Here is some technical discussion from the usenet newsgroup sci.electronics.design:
I'm looking into a battery-powered small vehicle (tractor) design which will use a three-phase motor and VF controller, and I want to create the DC bus, or link, voltage from several 12V SLA batteries. I will need 360 or 720 VDC for 208/240 or 480 VAC motors.
I thought about using some of the inexpensive modified sine wave inverters available from Harbor Freight and other companies, where a 2000W inverter can be obtained for about $150. My first thought was to rectify the output and filter it, to get about 180 VDC and then use two or four in series for the higher voltages. But I think these inverters create their own DC bus using a switching power supply and then use high voltage transistors or IGBTs in a bridge which produces the modified sine wave.
I have searched for schematics but most of them are for simple inverters using a 60 Hz power transformer driven with a square wave, but that's not what I want. And other schematics are like the reference designs for UPS systems, which are probably similar to what I want, and they can have either modified sine wave or true sine (PWM) outputs. It's probably not hard to design and build my own, but if I can use off-the-shelf inverters then I'll be ahead of the game.
My next step is to open up one of the small inverters I have to see how they are made, and maybe discover and measure the DC bus voltage, but if anyone has the actual schematics for one of these, I'd appreciate a look at it. Otherwise, I'll let you know what I find.
=======================================================
> You could also look for literature on DC-DC converters. You need a
> honkin' big one, and if you're designing from scratch, and to a point
> the higher the input voltage the better off you'll be -- the currents
> are lower, and the transformer turns ratio is less. But you're getting
> into some pretty serious power levels, which means that the design
> of the supply isn't trivial. If you want to focus on tractor-building
> and not power supply design, then finding some existing thing and
> hacking it is the way to go.
I think off-the-shelf DC-DC converters at the power levels I'm looking at
will be rare and hugely expensive. I'm sure I won't find any for as little
as $100/kW as the Harbor Freight units are. And this is now a
proof-of-concept design, and targeted to people who have tractors and want
to do a conversion. See my thread(s):
http://www.mytractorforum.com/showthread.php?t=226278
http://www.mytractorforum.com/showthread.php?t=226645
And I have an idea for a simple DC-DC converter. I can use a nominal 500VA
toroid and rewind it with less turns of much heavier wire, and drive it with
a 600 Hz square wave. It should be able to produce about 5 kVA, and such a
toroid is about the size of your fist and costs only about $50 and is much
more rugged than ferrite cores for higher frequency magnetics. I think it
will work out about the same size and efficiency as a high-tech 5000W
switching supply, but it can be much simpler to build.
> See the comments about just using battery voltage -- they're well
> meant and meaningful. In particular, keep in mind that if you rip
> all the wire out of a motor and rewind it with thicker wire, filling
> the space just as efficiently as it was filled before, you end up
> scaling the current up exactly as much as you scale the voltage
> down -- so all the efficiencies, maximum torques, etc., stay the
> same, just at a different voltage vs. current.
Actually, I did that very thing about 8 years ago. I pulled the original
windings out of a single phase 120V 3/4 HP fan motor, and rewound it with
much heavier wire (something like #18 or #16). And I converted it to a 6
pole three phase motor. It was designed to run on about 8 VAC so that a 12V
battery could drive it directly, and I wanted to be able to overclock it by
about 4x. I used a 2HP VF motor controller running on 240 VAC single phase,
and I used two step-down transformers to match the motor voltage. It ran OK
at 180 Hz with a fan load. And I made another motor that I drove using a
three-phase bridge and modified sine wave (rectangular wave) directly from a
12V battery, using a PIC controller.
Here is a picture of the stator:
http://www.pstech-inc.com/pictures/Motor_winding_320x240-01.jpg
It's probably not filled as tightly as the original windings. I wound by
hand rather than making belts externally and then fitting them in the
lamination slots.
My ultimate goal is to make a low-voltage motor that can be overclocked up
to 6x (360 Hz), and theoretically get 6 HP from a 1 HP frame. I have some
motors that I got cheap with the intention of rewinding them, but it's
difficult to do so with the heavier wire, and I am hesitant to rip out the
windings from these motors since they are new and unused. So, first I want
to use them as-is, and maybe overclock the 1.5 HP 240/480 4 pole motor to
2x, which is 3600 RPM, same as the gas motor in my tractor, and I was going
to use a 480V controller so I can get as much as 3 HP. The gas engine is
about 7 HP peak, while the induction motor should be able to produce 2x to
3x normal as peak, so it should be about equivalent. This is the tractor I
may use for this project:
http://youtu.be/mF24Sict-MM
> Of course, if you string 30 12V batteries in series then you have
> your 360V DC rail right there. Just don't lick your fingers and
> touch them to the end points to see if the thing is live.
30 batteries will be a problem. I'm looking at using 100 A-H SLA batteries,
so two of them with 12V-120V inverters should give me the 360V I need, and
2400 W-H which may provide up to one hour running time at 3-4 HP average.
Then I can go with 4 batteries to get 2 hours or twice the power.
I'm really focused on using a three-phase induction motor rather than the
more expensive and fragile BLDCs and PM or series wound brushed DC motors
that are the usual technology.
=====================================================
> Essentially there is a 12VDC to 170VDC converter in there that turns
> on and off. This is followed by a full-bridge that switches output
> polarity at mid-point of the off-phases. That's pretty much it.
I found a 150 VDC bus in the 175 watt unit I took apart. And the switching
frequency is 43 kHz. The transformer is quite small, a little over 1" cube.
> Mind that super-cheapo versions sometimes save a few pennies or
> "boost efficiency" by only generating 160VDC and roaching that on
> top of the incoming 12VDC. Then the input and output are not
> isolated. This has resulted in some "interesting" experiences for
> some folks. "Yo, dude, why's there smoke comin' out of your camper?"
I had thought it was isolated. So that makes it riskier to use two of these
in series from separate batteries, as at least one of the batteries will be
about 150 volts above the other. I measured the resistance from the input to
the output and there is no direct connection. But from the schematic you
provided, that's probably because the output is bridged by MOSFETs, which
have high resistance until turned on.
> Here ya go:
>
http://www.analogzone.com/dearden_042307.pdf
That's very helpful. It looks like the 150V bus could be isolated by
removing some of the output components and the connection of the diode
bridge to battery +12V. But that defeats the purpose of using an
off-the-shelf component. I think I can make a simple DC-DC converter that
can produce, say, 1200 watts at 180 VDC for each battery, with isolation.
Then I can use two or four in series for 360 and 720 VDC bus voltages, for 3
HP and 6 HP. I could make them as modular components designed for series and
parallel connection, and have a way to produce a regenerative charging
current during dynamic braking (although that might not be very useful for a
tractor unless it is used for frequent up/down trips rather than, say, lawn
mowing where it needs power at all times. I could also add ultracaps for
surge power and energy storage, so the the battery would see only average
current within its more efficient range, but they are expensive.
==================================================
> I have a AC-Delco 2500W Msine inverter. No schematics, but it has
> 2 front end converters ~170VDC stacked for 340ish VDC.
> Which is then switched at 60 hz into AC. There is a common ground
> point, so there is +/- 170V to the output.
> Looks like this...
> <http://www.sportsmansguide.com/net/cb/ac-delco-2500-watt-inverter.aspx?a=421592>
> Must have been a school project, uses a TL494!
That particular model was sold out, but they had others from 200 watts up to
6000 watts. The 900 watt model was the lowest price per watt, at $0.07. Most
of the others were about $0.10/watt, which is about what the Harbor Freight
units sell for. It does seem to be better to design this from scratch, and I
think it should be possible to make it for less than $0.05/watt.
A 200 VA toroid kit is just $63:
http://www.toroid.com/standard_transformers/transformer_kits/transformer_kits.htm.
This is a toroid core prewound for 120 VAC 60 Hz, and has 0.23 V/turn. I
would just remove the primary winding and replace it with something like 10
turns of much heavier wire (probably actually four or more wires in
parallel), which would be about 23 VAC at 600 Hz. Maybe with a center tap
for 12-0-12. Then I would wind a secondary with 140 turns, for 320 VAC. A 12
volt square wave using a center tap design with just two power transistors
in push-pull should produce a square wave output with 320 volts peak, which
can be rectified to 320 VDC for the link voltage. This should provide up to
2 kVA, more than enough for a 2 HP motor. The battery draw would be 167
amps, however, so I should really design for about 1000 watts average for
each module.
In that case, I could even use the smallest (80 VA) toroid kit, which is
just $52, and should provide at least 800 watts for 1 HP, and a more
realistic 87 amps draw. I'm sure I could get the unwound cores for much
less, and even have them custom wound, for under $100. Then a pair of husky
MOSFETs like IRFB3077 at $3.50 each, a 5A 600V bridge and 470 uF 450 V
capacitor, controlled by a PIC or SG3526, and a few minor parts, and I'll
have a 1 kVA module that can be connected in series or parallel with its own
battery for the voltage and current needed. The whole thing should fit in a
2" x 4" x 6" metal box which will be a heat sink for up to 100 watts if it
is only 90% efficient.
Well, maybe that does come out as more than $0.10/watt. But if I made
thousands of them, the cost would go way down. And the biggest expense will
be the batteries, at about $200 each:
<http://www.homedepot.com/Electrical-Alternative-Energy-Solutions-Solar-Power/h_d1/N-5yc1vZbm18/R-202548459/h_d2/ProductDisplay?catalogId=10053&langId=-1&storeId=10051>
I couldn't find actual technical specs, but this seems to be the battery:
http://www.exide.com/Media/files/Do...nload/ExideRoad Force Heavy Duty Brochure.pdf
It seems like a good battery, but actual A-H capacity seems to go down
sharply even at an 8 hr. rate compared to 20 hr. and 100 hr. rates. So I
probably need to design for no higher than a 4 hour rate for continuous
duty, so a 100 A-H battery would be good for only 25 amps. Yeah, battery
technology is really the kicker for practical EVs. Here is a pretty good
reference:
http://www.windsun.com/Batteries/Battery_FAQ.htm
Thanks for the reality check.
=====================================================
I did a little more investigation, and found some Lithium batteries designed
for EVs. Here is a 100 A-H battery for $150:
http://www.ebay.com/itm/NEW-CALB-10...Parts_Accessories&vxp=mtr&hash=item256c18331c
But it's only 3.2 volts so four of them are needed for the same W-H capacity
as the Lead-Acid. So about 3x the cost. Here is the spec:
http://en.calb.cn/Product/?id-115.html
But the good news is that it holds 3 volts even at 100A draw (1C), and can
even handle 5C surges. Here is more information from a US distributor:
http://www.emotorwerks.com/emw3/product/calb-battery-100ah-cells/
Their home page
http://emotorwerks.com/cgi-bin/index.pl has some interesting
information. They show about $10,000 for a complete system for an ordinary
car, for 100 mile range with a 500 pound battery pack, and an expected
lifetime of 200,000 miles (about 2000 charge/discharge cycles). A typical
small car is probably about 40 HP, so a 4 HP tractor might be only about
$1000 (although $2000 is probably more like it, since it probably does not
scale linearly). So 10 batteries will give 3000 W-H or 4 HP for one hour,
and cost $1500. The other components could probably be purchased/built for
$500.
Things may become even more interesting, and EVs more competitive, when
gasoline passes the magic $5 point (probably within a year), and if battery
technology can cut the present costs in half (which I think is very
realistic).
And I'd also like to put a datalogger in my controller to determine how much
power and energy are actually needed for various lawn tractor needs. I have
a feeling that basic lawn mowing on a flat surface might take only 1 to 2
HP. And hills may average out if some regeneration is possible. Of course
there will be some tough jobs that require raw power, and for that a diesel
engine may be the way to go. I just bought a diesel lawn tractor with a 15
HP Kubota engine for $1000, and you can see it in action at
http://youtu.be/6KbZc0WLq4Q, or go for a ride with me at
http://youtu.be/BgXon0J15as
===================================================
These are the sorts of technical questions that must be discussed for EVs. I learned a lot from my research, and there is still much to learn. EVs are very impressive for "short" time (pun acknowledged) demonstrations like the weighted sled pulling. But practicality will be determined by the ability to do the usual work of mowing a lawn or clearing snow and such, which involves a certain amount of power over a certain period of time.
Has anyone actually measured how much power it takes to mow an average lawn with various size mowers? This is difficult to do with a gas or diesel engine, but a battery powered vehicle can easily record the average and peak current draw (and power), as well as total energy usage. It can probably be estimated by knowing how much fuel is required for a certain job, and then estimating the efficiency (probably 40% or less) and calculating the BTUs (or kW-hr) of the fuel used.
