High power DC/DC converter 12 V to 360 or 720 V 10kW

[Paul E. Schoen]

I am planning to make a circuit to convert 12 VDC or 24 VDC (and possibly
48 VDC) battery power to 360 VDC or 720 VDC to drive the DC link of a three
phase VF motor controller. I'm sure it could be done with high frequency
(20 kHz-200 kHz) magnetics, but I would like to try a somewhat lower
frequency. Here is my plan:

I will use a nominal 1 kVA (60 Hz) toroid core, and remove the 120/240 VAC
windings. I estimate about 0.4 volts per turn. I will drive the transformer
with 600 Hz, so that would be about 4 volts per turn. If I use a center tap
push pull driver on 24 VDC, I think about 8 turns on each leg would be
about right. For 10 kW, I would need close to 400 amps at 24 VDC. I'll use
#10 AWG wire with about 8 in parallel for 50 amps each. For the secondary I
will try 160 turns of #14 for 20 amps at 480 VAC, into a full wave bridge
and capacitors for 720 VDC. If I run it on 12 VDC I should get 360 VDC for
the 240 VAC VF controller.

Now, for the drive, I will try using a PIC or an SG3526 to generate a
simple square wave with some dead time. I may add a capacitor across the
primary to reduce high voltage switching spikes. The current sinking
components will be power MOSFETs. I may use IRL2203N (30 V, 100 A) for my
initial prototype (which will be just 0.75 kW and 12 VDC). For this I
figure about 60 amps input, shared by the two MOSFETs. At an ON resistance
of 0.007 ohms, the power should be about 25 watts. For my final product, I
will use something like STP140NF55, and about four in parallel. They are 55
V, 80 A, and 0.005 ohm. Not bad for $1.56 each. What I will have is a
battery powered three phase 15 HP motor controller.

What I am unsure about is high voltage spikes and high current surges
resulting from using a square wave on the transformer. I tried modeling
this using Tina, and it seemed to be OK, but it helped to add a capacitor
across the transformer primary and some inductors in series with the legs.
I know that uncontrolled transients with this much power can quickly turn
quite ugly, so I want to be sure everything will be OK.

Any insight and comments will be appreciated. Thanks.

Paul E. Schoen
www.pstech-inc.com

 

[[email protected]]

Just a quick note.
Your trannies will see 48V on their drains when the other "side"
switches in a 24V system . For a 12V system the IRL2203N will probably
be ok but not for a 24V system.
This is assuming I have not misinterperated what you are doing.
Cheers
Rob

 

[Rich Grise]

I am planning to make a circuit to convert 12 VDC or 24 VDC (and possibly
48 VDC) battery power to 360 VDC or 720 VDC to drive the DC link of a
three phase VF motor controller.

The only suggestion I really have is to start with the 48V - the switching
transients will be higher, but you'll only have to deal with 1/4 the
current.

Good Luck!
Rich

 

[John Larkin]

I am planning to make a circuit to convert 12 VDC or 24 VDC (and possibly
48 VDC) battery power to 360 VDC or 720 VDC to drive the DC link of a three
phase VF motor controller. I'm sure it could be done with high frequency
(20 kHz-200 kHz) magnetics, but I would like to try a somewhat lower
frequency. Here is my plan:

I will use a nominal 1 kVA (60 Hz) toroid core, and remove the 120/240 VAC
windings. I estimate about 0.4 volts per turn. I will drive the transformer
with 600 Hz, so that would be about 4 volts per turn. If I use a center tap
push pull driver on 24 VDC, I think about 8 turns on each leg would be
about right. For 10 kW, I would need close to 400 amps at 24 VDC. I'll use
#10 AWG wire with about 8 in parallel for 50 amps each. For the secondary I
will try 160 turns of #14 for 20 amps at 480 VAC, into a full wave bridge
and capacitors for 720 VDC. If I run it on 12 VDC I should get 360 VDC for
the 240 VAC VF controller.

CT-PP will be copper inefficient, and getting rid of heat from the
transformer will be, umm, important.

Now, for the drive, I will try using a PIC or an SG3526 to generate a
simple square wave with some dead time. I may add a capacitor across the
primary to reduce high voltage switching spikes. The current sinking
components will be power MOSFETs. I may use IRL2203N (30 V, 100 A) for my
initial prototype (which will be just 0.75 kW and 12 VDC). For this I
figure about 60 amps input, shared by the two MOSFETs. At an ON resistance
of 0.007 ohms, the power should be about 25 watts. For my final product, I
will use something like STP140NF55, and about four in parallel. They are 55
V, 80 A, and 0.005 ohm. Not bad for $1.56 each. What I will have is a
battery powered three phase 15 HP motor controller.

What I am unsure about is high voltage spikes and high current surges
resulting from using a square wave on the transformer. I tried modeling
this using Tina, and it seemed to be OK, but it helped to add a capacitor
across the transformer primary and some inductors in series with the legs.
I know that uncontrolled transients with this much power can quickly turn
quite ugly, so I want to be sure everything will be OK.

Any insight and comments will be appreciated. Thanks.

Paul E. Schoen
www.pstech-inc.com

Where are you going to get 12 volts at 1000 amps?

John

 

[Paul E. Schoen]

The only suggestion I really have is to start with the 48V - the
switching
transients will be higher, but you'll only have to deal with 1/4 the
current.

Good Luck!
Rich

I will probably make the first prototype for 12 VDC source and 360 VDC
output at nominal 1 HP (750 W). I already have the 240 VAC VF controller,
IRL2203N transistors and a 12 VDC SLA battery, which should handle the
50-60 amps. I also have several 240 VAC motors of 1 to 2 HP.

Once I prove the concept, and work out any unforeseen bugs, I will be able
to scale up the design to the next step, which would be 24 VDC to 720 VDC,
and about 4 HP (3 kW). I will need to obtain or make a 480 VAC VF
controller. I found some on eBay for under $100, but the one I was most
interested in (7620803063) has a date of 1988 and I wonder how good it
would be. I want a 480 VAC controller to see if I can double the HP of the
240 VAC motor by running at 120 Hz. I might use this system to make an
electric powered lawn tractor, which will also be a "vehicle" for testing
other concepts.

My final goal is a controller and motor system for an electrically powered
car. My original idea was to rewind a motor for low voltage and run it
directly from a battery pack of 48 or 72 VDC. I figure about 10-15 HP per
wheel should be OK. I was going to use 3 HP 600 or 900 RPM motors and then
push the HP to 4x or 6x, and use two motors on the rear wheels. That may
prove to be the best final solution, but I wanted to try some ideas with
standard motors first, before investing in rewinding. I did rewind a small
3/4 HP motor for 8 VAC with 12 poles and it runs at least up to 1800 RPM at
180 Hz, and I will do some experiments with my low voltage VF drive using
that motor before I rewind a larger one.

I'll let you know as I progress. I appreciate your comments and
suggestions.

Paul

 

[Ignoramus8797]

would be. I want a 480 VAC controller to see if I can double the HP of the
240 VAC motor by running at 120 Hz.

You are a fun guy!

i

 

[Fritz Schlunder]

I am planning to make a circuit to convert 12 VDC or 24 VDC (and possibly
48 VDC) battery power to 360 VDC or 720 VDC to drive the DC link of a three
phase VF motor controller. I'm sure it could be done with high frequency
(20 kHz-200 kHz) magnetics, but I would like to try a somewhat lower
frequency. Here is my plan:

I will use a nominal 1 kVA (60 Hz) toroid core, and remove the 120/240 VAC
windings. I estimate about 0.4 volts per turn. I will drive the transformer
with 600 Hz, so that would be about 4 volts per turn. If I use a center tap
push pull driver on 24 VDC, I think about 8 turns on each leg would be
about right. For 10 kW, I would need close to 400 amps at 24 VDC. I'll use
#10 AWG wire with about 8 in parallel for 50 amps each. For the secondary I
will try 160 turns of #14 for 20 amps at 480 VAC, into a full wave bridge
and capacitors for 720 VDC. If I run it on 12 VDC I should get 360 VDC for
the 240 VAC VF controller.

Unfortunately your transformer will likely start smoking before you even
hook up your load. If you 10X your frequency, that should allow you to 10X
your applied voltage before hitting your saturation limit. Unfortunately,
the transformer won't be saturation limited, it will be thermally limited
long before you apply 10X the volts/turn.

If I understand my magnetics correctly, 10X-ing the applied voltage and
10X-ing the frequency should 10X the core hysteresis loss. Core hysteresis
loss is basically independent of load power.

That won't be your real problem though. The real problem will be core eddy
current loss. If I understand things correctly, all else being equal,
10X-ing the volts/turn should 100X (a factor of 10 squared, from the formula
V^2/R = Power) the core eddy current loss. Core eddy current loss is also
more or less independent of load power.

So, all else being equal, the net result will be your total core loss will
likely increase by some factor between 10 and 100.

There is a small advantage to using square wave excitation compared to
sinusoid excitation on the transformer however. I assume your 0.4V/turn
figure is an RMS AC value. For an equivalent RMS voltage value, a full duty
cycle square wave will subject the core to less volt seconds/cycle than the
equivalent sinusoidal RMS voltage. Nevertheless, this small advantage will
not keep your transformer from smoking, the above still applies.

You can still rewind the transformer for more power than the original 1kVA
if you use 600Hz, but 10X-ing the power level is not realistic (at least not
without much improved cooling).

Now, for the drive, I will try using a PIC or an SG3526 to generate a
simple square wave with some dead time. I may add a capacitor across the
primary to reduce high voltage switching spikes. The current sinking
components will be power MOSFETs. I may use IRL2203N (30 V, 100 A) for my
initial prototype (which will be just 0.75 kW and 12 VDC). For this I
figure about 60 amps input, shared by the two MOSFETs. At an ON resistance
of 0.007 ohms, the power should be about 25 watts. For my final product, I
will use something like STP140NF55, and about four in parallel. They are 55
V, 80 A, and 0.005 ohm. Not bad for $1.56 each. What I will have is a
battery powered three phase 15 HP motor controller.

The center tapped push pull topology is not a very good choice for such
large current levels. Any leakage inductance on the primary will contain
tremendous amounts of energy at the moment of MOSFET turn off. This energy
must go somewhere, and it will have to avalanche the MOSFET in that topology
(unless you use a couple of huge RCD snubbers across each half of the
primary). The energy will likely destroy the MOSFETs.

The H-bridge topology would be much better suited to this task. In
particular, the H-bridge topology allows energy stored in the leakage
inductance of the primary to be returned to the DC input rail.

 

[EE123]

I am planning to make a circuit to convert 12 VDC or 24 VDC (and possibly
48 VDC) battery power to 360 VDC or 720 VDC to drive the DC link of a three
phase VF motor controller. I'm sure it could be done with high frequency
(20 kHz-200 kHz) magnetics, but I would like to try a somewhat lower
frequency. Here is my plan:

I will use a nominal 1 kVA (60 Hz) toroid core, and remove the 120/240 VAC
windings. I estimate about 0.4 volts per turn. I will drive the transformer
with 600 Hz, so that would be about 4 volts per turn. If I use a center tap
push pull driver on 24 VDC, I think about 8 turns on each leg would be
about right. For 10 kW, I would need close to 400 amps at 24 VDC. I'll use
#10 AWG wire with about 8 in parallel for 50 amps each. For the secondary I
will try 160 turns of #14 for 20 amps at 480 VAC, into a full wave bridge
and capacitors for 720 VDC. If I run it on 12 VDC I should get 360 VDC for
the 240 VAC VF controller.

Now, for the drive, I will try using a PIC or an SG3526 to generate a
simple square wave with some dead time. I may add a capacitor across the
primary to reduce high voltage switching spikes. The current sinking
components will be power MOSFETs. I may use IRL2203N (30 V, 100 A) for my
initial prototype (which will be just 0.75 kW and 12 VDC). For this I
figure about 60 amps input, shared by the two MOSFETs. At an ON resistance
of 0.007 ohms, the power should be about 25 watts. For my final product, I
will use something like STP140NF55, and about four in parallel. They are 55
V, 80 A, and 0.005 ohm. Not bad for $1.56 each. What I will have is a
battery powered three phase 15 HP motor controller.

What I am unsure about is high voltage spikes and high current surges
resulting from using a square wave on the transformer. I tried modeling
this using Tina, and it seemed to be OK, but it helped to add a capacitor
across the transformer primary and some inductors in series with the legs.
I know that uncontrolled transients with this much power can quickly turn
quite ugly, so I want to be sure everything will be OK.

Any insight and comments will be appreciated. Thanks.

Paul E. Schoen
www.pstech-inc.com

As John Larkin cogently pointed out: 12V at 1000 amps
That is excellent as it would give you about 83 % conversion
efficiency.
While not state of the art, it was a good point. Leading to:
mucho problems.
Without proper insulation, you could turn your golf cart into a
4 wheel MO BYLE BBQ pit!

I designed inverters at Ford Motor as part of their electric vehicle
program
1kW 2kW 5kW with about 85 % efficiency in the early 1990's.
Current state of the art is about 92 % efficiency.

Due to mistakes made at various manufacturing stages, we saw a lot of
components go Fzzzzzzzzzz and up in smoke, sometimes with serious
damage!

Also, during the development of the 5kW inverters, EMI issues really
surfaced.
A nearby Ford EMI lab told us that we were generating fields strong
enough
to "confuse" some the on board electronics on automobiles. They said it
took about
a month and some direction finding equipment to track us down before
the FCC
was able to find us. We had to build a shielded lab.

At this point, the manager had finally agreed that simulation was an
important
ingredient on the development process. But these always work in
simualtion.

Good luck,
Dave

 

[Paul E. Schoen]

As John Larkin cogently pointed out: 12V at 1000 amps
That is excellent as it would give you about 83 % conversion
efficiency.
While not state of the art, it was a good point. Leading to:
mucho problems.
Without proper insulation, you could turn your golf cart into a
4 wheel MO BYLE BBQ pit!

I designed inverters at Ford Motor as part of their electric vehicle
program
1kW 2kW 5kW with about 85 % efficiency in the early 1990's.
Current state of the art is about 92 % efficiency.

Due to mistakes made at various manufacturing stages, we saw a lot of
components go Fzzzzzzzzzz and up in smoke, sometimes with serious
damage!

Also, during the development of the 5kW inverters, EMI issues really
surfaced.
A nearby Ford EMI lab told us that we were generating fields strong
enough
to "confuse" some the on board electronics on automobiles. They said it
took about
a month and some direction finding equipment to track us down before
the FCC
was able to find us. We had to build a shielded lab.

At this point, the manager had finally agreed that simulation was an
important
ingredient on the development process. But these always work in
simualtion.

Good luck,
Dave

I have assembled the first prototype, and it seems to work OK. The toroid
core is actually 1 volt per turn, and about 2 kVA nominal at 60 Hz. I used
6 turns of #10 AWG for the two primary legs, and two secondary windings of
30 turns each of #15. I use two MT60N06 MOSFETs for switching, with
snubbers across each. I am using an SG3526 PWM IC. I have a FWB on one of
the secondaries, into a 1500 uF 450 VDC capacitor.

As a first try, I was able to get 81 VDC with a 15 VDC supply, at about 400
mA with about 20% PWM. Then I adjusted the PWM to near 100% and got 56 VDC
with 10 VDC at 1.3 amps input. I added a load of 900 ohms, and got 54 VDC
at 1.5 amps input. I'm running at about 450 Hz. The current goes up if I
reduce the frequency.

From this, it looks like there will be a minimum loss of about 13 watts. I
was able to get 3.4 watts of power out for an additional 2 watts input. I
will need to apply much heavier loads to get a true idea of the efficiency.
I thought most of the power was lost in the forward conduction of the
transistors, but with 0.016 ohms at 1.3 amps that is only 0.03 W. It is
probably lost in the transformer. If the transformer is rated for 2000
watts, 13 watts loss is better than 99% efficiency, and if it is rated at
10 KVA at this frequency, it is much better. At 600 watts, with 50 amps
being switched at 12 V, the forward drop in the MOSFETs would be about 40
watts, so efficiency would be about 93%. I will probably need better
MOSFETs or a lot of them in parallel to get my 10 kW.

Of course, higher currents will also result in copper losses. I have room
to add another set of primary windings and another set of secondaries. I
would expect the no load primary current to double with two in parallel.
With four secondaries, I should be able to get 260 VDC at 12 VDC. I may
need to remove one primary turn to get 320 VDC which is about minimum to
run a 208/240 VAC motor, but realistically I expect to use 24 VDC for 240
VAC, and 48 VDC for 480 VAC. It looks like the output voltage is just about
5 times the input voltage, which is the winding ratio of 30 to 6.

I will need to package this monster a bit more carefully to do higher power
testing. I think the concept is proven enough to proceed to the next step.
I'll let you know what sort of results I get.

BTW, what ever happened to Ford's electric car program?

Paul

 

[Glen Walpert]

On Fri, 26 May 2006 07:58:39 -0400, "Paul E. Schoen"

BTW, what ever happened to Ford's electric car program?

http://www.ford.com/en/vehicles/specialtyVehicles/environmental/fuelCell/default.htm

Like all other major auto manufacturers they have realized that
battery technology has a *long* way to go before even coming close to
the performance of the current generation of hydrogen fuel cell
electric vehicles, with their 65 kW fuel cells, 60 kW motors, and 400
kM range. While others are busy talking about how impractical fuel
cells are the auto manufacturers are busy developing them. I guess
all af the major auto manufacturers must be clueless .

 

[John Larkin]

On Fri, 26 May 2006 07:58:39 -0400, "Paul E. Schoen"



http://www.ford.com/en/vehicles/specialtyVehicles/environmental/fuelCell/default.htm

Like all other major auto manufacturers they have realized that
battery technology has a *long* way to go before even coming close to
the performance of the current generation of hydrogen fuel cell
electric vehicles, with their 65 kW fuel cells, 60 kW motors, and 400
kM range. While others are busy talking about how impractical fuel
cells are the auto manufacturers are busy developing them. I guess
all af the major auto manufacturers must be clueless .

Fuel cells are insanely expensive and short-lived. There's no
practical way to store a useful amount of hydrogen on-board a car, and
no way to manufacture the hydrogen that's not polluting, absurdly
inefficient, or both.

The thing that electric and hydrogen-powered vehicles almost always
have in common is that they're small, light, and very aerodynamic. So
it makes a lot more sense to just make small, light, slippery cars and
power them from gasoline engines, maybe with hybrid electric boost.

John

 

[John Larkin]

From this, it looks like there will be a minimum loss of about 13 watts. I
was able to get 3.4 watts of power out for an additional 2 watts input. I
will need to apply much heavier loads to get a true idea of the efficiency.
I thought most of the power was lost in the forward conduction of the
transistors, but with 0.016 ohms at 1.3 amps that is only 0.03 W. It is
probably lost in the transformer. If the transformer is rated for 2000
watts, 13 watts loss is better than 99% efficiency, and if it is rated at
10 KVA at this frequency, it is much better.

Bad, bad math. Sorry.

Of course, higher currents will also result in copper losses.

Sure will. Without copper losses, a megawatt transformer would fit in
your pocket.

I have room
to add another set of primary windings and another set of secondaries. I
would expect the no load primary current to double with two in parallel.

No, magnetizing current and core loss wouldn't change.

With four secondaries, I should be able to get 260 VDC at 12 VDC. I may
need to remove one primary turn to get 320 VDC which is about minimum to
run a 208/240 VAC motor, but realistically I expect to use 24 VDC for 240
VAC, and 48 VDC for 480 VAC. It looks like the output voltage is just about
5 times the input voltage, which is the winding ratio of 30 to 6.

As you increase primary voltage, you must increase primary turns to
avoid core saturation. More turns of thinner wire blows up the copper
loss. The only way out of that dilemma is to keep the pri turns
constant but increase frequency, which has a set of problems of its
own.

Mother Nature was in a very bad mood the day transformers were
invented.


John

 

[Charlie Edmondson]

On Fri, 26 May 2006 07:58:39 -0400, "Paul E. Schoen"




http://www.ford.com/en/vehicles/specialtyVehicles/environmental/fuelCell/default.htm

Like all other major auto manufacturers they have realized that
battery technology has a *long* way to go before even coming close to
the performance of the current generation of hydrogen fuel cell
electric vehicles, with their 65 kW fuel cells, 60 kW motors, and 400
kM range. While others are busy talking about how impractical fuel
cells are the auto manufacturers are busy developing them. I guess
all af the major auto manufacturers must be clueless .

Yep, pretty much so. The clue that they have is that there is big bucks
in researching hydrogen vehicles (Hey, the government is giving away
free money!) but that there is NO MONEY WHATSOEVER in actually producing
the expensive, short-lived, short range dangerous things for commercial
sale.

However, there might be a reasonable market for electric vehicles for
commuters, but these might take away sales from SUVs and other profit
centers, so they put them on a back burner.

Charlie

 

[Paul E. Schoen]

Fuel cells are insanely expensive and short-lived. There's no
practical way to store a useful amount of hydrogen on-board a car, and
no way to manufacture the hydrogen that's not polluting, absurdly
inefficient, or both.

The thing that electric and hydrogen-powered vehicles almost always
have in common is that they're small, light, and very aerodynamic. So
it makes a lot more sense to just make small, light, slippery cars and
power them from gasoline engines, maybe with hybrid electric boost.

John

I agree. My 1998 Saturn SL1 5 speed bare bones car averages 36 MPG and can
get 45 MPG on the highway. Ten years later and you probably can't get a
non-hybrid that will do as well, because the commercials push ZoomZoomZoom.
For the manufacturers, it's all about immediate profits, and they move so
slowly that they won't be able to come out with an economical vehicle to
meet present demands, due to fuel costs, for about five years. Also, people
will not really "demand" a lightweight, efficient vehicle until gasoline
reaches about $5/gallon. And there is the question of safety, which will
make it hard to reduce weight while there are still so many heavy cars,
SUVs, and trucks, being driven by increasingly aggressive, cell-phone
distracted drivers.

I am still considering adding rear wheel drive electric boost to my other
Saturn SW1, which has a failing motor. Maybe I'll go all electric, which is
probably practical for my usual 25 mile commute and other trips which
usually are less than 100 miles a day. About 500 lb of batteries should
give me that, and the energy cost will probably be about $2.00, compared to
about $10 for gas at 30 MPG.

Ultimately, I think hybrids will be designed that use ethanol or veggie
oil. Another possibility is to add a modular motor/generator pack to an
electric vehicle for long trips, so you leave the extra weight at home when
you are just commuting. There are many ideas floating around that can solve
our energy crisis, but we need public acceptance and government support to
make them work on a large scale. Eventually, I hope, battery technology
will make EVs viable for 90% of all transportation needs.

Paul

 

[Rich Grise]

Fuel cells are insanely expensive and short-lived. There's no
practical way to store a useful amount of hydrogen on-board a car, and
no way to manufacture the hydrogen that's not polluting, absurdly
inefficient, or both.

The thing that electric and hydrogen-powered vehicles almost always
have in common is that they're small, light, and very aerodynamic. So
it makes a lot more sense to just make small, light, slippery cars and
power them from gasoline engines, maybe with hybrid electric boost.

I'm hoping to meet a witch who knows some physics, and create a Zero Point
Daemon for unlimited free energy. ;-)

Cheers!
Rich

 

[Rich the Philosophizer]

Bad, bad math. Sorry.


Sure will. Without copper losses, a megawatt transformer would fit in
your pocket.


No, magnetizing current and core loss wouldn't change.


As you increase primary voltage, you must increase primary turns to
avoid core saturation. More turns of thinner wire blows up the copper
loss. The only way out of that dilemma is to keep the pri turns
constant but increase frequency, which has a set of problems of its
own.

Mother Nature was in a very bad mood the day transformers were
invented.

That's because Mother's energy is magnetic, and She doesn't like being
pushed around by cold, unfeeling electricity.

Thanks,
Rich

 

[John Popelish]

I'm hoping to meet a witch who knows some physics, and create a Zero Point
Daemon for unlimited free energy. ;-)

Be careful what you wish for. Such a thing would be built into a
terrible weapon long before it was put into any transportation
machine. Availability of unlimited free energy would be the the
beginning of a very quick end of the human race, and possibly the planet.

 

[Rich Grise]

Be careful what you wish for. Such a thing would be built into a
terrible weapon long before it was put into any transportation
machine. Availability of unlimited free energy would be the the
beginning of a very quick end of the human race, and possibly the planet.

No, you see, we'd create a daemon who would refuse to let his powers be
used for evil.

Take the Rich Grise/Insert Name Here power unit, and try to make a bomb,
and it would say, "**** You." Or maybe just azp the individual evil
person. ;-)

Cheers!
Rich

 

[John Larkin]

No, you see, we'd create a daemon who would refuse to let his powers be
used for evil.

Geez, you sure sound like a person who has no kids.

John

 

[krw]

Geez, you sure sound like a person who has no kids.

Don't blame Grease. It's hereditary.