Electric Vehicle Controller Design

Alternative Fuels of all sorts, Biodiesel, Electric, Human Power, Steam, etc.

Moderator: TMAX

Post Reply
Sharkey
Original Founder
Posts: 1364
Joined: Sat Oct 09, 2004 4:00 am
Contact:

Electric Vehicle Controller Design

Post by Sharkey »

This thread might be a bit on the heavy technical side for some of the regular forum users, I'm posting it to allow a couple of fellow EV'ers to have a place to discuss designing and building a homebrew PWM controller for speed control of a DC electric motor.

I've been working on a design for this controller in my spare time for the last couple of weeks. It's based on the LM3524 PWM voltage regulator chip. My application is a replacement controller for my E-20 Electrak, which uses contactors and series resistance to control motor speed. The output stages of the new controller will be the MOSFET stack salvaged out of an ancient Heart Interface DC-to-AC inverter.

Recently, a topic thread appeared in the DIY Electric Car forum using the LM3524 chipset. The user lazzer408 has done some decent research on using the chip to build a basic controller, but I have some reservations about parts of his design. Rather than contradict him over there, I'll use this thread to refine my own findings, and allow others to throw in their ideas as well.

First, and because I haven't had time to draw up a postable drawing of my own circuit, here's the schematic from the first page of posts at DIY:

Image

I'll begin by critiquing one part of the circuit, and add more as I have time.

The circuit, as shown, connects the throttle pot signal to the input fo the error amplifier, pin 2. Pin 1 of the error amplifier is connected to pin 9, the compensation output. This arrangement produces unity gain in the EA, and is probably acceptable.

In my own design experiments, I simply bypassed the error amplifier entirely, grounding both the inverting and non-inverting inputs, pins 1 & 2. The purpose of the EA is to detect and amplify a proportional signal fom the output of the circuit to provide feedback to correct for variances in the operation of the circuit's supply or load. For example, in voltage regulator duty, the chip has to be able to sense the output voltage and adjust the PWM signal to keep it constant. Forcing the EA to unity gain means that it isn't amplifying, so all it's doing is buffering the signal. You don't lose anything by using it, but you don't gain anything, either.

The National Semiconductor LM3524 data sheet has this to say about the error amplifier:
The output of the error amplifier, or input to the pulse width modulator, can be overridden easily as its output impedance is very high (ZO 5 MO). For this reason a DC voltage can be applied to pin 9 which will override the error amplifier and force a particular duty cycle to the outputs. An example of this could be a non-regulating motor speed control where a variable voltage was applied to pin 9 to control motor speed.
When a manufacturer gives me a shortcut to circuit design, I'm usually inclined to use it.

My own experiments show that using the compensation pin as a throttle input works quite well, and the shutdown and current limit inputs still work as designed.

That's about it for now. It's a beautiful, sunny day, and I want to go out and get some of it. I'll post some photos of my MOSFET array and expound on other aspects of the circuit in later posts.
ysogrimm
Posts: 3
Joined: Mon Jul 21, 2008 6:26 am
Contact:

Post by ysogrimm »

Thank you for posting this and of course letting me know it was here. I really feel this will be a good yet easy to build, inexpensive controller for a whole lot less moneythan the commercial units. When you start planning for an electric vehicle conversion whether it is a car, truck, motorcycle, lawn mower etc. the price of the controller can be the project buster. Some have tried to save money by purchasing a smaller controller than what they needen and then it either reduced the power to the motor and enden up being very sluggish or the controller went up in smoke since it was too much load. The beauty of this design is that the power it delivers to the motor is via the mosfets and they can be paralleled to add as much power as you like without changing the rest of the controller much. So in other words, you only need to buy the number of mosfets, caps and freewheel diodes to get the amps and volt your project requires. Thanks again Sharkey!
A quest to electrify everything I drive
Sharkey
Original Founder
Posts: 1364
Joined: Sat Oct 09, 2004 4:00 am
Contact:

Post by Sharkey »

This particular project was approached "backwards", as I already had the MOSFET array in the junk box, and decided to cobble together something to control the motor on my E-15 tractor. This was probably ten years ago, maybe more. I did some fabrication on the heat sink (something else from the junk box), and drilled and tapped it and some aluminum bar as a spacer. All of the thermal junctions were isolated with heat transfer pads, which I bought as a sheet and cut to size.

After getting the power components put together, I breadboarded the LM3524 chip, and kind of clip-leaded it all together to control the PM motor in my E-12 Electrak. It all worked, but I got busy with other projects, and never built the control circuitry on a perf board. Eventually, I needed my proto board for something else, and shredded up the installed components, making a diagram of the successful circuit.

Fast forward to a couple of weeks ago. The E-15 is long gone. I spent a good deal of time late in the spring refurbing my E-20, which I had bought and basically put into storage for ten years. The E-20 needs controller help, so I broke out the completed MOSFET assembly and attempted to find the circuit diagram. No luck. In spite of my being very organized about this kind of thing, I am no longer able to find the schematic, so I'm starting that part of the project with only a dim memory of what I did back then, and a manufacturers data sheet (oh, and the chipset, I did find it in the parts drawer).

Anyhow, here's a pic of the power assembly. It's the active power handling part of an old Heart Interface DC-AC inverter. It has 16 BUZ71A MOSFET's (replacement part = NTE66), with a power handling capacity of 12 amps apiece, giving the assembly 192 amps continuous power handling.

Image

There's probably a lot more heat sink there than I really need, but since it was in the junk box, and was about the right size, I used it anyway. I figure that it will probably get clogged with dirt, dust, and grass clippings, so having some extra radiation area is not a bad idea.

Here's a second inverter power board, this one has 28 International Rectifier IRF530 MOSFETs, for a load current capacity of 420 amps total. This photo gives a look at the backside of the board and the PC traces there:

Image

Some day, I'll have to find a use for this. Electric drag tractor, anyone?

I've drawn up a schematic of these boards, and when I have a minute, I'll fire up Publisher and draw up something to post.
ysogrimm
Posts: 3
Joined: Mon Jul 21, 2008 6:26 am
Contact:

Post by ysogrimm »

I must say I like that heat sink. I keep toying with the idea of liquid cooling. It may not be all that practical but my heat sink consists of a one inch thick aluminum plate that came from a forklift. I have been thinking about drilling some holes length ways through it, tapping it for fittings and pumping coolant to a heater core with a small fan. Probably not necessary for my electric motorcycle but might be nice once I get a car converted. I suspect over heating would be the main cause of controller failures. Have you ever played around with liquid cooling Sharkey?
A quest to electrify everything I drive
Sharkey
Original Founder
Posts: 1364
Joined: Sat Oct 09, 2004 4:00 am
Contact:

Post by Sharkey »

Liquid cooling should work well, although it's a bit more complex than air, what with circulation pumps and radiators and such. The potential for keeping the power handling devices at a very even temperature is much better with liquid, but you don't want any pump failures!

I picked up the replacement LM3524 chips from town today. They were "samples" from National Semiconductor. Cost me about $8 shipped for two of them. I also put in a order to Digikey for repair parts for two flat panel VGA monitors for work, and slipped in two each of the TC4421 and TC4422 MOSFET driver chips. The difference between the two is that the TC4421 is an inverting driver, it requires a low input to drive the MOSFET gate high. This will address another shortcoming of the circuit above, but I'll leave the results until after I've done some experimenting.
ysogrimm
Posts: 3
Joined: Mon Jul 21, 2008 6:26 am
Contact:

Post by ysogrimm »

Hello. I have been playing around with the circuit and actually have it working on the bench running a 12v pump after making a few changes. I am getting ready to reconfigure the throttle connection as you mentioned since so far I have had it connected to the NI input. Once I have done that I will experiment with current limiting. I think I have all the parts to make it power my electric motorcycle with 72 volts instead of the 36 volts currently used. One thing I don't have that I think I need is some diodes to place across the motor (freewheel?). What I am considering is 12 of the MBR4250 Schottky rectifier diodes which are rated at 250 volts and 40 amps. That is the closest I can find which replace what comes on the 120 volt 400 amp Curtis controllers. Again I appreciate your prior help and look forward to any suggestions. Norm
A quest to electrify everything I drive
lazzer408
Posts: 2
Joined: Fri Jul 24, 2009 5:03 pm
Location: Illinois/wisconsin
Contact:

Post by lazzer408 »

I'm the one who created that file. You should at least -ASK- before you share someone elses work. I spent alot fo time on it.

I'm still happy to answer questions. =)
lazzer408
Posts: 2
Joined: Fri Jul 24, 2009 5:03 pm
Location: Illinois/wisconsin
Contact:

Post by lazzer408 »

(cant find edit button)

I have a MUCH better design in the works right now. With the following.

-Adjustable throttle ramp.
-Open throttle/high throttle detect and disable.
-Pulse-by-pulse current limiting.
-Minimal part count and cost.

The 352x's biggest flaw was it's inability to reach 100%. The NTE1720 replacement would. Apperently the NTE doesn't have proper blanking which allowed it to reach 100% on my test setup. Go figure.

Maybe if I have time more time soon I'll hop in here and give some pointers.
Post Reply

Who is online

Users browsing this forum: Google [Bot] and 10 guests