Havn't followed all, but won't a IR IC do it?

http://tahmidmc.blogspot.com/2013/01/using-high-low-side-driver-ir2110-with.html

 

only the 3 hi-side, lo-side can be one supply

 

Havn't followed all, but won't a IR IC do it?

http://tahmidmc.blogspot.com/2013/01/using-high-low-side-driver-ir2110-with.html

Yes, but he needs 4A of gate drive, more than a 2110 or most other drivers can manage..

 

So to keep the 3 Phases separate from a single power supply I need to use 3 separate flyback smps to provide 12v

No not a SMPS but a DC-DC power supply like ones Murata makes. I didn't go back to read what voltages your working with so this is just an example of what is needed. You can look to see in the catalog for ones that meet you voltage input with a 12Voutput.
https://www.digikey.com/en/products...3ic0txmiJGXpjFbyqJjv7bR8Wft3plwhoCZz0QAvD_BwE

 

It needs to be around 12V @ 4A

Why such a high output current? I must be missing something again. Don't you only need enough current to turn on/charge the gate capacitance? Not to match the driver output?

 

Why such a high output current? I must be missing something again. Don't you only need enough current to turn on/charge the gate capacitance? Not to match the driver output?

The way I understand it, you need a high current pulse to charge the gate capacitance as quickly as possible. This to keep the switching efficiency also as high as possible, and thus generate the least amount of heat.

And now that I think of it, the gate charge process should be so fast that it would probably be more meaningful to think about it as energy being pumped into it (that is, in terms of joules), instead of power (watts).

 

@Irving I think We can move forward with this driver It is a 4A/6A Device and placing an external power supply wouldn't be needed, Input to Output Isolation barrier is also there.
I wish to use Line Buffer (Signal coming from MCU to Gate Driver will be buffered first then it will move to another PCB where the Gate driver and power Mosfets are placed.

 

@Irving I think We can move forward with this driver It is a 4.5 A/8A Device and placing an external power supply wouldn't be needed, Input to Output Isolation barrier is also there.
I wish to use Line Buffer (Signal coming from MCU to Gate Driver will be buffered first then it will move to another PCB where the Gate driver and power Mosfets are placed.

That's a good find though I've not looked at the data-sheet in great detail. One thing that's immediately obvious is the lack of detail regarding sizing of the bootstrap capacitors. Maybe that's in a separate application note? There is some useful info on pages 24 - 27 regarding gate current, gate charge, switching time and gate resistor. Note that this device is capable of sourcing typically 4.5A @25degC but with a minimum of 2.6A at lower temperatures.

A crude simulation shows that, with a gate set up as shown here

You can drive 4 MOSFETs easily.
I'll post a simulation later, but you don't need a 4A power supply on the low side because you only need 4A while MOSFETs are switching. A low-esr, high-ripple safe, 10uF capacitor across the supply provides sufficient charge to switch the MOSFETs on (about 300nS) and then recharges during the rest of the 20uS cycle. So you only need around 300mA from the PSU.

Be sure to follow closely the PCB layout guidelines - especially the bit about short, low-inductance, gate drive lines.

 

The way I understand it, you need a high current pulse to charge the gate capacitance as quickly as possible. This to keep the switching efficiency also as high as possible, and thus generate the least amount of heat.

And now that I think of it, the gate charge process should be so fast that it would probably be more meaningful to think about it as energy being pumped into it (that is, in terms of joules), instead of power (watts).

You may be correct. But it has been my understanding that the bootstrap is only to keep the highside gate on, the high current part comes from the initial signal TO turn on being either Vcc or Vdd depending on the way the driver pins are called out. Then as the D-S voltage gets higher than the ~10V needed the bootstrap kicks in. And at that time the bootstrap current is pretty low, since the gate is fully charged.

I'm learning as we go, so many of my posts are as much for my education as they are for the TS.

 

I'm learning as we go, so many of my posts are as much for my education as they are for the TS.

Likewise...

 

OK, finally got there after a few crashes... think this one of the most complex simulations I've done, but fun getting it to work...
Let me say at the outset its a simulation not reality, track lengths/widths, temperature, tolerances, etc, are not modelled. This took an hour to simulate 500uS of operation of which I show here a few choice snapshots...

Firstly the circuit.... I've loosely modelled the driver IC on the LH side with M9 - M12. V4 is the 15v supply to the driver with C2 across it to provide bulk current sourcing... D9, R21, C1 are the bootstrap components that deliver the high side Vdd at V(bs, its absolute voltage relative to V(Load). 8 MOSFETs on the RH side model the upper and lower quad switchers driving the load resistor Rload. For simplicity I've shown this connected to the centre tap of the 60v supply, with Rload set to give I(load) as +/-300A. V(load) swings therefore between 0 and 60v. This is not a simulation of driving the motor (yet) as its purely resistive.

The 4 driving signals h_hi, h_lo, l_hi, l_lo (V1, V3, V5, V6) generate a 50kHz output square wave at the load consisting of upper and lower MOSFETs alternately on for 19.5uS with a 500nS deadband beween them to avoid shoot-through.



Now to the traces...

The top trace shows one cycle of the load voltage and current, plus M8 source and M1 drain currents showing that each upper and lower MOSFET contributes 25% of the load current... total switching + I2R losses are ~65W across all 8 MOSFETs (trace not shown).

Second trace shows the low side driver signals l_hi, l_lo to the driver MOSFETs, plus the resulting ldrv voltage, and a representative single gate drive. Third trace shows the same, but for the high side driver (h_hi, h_lo), referenced to V(load) rather than ground.

The fourth trace shows the supply voltages for the lower (ref ground) and upper (ref v(load)) drivers showing that C1 & C2 hold the voltages up well.

Lastly the 5th trace shows the currents from the 15v lower supply, the current charging C1 through D9 when V(load) is driven low, the bulk turn-on currents supplied by C1 to hdrv (3.7A approx) and V4+C2 to ldrv (0.6+3.1A approx). The lower supply current peaks at ~600mA and the rms value is <100mA so a single 15V, 2A supply should be sufficient for all 3 phases.
The average current through D9 is <100mA over the 40uS cycle, around 4nC charge, and C1 shows a 0.4v drop on discharge, which on a 10uF capacitor is 4nC!

 

OK, finally got there after a few crashes... think this one of the most complex simulations I've done, but fun getting it to work...
Let me say at the outset its a simulation not reality, track lengths/widths, temperature, tolerances, etc, are not modelled. This took an hour to simulate 500uS of operation of which I show here a few choice snapshots...

Firstly the circuit.... I've loosely modelled the driver IC on the LH side with M9 - M12. V4 is the 15v supply to the driver with C2 across it to provide bulk current sourcing... D9, R21, C1 are the bootstrap components that deliver the high side Vdd at V(bs, its absolute voltage relative to V(Load). 8 MOSFETs on the RH side model the upper and lower quad switchers driving the load resistor Rload. For simplicity I've shown this connected to the centre tap of the 60v supply, with Rload set to give I(load) as +/-300A. V(load) swings therefore between 0 and 60v. This is not a simulation of driving the motor (yet) as its purely resistive.

The 4 driving signals h_hi, h_lo, l_hi, l_lo (V1, V3, V5, V6) generate a 50kHz output square wave at the load consisting of upper and lower MOSFETs alternately on for 19.5uS with a 500nS deadband beween them to avoid shoot-through.

View attachment 241906

Now to the traces...

The top trace shows one cycle of the load voltage and current, plus M8 source and M1 drain currents showing that each upper and lower MOSFET contributes 25% of the load current... total switching + I2R losses are ~65W across all 8 MOSFETs (trace not shown).

Second trace shows the low side driver signals l_hi, l_lo to the driver MOSFETs, plus the resulting ldrv voltage, and a representative single gate drive. Third trace shows the same, but for the high side driver (h_hi, h_lo), referenced to V(load) rather than ground.

The fourth trace shows the supply voltages for the lower (ref ground) and upper (ref v(load)) drivers showing that C1 & C2 hold the voltages up well.

Lastly the 5th trace shows the currents from the 15v lower supply, the current charging C1 through D9 when V(load) is driven low, the bulk turn-on currents supplied by C1 to hdrv (3.7A approx) and V4+C2 to ldrv (0.6+3.1A approx). The lower supply current peaks at ~600mA and the rms value is <100mA so a single 15V, 2A supply should be sufficient for all 3 phases.
The average current through D9 is <100mA over the 40uS cycle, around 4nC charge, and C1 shows a 0.4v drop on discharge, which on a 10uF capacitor is 4nC!

View attachment 241913

@ Irving , You are definitely a great teacher and a very supportive member, Thank you for taking your time in writing and specifying all these calculations, Data and Graph.

Let me clear things up, I used Pspice for calculating the Gate current as done by you in the previous chat which showed a total of 2 A Per Mosfet at 50 Khz so I thought of lowering down the frequency to about 20 Khz max for our initial test results.
And I came up with the following graphs on the gate with 5 Ohm Gate resistance.
So this Positive peak specifies the charging of Mosfet which is about 1.6 Amp for 1 uS and during turn off the negative current peak goes to a similar - 1.6 Amp.
So According to this, we would need 1.6 * 4 =6.4 Amp of Current from each driver.

 

Can you show me the circuit? You will see in the latest simulation I've changed the gate resistor configuration to stay within the limits of that driver chip, and it seems to work well.

 

Can you show me the circuit? You will see in the latest simulation I've changed the gate resistor configuration to stay within the limits of that driver chip, and it seems to work well.

 

Rerun your simulation with correct parameters for the pulse - tr & tf 5nS, PW 1uS, Period 20uS, using this transistor model:

.model AUIRFS4010Pbf VDMOS(Rg=2 Vto=3.9 Rd=2.05m Rs=0.0m Rb=1.6m Kp=186 Cgdmax=4n Cgdmin=1n
+ Cgs=4.51n Cjo=3.7n Is=200p tt=72n ksubthres=.1 mfg=International_Rectifier Vds=100 Ron=3.9m Qg=143n)

 

@Irving

 

Yes, but as you can see from the bigger simulation I've tuned the gate arrangement to match the driver and the switching response and losses are fine. So there's no need to find a 'better' solution as yet.

Do you have data on the motor, eg coil data, torque, etc.?

 

Yes, but as you can see from the bigger simulation I've tuned the gate arrangement to match the driver and the switching response and losses are fine. So there's no need to find a 'better' solution as yet.

Do you have data on the motor, eg coil data, torque, etc.?

Yes I just got the data for the motor, It is a 6 Phase machine with the following ratings


DC Voltage input - Min 60v
Nom 75v
Max 100v
Output Phase Current (RMS Per Phase)
Min 55A
Nom 104A
Max 212 A
Fundamental Frequency 690Hz

 

Great, but need info on phase inductance and resistance.

6-phase motor? That raises some interesting issues... not least how many half-bridges do you need?

The only 6-phsse motor I've ever come across was for the Segway personal transport where it was a safety feature (early models only I believe)

 

Great, but need info on phase inductance and resistance.

6-phase motor? That raises some interesting issues... not least how many half-bridges do you need?

The only 6-phase motor I've ever come across was for the Segway personal transport where it was a safety feature (early models only I believe)

Right Now Team is stuck with the Motor development part and I have to go with the maximum achievable specs for the controller.
Updates-
PCB are here now.
Some Components are arriving soon for initial testing.

Now I need Some clarification on PWM Switching Method.
*How Should the Gate driver be feed with PWM?
For Half Bridge
and In terms of H-Bridge (If Driving A Brushed Motor )