Problem with IRS2092 Deadtime

[fanmade]

Hi all,
I'm working on a 300W Class-D amp with IRS2092 and IRFB4227.
Should deliver up to 300W stereo. The FETs where driven by a BJT totem pole and the amp sounds fairly good.
Theres just one problem: The MOSFETs getting hotter than they should in idle mode.
The Pics show more:

(1) CH1 Low Side Gate Signal CH2 Drain Low Side Rising Edge
(2) CH1 Low Side Gate Signal CH2 Drain Low Side Falling Edge
(3) CH1 High Side Gate Signal CH2 Low Side Gate Signal Rising Edge
(4) CH1 High Side Gate CH2 Low Side Gate Falling Edge
(5) PWM Output Signal Idle (250kHz)
All referenced to -B

In (4) the High Side Gate have enought dead time but in (3) the it isn't enought and causes shoot through.
It seem's the DT isn't symmetrical on High and Low side of the IRS, I dont know what to do because already using max setting(105ns)

Supply is +-40V

thank you for help

[JensRasmussen]

Originally Posted by fanmade

Hi all,
I'm working on a 300W Class-D amp with IRS2092 and IRFB4227.
Should deliver up to 300W stereo. The FETs where driven by a BJT totem pole and the amp sounds fairly good.
Theres just one problem: The MOSFETs getting hotter than they should in idle mode.
The Pics show more:

(1) CH1 Low Side Gate Signal CH2 Drain Low Side Rising Edge
(2) CH1 Low Side Gate Signal CH2 Drain Low Side Falling Edge
(3) CH1 High Side Gate Signal CH2 Low Side Gate Signal Rising Edge
(4) CH1 High Side Gate CH2 Low Side Gate Falling Edge
(5) PWM Output Signal Idle (250kHz)
All referenced to -B

In (4) the High Side Gate have enought dead time but in (3) the it isn't enought and causes shoot through.
It seem's the DT isn't symmetrical on High and Low side of the IRS, I dont know what to do because already using max setting(105ns)

Supply is +-40V

thank you for help

Hi,

I would make a different gatedrive circuit, you have some nice drivertransistors and then you add 4R7 in series with both turn on and off.

Why not try and make you gatedrive a little different, start maybe with a 1N4148 in parallel with 4R7 to make turn off faster than turn on?

A base resistor in you gatedrive transistors might also be worth it ;-)

Best regards
\\\Jens

[nigelwright7557]

The deadtime on the 2092 works fine so it has got to be your driver that is the problem.
I use a TC4420 driver chip and it works with upto 2 pairs of IRB4227.
I also use 10R gate resistors.

[Eva]

The gate drive buffers need base resistors to control reverse behavior of the NPN/PNP when di/dt spikes show up at the gate (through Ld) and drive it temporarily outside the gate drive rails. Also, some resistive drive impedance is needed at high frequencies to damp potential layout resonances (MOS based gate driver outputs are resistive but always have capacitance and body diodes to the rails).

Speed up turn off with diodes and optional resistors as recommended, turn off di/dt must always be higher than turn-on di/dt to avoid "regenerative" cross-conduction during any overlap condition in switching. Don't make it faster than needed, gate drive "commands" EMI production. EDIT: IRFB4227 has 2.2 ohm of built in gate resistance, with a good layout it may not need more, just the diode.

(1) and (2) show cross-conduction further helped by the 20ns/50Mhz resonance. This resonance should be damped with RC(L) for optimum performance and minimum interference between channels. It's d-s capacitance of the MOSFET that is turned off (240pf typ. for IRFB4227) resonating with all the parasitic inductances that close the loop (TO-220 leads typ 12nH d-s, PCB tracks, supply capacitors typ 1nH per mm of lead spacing).

[luka]

I would also use different gate drivers, something like ZXGD3002E6, and resistor to this driver from IRS, 18v zeners are not really needed

[fanmade]

@Eva
OK i will add Base Resistors and Speed Up turn-off diodes, lets see if it improves switching timing.

Where i have to add the damping RC network because of the cross-conduction resonances?
Are the 1n 10R at the PWM Output bad dimensioned?

@luka
The ZXGD3002E6 look very nice, I think i will test it in a new desing later.

[luka]

Originally Posted by fanmade

@luka
The ZXGD3002E6 look very nice, I think i will test it in a new desing later.

Oh and VERY cheap

[nigelwright7557]

[QUOTE=fanmade;2821499@luka
The ZXGD3002E6 look very nice, I think i will test it in a new desing later.[/QUOTE]

The TC4420 costs a little more but is available in through hole.

[luka]

Originally Posted by nigelwright7557

The TC4420 costs a little more but is available in through hole.

how much would be a little more?
and smd is good, very small footprint, it takes no more then smd resistor, also faster then tc

[nigelwright7557]

Originally Posted by luka

how much would be a little more?
and smd is good, very small footprint, it takes no more then smd resistor, also faster then tc

86p its a larger package so should dissipate a bit more power.
I use them for driving 2 pairs of irb4227.