Mosfet driver for POWER MOSFET SPW47N60C3

[Zdenko]

Dear People,

can anyone help me with this problem?

I need to drive a Power mosfet for 600V, 47 A and it have cca 12 nF
input capacitance.

Switching frequency should be a 200khz and with very small rise and
fall time (cca 50 ns). Everything is on high side driver and should be
very fast. I will use appropiate optocoupler if its for low side
driver too.

Can you give me a right sign for driver?

Regards

Zdenko

Mobile for quick sms:+385 91 5012417

 

[Winfield Hill]

Zdenko wrote...

I need to drive a Power mosfet for 600V, 47 A and it have cca 12 nF
input capacitance. Switching frequency should be a 200khz and with
very small rise and fall time (cca 50 ns).

To evaluate switching speed it's Qgd that matters, more than Ciss.
Qgd is often called the Miller gate-to-drain charge, because it's
the gate charge required as the MOSFET's drain voltage swings from
fully on to off (or back). This corresponds to figure 11 in the
data sheet, and is specified as 121nC typical, which is not bad for
a 415W high-voltage FET. APT's version of the same part highlights
"Low Miller Capacitance" on the front page of the data sheet.

The charge formula tells us you'll need i = Q/t = 121/50 = 2.4A
of gate drive to switch this part in 50ns. I keep an assortment
of FET driver chips to handle different size FETs. For example,
the TC4427 is a dual driver rated at 1.5A, the TC4424 is a 3A
dual and the TC4420 is a single 6A driver. But I'd recommend
the 6A part for your big FET. I'd use a small say 1.5-ohm gate
resistor and I'd keep the FET-driver to gate leads very short.
Keep the return path from the source to the driver GND pin short
and immediately parallel to the gate lead to minimize inductance.
The same holds for the 0.1uF ceramic cap bypassing the driver IC.

I'd start my testing at low voltages and without much load. I'd
evaluate the gate and source voltages during switching to check
the quality of my low-inductance FET-druve wiring. For example,
I'd run the high-side FET with its opto-coupler, but without the
400V on it's drain, and without the low-side switch. In fact I'd
ground the FET's source (to keep gate and source voltages within
range for my scope) and I'd use a temporary resistive load to HV.
I'd pulse the FET on for short pulses (to avoid overheating the
resistor) and carefully scope the lead-inductance spikes, which
will grow higher as the load current is increased. Finally I'd
use an inductive load to evaluate high-current turn-off spikes.

Everything is on high side driver and should be very fast. I
will use appropiate optocoupler if its for low side driver too.

Be careful to maintain enough time delay between turning one FET
off and the other one on! We don't want any high shoot-through
currents at 600V.

Thanks,
- Win

 

[Spehro Pefhany]

On 15 Jul 2003 04:05:12 -0700, the renowned Winfield Hill

The same holds for the 0.1uF ceramic cap bypassing the driver IC.

Win, I wonder if it's time to start using 1uF caps for this sort of
thing? I happened into a bunch of reels of SMT parts, one of them was
4,000 0805 caps of 1uF/16V- and I notice they are not expensive at all
by the reel, less than 3 cents each. 50nsec * 6A/0.1uF = 3V.

Best regards,
Spehro Pefhany

 

[Winfield Hill]

Spehro wrote...

Winfield wrote:



Win, I wonder if it's time to start using 1uF caps for this sort of
thing? I happened into a bunch of reels of SMT parts, one of them
was 4,000 0805 caps of 1uF/16V- and I notice they are not expensive
at all by the reel, less than 3 cents each. 50nsec * 6A/0.1uF = 3V.

Hey, what we need is part numbers and ordering addresses!

With the 1.5-ohm resistor, I had in mind gate currents closer to
3A... But the relevant issue is supplying the gate charge. For
a 0.1uF cap this works out to a q/C = 121nC/100nF = 1.2V change
during the 50ns drain swing. The spw47n60's total gate charge
is spec'd at 320nC max at dVgs = 10V, which would drop the puny
cap 3.2V for the entire gate event. Another relevant issue is
the capacitor's esr, which could cause the driver to suffer badly
during a 3A pulse. Yep, it appears that 0.1uF is too small.

Thanks,
- Win

 

[Spehro Pefhany]

Spehro wrote...

Hey, what we need is part numbers and ordering addresses!

Panasonic ECJ-2VF1C105Z Y5V are 2.5 cents each 1uF/16V @4K quantity.
The price quadruples if you need 25V in 0805, but in 1206 less than
double.

With the 1.5-ohm resistor, I had in mind gate currents closer to
3A... But the relevant issue is supplying the gate charge. For
a 0.1uF cap this works out to a q/C = 121nC/100nF = 1.2V change
during the 50ns drain swing. The spw47n60's total gate charge
is spec'd at 320nC max at dVgs = 10V, which would drop the puny
cap 3.2V for the entire gate event. Another relevant issue is
the capacitor's esr, which could cause the driver to suffer badly
during a 3A pulse.

I don't see an ESR spec for the series:

http://rocky.digikey.com/WebLib/Panasonic/Web data/ECJ Series.pdf


Best regards,
Spehro Pefhany

 

[Paul Burridge]

Dear People,

can anyone help me with this problem?

I need to drive a Power mosfet for 600V, 47 A and it have cca 12 nF
input capacitance.

Switching frequency should be a 200khz and with very small rise and
fall time (cca 50 ns). Everything is on high side driver and should be
very fast. I will use appropiate optocoupler if its for low side
driver too.

What's the intended application, actually?

 

[Spehro Pefhany]

Carrying on with this theme...

16V and 10uF (1206 and 1210 sizes):

NMC1206Y5V106Z16TRPLPF (1206/ 16V/ Y5V/ 10uF/ +80,-20%) Nippon $0.115 mpq=2K/reel

1210YG106ZAT2A (1210/16V/Y5V/10UF/+80-20%) AVX $0.116 mpq=2.5k/reel


for 22uF, 1210 and 10V:

1210ZG226ZAT2A (1210/10V/Y5V/22uF/+80,-20%) AVX $0.155 mpq=2k/reel

C3225Y5V1A226Z (plastic tape) TDK $0.165 mpq=1k/reel

They look like they can be used instead of electrolytics for many
applications (I'm thinking SMPS rather than bypassing here).

Not bad! 10uF/22uF is getting up there. For very low consumption
circuits, a handful in parallel could even be considered for mains
frequency filtering in special circumstances (low height, long life at
high temperature, cost not very important).

OTOH, they might be low enough ESR to cause stability problems with
some linear regulators if used on the output.

Best regards,
Spehro Pefhany