Mosfet and its driver 1

[thomaslt]

Hi everyone, I am new on here..

I am designing a pulsing circuit for one of my RF amps.
Pulsing/DC supplies is not really my background.

These are my specs:
- Vd = 33V
- Id = 6A (ie load = 5 ohms approx)
- Need the switch on the drain of the amplifier (the amplifier is connected to GND after that: the load is grounded)
- pulsing frequency = 600kHz
- duty cycle = 35%
- pulse timing signal: 0/5V
- tr/tf = 70ns

I established so far I need a P-channel MOSFET (eg Si4401BDY), source connected to the 32V supply, drain connected to my load.

Would anyone be able to give me advice on driving the P-channel mosfet gate?
- I cant connect directly from a CMOS buffer, as it's higher than the CMOS Vdd
- DO I need a kind of level shifter?
- But if so, how will this supply enough peak current into the P-channel gate?
- Or shall I use more like an integrated IRS2117 to drive an NMOS?

Thanks in advance!
Tom

 

[MacGyverS2000]

this sounds a lot like a number of recent postings for the same thing...

Dan - Owner

http://www.Hi-TecDesigns.com

 

[Henneshoe]

Attached is a circuit that has been used for this type of thing many times. R1 and R2 are chosen so that when the NFET is on, the voltage divider creates the desired VGS on the PFET. Larger values for R1 and R2 will give you a lower power consumption, smaller values will make the PFET turn on faster. The NFET needs to be rated for a cmos compatible gate, and VDS > 33VDC.

 

[benta]

Never understood people who muck around with this kind of stuff, when there's off-the-shelf products available that work much better.
Go to Infineon's website, click on "Automotive ICs", select "Smart High- and Low-Side Switches", choose "PROFET" and find your device.

Benta.

 

[Henneshoe]

1 Profet for rated load (BTS428L2) is priced at $1.347 @ quantities of 2500 (Digikey).

1 generic Pfet at rated load (STMicro STD10PF06T4) is priced at $0.29 @ quantities of 2500 (Digikey).
1 generic Nfet at rated load (Vishay SI2318DS-T1-E3) is priced at $0.17 @ quantities of 3000 (Digikey).
2 0603 resistors (Panasonic ERJ-3GEYJ243V) are priced at $0.003 @ quantities of 5000 (Digikey).

Even with adding $0.02 per component for assembly my circuit costs 40% of what your circuit costs. I think that is why my boss pays me to "muck around" with this stuff.

 

[benta]

Well, then, let's see how your resistor drive handles the 70 ns switching times wanted by the OP. I don't think you'll achieve that with 0603 resistors, to be honest.

Benta.

 

[DCLabs]

Thomaslt,
I am with Benta. Go with a driver. The IRS2117 you mentioned would work with IRZ44 N channel MOSFET. You will need an VCC less that 25V though.

Dean Cooper
dclabseng.com

 

[jimkirk]

However you drive it, just note that in Henneshoe's circuit, the source and drain of the P channel MOSFET have been swapped. If you build it as shown, the bulk diode will be forward biased, applying about 32 volts to your load.

 

[thomaslt]

Hi guys

Thanks for all the comments.
Yes I have seen the NMOS/PMOS option around, just didnt manage to make it work on a breadboard, but I am thinking I had some grounding issues (high current in the load).

Benta, that's a great pointer, the Infineon circuits are very interesting; although I have seen the Tr/Tf on the datasheet is in the order of 100-500us. I need more 100ns, I'll have a look if there's a similar type of IC around.

Cheers
Thomas

 

[DCLabs]

The rise and fall times of the MOSFET have more to do with the gate to source capacitance and gate resistor than the tr/tf of the driver.

Most drivers switch in the nS range.
IRS2117 Tr/Tf 75-135 ns
HIP2101 Tr/Tf 10nS w/ Cl = 1000pf



Dean Cooper
dclabseng.com

 

[Henneshoe]

Benta, good point. 70nS tr/tf at 6A is a tough requirement for the cheap resistor divider driver I described. It would need to have R1 and R2 = around 80 ohms (Cg = 850pF). They would also burn about 3 watts per resistor so a 0603 part would not work.

thomasIt, if the driver recommended by DCLabs doesn't work for some reason, and you have not found another option then you may want to try a PNP BJT instead of a PFET. They are a current mode device, so you need to drive them a bit differently, but with a BJT you do not have gate capacitance so they are inherently faster.

 

[IRstuff]

While a BJT might turn on faster, unless you control the collector saturation, it might turn off considerably slower. You might need to add a Schottky clamp across the collector and base to keep the collector out of saturation.

Note, also, that the base current for the PNP would be a constant current drain, so long as the transistor is on, so if the duty cycle is high, the power dissipation adds up. If you assume a Beta.sat=10, then you're looking at 0.6A base current, which, with the 33V supply, results in nearly 20W of dissipation.

TTFN

FAQ731-376

 

[electricuwe]

Forget about the ProFET. Such a device is intended for a totally different application than the one described here.

You can not expect a device with switching times in the 100 µs range to switch at 600 kHz.

But I'd also like to see how Henneshoe can drive a Mosfet at 600 kHz with two resistors, even if the rating is increased.

In this application you can expect the driver to cost at least the same as Mosfet.

 

[jimkirk]

This

http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1003,C1142,C1041,P2382

will let you drive an N channel MOSFET (amp for amp cheaper than the equivalent P channel device) on the high side at high frequency.