IR2110 getting hot while connected as half bridge

[sam781]

I've made the the attached circuit its giving the output but the MOSFET driver is getting hot after powering on. I've checked by changing C1 to 5uF and removing D2 and D3 but same result.

 

[barry]

1) How hot is 'hot'?
2) Does the circuit work, or is your chip damaged?
3) How fast are you toggling?

 

[sam781]

1) How hot is 'hot'?
2) Does the circuit work, or is your chip damaged?
3) How fast are you toggling?

1) After powering on... It starts working. Both the outputs (HO & LO) seem ok. It started becoming hot gradually. After 2/3min it seemed that the need to be be shut down otherwise it would be burnt. Then I switched off to safe it. Output remains ok till then. I did such experiment several times by sensing the heat with my finger.

2) Circuit works

3) PWM frequency 50kHz

 

[D.A.(Tony)Stewart]

What was load impedance used for test?
Have you verified your input and output pulse delays for dead-time control?

 

[Audioguru]

The Mosfets have a very high gate-source capacitance which takes a lot of power from the IR2110 driver to charge and discharge at your fairly high frequency. The diodes you use parallel with the series gate resistors are not shown on the typical circuit on the datasheet of the IR2110 so maybe they also cause too much heating.

 

[FvM]

The principle topology looks o.k., also the diodes for slow on/fast off operation. If it's appropriate to use no gate resistor for switch-off depends on your circuit layout. You can also estimate the IR2110 power dissipation from IRF3205 gate charge data. As far as I see it's in an acceptable range. 50 kHz isn't too fast.

This leads to the conclusion that the problems are brought up by circuit properties that can't be seen from the schematic. Either heavy ringing of the half-bridge output node that might pull Vs too far below COM. Or a ringing control signal that causes multiple switching edges.

 

[sam781]

What was load impedance used for test?
Have you verified your input and output pulse delays for dead-time control?

No load and 1k ohm resistor. Just for testing...

Didn't vary. Large dead zone set as a safe guard for testing.

The Mosfets have a very high gate-source capacitance which takes a lot of power from the IR2110 driver to charge and discharge at your fairly high frequency. The diodes you use parallel with the series gate resistors are not shown on the typical circuit on the datasheet of the IR2110 so maybe they also cause too much heating.

Input capacitance of IR2110 is high. As per the datasheet, Ciss= 3247pF at 1MHz (Am I correct? Is it Cgs?)

Diode is for fast de-charging. I removed the diode and found the same result.

The principle topology looks o.k., also the diodes for slow on/fast off operation. If it's appropriate to use no gate resistor for switch-off depends on your circuit layout. You can also estimate the IR2110 power dissipation from IRF3205 gate charge data. As far as I see it's in an acceptable range. 50 kHz isn't too fast.

This leads to the conclusion that the problems are brought up by circuit properties that can't be seen from the schematic. Either heavy ringing of the half-bridge output node that might pull Vs too far below COM. Or a ringing control signal that causes multiple switching edges.

Tired removing the gate diode.

Not getting "heavy ringing". :sad:

 

[cdez]

Hello,
I am going to build a circuit almost same as you with IR2110 and IRF3205 so I am interested by your problem, I would'nt like to have the same issue

Where did you bought your IR2110 ? Maybe it is a counterfeit ?

 

[dick_freebird]

Ciss I think doesn't tell the whole story, the Miller charge
often is more significant. It can also keep the output in a
high dissipation region during turnoff while you wait for the
drain to slew. Look at the LO output (since this is easier)
and see if there's a long "porch".

 

[kappa_am]

Put a 6.8 Ohm resistor in series with parallel diode. it may solve the problem.

 

[sam781]

I've changed the driver and its working. Its confusing that the old IC gives the output but becomes hot w.r.t time, new one is not getting hot. If the old one did not give the correct output I would change it much earlier.

Where did you bought your IR2110 ? Maybe it is a counterfeit ?

Yes, you are correct. I bought a lot of faulty IR2110. The problem was it was giving the output. Current one is not getting hot. I think you faced the same problem. I bought from Patuatuli. Do you have any suggestion?

 

[cdez]

What is Patuatuli ? Didn't find anything about it.
I'm expecting to buy my IR2110's from ebay from China

 

[D.A.(Tony)Stewart]

Monitor the current with a small 50mV shunt resistor on high or ground side.

See what effect removing the HO gate resistor to ground has. When it bleeds the charge pump boost cap, the LO PWM must restore the voltage and this represents a charge dump current that heats up the lower MOSFET. Remove it.

Then when running cool, load the output with pull up and pull down and measure the voltage drop on opposite MOSFET to load resistor.
This drop in voltage from Vcc or rise from Gnd over the load current is the RdsOn value and compare with spec. To make it easy use a 1 Ohm resistor with a small cap and look at the step waveform voltage drop and use R ratio to compute RdsOn. Note that the ESR of the Boost Caps is in parallel which are in turn in series with the gated HO drive pulses.

 

[sam781]

Thanks everyone for helping me...

Now, I've made H-bridge configuration using the one pair of previously given circuit (without the gate diode). Out puts of two HOs are okay. But LO (cyan color) of one driver is distorted other LO (yellow) is ok.

See below -


My opinion is, it could be due to gate to source capacitance of the power MOSFET (IRF3205). But all the four MOSFETs are same. :sad:

 

[sam781]

I've changed the Driver and its okay now... :smile:

 

[sam781]

Now facing problem while connect Transformer with the bridge output. MOSFEs are getting hot w.r.t time. I've made ferrite core transformer - dimension calculation taken from here - https://www.edaboard.com/threads/254847/#post1130295

Battery voltage=13.8v (max)
Required output voltage to make ac 230v (rms) = 455v
So, Vp=13.8=14v
Vs=455v

core= EDT39
Tern per volt=1e8/( 4*f*B*Ae)= 1e8/(4*50000*1600*1.25)=0.25

Hence, primary terns=14x0.25=3.5
Secondary terns = 455x0.25=114

Any idea why it is getting hot? How can identify and resolve this problem?

 

[kappa_am]

observe voltage between drain and source, how much is it?

 

[sam781]

Drain to source voltage of each MOSFET is aounnd 0.8v (when measured using DMM selected as AC volt) and 1.6 (DMM selected as DC volt)

Output waveform from bridge wihtout connecting to transformer's primary (just 1k resistor is connected to the output terminals)


Output waveform from bridge, output is connected to transformer's primary (one 1k resistor is also connected to the bridge output terminals)

 

[sam781]

Sorry, drain to source voltage is 6.1v (DMM selected as DC)

- - - Updated - - -

Drain to source output waveform for both the low MOSFETs are here -

 

[KlausST]

Hi,

Why R3, R4 are 1k?
Try 10k or higher.

Then you don´t need that huge 100uF capacitor for bootstrap voltage (although not wrong... You need well chosen electrolytic capacitors for 50kHz)

********
Picture in post#19:
It seems the transformer is saturating.
timing of scope seems to be 5us/div. If so ... saturation begins after 6us (rapidly increasing current, high power dissipation) .
twice 6 us = 12 us --> try PWM freq with 80kHz or more.

Klaus