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

[P-hack]

Originally Posted by cts_casemod

If we want to be really scientific, yes for a 50% duty cycle and an output of 200Amps one has to transfer 400A during half the time. But the average continuous current is still 200A. In fact 200Amps AC.

Your boost explanation so far makes sense, except here, I see the voltage alternates on the inductor, but the current is fairly stable. There's some current ripple, but the whole point of the inductor is to maintain current, no (like a cap reverses current to maintain voltage)? If you are drawing 400 amps at whatever power level from the battery, the boost inductor needs to be able to handle ~400 amps it seems (the two are in lock step current wise in ltspice, I think that is kinda how series current works).

Edit: would be nice if I was wrong.

[cts_casemod]

Originally Posted by P-hack

Your boost explanation so far makes sense, except here, I see the voltage alternates on the inductor, but the current is fairly stable. There's some current ripple, but the whole point of the inductor is to maintain current, no? If you are drawing 400 amps at whatever power level from the battery, the boost inductor needs to be able to handle ~400 amps it seems (the two are in lock step current wise in ltspice, I think that is kinda how series current works).

Edit: would be nice if I was wrong.

Most of the losses on an inductor are associated with the ripple. So the smaller the ripple, the less I2R and core losses. It might be worth to make a simulation with larger values of inductance and see how the current and efficiency change.

Regarding the current: Say a 12V light bulb feed with PWM with a 50% duty cycle, and an input voltage of 24V.

Will the bulb burn? Depends.
If the switching frequency is 1Hz, it will surely burn!

If the switching is fast enough the filament temperature will average due to thermal inertia. This is also true for magnetic fields and the wiring on the inductor.

Same with AC current. RMS reading is 13Amps, but peak current is 18A.

[Astro]

Originally Posted by cts_casemod

Does this sound correct?

Assuming a 50% duty cycle square wave and a battery (Vs) voltage of say 20V and a current of 100A.

SW1 closed

Battery Volts = 20V
Inductor Volts = 20V
Capacitor Volts = 40V
Battery Amps = 100A
Inductor Amps = 100A
Capacitor Amps = 0A

SW1 open

Battery Volts = 20V
Inductor Volts = 20V
Capacitor Volts = 40V
Battery Amps = 100A
Inductor Amps = 100A
Capacitor Amps = 100A

So the capacitor is supplied double the battery voltage for half the time at battery current.
D1 is there to prevent the capacitor being shorted when SW1 closes.
The battery delivers a constant 100A at 20V.
The inductor charges and discharges to 20V at 100A.
As long as the load placed on the capacitor discharges at an average current of half the battery current or less then the capacitor will supply twice the battery voltage.

[cts_casemod]

Originally Posted by Astro

Does this sound correct?

Assuming a 50% duty cycle square wave and a battery (Vs) voltage of say 20V and a current of 100A.

SW1 closed

Battery Volts = 20V
Inductor Volts = 20V
Capacitor Volts = 40V
Battery Amps = 100A
Inductor Amps = 100A
Capacitor Amps = 0A

SW1 open

Battery Volts = 20V
Inductor Volts = 20V
Capacitor Volts = 40V
Battery Amps = 100A
Inductor Amps = 100A
Capacitor Amps = 100A

So the capacitor is supplied double the battery voltage for half the time at battery current.
D1 is there to prevent the capacitor being shorted when SW1 closes.
The battery delivers a constant 100A at 20V.
The inductor charges and discharges to 20V at 100A.
As long as the load placed on the capacitor discharges at an average current of half the battery current or less then the capacitor will supply twice the battery voltage.

Yes.
The capacitor is initially charged to 20V, rather than 40, by the diode. This is where the pre-charge goes. From here the boost converter charges the output capacitor in pulse by pulse current limiting, which also means that if the output is too large it will simply throttle back the output voltage.

I like to say that the battery current is twice the output current for half the duty cycle, which is about the same and its also valid for the switch, inductor, diode and output capacitor, which can now supply the nominal output, but your principle is correct.

Technically since this is done at HF the value averages, and its the reason a capacitor is needed at the input of the converter, to average the battery current.

For continuous mode, the minimum inductor current is the output current (200A) and the maximum current is the output added to the ripple. So the smaller the ripple the smaller the current with minimum being the load current.


From this the interesting thing is the RMS current


For an output of 200Amps and a 30% ripple, the RMS value comes as 60Amps, again backing up my last post that only the ripple current contributes toward the losses.

Here's the link:

Boost Switching Converter Design Equations

The formulas to calculate the inductor current in function of the inductance are also there.

[P-hack]

So, boost inductors of low inductance value like large currents to stay out of discontinuous mode. And a boost isn't technically needed for a v/hz below battery voltage, but I don't know how hard that makes the inverter controllers job having multiple bus voltages. But a smaller value inductor for a given current carrying capacity will mean less losses.

For most of my driving, I pulse and glide, so an all-or-nothing booster would be quite manageable, but with a cpu it could do that on a much smaller scale, i.e. run the booster %10 of the time, or 10/100 of a second or more depending on demand to optimize the inductor to the smallest "continuous boost mode" size. Is that sort of what you are thinking CTS? I would like to see it rated for continuous duty cycle though, direct drive vehicles will need boost for sustained high speed.

[cts_casemod]

Originally Posted by P-hack

So, boost inductors of low inductance value like large currents to stay out of discontinuous mode. And a boost isn't technically needed for a v/hz below battery voltage, but I don't know how hard that makes the inverter controllers job having multiple bus voltages. But a smaller value inductor for a given current carrying capacity will mean less losses.

For most of my driving, I pulse and glide, so an all-or-nothing booster would be quite manageable, but with a cpu it could do that on a much smaller scale, i.e. run the booster %10 of the time, or 10/100 of a second or more depending on demand to optimize the inductor to the smallest "continuous boost mode" size. Is that sort of what you are thinking CTS? I would like to see it rated for continuous duty cycle though, direct drive vehicles will need boost for sustained high speed.

Correct. But in practice higher inductance is better as it allows continuous current down to a few amps and a very low ripple while under heavy use. It also allows a smaller igbt peak current. Lower inductance is best for small size, but the penalty is reduced efficiency at low loads.

The losses are proportional to the ripple. So while a larger inductor has a higher resistance, it has less core losses and vice versa. One has to find a compromise and needs to have in account the average current, rather than the peak current. I'm cool with having 200Amps output, but how many seconds will I use the inductor at these levels during a charge cycle? So I would say the sweet spot on efficiency should be calculated to match the motor nominal power. On a 18HP at 208V that would be less than 70Amps. And let me mention, if I haven't, that the gains are larger should the output be boosted to 450V at 70Amps that to 300v at 150Amps. In fact the sweet spot for a 208V 4 Pole motor is sqroot(2)*415 = nearly 600V for full torque up to 2900RPM and of course, at such power levels the motor would have 36HP nominal with no loss of performance or operational time limits other than the battery running out!

Continuous boost is needed for high speed, correct. But not at 200Amps. Say 70-90Amps, which would draw ~ 200Amps from the battery (40HP). If the motor only has a duty cycle of 15 seconds above this, it doesn't seem right to design a boost converter which is any different.

[P-hack]

ran some simulations for 36kw @ 510v/70a. (510 because svpwm, hmm sounds like hubeners pack)

freq: 20k
battv: 144
batt/inductor avg A: 260 (160min/355max, triangular)
min inductor to avoid discontinuous mode: ~ 26uh
peak/avg switch current ~355/190
duty cycle %66

I think this is how the prius (mgr perhaps) 25kw boost does it too, doesn't run all the time, beefy inductor with few turns and large conductor, beefy igbt switch, available at a junkyard near you (The whole inverter can be something of a 50kw-75kw bargain with a custom logic board.)

It would allow operation at v/hz < battv/510 with minimal losses, or something like that, below a certain rpm.

[cts_casemod]

Originally Posted by P-hack

ran some simulations for 36hp @ 510v/70a. (510 because svpwm, hmm sounds like hubeners pack)

freq: 20k
battv: 144
batt/inductor avg A: 260 (160min/355max, triangular)
min inductor to avoid discontinuous mode: ~ 26uh
peak/avg switch current ~355/190
duty cycle %66

I think this is how the prius (mgr perhaps) 25kw boost does it too, doesn't run all the time, beefy inductor with few turns and large conductor, beefy igbt switch, available at a junkyard near you (The whole inverter can be something of a 50kw-75kw bargain with a custom logic board.)

It would allow operation at v/hz < battv/510 with minimal losses, or something like that, below a certain rpm.

That looks more meaningful than the 50% duty.

If one was to boost accordingly to the RPM it would make the inverter software very neat. Basically the V/Hz ratio and hence torque would be pretty much constant at all revs, which would save you of the nonlinear torque curves above nameplate.

Here is how I think this could be implemented:

Set the V/Hz ratio to 4V/Hz, for example. Run the booster regardless. If the result was under the battery voltage, the inductor would never run as the comparator would keep the IGBT off and only battery voltage would be used.

As soon as the output voltage was lower than the target (and this could be because the V/Hz ratio was too low or the battery was sagging too much) force the booster to regulate the output voltage. This could be to anywhere between 150 and 600V, depending on motor frequency.

This is transparent to the controller. Normally one would simply loose torque above nameplate, now the output becomes linear. 50% throttle = ~ 50% torque.

One less programming nightmare!
The only thing the boost controller would require is a signal from the motor encoder/tachogenerator.

[MPaulHolmes]

This is a very neat idea. From a field oriented control perspective, you know you are commanding too much torque for a given speed when the PI loops for Id and Iq are no longer converging. The radius of (Vd,Vq) is bigger than Vmax. So, you could boost until convergence starts to happen again. And always attempt to reduce boost as long as convergence is happening.

For what I have planned (the board is almost done), the boost stage would just be a 4th IGBT half bridge that's controlled by the same microcontroller. It would be bolted right next to the other 3 IGBTs. It would be connected to the same ring capacitor. It would look identical to a 4 phase motor controller. The only difference would be that you hook up the "4th phase leg" to the inductor rather than to the motor.

You would have to move to buck mode when in regen though.

[cts_casemod]

I found this page while looking for some inductors for my charger.

Have a look at this series:

E Series : CWS Coil Winding Specialist, manufacturer of transformers, inductors, coils and chokes

Apparently can do custom wound inductors as well. Might be worth a look.