I see a half bridge for each phase on the driver cards. For 24v vcc, a wave is created between 0v and 24v. It takes the 12v level as the midpoint.
Isn’t it possible to take all the ends of the coils out and drive each coil with its own 4 mosfet full bridge?
In this way, we can create +24v and -24v waves with a center of 0v.
I wonder why this method is not used. Is it the cost? Or is it not as useful as I thought?
Hi @serkan,
Its a perfectly reasonable question. The guys have made some valid points already and I agree with their explanations. I’d just like to add one more visual interpretation of their explanations.
Say you are comparing a half bridge and a full bridge on one phase of the motor, for example phase A.
The main argument for using this architecture would be to enable using the double voltage range of your power supply. For example if you have 24V power supply you should be able to have -24V to 24V voltage range (so 48V range).
If we imagine our mosfets as switches there are only four states they can be in
| State | High side FET | Low side FET |
|---|---|---|
| 24V | ON | OFF |
| GND | OFF | ON |
| High impedance | OFF | OFF |
| Short circuit | ON | ON |
Of course the short circuit state is the one we always (try to) avoid 
In this case you would have two high side mosfets and two low side mosfets. You would still have the same four states but you can achieve all of them with multiple combinations of ON/OFF states of mosfets.
You have 16 combintions of ON/OFF states of mosftets in total but only 4 states of the phase A: 24V, GND, High impedance or short circuit.
| State | High side FET (left) | High side FET (right) | Low side FET (left) | Low side FET (right) |
|---|---|---|---|---|
| 24V | ON | ON | OFF | OFF |
| 24V | ON | OFF | OFF | OFF |
| 24V | OFF | ON | OFF | OFF |
| GND | OFF | OFF | ON | ON |
| GND | OFF | OFF | OFF | ON |
| GND | OFF | OFF | ON | OFF |
| High impedance | OFF | OFF | OFF | OFF |
| Short circuit | ON | OFF | ON | ON |
| Short circuit | ON | OFF | ON | OFF |
| Short circuit | ON | OFF | OFF | ON |
| Short circuit | ON | ON | ON | ON |
| Short circuit | ON | ON | OFF | ON |
| Short circuit | ON | ON | ON | OFF |
| Short circuit | OFF | ON | ON | ON |
| Short circuit | OFF | ON | ON | OFF |
| Short circuit | OFF | ON | OFF | ON |
Here are 24V combinations:
And the high-impedance one
And here are couple of short circuit states
So having the full bridge on the phases of the BLDC motor will not result in giving you 48V range [-24V, 24V]. It will only allow you to have some redundancy and potentially more current capacity as you will be able to pass the current through multiple mosfets at the same time improving their efficiency and producing lower thermal losses.
Another way of looking at this
For example if you take a pair of BLDC motor phases, say A and B. And they both have a half-bridge driving them. Then the relative voltage between the phases A and B is effectively driven by a full-bridge.
So adding additional mosfets to the phases A or B does not really change anything in terms of the voltage range. It is already [-24V, 24V].
I now understand your question better @serkan,
I agree with Antun’s answer of course, but I think the setups he has drawn are not actually connecting the phases independently to full bridges. In Antun’s drawing each phase is connected to the half bridge or full bridge at only one end, and the other end is connected to the common centre (wye) or to another phase’s half/full bridge (delta).
What I understand you were asking is why not connect each phase “between” the two half bridges of a full bridge (like in a DC motor, where the motor is between the two half bridges).
Actually I think the type of setup you describe does exist, for example like this 6 phase motor:
I think you could do the same thing with just 3 phases.
But SimpleFOC does not support such configuration at the moment.
Another thing that you see sometimes is the six phase wires coming out of the motor, but still only using 3 half bridges. The other ends of the phases can be configured as either delta or wye configurations, to change the winding type of the motor.
Lets start with the three individual coils.
For ease of explanation, Lets say that each coil has a positive and negative end. Lets also assume that standard polarity across the coil generates a northern magnetic field towards the magnets while a reverse polarity generate a southern pole towards the magnets.
To form a delta motor winding, we wire one coils positive to another coils negative.
Now, we have three nodes in the form of a triangle. When we apply VCC to one node, and 0v to the other nodes, we have one winding that has a standard polarity and another winding that has a reverse polarity.
So one coil sees +24v while the other sees -24v.
So with a standard half bridge setup and a delta winding, we can already achieve a -24v across a motor winding without needed a full bridge.
The only benefit I can see from separating the coils is that you could drive all three coils at once. Given the construction of a bldc motor, I doubt this would come in much use however.
Does this make sense?
It’s not useful at all, because you can already create +24V and -24V across the phases (given a delta winding for the motor), this happens when 1 phase has a PWM duty cycle of 100% and the other phase has a PWM duty cycle of 0%.
If your motor has a Wye winding, rewinding it to Delta would make much more sense.
But where is the -24V coming from? In this scenario you have only 24V potential?
A sine wave between 0 and 24 volts is produced with 3 half-bridge drivers. The center of this wave is 12v.
Let’s consider a delta connected motor. Phase A is pulled to 24v with a 100% pwm rate, phase B is pulled to GND with a 0% pwm rate. In other words, 24v voltage is applied to the coil between terminals A and B.
What if each of the 3 coils in the motor had its own full bridge? Thanks to the 4 mosfet full bridge, we could give 24 volts in two directions to the coil with terminals removed from both sides.
Or we could give a sine wave to the coil that I mentioned in the previous example and is now electrically isolated from the others. The center of this wave is 0v and the wave is between +24 volts and -24 volts. It can be drawn as 48v from top to bottom.
Am I missing something? Or can I use 12 mosfets (4 full bridge x 3 phase) instead of 6 and run a 24 volt system as 48 volts?
I don’t have a motor, it would be nice to try. I just want to discuss if it is technically possible.
Does SimpleFoc produce a sine wave between 0 and 24v as shown in the figure above?
The figure below shows the coil between 2 half bridges (like a dc motor). With this setup, wouldn’t it be between -24 and +24?
(The dark dots are mosfets.)
SimpleFOC produces phase to phase voltage like the bottom figure, between +24V and -24V.
Here’s a sample schematic showing the MOSFETs and motor coils, in a delta configuration. You can see clearly that the structure of the coils and MOSFETs is very similar to your half bridge.
To actually achieve the phase to phase peak voltage equal to the supply voltage, you have to be smart with the PWM. The technique is called Space Vector PWM / Third Harmonic Injection. Each phase does not actually get sent a sine wave, the PWM waveforms on each phase, over a single 360degree rotation, looks more like this:
Different implementations may have slightly different waveforms, with different advantages. The waveform pictured is using the pulled to GND technique, which has the advantage of only 2 phases switching at a time.
To get the phase to phase voltage, simply subtract the voltage of one phase from the voltage of another. When you do this, you will see you get a sine waves with peaks of + or - the supply voltage.
This is an interesting discussion, for sure!
I think this is a completely orthogonal topic. The 3rd harmonic injection is a modulation technique to boost the voltage, and can be used with any winding configuration.
I think it makes more sense to examine the situation under block (6-step) commutation - a more simple commutation technique where we don’t need the PWM to generate waveforms, here shown for a normal 3-half bridge configuration and a motor with wye winding:
@serkan 's configuration would be this:
where L1, L2 and L3 are the motor windings, neither wye, nor delta, but each winding connected to its own full bridge.
If we compare this to the wye motor image above, and look at the block commutation pattern, then we can compare the voltages through the windings, looking at the yellow highlighted sector 3 for example:
3 half bridges:
Phase A (blue): Q1 OFF Q2 OFF = High Z
Phase B (green): Q2 ON Q3 OFF = 24V
Phase C (red): Q4 OFF Q5 ON = GND
The effect:
Phase A: High Z
Phase B: 24V (from VBUS to star point)
Phase C: -24V (from GND to star point)
3 full bridges:
Phase A (L1): Q1, Q2, Q3, Q4 all OFF = High Z
Phase B (L2): Q5 ON Q6 OFF Q7 OFF Q8 ON = 24V
Phase C (L3): Q9 OFF Q10 ON Q11 ON Q12 OFF = -24V
The effect:
Phase A: High Z
Phase B: 24V (Q5->Q8 from left to right)
Phase C: -24V (Q11->Q10 from right to left)
So we see the effect is the same, as @Andrew and @Antun_Skuric have already said.
What will be different is the amount of current that flows through the windings. Assuming we use the same windings in each configuration, the motors will exhibit different resistances:
In practice, I’ve never seen a motor that uses 3 full bridges. Using the 3 full bridges would need more control wires and need double the hardware in terms of the FETs and gate drivers.
Actually I think my conclusion is correct but my description of what’s going on with the phase voltages is still wrong. I’ll edit it tomorrow when I’m less tired…
