Problem with the mosfets of my ESC

[joelsrf]

Hi there, i´ve dimensioned the next circuit to control a bldc motor which would require 32A or more with load and instead of being 12 V is 24V.  I have changed the mosfets and some of the BEMF resistor so that the arduino has a maximun 5V entry. Well, today i have done a test but my mosfets had burned. I´m attaching the datasheet of the mosfets that i have used. I don´t know what could happen to burn the mosfets (they are able to handle till 46 A, +-30V).  If anyone can help me i will be grateful.

[joelsrf]

Eschematic:

[wraper]

i´ve dimensioned the next circuit to control a bldc motor which would require 32A

they are able to handle till 46 A, +-30V

Typical R_DS(on) is 0.082 Ohm, P=R*I² -> 0.082Ohm*32A² = 84W power dissipation. Of course it will be less in average when switched but still not reasonable. Not to say even if they were appropriate MOSFETs with appropriate cooling, it's still easy to kill them if not driven properly.

[Siwastaja]

Current sense and control circuitry is completely missing. It's absolutely essential. I recommend low-side shunt resistor on two phases + sense amplifiers into the analog comparator pin of the AVR.

Also post photo of layout. Placement of Q1, Q2 and C10 is essential (same for other phases, too). Assuming C10 is an MLCC, consider adding some electrolytic bulk capacitance, too.

Calculate MOSFET power dissipation and resulting die temperature. If you ignore switching loss at first, it's a simple calculation. If unsure about your RthJ-A, look at different MOSFET datasheets for example values with different kinds of PCB layouts, compare to yours. Remember that Rds_on at higher temperatures is like 1.5 times the Rds_on at 25degC.

[Manul]

I will add that there is no resistor in series with bootstrap diode. High peak current might kill the diode. Fix that.

Your driver IC is quite weak. 100R gate resistor certainly does not help. You will have increased switching losses.

IR2101 has independent high and low drivers, it is essential to confirm, that your software is switching the mosfets correctly and with correct deadtimes, because otherwise there might be shoot-through. It is common for software to cause all kinds of disasters, don't overlook that.

[Benta]

Couple of other things:
There's no dead time control at all (except in FW, I expect, no way to know).
I don't know the the MCU used, but of the ones I've worked with default to "high" I/Os during booting = all MOSFETs full on (pull-ups, or even worse floating).

This design is not good, and the schematic style is almost unreadable. Why make blocks connected with labels? What's wrong with a connection? No one wants to search a schematic for text labels, when following a wire is much easier.
CAD packages use hierachical sheets for complicated schematics, but that's not the case here.

[rob77]

instead of being 12 V is 24V.

why do you think it will work @ 24V ? 
IR2101 has a recomended VCC range of 10-20V and has a absolute maximum rating of 25V.
MP2307 has a 4.75V to 23V Operating Input Range , absolute max rating 26V

you can't double the voltage without making sure all components will be happy with it.

[Benta]

Lots and lots of troubles with this design.
@joelsrf: it's back to drawing board

Welcome to the Forum, by the way, don't be discouraged. We're generally friendly.

[rob77]

you could have kept the thing at 12V and use 24V only to power the half bridges. ir2101 powered with 12V can handle a half bridge powered by higher voltage (upto 600V) no need to change everything to 24V

[Benta]

True.
You just need to place the bootstrap diodes differently. Supply voltage to the MOSFETs could be anything.

[Siwastaja]

Couple of other things:
There's no dead time control at all (except in FW, I expect, no way to know).

Indeed; most bootstrap gate drivers seem to do internal deadtime (or at least very least cross-conduction prevention with zero-ish deadtime), so I didn't even bother to check; but this ancient part does not mention about it in the datasheet so I'm assuming it does not have such protection.

For the OP, I would suggest any more modern bootstrap driver with cross-conduction prevention logic. My gripe with these parts is, they often give too much deadtime, but it will be only a small issue, if an issue at all.

One with internal deadtime plus IN/EN inputs (instead of HIN/LIN) would be easiest to use: deadtime just works and you can't even ask the part to cross-conduct. Software is also easier to write: wire all EN pins together and connect to GPIO for global enable control, and then the three PWM outputs to INputs.

OTOH, the HIN/LIN interface provides you with software-controlled deadtime because you can let the MCU timer peripheral do it - of course assuming you have it in the MCU and configure it correctly.

[inse]

Why did you choose 650V MOSFETs if you only have 24V?
Btw, the rated 36A Id is only at 25°C case temperature.
I'd select a MOSFET with lower Rdson.

[joelsrf]

One guy told me this too:

"I see things wrong.
------------------------------------------------
With no load too much heat, the problem is "shoot through". You need to turn off a MOSFET, wait 200nS then turn on the other part! (on both edges) The IR2101 allows you to turn on both the top and bottom MOSFET at the same time. You must add some "dead time".
------------------------------------------------
100 ohms on the Gate? Try 2 to 10 ohms. Your AC losses will be too high with a slow 100 ohm resistor.
------------------------------------------------
You are using 650V MOSFETs. That is OK but 30V through 100V would be much cheaper. Look for 50V parts.

You are using 95m ohm MOSFETs that are rated for 35A. For the moment forget about AC losses and look at only DC losses. 35A x 35A x 0.095 = 113 watts. You will need a very large heat sink. You will lose 3.3V across the MOSFET.

I do not know this motor. A motor rated at 35A run current probably will need 100A to start up.

I searched digikey.com for MOSFETS, N, 75 to 200A, 30V to 100V in a TO-220, <2.5 gate turn on voltage.
I randomly picked PSMN4R3-30PL,127 which is 100A 30V 4.3m ohm, It looks like it will handle 100A at 3.5V of gate. (5V drive is better) DC loss = 5.5 watts. Needs a heat sink but better.

PSMN1R1-30PL,127 is 30V 0.0013 ohms. 200A Now we are at 1.5 watts. small heat sink.
--------------------------------
You defiantly have shoot through. You need low voltage, high current MOSFETs. (low Gate turn on voltage)"

[wraper]

You are using 650V MOSFETs. That is OK but 30V through 100V would be much cheaper. Look for 50V parts.

It's not about cost. With lower voltage rating, you get lower channel resistance, and with that orders of magnitude less heat dissipated. The first parameter you should look for is R_DS(on) (drain-source ON resistance), not current rating which is pretty much a useless parameter without taking into account R_DS(on). Of course you can get extremely beefy high voltage MOSFETs with low enough R_DS(on), or connect multiple in parallel. But then you will get extremely high gate capacitance/charge that make extremely difficult to drive the gate fast enough.

[joelsrf]

Well, first of all thanks to all of you.
I have done a phase of the squematic with the components that I have rally used. I´m attaching the file behind.
Apart from the notes that  I take on the photo I would like to be recommended about this items.

1. The resistance that goes to the driver is 100 ohm but some of you told me that it isn´t the most apropiate value. Which value would you recommend me?

2.I tried some days ago this circuit modifying the value of the C13 capacitor. I just try the circuit with a program that putted 'HO' (HIGH or LOW) and 'LO' (HIGH/LOW). Well, I putted some capacitor with more value, 10 uf and 33uf for example. Firstly the voltage of the high side MOSFET in the gate was about 24 V more or less, but instantly decreased, the lower the value of the capacitor, the faster it fell the voltage on the gate. This can lead on what someone told me, the program really turns off the high side MOSFET but physcally it is really slowly decreasing his value instead of putting off instantly. This can lead into a short circuit between the two MOSFETs.

3. Which MOSFET would you recommend me? Do you think the ones that the guy told me are okay for what I´m trying to do?

4. He also mentioned something about the peak current when the motor tries to start spinning. I didn´t take that into account and the MOSFET  that I had used wheren´t able to handle that so they burned.

Tell me what you think, thanks!

[Siwastaja]

4. He also mentioned something about the peak current when the motor tries to start spinning. I didn´t take that into account and the MOSFET  that I had used wheren´t able to handle that so they burned.

This is only applicable to very small or inefficient motors. Oversizing parts until cows come home is not likely a good solution for your use case, you should just add current control, in which case rated motor stall current does not matter at all. Motor current and hence MOSFET current will be fully under your control under all times. Because any motor type (of those you would use) has a lot of inductance, current changes quite slowly. This makes control easy: just sense current at least once every PWM cycle, and lower the duty cycle and/or terminate current cycle early if necessary, and current starts to drop again. Motor current is basically a triangle wave where it increases during MOSFET ON time and decreases during OFF time.

Current control is just a way to sense motor current instantaneously, and terminate PWM cycles early if limit is exceeded. You could use a shunt resistor between lower MOSFET Source and GND wired directly to Analog Comparator pin of the AVR, but for lower power loss, you really want to use small shunt resistor value, and amplify the small voltage by a thing called current sense amplifier IC.

[rob77]

Well, first of all thanks to all of you.
I have done a phase of the squematic with the components that I have rally used. I´m attaching the file behind.
Apart from the notes that  I take on the photo I would like to be recommended about this items.

1. The resistance that goes to the driver is 100 ohm but some of you told me that it isn´t the most apropiate value. Which value would you recommend me?

2.I tried some days ago this circuit modifying the value of the C13 capacitor. I just try the circuit with a program that putted 'HO' (HIGH or LOW) and 'LO' (HIGH/LOW). Well, I putted some capacitor with more value, 10 uf and 33uf for example. Firstly the voltage of the high side MOSFET in the gate was about 24 V more or less, but instantly decreased, the lower the value of the capacitor, the faster it fell the voltage on the gate. This can lead on what someone told me, the program really turns off the high side MOSFET but physcally it is really slowly decreasing his value instead of putting off instantly. This can lead into a short circuit between the two MOSFETs.

3. Which MOSFET would you recommend me? Do you think the ones that the guy told me are okay for what I´m trying to do?

4. He also mentioned something about the peak current when the motor tries to start spinning. I didn´t take that into account and the MOSFET  that I had used wheren´t able to handle that so they burned.

Tell me what you think, thanks!

C13 is part of the charge pump to provide gate voltage above source.. it has nothing to do with turning the mosfet off.  you can use any value significantly higher than the gate capacitance of the mosfet, it's charged on every "LO" to Vcc and then it's "lifted" and provides the gate voltage during "HI" for the upper mosfet.  even 100nF must be enough for most of mosfets

regarding the 24V... not using 12V because you would need a buck converter ? WTF ? so to save a buck converter, you will rather operate the circuit on the edge without any safety margin ? IR2101 has an ABSOLUTE MAXIMUM rating of 25V for VCC... that means if you power it with 24V and there will be a spike on the rail it will go "KA-BOOM" most probably taking the mosfets with it.

another point on 24V to the IR2101 is the max gate voltage of mosfets... most modern low RDSon mosfets will have a max Vgs of 20V...

[joelsrf]

Well, so first of all, would you recommend me to put a buck converter and supply de IR2101 with 12V or to change that driver for another one with internal deadtime for switching and ensure that both MOSFET are never 'ON' at the same time. If you would recommend me the 2nd option, which driver would you recommend me?
Thanks.

[rob77]

Well, so first of all, would you recommend me to put a buck converter and supply de IR2101 with 12V or to change that driver for another one with internal deadtime for switching and ensure that both MOSFET are never 'ON' at the same time. If you would recommend me the 2nd option, which driver would you recommend me?
Thanks.

apparently you're not getting the point... there is NO option to run ANY gate driver at @24V... so it's not a "OR"...  it's "AND" , so both buck converter to make VCC suitable for the gate drivers AND change the driver to one with deadtime.

if you want a gate driver which is in stock and has dead time, then IR2108 for example...

https://www.tme.eu/en/details/ir2108spbf/mosfet-igbt-drivers/infineon-technologies/