Problem with IRS2104 (half-brifge MOSFET Driver)

[kappa_am]

Hi all,
I like to drive some switches that half of them have complementary state to other half ( some thing like a full-bridge inverter). in other word if s1=ON, s'1=Off. the situation is like half-bridge operation; so I decide to use a half-bridge MOSFET driver. since charging path of boot-strap capacitor is not secured, I have used two separate DC source to drive FETS. But there is a problem, some times both outputs of driver have same state (both ON or OFF).!!! I am confused what's going wrong in the Circuit. the schematic of the driver is attached bellow. I would be gratefull if you share your comments.

Thank you.

 

[BradtheRad]

The N-mos gate is referenced to its source terminal (more negative). The source terminal needs to see a definite path to ground, in order for the mosfet to operate.

A mosfet and its controller U1 ought to have the same ground connection.

Q1 is at the high side so its gate may need extremes of voltage. Make sure it is turning fully off and fully on. It might help if you connect a 10k resistor from its gate to 0V ground.

 

[kappa_am]

Thank you for your reply.
I have tried 10K resistor, but the problem is still exist. I removed MosFETs and replace them with 2 LEDs. sometimes high side LEDs remains ON even when the low side LED gets ON. I think the problem is IRS2104. I have a question; could I use a isolated DC source instead boot-strap capacitor in configuration of half-bridge driver?
if yes what issues I have to take into consideration?
If no what's your suggestion to drive this kind of circuit?

Thank you

 

[kappa_am]

I found out the problem is because separate ground of the DC source of high-side and Low-side.
so I think I could not use a half-bridge driver in my circuit. what can I do to avoid implementing shoot-through and dead-time protection, on PCB? is there a driver which uses separate DC sources for high-side and low-side ?

thank you

 

[BradtheRad]

I have no experience with driver IC's such as you are using, so I can't know which to recommend.

One option is to use a P-device at the high side. Pull the gate low (with an NPN transistor) to turn it on.
Use a pull-up resistor to turn it off. The resistor must be connected to the supply, because the gate is referenced to the positive terminal.
Of course this may complicate the ground paths further.

 

[FvM]

The schematic doesn't show high- and low-side transistors rather than low-side switches with separate grounds. But how is the relation of G1 and G2? Why are you thinking about possible shoot-through?

You are probably right that IRS2104 can't be used, but you should tell the full story to get meaningful answer.

 

[kappa_am]

the schematic just shows the driver and its connection to FETs. the real schematic looks like a half-bridge, But Source of the high-side switch and drain of low-side switch isn't directly connected to each other. since the switches are connected to a large capacitor there is a possibility of shoot through.
the power section looks like below picture.
G1 and G2 is ground of two separate DC sources which I use to fire MOSFETs

Thank you

View attachment fc.bmp

 

[FvM]

Similar to multilevel converters, the topology doesn't like suited for level-shifting bootstrap drivers. You should use isolated DC/DC (which you apparently already have) and opto-isolated gate drivers as long as the isolation is sufficient, or inductive isolated drivers.

 

[kappa_am]

thank you. I think so.
what about shoot-through problem?

 

[FvM]

Ther may be a problem that G2 potential falls below G1 which causes problems for the drivers chip. With optoisolated gate drivers, this shouldn't be a problem. Correct control waveforms must be assumed of course.

 

[D.A.(Tony)Stewart]

The IR2104 and re higher voltage S version have well controlled dead times by controlling the rise time to be slower than the fall time. 520ns dead time.

Meanwhile the opto input is very slow rise-time and also saturation recovery time. tPLH=7us max , tPHL =1us max @25C

You must choose fast or low capacitance FETs to control the load dead time or increase the gate dead time by the required amount such as from 0.5 to 1 or 2us.


Please report back with measured times or with photos.I hope you have very short ground leads.(low ESL)

 

[kappa_am]

how about using optocoupler along with IRS2104 but just a driver for each MosFET. I mean the low-side of driver get used for firing MosFET. It looks naive but I think it can solve the problem. In this way i think the shoot-through problem could be solved. what's your opinion?

 

[FvM]

Why IRS2104? You can have single channel drivers or optocouplers with built-in gate driver.

 

[kappa_am]

because IRS2104 has in-built shoot-through protection. therefore there is no need to consider it in programing or building extra circuit to deal with that.
IRS2104 typically has 520ns dead-time turn-off to turn-on at low side. and maximum 170ns turn-on rise time. my switches (IRF740) need about 50ns to turn on and about 80ns to turn off.so this dead-band could provide shoot-through protection. furthermore, the price of the IRS2104 and a single channel driver like IRS2117 are almost same. what's your opinion?

 

[D.A.(Tony)Stewart]

Your schem. indicates two IRF74D, which may have intended to be the common IRF740 . This part is very fast but reverse body diodes rise time can exceed the 0.5uS of dead time in the gate driver signals with 10A load.

Where are your measurements?

 

[kappa_am]

I haven't proper instrument in home to measure the times precisely. but the diode problem could be solve by using another anti-parallel fast diode.