Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[Astro]

Originally Posted by cts_casemod

Why don't you guys use the flywheel and a reluctance sensor as an encoder? That would cut a significant portion of the costs.



The input capture peripheral and a free running timer can be used to measure the time between each pulse

@cts_casemod Out of curiosity what was that flywheel out of? The flywheel from my little four cylinder 1.2L looks about two thirds the thickness of the pictured flywheel.
I was considering removing the ring gear to reduce the weight as i thought it was quite heavy but looking at the pictured one i think mine would be light by comparison.

Then again, mine seems to have more diameter beyond the clutch friction surface. But then my clutch may be a smaller diameter.



I was talked out of removing the ring gear by a co-worker who builds and races cars. He said he had seen flywheels that had been lightened by removing some of the diameter basically grenading under high rpms. He said it depends on the material they are made of but best to leave them alone and just go for an after market lightened flywheel.

I think i will leave it as is, at least it will allow me the option of putting the starter motor back in.

Back on topic. Pauls AC controller with it's regen should allow me to recover some of the energy from the flywheels inertia.

[Astro]

Originally Posted by cts_casemod

Thanks,
The water pump is a good one, with the complementary temperature sensor input

Should allow the cooling system to adapt automatically to various driving environments. Stop/go traffic, cruising, racing, etc.

Originally Posted by cts_casemod

Load indicator can be taken directly from the throttle, since the user commands torque...

Sounds good.
I was pondering the "user commands torque" bit as i am aiming to replicate the normal ICE driving environment as much is possible.
Just wanted to check my understanding of how i think the throttle will behave against some EVers actual experience.
With an ICE, giving 30% throttle gives more aggressive acceleration than say 10% throttle. So the EVs torque based throttle would replicate that.
But as the cars speed increases and the engine RPM increases, in an ICE motor the amount of fuel mixture delivered to each cylinder on each cycle reduces. So with a fixed throttle input of 30% the torque decreases as the RPM increases until an equilibrium is reached where the torque required to overcome wind/rolling resistance is balanced by the torque supplied by the motor. So now the car is cruising at a fixed speed and the throttle is still at 30%.
With an electric motor setup the 30% throttle will give a similar to ICE initial acceleration but holding the throttle at 30% will continue the torque delivery. The torque will not decrease with increased RPM because the controller is working hard to deliver the commanded torque so at some point you need to reduce the throttle input to decrease the torque delivered and achieve a steady speed cruise.
So some sort of RPM feedback to bias the throttle would be needed to have the EV throttle behave just like an ICE one?

Originally Posted by cts_casemod

I could do that. but at this point it starts to make sense to isolate the boards, perhaps. I want to have a relatively small board. This would allow the code to be changed as per user needs. Would 8 outputs and 8 inputs (digital) be enough?

Should be heaps. Another thought would be to have some extra outputs on the display PCB to drive the unused lights/indicators in the instrument cluster so you could convey important information in a way that is easily noticed by the driver.

[cts_casemod]

That's the flywheel from a 1.4/1.6L VW Polo.

If memory serves me right it weights about 14KG itself and 18KG with the clutch and pressure plate installed. Add another 3KG to the crank adapter. The motor rotor weights about 25KG.

I removed the ring gear because of clearance issues with my adapter, when I changed from steel to aluminum. To be fair never had any sort of issues with the original one, but my motor is capped at 3800RPM.

Astro,

I drive 10 miles before the coolant pump kicks in so although PWM seems needed, circulating the coolant once a minute does the job just nicelly.

On an ICE the fuel decreases with RPM because for the same throttle opening less air can go into the engine as the admission stroke is shorter. Unless is turboe'd, of course.

On a EV startup torque is quite considerable, but its not linear either, specially as the motor goes into field weakening. Battery sag also plays a factor.

So in reply to your question, yes just like a petrol you get a time when the torque at 30% is just enough to overcome friction, rather than continuing linear delivery. There is some customization one can do with look up tables on my controller, in fact I just had a dead spot on the 10-40% Regen that I just fixed. I'm not sure how Paul setup will work, but technically he can do something similar.

In practice you just need to drive it... like an ICE, except with much more low end torque and limited gear changes

I have that too, although its part of a separate project, done with an arduino mega.





That was an earlier version, that drove the gauges and replaced the odometer. No photos of the newer version sadly.
there is also a Texas instruments fuel gauge to report battery status

[Astro]

Love the instrument cluster, looks like it came out of an aircraft.

Originally Posted by cts_casemod

That's the flywheel from a 1.4/1.6L VW Polo.
If memory serves me right it weights about 14KG itself and 18KG with the clutch and pressure plate installed. Add another 3KG to the crank adapter. The motor rotor weights about 25KG.

I don't know how much my motor rotor or shaft to flywheel adaptor weigh. But the flywheel itself is 7.4kg and with the clutch and pressure plate assembly it goes up to 11.2kg.
So the flywheel is lighter but the clutch and pressure plate weigh about the same as yours. So i don't think i will mess about lightening the flywheel.

Originally Posted by cts_casemod

I removed the ring gear because of clearance issues with my adapter, when I changed from steel to aluminum. To be fair never had any sort of issues with the original one, but my motor is capped at 3800RPM.

How much lighter is an aluminium one?

Originally Posted by cts_casemod

In practice you just need to drive it... like an ICE, except with much more low end torque and limited gear changes

Whoohoo.. i can't wait. Previously i drove a 5.7litre V8 to and from work, the EV is going to be like being allowed to drive a go-cart to work. Hopefully a very quick go-cart.

The very quick bit will be due to Paul's controller and the oversized motor. I have no idea how much punishment the little Barina's gearbox will take. Hopefully the clutch will slip before the gearbox goes bang.

I am trying to decide where to mount Paul's controller. I have two options. The first is above the motor. This would allow air flow from the front grille to blow over both the motor and the controller. Or i could mount it forward of the motor where it will get more of the air flow but then the motor will get less.
Which component is likely to need more cooling, the motor or the controller?

I am leaning towards putting the controller forward of the motor not for the air flow but because it will free up a bigger area in the engine bay if i ever need to add some cells in there.

[cts_casemod]

My adapter plate was changed to aluminum. The flywheel is still the same.

I would avoid having the controller in front of the motor for cooling reasons, unless you have a water cooled motor.

[e*clipse]

Yea! Some progress with the resolver driver!

It's working! Just a couple of problems - an incorrect resistor and a cold solder joint. Fixed those, and the board works great!

Here's a pic of the test:

The flat ribbon cables are the input/output from the circuit, the red/black cable provides 24V to power the driver amplifier. The larger ribbon cable on the right is for all the sensor I/O of the MGR. Right now the test cable is connected to the MGR rotor sitting on my desk. It includes the resolver carrier signal, the sin and cos return signals, and a temperature probe. The smaller one is just for testing; it includes all the output signals and 5V board power. This one plugs directly into the QEI interface terminal on Paul's board.

Here's a close-up of the board itself. I've got to say, it would have been a LOT easier to debug on a through-hole circuit board. One big thing I'm going to change on my next board is to add some test-loops for places to attach the 'scope probes.


Here's a fuzzy shot of the output from my "new" scope. This is the carrier signal, running at 62.5kHz, with a p-p voltage of 9.6V.


I need to get my old computer running with the PC scope, then I'll post some pics of the output signals.

- E*clipse

[cts_casemod]

Originally Posted by e*clipse

Yea! Some progress with the resolver driver!

It's working! Just a couple of problems - an incorrect resistor and a cold solder joint. Fixed those, and the board works great!

I need to get my old computer running with the PC scope, then I'll post some pics of the output signals.

- E*clipse

That's class, looking forward for that!

[cts_casemod]

The other day I was thinking about system isolation.

Normally one does away with linking the HV and LV grounds for Isolation purposes, but that got me thinking what would happen if a fault on the motor windings does occur.

The motor is earthed to the chassis, so the chassis will become live. The isolation means a standard RCD cannot be used to detect the fault.

The other thing is capacitive coupling. There may be a difference of potential of a few KV due to static charge accumulation that may blow the isolators or other devices that have linkage between the power rails.

Has anyone considered this?

[MPaulHolmes]

That thar looks like a sine wave, e*clipse!!

I was thinking of adding a leakage current monitor circuit, with a known large resistance between pack ground and chassis. And if the resistance drops below a particular value, then it would flag an alarm.

[thingstodo]

Originally Posted by MPaulHolmes

I was thinking of adding a leakage current monitor circuit, with a known large resistance between pack ground and chassis. And if the resistance drops below a particular value, then it would flag an alarm.

First I guess I should ask how small a current you would be allowing. If it's under 1 mA, then no one should get a shock from it but it may tingle a bit

I would like to be able to lean on the car and check tightness on the battery terminals without needing gloves. Perhaps a momentary test for leakage on startup?