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

[MPaulHolmes]

20kW at 72v is almost 300 amp. You could probably use 300amp IGBTs, which are quite a bit cheaper than the 600amp modules.

[learnerguy]

No changes in the pcb board ? I want to make it as cheap as possible.

[MPaulHolmes]

You could use 4.7uF 1206 package capacitors rather than 22uF. You could leave off the comparator for the 3rd current sensor and also leave off the 3rd current sensor.

[thingstodo]

Originally Posted by learnerguy

Hi all,

Can you please guide me with making a 10 to 20 Kw controller for use in motorcycles.

What parts do I have to downgrade and will it be ecnomically viable ?
I am following Paul's instructions on instructables .

My un-informed opinion, and a few un-informed assumptions
- 20 kw = 26.8 hp (I need this for torque)
- 5,000 rpm for a decent small AC motor (it should be higher)
- 2 pole motor (3600 rpm at 60 Hz)
- 72V pack

72V pack is about 50 VAC or 50V rms
20,000W / 50V = 400 amps. For short periods of acceleration. Maybe 5 HP for cruising is under 90 amps. That's still over 2C on your battery pack if it is 40 A-h. Less than 30 minutes of cruising ... much better in town!

I chose a motorcycle that I have heard of - a yamaha XT660 - and used the gear ratios. !st gear gets you about 100 lbs pushing the bike and you down the street. Decent for acceleration. 5th gear is 5000 rpm for 80 mph. Also decent.

The controller is only putting out 100 Hz at 5000 rpm. It can put out much more. 400A IGBTs should be fine since it will take .. not much time ... likely under 2 seconds? .. to accelerate through any one gear. A decent chill plate would get the heat away from your IGBTs. Or you could use a big set of fins ... depends on what speed you would normally drive. air-cooled does not do well under 20 mph.

The parts that Paul is recommending for the board are in the ?$125 - 150? range if memory serves. That is not a big portion of the cost. The ring cap and the IGBTs are expensive. The chill plate .. is sort of expensive depending on if you do it yourself or have to pay someone.

Paul - does smaller DC/DC for the separate IGBT drivers drop significant price? Or maybe smaller caps? I wouldn't think so.

My opinion - leave the controller board the same - reliable and tested - trim cost by using smaller power electronics with lower max voltage, a lower voltage rated cap, and maybe leave the 3rd phase off the current as Paul suggested. There are not many spare pins on the controller for doing pre-charge and interlocks. You may want to use an arduino or some other separate MPU to annunciate any problems and do the interface with the user.

Are you planning to do a production run? Is that why you want to minimize costs? If you are making under 10 I would expect any changes may end up costing instead of saving.

YMMV ... and it's only my opinion ..

[johkim]

Hello!

I am completely new to the EV community, and have been reading a lot of information on EVs online for the last few days. I have an interest in cars in general, and currently drive a modded VW GTI

Not to bore you with too much of my own story, I was just wondering where I would go to learn more about inverters before coming back to this thread and also Paul's instructables. I am by no means a ME or an EE, just an ordinary guy (my background is in civil engineering!), but I dream of one day making my own EV hotrod.

(the reason why I'm asking about inverters is that AC motors, and batteries are something that I understand to some degree, including BMS, at least enough to read and learn, but inverters are quite a bit harder!)

Sorry if I've posted in the wrong section! I just got stuck here hahaha

[learnerguy]

Not a production run but an experimentation for production run. The potential is very high when it comes to high end electric motorcycles and low end cars.

Thanks for the run time and your advice.

[thingstodo]

Originally Posted by johkim

Hello!

I am completely new to the EV community, and have been reading a lot of information on EVs online for the last few days. I have an interest in cars in general, and currently drive a modded VW GTI

Not to bore you with too much of my own story, I was just wondering where I would go to learn more about inverters before coming back to this thread and also Paul's instructables. I am by no means a ME or an EE, just an ordinary guy (my background is in civil engineering!), but I dream of one day making my own EV hotrod.

(the reason why I'm asking about inverters is that AC motors, and batteries are something that I understand to some degree, including BMS, at least enough to read and learn, but inverters are quite a bit harder!)

Sorry if I've posted in the wrong section! I just got stuck here hahaha

I have not seen a primer for three phase AC inverters.

The wikipedia article is OK as far as it goes

https://en.wikipedia.org/wiki/Power_inverter

My poor and disjointed summary:
- The inverter turns on 2 or 3 transistors (IGBTs) in different combinations to get current running through the motor.
- How these pulses are controlled is the magic behind Paul's controller. A bunch of math is used to figure out the angle between where the rotor is and where the voltage applied to the motor should be in order to apply torque.
- How much torque you want is controlled by how hard you push the throttle.
- The controller makes that happen by varying the applied voltage and the phase angle that the voltage is applied at relative to what angle the rotor is at now. That is calculated ... hmm ... at least 8000 times per second but I think it is more like 20,000 times per second.
- The motor position is measured by the encoder (or a resolver that pretends it is an encoder). 64 pulses per rotation is OK. More is better. My testing started out with 1000 pulses per rotation and ended up with 64 pulses per rotation.

Paul has over-current measurements that shut down the controller in the case of a problem .. sort of hard-wired into the enable circuit that drives the transistors. The temperature sensors ... if I remember .. (doubtful) .. I think those are inputs to the controller.

Many plans on the internet have been published by people with limited experience. It is quite easy to burn up IGBT(s) by turning on the positive transistor before the negative transistor finishes turning off. Ask me how I know! These types of mistakes are expensive ... and it is not usually obvious what you did wrong, or what part of your code is responsible ... or if it was a bad solder joint ... or ... you get the idea.

Paul has done a good job of ensuring that these types of things do not happen - traces well separated, ground and power planes used, inductive loop areas minimized - and of shutting things down immediately if .. sh*t happens .. and there is a lot of current somehow going through one or more of the IGBTs.

I can point you to a few videos that begin with the three phase motor and continue on to the inverters, field oriented control, the theory involved, the math, etc. It is not like a video night. Several of the videos I had to watch maybe 10 or 12 times to get through all of the details ... and the info still does not make sense to me when I go through it! I end up going through it a couple of times each time I need to revisit a small chunk of it.

Paul sent me the links. If there is any interest, I can post them

[freebeard]

https://www.youtube.com/watch?v=cdiZUszYLiA

[MPaulHolmes]

I think on the inverter board. IT's always nice for the micro to have a default value for something if a cable comes loose.

[thingstodo]

Paul - I think this was discussed briefly a while ago, but I don't remember the result:

- do you have any code kicking around for simple voltage output based on frequency output (instead of torque?)

I'm looking for a 'limp home' mode. It seems that the encoder is like anything else in a car - it fails sometimes. I know that it is not efficient to run in Volts/Hz, and that the motor will heat a bit more, and that the torque will be reduced ... but it should get me home instead of sitting on the side of the road.