I have problems understanding bootstrap circuit application note

[Futterama]

Hi forum,

I'm building a high-current 3-phase brushless motor controller and have found a high & low-side gate-drive IC and N-channel MOSFETs for the circuit (see bottom for links).

Circuit details:
The gate-drive IC uses bootstrap operation and can supply 4.5A source/sink gate current.
VCC = 15V
VDC (voltage to the load) = max 42V (10S LiPo cells)
Max load current = 200A
MOSFETs in parallel to handle high current = 4

I have read the datasheet and this application note AN-6076:
https://www.electro-tech-online.com/custompdfs/2013/08/AN-6076.pdf

I think I have come to the conclusion that the circuit in figure 19, page 7 would be the best way to design my circuit (attached).
I'm trying to follow the AN when calculating bootstrap capacitor and gate resistor values but the AN has a bit bad english and the calculations makes little sense to me.

If I was to calculate the gate resistor, I would calculate the resistor value needed to limit the current to 4.5A at 15V = 3.33ohms. But when calculating it using the formulas in the AN, I get 57.7ohms.

Using the AN to calculate the switching time gives me ~95ns (4 MOSFET gates in parallel with each 107nC max and 4.5A drive current). But later there is another calculation where the sink and source current capability of the gate driver is calculated and they use a 1.5 factor. So now the switching time is suddenly ~143ns as the gate-drive current is "reduced" to 3A.

The switching time is not really important to me (of course it should be as low as possible) but I only calculated it so I could work with the rest of the calculations in the AN.

Besides the gate resistor and bootstrap capacitor values, I don't quite get the rule about sizing the zener diode for the Vs clamp. The AN gives me this:
"The Zener voltage must be sized according to the following rule: VB - VS < VBS,ABSMAX"
Does this mean the zener voltage must be lower than the largest negative voltage spike the drive IC can handle on the Vs pin (which should be 6-VCC = -9V)?

I haven't even begun to calculate the bootstrap capacitor value. I have been looking through another AN and was overwhelmed by the calculations and it seemed hard for me to figure out all the values in the calculations: https://www.electro-tech-online.com/custompdfs/2013/08/AN-9052.pdf

When calculating, I use a spreadsheet where I use 1 cell for each input value with a good explanation of the value and 1 cell with the calculation. I wish the AN's had such a spreadsheet attached.

Any help with these calculations for my circuit would be much appreciated


Gate-drive IC:
Fairchild Semiconductor FAN7190_F085
Datasheet: https://www.electro-tech-online.com/custompdfs/2013/08/FAN7190_F085.pdf

MOSFETs:
Fairchild Semiconductor FDMS86500DC
Datasheet: https://www.electro-tech-online.com/custompdfs/2013/08/FDMS86500DC.pdf

 

[ronv]

Hi forum,

If I was to calculate the gate resistor, I would calculate the resistor value needed to limit the current to 4.5A at 15V = 3.33ohms. But when calculating it using the formulas in the AN, I get 57.7ohms.

Here is my 2 cents:

I would calculate the gate resistor for say 2 amps. - 7.5 ohms.

This should give you a turn on time of about 214ns - 428NC/2=214

I'm not sure the FETs are big enough with just the PCB heatsink, but maybe you are counting on the duty cycle?

Besides the gate resistor and bootstrap capacitor values, I don't quite get the rule about sizing the zener diode for the Vs clamp. The AN gives me this:
"The Zener voltage must be sized according to the following rule: VB - VS < VBS,ABSMAX"
Does this mean the zener voltage must be lower than the largest negative voltage spike the drive IC can handle on the Vs pin (which should be 6-VCC = -9V)?

Yes, but I think I would use Fig. 18 then it is not so close.

A rule of thumb for Cboost is 10 X FET capacitance. So 4X7800 pfX10 = .3 ufd.

I'm not sure the FETs are cool enough with just a PCB heatsink, but maybe you are counting on the duty cycle?

 

[Futterama]

Hi ronv, thanks for replying.

So you would calculate the gate resistor the same way I would. I think 4.5A is too much, it is the maximum rating for the drive IC, it has a minimum rating of 3.5A so if our method of calculating is the correct way, I will use 3.5A to minimize switching time.

I will mount a big heatsink with fan on top of the MOSFETs. I have found some gap filling thermal material to fit between heatsink and MOSFETs to take up any gaps and to electrically isolate the exposed "Dual Cool" pad on top of the MOSFETs from the heatsink.

Figure 18 is almost the same as figure 19, just without the zener as far as I can see. I guess the zener is just there to "delay" the schottky diode so the voltage will need to be more than a few volts lower than ground before it conducts. I will probably not use the zener.

I have been using the examples in the AN-9052 to calculate the bootstrap capacitor value and it gives me a minimum capacitance of 101nF and using the margin it gives me 862nF.

 

[ronv]

Did you take into account 4 FETs in parallel? I think it should be bigger. What numbers did you use?

https://www.silabs.com/support/Pages/bootstrap-calculator.aspx

 

[Futterama]

Now you mention it, I think I forgot to multiply the gate to source leakage current with 4.

 

[crutschow]

..................................

I will mount a big heatsink with fan on top of the MOSFETs. I have found some gap filling thermal material to fit between heatsink and MOSFETs to take up any gaps and to electrically isolate the exposed "Dual Cool" pad on top of the MOSFETs from the heatsink.

............................

You cannot use thermal grease to reliably isolate a pad from the heatsink. If you need electrical isolation, then you need to use an insulating pad washer.

 

[Futterama]

It is not thermal grease I'll be using, it is a Gap Pad:

https://www.bergquistcompany.com/thermal_materials/gap_pad/gap-pad-5000S35.htm

 

[ronv]

What is the lowest frequency you will run at? And, what is the maximum on time or duty cycle

 

[Futterama]

I haven't decided these things yet, but I assume 16 or 32kHz and duty cycle can be 100% but I would guess the motor would not run any slower than 1000rpm at 100% duty cycle (this would only happen during fast startup) and as far as I can see this will give a maximum ON-time of 30ms before capacitor refresh.

 

[ronv]

That's always a problem with motors and high side drivers.

Take a look at the HIP4080 or 81. They have a charge pump so can work at DC.

https://www.electro-tech-online.com/custompdfs/2013/08/fn3178.pdf

 

[Futterama]

Will the capacitor be unable to hold the gate voltage in the safe area for 30ms???

I have attached a screenshot of my calculations with the formulas from AN-9052 page 3.

 

[Futterama]

According to a capacitor calculator I found online, and using 15V and the leakage current to calculate the RC resistor value, a 230nF capacitor will discharge from 15V (optimistic because the capacitor will not hold this voltage due to forward voltage drop in the bootstrap diode) to 10V in 24.5ms so to answer my own question, NO, the 230nF capacitor will not hold the gate voltage within the safe area for 30ms...

 

[ronv]

Nothing good ever happens when undervoltage circuits have to work, and those FETs don't look cheap.

 

[Futterama]

Nothing good ever happens when undervoltage circuits have to work, and those FETs don't look cheap.

Can you elaborate?

They are only $2.55 each

 

[Futterama]

If I keep the duty cycle at 99% (and use 32kHz) when the motor spins at lower speeds (max motor speed is 30.000rpm) the bootstrap capacitor will be recharged every 31.25µs. If I then use 4.7ohms gate resistor, which doubles as bootstrap resistor, a bootstrap capacitor of 1µF will recharge from 10V to 14.99V in just under 30µs. So I just need to make sure the capacitor is refreshed using the MCU to control the duty cycle at lower motor speeds or just settle with 99% duty cycle as a maximum.

 

[ronv]

Maybe we should look at the whole system. Have you picked a motor controller chip? If not look at something like this:
https://www.mouser.com/ProductDetai...=sGAEpiMZZMv0DJfhVcWlK4gwEHe5ABs0XVwFf8xkJC8=
Do you have a link to the motor?
Is it controlled overall by a micro?

 

[Futterama]

ronv, I was not planning to use a motor controller chip, I was planning to use a PIC, with hardware PWM and multiplexed PWM outputs, and the mentioned MOSFET drivers, so yes, overall control by µP. The controller you link to looks very simple to use, but it concerns me that the high-side gate voltage is only 5.8-7V and the gate output current is below 100mA. So I don't think it is suitable for my high-current application.

The end application is an ESC for a largescale RC car. The motor is this:

https://www.skyrc.com/index.php?route=product/product&path=39_48&product_id=141

I got the motor already. The motor is sensorless but I plan to add hall-effect sensors, I have had the motor disassembled to make sure there is room for the sensors, I just haven't got around to find the correct sensors yet (I guess it will be trial and error regarding hall-effect sensitivity).

The motor is not wound like the 1:10 motors I have seen, it looks more like how an AC motor is wound. The rotor is also 4-poles where the 1:10 motors are 2-pole, so I'm also a bit confused about the commutation sequence and necessary number of hall-sensors.

 

[Futterama]

Whole system:

 

[ronv]

I could be wrong but I think the bottom FET needs to be on to charge the cap & I think this only happens when that particular coil is energized. So the duty cycle is the motors slowest speed not the pwm frequency.

 

[Futterama]

You are correct, the bootstrap capacitor is charged through the bottom FET.

When the top FET turns OFF, the coil will create a voltage opposite of the supply it just received, and the current from this will flow through the freewheeling diode in the FET. If active freewheeling/synchronous rectification is used, the bottom FET will be turned ON when the top FET is turned OFF but only when using a PWM duty cycle below 100%. If the PWM duty cycle is 100%, THEN the motor speed will be the controlling factor of capacitor refresh. But if I was to use a 1µF bootstrap capacitor, then a fully charged capacitor could hold the top FET turned ON for ~100ms but it would only take ~30µs to recharge the capacitor again from 10V to 15V. So I just have to make sure the bottom FET is ON for 30µs every 100ms or less.

So I can use 100% duty cycle whenever the motor is running faster than 60rpm or less (I haven't figured the commutation sequence yet so it is probably even slower).

Since the capacitor recharge time is so short, I can also consider to increase the capacitance so there is even less chance of hitting a low capacitor voltage condition (which will trigger the under-voltage lockout in the MOSFET gate driver which will prevent the high side MOSFET from operating in it's linear region, but I haven't figured out what will happen to the rest of the circuit if that happens).