Synchronous rectification How the F*k does it work for regen

[Arlo1]

OK guys I understand Synchronous rectification or Active rectification works using mosfets or Transistors instead of diodes but what about in a motor controller??

IM asking because my friend Doug and I are talking about getting a hybrid setup running to get a car home. And I have been really thinking about it lately and I NEED TO KNOW!
Here is my problem I don't understand how a motor can produce power to charge a batter at low rpm. I mean if you are at say 50% of your top speed you can still regen but how what is the trick the mosfets do to make more the pack voltage when the peaks of the sine waves are at less then 1/2 of pack voltage???

 

[major]

.... I don't understand how a motor can produce power to charge a batter at low rpm. I mean if you are at say 50% of your top speed you can still regen but how what is the trick the mosfets do to make more the pack voltage when the peaks of the sine waves are at less then 1/2 of pack voltage???

Hi Arlo,

Here is the way I explain it without the use of diagrams or math. I hope it helps.

For the basic understanding look at a simple single quadrant controller on a PMDC motor. It has a series switch (mosfet) and a reverse biased diode across the motor (FWD). It uses the inductance of the motor and with PWM operates as a buck converter. The average motor voltage is lower than the battery (source) and average motor current is greater than battery current. During the PWM off-periods the FWD conducts. If the main switch is left off long enough, the FWD current decays to zero and the FWD is reversed biased again (stops conducting). The motor is in a coast mode with zero current.

The freewheeling diode can be replaced with switch and complementarily controlled, off when the main switch is on; on when the main switch is off. The circuit functions the same as with the FWD for conditions where motor voltage is less than battery. Call this the normal motor operation; motor voltage is always lower than battery voltage. Current in the motor flows from motor positive to motor negative.

For regeneration to occur the motor voltage must be greater than battery and current must reverse in the motor and flow from motor negative to motor positive. This is accomplished by leaving the switch which replaced the FWD diode on long enough for the current to reverse in freewheeling loop shorting the motor. Once the current in the motor has reversed, when that switch is opened, the motor coils and field will induce a voltage on top of the terminal voltage. In other words, it becomes a boost converter. The generated voltage in the motor is now the source and the battery becomes the load. The inductance is in series with source now and the switch (former FWD) now shorts the source and inductor. Current flow is reverse from the motor mode and torque is reversed providing braking against rotation.

So by replacing the FWD with a switch, the controller is capable of two quadrant operation; forward rotation motoring and forward rotation braking.

 

[Arlo1]

OK so... What does it do with the mosfets to make this happen?

 

[major]

OK so... What does it do with the mosfets to make this happen?

It is all in the timing. The mosfet which replaced the FWD must conduct long enough for motor current to reverse. If you just left it on it would short the motor until rotation completely stopped. Before that happens, you switch off that mosfet and turn on the other which conducts the voltage spike back to the battery, so to speak.

 

[Arlo1]

OK so... What does it do with the mosfets to make this happen?

It is all in the timing. The mosfet which replaced the FWD must conduct long enough for motor current to reverse. If you just left it on it would short the motor until rotation completely stopped. Before that happens, you switch off that mosfet and turn on the other which conducts the voltage spike back to the battery, so to speak.

Ok this is starting to make sense thanks.

 

[7circle]

Found this pic with typical circuit with FETs and diodes and even ds30pic
But no phase current sensors shown.

So with motor spinning, magnets induce AC voltage curve.
At this RPM the curve is 100Hz
The phase to phase with peak to peak voltage at only 1/2 the battery voltage.
Not enough to get diodes to flow.

As the controller can measure and update its control stuff 10000 times a second,
imagine a time span of 100uSeconds.... the voltage looks like it is barely changing.

So to get a "Boost" from the motor inductance
the controller can get the current to flow by turning two of the bottom or top MOSFETs on.
With only 1V enduced by the magnets there will be current flow.
It will just cause a current flow RAMP UP in the motor inductance 1/10th compared if it was 10V.

The winding inductance will store more and more magnetic energy as the current ramps up.
When the controller triggers enough current by predicting it or using phase current sensors.

So say the current gets to 10A in 10uSec and Vph-ph is at 40V.
The power starts at 0W and reaches 400W.
At 10A the motor torque is say negative 10Nm and it brakes the motion.
Then the controller turns off say BOTH MOSFETs
but the inductance, like momentum, will keep pushing the current
it wil be forced to keep flowing at 10A. This causes "free-wheeling diode"(FWD) and top and bottom diode set will conduct.
The battery is at 80V...
So the motor inductance will add 42V to magnetic enduced MMF causing 40V.
The winding terminals will jump up to 82V to allow the current to keep flowing.
With 1V in each diode. Loosing 20W at 10A.

As energy is tranferred out of the motor at 820W
part is the inductor at 420W and the moving magnet is still 400W giving 10Nm.
This energy loss in the inductor allows the current to ramp down.
So as the current ramps down the inductance of the winding is producing 42V and the moving magnets are generating 40V.

So in 10uSec it ramps back down to 0A.And 0Nm braking
and 0W power flow.
You can see this is causing a torque ripple.

But if the controller Shorts the two phases again after only 5uSec the current will be at 5A.
The controller can thold the same FETs on again for another 10uSec and the current will RAMP UP to 15A. and 15Nm braking. The controller can monitor the phase currents and let is slowly move up in jaggared steps untill it reaches 100A into the rails of the of the MOSFETs.
100A givivg 100Nm braking
Diode losses are now 200w
Just like it can control the torque in motor mode it does it in reverse for generator mode.

But thit doesn't explain the "Syncronous" stuff.
This suggests the "Syncronous Boost Converter" topology is applied to regen.
Like Major posted the losses in the diodes can be by-passed by using the parallel MOSFET.
So with twof ets conducting to the rails and say 1mOhm power is 100x100 x 0.001 is 10W

The controller needs to tracking the motor phase voltages
Getting the controller to make a smooth feeling and quiet motor is all about shaping the currents in each phase to the sinusoid fundamental at the operating RPM.

The field orientated control can be used to get the best of the magnet induced voltage curve.

I was supprised a wiki or similar didn't pop up easy in a search.

Good question Arlo.

(edit: some typos ... oh once controller gets current flowing in all three phases, torque comes from each winding. So using syncronous MOSFET rectification just saves a the difference of better MOSfET losses over the freewheeling diodes. If they were Power BJT's or IGBTs it just be the top rail diodes that can be by-passed)

 

[Arlo1]

Thanks 7 its tough to imagine. I think the trick here is the windings are shorted when 2 upper or lower mosfets are on. Causing the motor to store energy in the windings then you open 1 or 2 of them and the current keeps flowing but to the caps/battery instead and repeat. It would be ideal for an animation because its quite hard to understand.

But understanding will help with better controller development and the possibilities of making more efficient generators etc.

 

[7circle]

Yeah its hard to think about everything in little tiny steps the way the controller needs to.... to stay in control. And then the effects of circuit dynamics.

Animation would be great.
Or simulator.

I edited above post to add clariry ... [] :?