Another DIY Inverter - 125kVA

AC, DC, amps, volts and kilowatt. It's all discussed in here
Post Reply
tesla500
Noobie
Posts: 10
Joined: Thu, 03 Nov 2011, 23:52
Real Name: David Kronstein

Another DIY Inverter - 125kVA

Post by tesla500 » Fri, 04 Nov 2011, 19:02

Hi Guys,

I just discovered this forum, I had no idea so many others were working on DIY AC inverters. I've been working on one myself for the last while. I've got a prototype running, see the following video:

AC Drive System for EV

I had lost interest in the project soon after getting that far, but I've gotten back into it now and I'm determined to finish it!

I've settled on APTGT600A60G IGBTs, which should be good for about 300Arms, maybe 350A, output at 360V bus voltage. These ones used to be about $360 each, but they recently dropped to $240, so I've got 6 on order.

I've got the schematics pretty much done, see the links below. Let me know if you'd like the Altium sources.

Control Board (PDF) (sorry for the weird page ordering)
Power Board (PDF)

Here's a CAD drawing of the layout:

Image

The control board will mount over the top, with a cutout to access the power terminals. I've tried to get the IGBTs in the middle to allow the best heat spreading over the base, as opposed to having them at one end. This may allow air cooling to be practical.


I'm trying to finalize on the IO, here's a list of what I've got at the moment:

Analog x3     Throttle, regen, auxiliary
Throttle enable (pedal zero switch)
Regen enable (pedal zero switch)
Aux digital
Forward
Reverse
Cruise control (On, Off, Set speed, speed up, speed down)
Drive, Neutral, reverse LEDs
Overtemperature LED
Analog gauge driver
RS232
CAN
Variable speed fan drives x2 (4A 0-12V)

And I'm probably going to add a relay output to activate the brake light during regen braking.

Is anything missing? Any features you'd like to see?


Regards,
David

User avatar
woody
Senior Member
Posts: 1715
Joined: Sat, 21 Jun 2008, 02:03
Real Name: Anthony Wood
Location: Mt Colah

Another DIY Inverter - 125kVA

Post by woody » Fri, 04 Nov 2011, 20:01

Welcome!
Awesome work!

Going by your accent I guess you are in North America somewhere?

(Crazy) ideas follow.

By analogue gauge driver, do you mean 1 pulse per rotation for tacho and/or 1000 pulses per mile for electronic speedo/odometer?

Maybe some coloumb counting on the DC bus input and a fuel gauge output?
Temperature gauge output for existing temperature gauge?

Different Regen modes selectable (for different drivers) e.g. strong pedal off regen, separate pedal regen, petrol-like regen, diesel-like regen...

More relay outs for e.g. reverse (to activate the reversing lights).

Some visual feedback on the cruise control speed - e.g. alternate the speedo between the set speed (1/2 second) and actual speed (full second)

Idle mode (for automatic gearboxes)
Planned EV: '63 Cortina using AC and LiFePO4 Battery Pack

User avatar
Richo
Senior Member
Posts: 3737
Joined: Mon, 16 Jun 2008, 00:19
Real Name: Richard
Location: Perth, WA

Another DIY Inverter - 125kVA

Post by Richo » Fri, 04 Nov 2011, 20:41

Acc+On+Start (key switch)
Main Contactors x 2
+12V
GND
Crash Switch
Speed sensor
Motor temp sensor
DC/DC Enable

I think the contactors should be at the source point ie the battery box.
If you want an additional contactor in your controller is up to you.
But you should still have control of the external main contactor(s).

Do you have IGBT's where in the control lines are recessed?
This means your main PCB won't reach the pins.
Hence the cutout in your PCB.

Nice job! Image
So the short answer is NO but the long answer is YES.
Help prevent road rage - get outta my way!

tesla500
Noobie
Posts: 10
Joined: Thu, 03 Nov 2011, 23:52
Real Name: David Kronstein

Another DIY Inverter - 125kVA

Post by tesla500 » Fri, 04 Nov 2011, 21:46

woody wrote: Welcome!
Awesome work!
Going by your accent I guess you are in North America somewhere?

Thanks! Yes, I'm in Vancouver, BC, Canada
woody wrote:(Crazy) ideas follow.

By analogue gauge driver, do you mean 1 pulse per rotation for tacho and/or 1000 pulses per mile for electronic speedo/odometer?

I was thinking 0-5V to drive pretty much any analog gauge, you could select which parameter is displayed and set scaling in software. This could also produce a pulse train for driving another type of gauge like you mention.
woody wrote:Maybe some coloumb counting on the DC bus input and a fuel gauge output?

With the 3 phase current sensors you can compute the DC bus current, probably to an accuracy of a few percent. It wouldn't' include other loads like heater, DC-DC, etc, or battery charger current though.
woody wrote:Temperature gauge output for existing temperature gauge?

The analog gauge driver may be able to do that, depending on what type of gauge it is. Maybe add a 2nd gauge driver?
woody wrote:Different Regen modes selectable (for different drivers) e.g. strong pedal off regen, separate pedal regen, petrol-like regen, diesel-like regen...

Good idea, easy to do in software, just need some way to switch modes. Perhaps the auxiliary analog input connected to a rotary selector switch that can switch in fixed resistors to select the user profile?
woody wrote:More relay outs for e.g. reverse (to activate the reversing lights).

Right, forgot about reverse, I'll add that, and room for a spare or two
woody wrote:Some visual feedback on the cruise control speed - e.g. alternate the speedo between the set speed (1/2 second) and actual speed (full second)

Good idea. Perhaps just "flash" the set speed on the speedometer while changing it? I plan to support a rotary encoder (like a mouse scroll wheel) in software as an option instead of the speed up/down buttons. With good regen, you could basically drive entirely with that control :)
woody wrote:Idle mode (for automatic gearboxes)

Easy to do in software.
Richo wrote: Acc+On+Start (key switch)
Main Contactors x 2
+12V
GND
Crash Switch
Speed sensor
Motor temp sensor
DC/DC Enable

Acc+On+Start (key switch)      -How would this work? I was thinking of accessory mode whenever 12V power is applied, and a separate "On" signal to enable traction output.
Main Contactors x 2           -Good idea
Crash Switch               -Good idea
DC/DC Enable               -Would this be a contactor driver, or just a logic signal? Would it have to take care of precharging the DC-DC as well?

Already have these, just forgot to note them:
+12V
GND
Speed sensor - 3 inputs, quadrature encoder for asynchronous, or 3 phase hall sensors for synchronous motors
Motor temp sensor - Thermistor
Richo wrote:I think the contactors should be at the source point ie the battery box.
If you want an additional contactor in your controller is up to you.
But you should still have control of the external main contactor(s).
I'll add those.

This brings up the question of system precharge architecture. How should this be handled? There are numerous things that may need precharge: Inverter, DC-DC converter, Air conditioning/heat pump inverter. Should each be required to take care of precharging themselves, or should there be a single precharge system for everything?
Richo wrote:Do you have IGBT's where in the control lines are recessed?
This means your main PCB won't reach the pins.
Hence the cutout in your PCB.
Yes, the PCB is too high to reach the IGBT gate drive pins, due to the bus bars and spacers. I was planning to run twisted pairs straight from the gate drive pins to the control board.
Richo wrote:Nice job! Image

Thanks!

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

Another DIY Inverter - 125kVA

Post by bga » Fri, 11 Nov 2011, 16:46

Hi David,

Verrry impressive effort!

I hope to be following your lead with the motor controller shortly.

It's good to hear that AN908 is workable, although the code examples will need a lot of peripheral improvement to make them into a deployable system.

I chickened out on the individual transistor idea early on and went for 1/2 bridge 1200V powerex modules (CM300DY type), although there are better three phase multi-modules for 600V around now.
I would certainly agree that these modules will greatly simplify the implementation of the controller power section.

Your comments about voltage regulation and field weakening on the controller are along the lines that I was considering. My precise thought was to regulate field current (~voltage) to control the slip speed so that the field strength is modulated by the torque demand, subject to the transistor, battery and line voltage limits.

Did you achieve the third harmonic by over-modulating the gate PWM drive signal so that it flat-tops at 100% PWM, producing a 3rd harmonic signal?
I was proposing this as the means of allowing the abovementioned field modulation to produce over-drive at high current and high speed as a continuous function.

One other thought was frequency modulation of the PWM carrier. This is probably not an issue with a 50V bus, but will be an issue at low speed and torque on a high voltage bus with slow brute IGBTs. I believe that the DSPIC30F should have no problem doing this as a continuous function so that the minimum on-time requirement of the power transistors isn't violated
[our local electric passenger trains seem to something like this as a mode change at various speeds with the effect being like a very smooth gear changes]

From my own cynical perspective, I would make the control inputs and outputs transient and short-circuit proof so that the inevitable 12V crossed wires don't cause any catastrophic failures.

Regen; I think it makes sense for this to be moderate, like first or second gear so that the car can be effectively slowed in creeping traffic without brake being needed. A coasting (dead) zone in the the throttle would probably be helpful.

An anti-rollback feature sounds like a good idea. I was thinking along this line, using the motor's high resolution shaft sensor (say 500+ PPR) and then driving the motor when reverse rotation of the shaft is detected. The effect would be that the vehicle will slowly creep backwards on the hill until the accelerator is pressed. The creep speed should be in the order of a few metres per minute, so that brake activation will cause the system to relax.
A dash warning and delayed buzzer would be a good idea so that the motor doesn't get cooked by the position holding current if it is activated by too little brake pressue.

As a thought, suppose reverse is selected at speed, what should happen?
Above about 10kph, it should refuse to select and warn with a buzzer.
Below that speed, it would be OK to engage and allow the anti-rollback behavior to stop the car, athough some control over the torque may be needed to make the deceleration not to violent and prevent a loud clang from the diff.

Cheers
It's not the end of the world, but I can see it from here.

User avatar
Richo
Senior Member
Posts: 3737
Joined: Mon, 16 Jun 2008, 00:19
Real Name: Richard
Location: Perth, WA

Another DIY Inverter - 125kVA

Post by Richo » Fri, 11 Nov 2011, 21:05

bga wrote: An anti-rollback feature sounds like a good idea. The effect would be that the vehicle will slowly creep backwards on the hill until the accelerator is pressed.

You mean forward.
Even an auto creeps forward up a hill.
Backwards would be dangerous - unless you are in reverse...
bga wrote:
As a thought, suppose reverse is selected at speed, what should happen?
Above about 10kph, it should refuse to select and warn with a buzzer.
Below that speed, it would be OK to engage and allow the anti-rollback behavior to stop the car, athough some control over the torque may be needed to make the deceleration not to violent and prevent a loud clang from the diff.

Even at 4kph changing direction would be a bit of a rough ride.

To change direction the brake should be on and a speed less than 4kph (my pref would be 0kph for more than 0.5secs)
So the short answer is NO but the long answer is YES.
Help prevent road rage - get outta my way!

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

Another DIY Inverter - 125kVA

Post by bga » Fri, 11 Nov 2011, 22:02

Richo wrote:
bga wrote: An anti-rollback feature sounds like a good idea. The effect would be that the vehicle will slowly creep backwards on the hill until the accelerator is pressed.

You mean forward.
Even an auto creeps forward up a hill.
Backwards would be dangerous - unless you are in reverse...

Definitely backwards.

On steep hills, most auto gearboxes will creep backwards in drive, the difference is that it only moves slowly, so the small amount of backwards motion between brake release and accelerator activation is not significant. There's always the handbrake.

The real issue with trying to implement absolute position holding is that there is no way the controller can readily determine whether it's straining against the brakes or simply holding position against an incline. Either way, sitting stationary with high current in the windings will quickly overheat the motor, so must be avoided.

As an extension, it will probably be necessary to apply a bit of torque when stopped, say a few Nm, so that the drive stays loaded in the going direction and doesn't roll back on the diff and gearbox backlash when the brakes are released.
This is also good for a quick takeoff, as the motor can torque up really quickly without causing the drive and gears any problems.
Richo wrote:
bga wrote:
As a thought, suppose reverse is selected at speed, what should happen?
Above about 10kph, it should refuse to select and warn with a buzzer.
Below that speed, it would be OK to engage and allow the anti-rollback behavior to stop the car, athough some control over the torque may be needed to make the deceleration not to violent and prevent a loud clang from the diff.

Even at 4kph changing direction would be a bit of a rough ride.

To change direction the brake should be on and a speed less than 4kph (my pref would be 0kph for more than 0.5secs)
Like I said, there are some behavior issues to address so the transition isn't violent or unpredictable.

Enforced waiting periods may be the difference between getting and not getting rear-ended, so should be carefully considered.

I think that there are a lot of creative opportunities in this area alone. It's all in the software, we don't even have to wait for the glue to set before trying the next iteration.
It's not the end of the world, but I can see it from here.

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

Another DIY Inverter - 125kVA

Post by bga » Fri, 11 Nov 2011, 22:10

Speaking of letting the glue set,

I built a biped robot some years ago that used a PC on the end of a serial lead to to do the dynamic balance and step control on the machine. It would be relatively simple to implement a remote control protocol over a serial line so these algorithms could be tested on a PC with decent compilers before programming them into the controller.

A comms failure or disconnect could switch back to the current working algorithm on the controller so the PC becomes optional and weird behavior can be quickly overcome by simply unplugging the PC.
It's not the end of the world, but I can see it from here.

tesla500
Noobie
Posts: 10
Joined: Thu, 03 Nov 2011, 23:52
Real Name: David Kronstein

Another DIY Inverter - 125kVA

Post by tesla500 » Sat, 12 Nov 2011, 11:38

AN908 really only implements the motor control, all the peripheral stuff is up to the user.

Using many discrete transistors may be cheaper, at least for low quantities, but you've got hundreds of connections, and if any one fails, the inverter will likely fail. For that reason I discounted using that approach an went with modules.

For the control of direct (field producing) and quadrature (torque producing) current, I was initially going to just have constant direct current (~voltage) as you were thinking. I had 1/f field weakening, but I noticed that the if I set the direct current correctly for operation at high speed (not clipping the output waveform), the inverter was poorly utilized at lower speeds (the voltage was far from clipping when it should be able to clip). This was reducing the maximum torque at lower speeds, so I implemented the feedback loop to regulate the direct current. I do plan on reducing the direct current at low torque demands to keep the current low. I'll probably make an automated test setup that can determine the optimum amount of direct and quadrature currents for every motor speed and torque command.

The 3rd harmonic was already taken care of by AN908, and is always on.

I'm not planning any spread spectrum modulation at the moment, but I was planning to have dynamic switching between 18khz and 9kHz. The inverter would operate at 18kHz at lower power levels, and switch to 9kHz at higher power levels as needed to stay within the safe power dissipation limits on the IGBTs. I haven't heard of any minimum on time requirements for the IGBTs I'm using. That requirement may exist for IGBT modules with integrated drivers.

All the IO is protected against short circuit to +12V or ground.

Since many people have different preferences for regen and pedal response, I'll make user adjustable pedal tables, so the user could create whatever feel they want.

I like the idea for the anti-rollback and slight constant torque to absorb backlash. I was thinking something like each encoder tick going in reverse ratchets up the torque slightly, and the torque will slowly "bleed" away towards zero so it would eventually stop producing torque when you're fully stopped.

Switching to reverse at speed would not damage the inverter, it would behave just like using regen (negative torque). But for safety, it should probably not be allowed to switch into reverse above some set speed, like you suggest.

David

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

Another DIY Inverter - 125kVA

Post by bga » Mon, 14 Nov 2011, 19:03

tesla500 wrote: AN908 really only implements the motor control, all the peripheral stuff is up to the user.lication.
Agreed. At last it turns the tricky core into more or less a commodity, leaving us the I/O, management and configuration sections to do. Just as well that the computrons haven't all been eaten by by the motor control algorithm.

Along the same lines, I encountered a project in Circuit Cellar that used AN908 for a motor controller. Here
Using a DSPIC and Powerex semis, it was only a bench test at the time of publication. I did compare their source code to the Microchip sample and found only minor differences apart from the USB stuff, which may have been the real project.

The field drive is one thing that I have been curious about, I haven't been overly concerned yet, as there are only a few variables in the solution, so shouldn't be tooo difficult to solve acceptably. I am certainly happy to hear of other experinces with this part of the problem.

Mapping this space should be illuminating.

My deveopment plan to start with a small motor on low voltage with minimum filter cap so that it's hard to blow up anything. I need to make a gate driver logic error protector so that software crashes don't take the motor or controller out. I like the idea of using the current sensors to provide another limit on this, more in the millisecond area than the likes of desaturation protection -- I shall implement as part of the test setup.

The minimum on-time requirement was driven by the relatively slow switching speed of the IGBTs I am using, probably ok at 15khz. The issue I was considering was the low voltage, low torque, such as parking lots, where the PWM on-time is likely to be shorter than the turnon-turnoff time of the transistors and their gate drivers, so they won't saturate well and will get hot. I think the usual way to address this is to lower the PWM frequency so a lower duty cycle can be achieved with the same on-time, at the expense of more harmonics. Probably something similar occurs at near 100% duty cycle, but is easily managed by going to 100% from 99%.
Last edited by bga on Mon, 14 Nov 2011, 08:08, edited 1 time in total.
It's not the end of the world, but I can see it from here.

tesla500
Noobie
Posts: 10
Joined: Thu, 03 Nov 2011, 23:52
Real Name: David Kronstein

Another DIY Inverter - 125kVA

Post by tesla500 » Wed, 23 Nov 2011, 09:25

bga wrote:Agreed. At last it turns the tricky core into more or less a commodity, leaving us the I/O, management and configuration sections to do. Just as well that the computrons haven't all been eaten by by the motor control algorithm.

Along the same lines, I encountered a project in Circuit Cellar that used AN908 for a motor controller. Here
Using a DSPIC and Powerex semis, it was only a bench test at the time of publication. I did compare their source code to the Microchip sample and found only minor differences apart from the USB stuff, which may have been the real project.

The field drive is one thing that I have been curious about, I haven't been overly concerned yet, as there are only a few variables in the solution, so shouldn't be tooo difficult to solve acceptably. I am certainly happy to hear of other experinces with this part of the problem.

Mapping this space should be illuminating.

My deveopment plan to start with a small motor on low voltage with minimum filter cap so that it's hard to blow up anything. I need to make a gate driver logic error protector so that software crashes don't take the motor or controller out. I like the idea of using the current sensors to provide another limit on this, more in the millisecond area than the likes of desaturation protection -- I shall implement as part of the test setup.

The minimum on-time requirement was driven by the relatively slow switching speed of the IGBTs I am using, probably ok at 15khz. The issue I was considering was the low voltage, low torque, such as parking lots, where the PWM on-time is likely to be shorter than the turnon-turnoff time of the transistors and their gate drivers, so they won't saturate well and will get hot. I think the usual way to address this is to lower the PWM frequency so a lower duty cycle can be achieved with the same on-time, at the expense of more harmonics. Probably something similar occurs at near 100% duty cycle, but is easily managed by going to 100% from 99%.



Thanks for the link, I haven't seen that article before.

Mapping the response and efficiency of the system will be interesting.

Starting with low voltage is a good idea. I've got hardware overcurrent and overvoltage protection. Desat detection may not be needed, the current sensors should respond within 5-10uS.

I came across one situation on the prototype where the software hit a breakpoint, and I had programmed it to set the PWM to 0 in this situation. Since the rotor still had flux through it, the motor still had back EMF, and this caused current flow though the inverter, which was effectively shorted due to the 0 duty cycle drive. The hardware overcurrent protection shut down the gate drives as it should, but this basically regulated a high current regeneration, which charged the DC bus up dangerously high. Luckily nothing blew up, but after that I added hardware overvoltage protection to turn off all the gate drives if the bus voltage gets too high.

Minimum on time is actually only a problem near 0 or 100% duty cycle. In the modulation scheme used in AN908, if the inverter outputs low voltage, like when moving slowly in a parking lot, the duty cycle is near 50% on all IGBTs, so there's no problem. Unless the datasheet specifically forbids doing it (usually for IGBTs with internal drivers), there's no problem turning the IGBT off before it fully turns on. This is not any more stressful than a complete on-off switch.

David

Stiive
Groupie
Posts: 295
Joined: Wed, 31 Dec 2008, 08:26
Location: Melbourne

Another DIY Inverter - 125kVA

Post by Stiive » Sat, 19 May 2012, 17:19

Hey mate,
Is this project still underway?

I started a thread over on DIYElectricCar looking for people who are designing their own controllers. (http://www.diyelectriccar.com/forums/sh ... 74151.html)

Have you designed the code yourself? Or using the brains from an industrial VFD?

Is this working yet?
Rgds,
Stiive

ACEVS4US
Noobie
Posts: 7
Joined: Wed, 27 Jun 2012, 02:32

Another DIY Inverter - 125kVA

Post by ACEVS4US » Wed, 25 Jul 2012, 16:38

Hi David
I am also building a vector driven 3 phase motor controller, so I was really interested to find this thread.

My plan is to power a 30kw continuous custom wound industrial ABB motor.

I,m using the Mitsubishi cm400dy dual (half bridge) modules which have an Ic continuous rating of 400A and a max voltage of 600V.

I was planning of using Atmel as my processor, but after seeing microchip code in action I've changed my mind. The Atmel code, in hindsight, was pretty average because the loop only measured shaft speed and not currents to estimate rotor position. I would have needed to implement the rest myself which takes up precious time.

So far, I've built a gate drive circuit (PCB) with all the required fault protection and PSU isolation. I decided early on to keep the gate drive stuff separate from the main processor board and power board which I have not yet implemented. The gate drive board has an auxilary ATmega32 processor whose sole job is to monitor the fault conditions and shut the IGBT's down if a fault condition occurs. Anyway, I've
been able to test one phase of the inverter using a couple bar heater elements - so far so good. I'll try to post a pic of this soon.

I do have some questions about your design.

Have you thought about including an input from the airbag deploy signal? I think this is pretty important in the event of an accident. If the airbag gets deployed the main contactor should be tripped to stop emergency services from getting electocuted.

I noticed that you have quite a bit of capacitance (electrolytics) on your power board. You might not need as much (or any electrolytics) as you've got. I found this paper to be very useful.

Selecting Film Bus Link Capacitors For High Performance Inverter Applications

In the picture of your inverter power board layout, does the dc bus (copper) just fold directly underneath the power board? - and if so, how do you attach your dc link caps to it.

Did you select your microchip processor based on the MIPS or just extra IO. It seemed to me after looking over the code that if you want a lot of rotor flux angle estimations in a revolution, then the MIPS of the processor might become a factor. I think from memory microchip uses a 1ms sample time of the currents, which at 3000rpm gives 20 samples per revolution (about 3 estimations per sector). I don't have the experience to know whether this is enough or not - any thoughts?

It would be great if I could get a copy of the Altium files because it would save me some time drawing up schematic components etc.

c(dot)young(at)irl.cri.nz

Cool project

Chris


Last edited by ACEVS4US on Sat, 28 Jul 2012, 05:24, edited 1 time in total.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Another DIY Inverter - 125kVA

Post by Johny » Wed, 25 Jul 2012, 16:47

Chris you might want to camouflage your email address a bit before the web trawlers find it.
Eg. c DOT young AT irl DOT cri DOT nz
(Remove spaces)

ACEVS4US
Noobie
Posts: 7
Joined: Wed, 27 Jun 2012, 02:32

Another DIY Inverter - 125kVA

Post by ACEVS4US » Wed, 25 Jul 2012, 16:53

ta

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Another DIY Inverter - 125kVA

Post by Johny » Wed, 25 Jul 2012, 16:55

Chris. I agree that using electro for a current (contemporary) design is a bit old - especially for a car where temperatures may quite elevated. I have read that paper as well (among others but that's the best one) and will be swapping out the two series 5600uF electros in my industrial controller for a single Film 330uF 1100V (my DC bus is 600V). Estimated controller power after upgrade will be about 77kW - 140 motor Amps. I would think somewhere around 500 to 800uF would be sufficient for a 120kW/300Amp - assuming good, low impedance connections to the DC bus.

Post Reply