching PIP-4048/5048 inverter-chargers.
If you appreciate my work, you can buy me a coffee.
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

I guess you did pay attention in class. The answer is spot on.
I was a bit worried when I first saw this because I thought two transistors were turning on at the same time. The bus structure and bus capacitor are extremely low impedance so there are a few thousand amps flowing to cause that big a drop. Consider the 400A ripple on the first picture is only 5 or 10v and the spike is 60v. The spike width is not long enough to trip the desat protection.
Overall I am quite happy with the response from the bus structure.
I would be confident of running this up to 600v with 700v devices.
I am planning to run up to about 500v bus.
I need to contact tritium to find out what is the maximum voltage the 50kw DC chargers can do. I dont want to end up finding out too late that I can't use them.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: Francisco's Electric Pajero

Post by weber »

francisco.shi wrote: Tue, 21 Apr 2020, 21:20 It turns out there is nothing wrong and there is an unavoidable reason for the behavior.
Points for anyone who can explain it.
Will post the answer in a few days. :lol:
I haven't read any other response yet. :) It's due to high shoot-thru current due to the reverse recovery time of the low-side antiparallel diode. Sure, it can't be avoided by adding dead-time, but it can be avoided, or at least minimised, by "soft switching" or "resonant" designs. But that's not easy.
https://www.pre-switch.com/single-post/ ... -Switching
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

Only one other answer and it was correct. The same as yours.
Yes. Soft switching would fix it but it adds a lot of complexity because the current is continuous during the switching cycle so it needs extra components. I did build a power factor corrected PSU front end years ago and it was amazing how effecient it was. It would draw 10A RMS from the mains with the heat sink facing down on the bench with no air flow and no fins and it would be skin warm. The heatsink was a bit of flat aluminium 190x95x6mm. It was a 3U block. It used a common IGBT and a fast diode and would run between about 15khz and 50khz.
What concerns me about the shoot thru due to the reverse recovery is the peak current. I have attempted to slow down the turn on so it will run lower current for longer. I am hoping this will reduce the stress on the IGBT and the diode.
Before
Before
20200423_121236.jpg (1.88 MiB) Viewed 3740 times
Before
Before
20200423_121236.jpg (1.88 MiB) Viewed 3740 times
Attachments
After
After
20200423_173643.jpg (1.86 MiB) Viewed 3740 times
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

I have increased the voltage of the battery pack after doing some testing to make sure everything is ok.
It has enough volts now to get to about 40kph. It now feels more like a real car. Still only running on one motor and about 75% of peak torque. I should get the next version of the gate drive in a few days so I will be able to test at full current. As it is now it accelerates at the same rate as the petrol Pajero in second gear (according to the accelerometer in my phone).
It feels quite good with a single motor. It would be quite drivable as it is.
Can't wait to get two motors running. :twisted:
Did an acceleration test on my brothers STi and with two motors it will have about 15% more acceleration then the STi in a nd gear.
I must say the STi feels really fast. I just hope I can make it up while he changes gears.
He has to do 2 gear changes to get to 100kph.
Here is a video of the acceleration test.
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

I have been working on rewinding the motor.
The motor is 8 poles.
The original winding was 5 turns per coil. 2 series coils per pole. Two poles in series and then 4 pole pairs in parallel.
I need to wind it to half the voltage so I need 2.5 turns per coil. So I rounded that down to 2 turns. Then to make winding a little easier I decided to put 4 poles in series instead of two. This means each coil is one turn.
So I will have two sets of windings per phase instead of 4. One winding goes on one half and the other winding on the other half. This will reduce the number of internal joins I have to make. The wire is getting pretty thick. Each half phase is 19mm2.
The plan is to have the windings such that the start of both half phases is on the same slot and the neutrals will be spread out. This will also help to avoid having to have a 120mm2 joiner in the middle. For now I just need two 60mm2 joints for the neutrals and the phase ends are joint on the outside (not quite).
The first attempt was to preform the coils and then put them in.
Preformed coils.
Preformed coils.
20200610_183835.jpg (1.7 MiB) Viewed 3328 times
This turned out to be really hard to wind. It took me almost an entire day to put the wire for half a phase.
Here is a picture with the preformed winding.
Winding with preformed coils.
Winding with preformed coils.
20200611_180726.jpg (1.49 MiB) Viewed 3328 times
So I gave up that idea and took out the wire and started again.
Here is the empty stator again.
Here is the empty stator again.
20200612_184114.jpg (1.85 MiB) Viewed 3328 times
This time I decided to get all the strands together and leave them straight and wind them in the stator as I put the wire in.
Half of one phase wound.
Half of one phase wound.
20200614_160259.jpg (1.7 MiB) Viewed 3328 times
This turned out to be much easier and quicker. It only took 40min to put half a phase.
One complete phase.
One complete phase.
20200614_173153.jpg (2 MiB) Viewed 3328 times
Note that the phase terminal for both half phases come out on the same slot. For this to work one half phase has to be wound with opposite polarity to the other half phase.
The reason for having two half phases is to do with the power stage.
The peak RMS current is 800A but I can not get 800A devices so I need to put two devices in parallel. But it is not as simple as just paralleling the devices. So to get around the IGBTs ON voltage mismatch I am using the windings as sharing resistor. So one half phase is connected to one IGBT and the other half phase to the other IGBT. This allows each IGBT to take exactly half of the current.
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

Two phases in one more to go.
Two phases in one more to go.
Two phases in one more to go.
20200615_210846.jpg (1.94 MiB) Viewed 3308 times
The last one will be the most difficult one because there is less space to put the wire and all the other coils get in the way.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: Francisco's Electric Pajero

Post by weber »

That's awesome. Good on you for having a go. :clap:
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
francisco.shi
Senior Member
Posts: 826
Joined: Mon, 17 Sep 2018, 16:30
Real Name: Francisco
Location: Brisbane, Strathpine

Re: Francisco's Electric Pajero

Post by francisco.shi »

I was hoping to get the last phase done today but got tied up with something else and only got half of it done. Hopefully tomorrow.
Post Reply