IR2110 low side mosfet heating up

[cdez]

hello,
I have build a half bridge motor controller with ir2110

I have made a first test, the half bridge worked good I accelerated decelerated a motor at 20khz with pwm coming from a microcontroler. The pwm is also inverted with the microcontroller with dead band delays of few microseconds for the second input.

But now I have a problem that appeared I don't know why: the low side mosfet heats up, and maybe the vcc-com capacitor.( I am not sure if the capacitor heats itself or it was heaten by the mosfet heatsink near it )
And also flowing current increases slowly, it changes all the time very weirdly.

(I don't remember all of this happening at the first test maybe I didn't test carefully.)

the motor has 0.08A without controler at 12v dc, with ~90% duty cycle using my controller it has 0.1A at beginning thens rises to 0.2A for no reason

mosfets are irfb7540, schotky diodes, 10 ohm gate resistors, blue caps 10uf and the bootstrap 100uf

here is a picture of my controller:


thanks

 

[KlausST]

Hi,

Sounds like cross conduction, but you say ..some microseconds of deadtime...
Please use a scope to test on this.

Other points....
Make sure your capacitors are specified for that frequency. Standard electrolytic capacitors won't do the job.
A schematic with values of the parets would help to help you.
If possible use solid copper planes for gnd and vcc.
Make sure the fet internal diodes have fast recovery time. If too slow use external schottky diodes in parallel.
Here they don't have much power dissioation, but they need to be specified for the pulse current.
Split power and gare drive paths

Klaus

 

[cdez]

I have reduced frequency to 5khz and set deadtime delay a bit more than 5 us ( before it was like 2 us )
And still the same issue flowing current gets higher because there is cross conduction and my generator is protecting from it (the generator is making switch noise regularly).

I suppose if I'd feed with a battery everything would burn.

The low side signal ( LO pin ) is really weird it's a pulsed signal but the low area is like 8 volts (not constantly, it changes all the time, sometimes its 0) and high is 12 so because of that LO is always ON and it causes cross conduction

I really don't understand why this is happening, because I remember that it was working fine when I first tested it.

My diodes are 1n5819
https://www.fairchildsemi.com/datasheets/1N/1N5817.pdf

 

[KlausST]

A complete schematic could help.

Klaus

 

[cdez]

Hello,
Sorry for the late,
I think I might have found the problem but I don't understand why.

I have put a 10ohm resistor at the + of my current source (12v source) so that in case of cross consuction current would be limited at 1.2 AMPS. (12/10)
After simulating that on LTSPICE I realised it makes the IR2110 behave weirdly: I think this resistor may have blow up the low side of my controller for a reason that I ignore ...
Without the resistor everything works well on simulation but with it it behaves weirdly.

In real life in my first test WITHOUT the resistor it worked fine, then I put the resistor, then it doesn't work properly anymore, with or without.

Here is the schematic:


low side input behaves weirdly:


the two 5v batteries are inverted pwm with delay coming from my controller
and the third one is DC 5v from my VCC pin to GND pin of my PIC


I feel so stupid now .. and I really don't understand why this resistor blowed up my controllers.. I thinked it was protecting..

 

[KlausST]

Hi,

the resistor limits current out of your 12V supply.
So every time you switch ON your load, the voltage across the resistor is about 11V, and at your load only about 1V. There is a RC constant with C2.
This surely is not what you want.

My recommendation:
Don´t use the 10R resistor.
Place an inductor (low uH , but saturation current > 12A) directely between upper mosfet drain to 12V. Place a diode in parallel to the inductor, cathode in direction to 12V.
See what happens.

BTW: I still think this is a layout/ wiring problem. Can you take a picture of you test circuit?

Klaus

 

[Vbase]

If the voltage to the power mosfets is always 12V and not more than 20V then it is better to connect Vb of IR2110 directly to 12V and Vs to COM (ground). This way you make sure that you always get solid drive to the mosfets.
If you are going to increase the voltage to the mosfets above 20V don't do it!!

 

[cdez]

Can you take a picture of you test circuit?

I can't actually I'm doing this at school, maybe next week

the circuit is simple, if you look at the first pictures I've posted, the 2 orange wires goes directly to the PIC18F pins (two different pins with inverted signals)
the red and the brown -> VCC pin and GND pin.
I have a 1k resistor on my VCC pin to limit current and not burn my PIC

since my PIC breakout board takes male inputs, I need to wire this through a breadboard, the controller is first wired to breadboard, then breadboard to the PIC, it's a bit messy ( wires are curved etc .. )


Then the big red wires at the middle goes to one side of the motor and black one to the other side, AND ground.

and the other big red goes to VCC (12V)

 

[KlausST]

Hi,

If the voltage to the power mosfets is always 12V and not more than 20V then it is better to connect Vb of IR2110 directly to 12V and Vs to COM (ground). This way you make sure that you always get solid drive to the mosfets.
If you are going to increase the voltage to the mosfets above 20V don't do it!!

I don´t think this could work.
because even on a 12V system the SOURCE pin of the upper FET travels form 0 to 12V. And the GATE of this fet needs to go about 10V higher than source_pin_voltage to make the FET conductive. This makes about 22V.

But maybe i´m mistaken.

Klaus

 

[Vbase]

Klaus, you are right, the mistake is mine.

 

[KlausST]

Hi,

i just reviewed your pictures.
No wonder there are problems. ..

There is a blue elctrolytic capacitor. It is not clear to me if it is for stabilizing the IR2110 voltage or to stabilize the power supply.
(Yes, you can say it is the same voltage. But the placement and the values are different.)
I recommend to use the blue one as stabilizer for IR2110.
And you need another electrolytic, larger, low_esr, high frequency type for load current stabilizing.

Let´s go to the picture with the bottom side (wiring)
* left top is the GND line. The connection to the FET is OK. If we use the blue capacitor as IR2110 supply stabilizer then the connection is good also. Wirinng of capacitor "+" is below...
(your orange line now carries pulsed currents. When switching the load you charge/discharge the capacitor. Therfore you have voltage drops on the orange line. This is not good as reference voltage for the IR2110)
* top middle there are three red lines. This is the connection to the load (why three?). They also carry pulsed current. Use only one. IMPORTANT: Connect all three joints on the PCB very solid. Use a thick solid wire.
(VS of the driver needs to be the very same as highside_fet_source_voltage. No routing to load and back!)
* right side red wire is the 12V wire. Here a thin red wire connects the blue capacitor. Instead of the red line use a 10R resistor in series with a standard diode. cathode to capacitor.
As said above add an extra large capacitor with short wiring (GND and 12V) directely to the FETs. Use thick short wires. This should carry the pulsed load current.
(the diode to the blue capacitor prevents then capacitor from discharge when heavy load is connected. And thus ensures a proper supply to drive fet (especially high side).

Please do the changes and see what happens.

Good luck

Klaus

 

[cdez]

Thanks for your review,

The two blue capacitors are for stabilizing IR2110/power supply ( IR2110 stabilizer ) , one stabilizes 5v supply (VSS VDD) and the other one (near the mosfet ) 12V supply ( VCC COM) These are 10 uf.

The black one is 100uf and it is the bootstrap capacitor.

So for the load supply stabilizer I will add another 100uf (+ 100nf in parallel )
and for the red line I 'll replace by resistor + diode as you said.