MOSFET heating problem H Bridge moter drive

Status
Not open for further replies.
U

usmanyousaf

Junior Member level 2
Joined
Nov 25, 2012
Messages
20
Helped
1
Reputation
2
Reaction score
1
Trophy points
1,283
Location
Karachi
Visit site
Activity points
1,425
HI
i am using IRF540 MOSFETS with Rds(on)=44m ohm. using IR2110 to drive the MOSFETs at switching frequency 5kHz. the H bridge is used to drive a permanent magnet DC motor with current rating 3A MAX while the MOSFET can handle upto 33A (according to datasheet). there is a dead band delay of approx 3us(Hard wired). Freewheeling diodes are used in the circuit. Bootstrap cap is 2.2uF. Gate resistors are 10ohms. and motor operation voltage is 24V DC from an SMPS.

i made 4 PCB's, same design same component values. two of them work fine while other two have heating problem..the MOSFETS heat too much..what might be the reason.

is the bootstrap cap inappropriate...??
i,ve changed Driver ic's many times but it didnt help.
is there any way to check health of MOSFETS using multi meter..??
 

FET do not really degrade. They either just stop working or go out in a puff of smoke...

Scope the waveform at the gate of the FET. Charging the gate capacitance too slowly will keep the FET in an ohmic region for a prolongued period of time. Charging it to a too low voltage will have the FET settle for a much higher RDSon than the 40mOhm you mentioned.
 

@ice-tea
thanks for your reply..what do you recommend..?? i,ve a working circuit of exactly same design and components value..!! how do i decrease charging time, by changing bootstrap cap..??
 

I recommend firstly scoping the waveform on the gate. Best place to start.
 

Hi ,

Add to that: use a 1k resistor between gate and source of mosfet to prevent accidental turn on of the mosfets. Usually it doesn't cause a problem at low voltage,however its better to make a safe design. Hoping it will also reduce the heating. Additionally, you should also use a 0.1uF ceramic/tantalum capacitors across the bootstrap as well as the filter capacitors.

Thanks.
 

Suggesting gate-source parallel resistors for integrated bootstrap drivers like IR2110 is a serial topic at edaboard.

I must confess that I wasn't yet a ware of a situation where these resistors would serve a purpose. IR2110 has push-pull outputs and undervoltage lockout. There might be a small Vcc voltage range where the output drivers don't work yet and a MOSFET with low threshold voltage (logic level FET) can be already partly turned on. So if people believe in this circuit detail, I don't feel prompted to contradict. But once the circuit is fully powered, the only effect of the resistors is in consuming precious bootstrap power.

I presume that the said 2.2 uF capacitor is already a ceramic type, otherwise a low ESR/ESL capacitor should be supplemented.

From the posted observations, there's no plausible explanation why some boards heat up, but I guess, the reason is simple nevertheless. Inspecting the waveforms thorougly would be my way.
 
FvM,

the 2.2uF is not ceramic type..but i will add a ceramic cap and see what happens. some friend suggested to add a ceramic cap of low value between source and ground to sink any spikes from power supply...should i try it..??

- - - Updated - - -

ADD to it: the MOSFETs on Faulty boards heat up even when load is disconnected. and MOSFETS on low side of H Bridge heat more as compared to those on high side.
 

anyone here with a solution/suggestion..?? the voltage waveform on gate is okay, enough to drive the FETs in saturation region..still FETs are heating up..!
 

i'm guessing you don't have low enough inductance capacitors across the power supply rails of your h bridge, and during the switching transition you're blowing the fets up due to the energy stored in the inductance dumped into the mosfet as avalanche over voltage.

ac couple your oscope to the power supply rail and don't let anything surprise you...
btw, you have to account for not only the motor current, but also the diode recovery current.
 

btw, you have to account for not only the motor current, but also the diode recovery current.
Click to expand...
There must be no diode current at all in a synchronously switched H-bridge, but a dead time of 3 us will force the full load current into the diode for reverse operated switches.

Of course all these considerations are guesses where we should have substantial measurements.
 

I am not using a diodo anti parallel with gate resistor, which according to some people helps turn of the FET quickly, can this be the problem.

but the question remains, why dont the TWO boards with same design work fine..=(

- - - Updated - - -

is it possible that just by chance all the MOSFET drivers are faulty..and FETs do not turn off/on properly..?
 

i have very clean Vgs wave form. no spikes no sinosoidal components, Vgs is little more than 10V.

- - - Updated - - -

i have very clean Vgs wave form. no spikes no sinosoidal components, Vgs is little more than 10V.
 

To get transistors excessively heated, there must be either unexpected conduction losses or switching losses. Both can be detected by exact measurements.
 

tell me a few methods to check conduction and switching losses...as mentioned earlier switching frequency is 5kHz and Ids is 2-3.5A and is fused at 4A...i do not expect that much conduction or switching losses in these conditions.
 

Losses are generated according to Eloss = ∫ Vds*Id*dt

You have to check the waveforms for regions where Vds and Id are both ≠ 0 at the same time.
 

that's a tough one.. you're setup sounds fairly good wiht lots of margin. i would not expect normal conditions to raise the temp much at all. i don't know the drive capability of IRF2100 but 1.5A peaks not too aggresive, should be OK.. hmm.. can your SMPS supply the pulsed current? how far away is it from your H-bridge? maybe you need bigger DC link cap near the Hbridge..

are you sure it is failing due to over temperature? high voltage on DC bus can also fail the mosfet fast and not much heat generated there. tough... do you have a piccture of the hardware?
 

Status
Not open for further replies.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…