AVR-Chat

At 04:29 PM 2/19/05 -0600, David Kelly wrote:
>On Feb 19, 2005, at 10:06 AM, Robert Adsett wrote:
>
> > That should be driving the power section off of more than 5 or 6 Volts
> > is
> > asking for trouble.  You want the gate drive voltage to be 10-15V above
> > it's reference.
>
>Datasheet reference please for your 10-15V recommendation?


Well take a look at the BUZ11's the OP is using as an instance.  The 
threshold voltage is 4 Volts.  RDSon is 40mOhm with a gate to source 
voltage of 10V.  That rises dramatically at 5V and it would be close to 
impossible to get reasonable currents through the device at that voltage, 
even 10 A takes the typical resistance well over 150m Ohms (it's actually 
off the chart, see figure 7, I don't think you could switch 10A on this 
device with a 5V gate).  Add in a little bit of noise and drops in the 
driver...

You also want to turn the FETs on quickly, that's usually easier to do with 
a larger supply voltage since a little droop from the initial current 
demand can be accommodated without falling below the threshold voltage.

Most drive circuits I've worked with use a 12-15 V gate drive supply for 
all of those reasons

>Rule of thumb which has worked well for me is to drive the gate with at
>least 2 volts more than the source. And remember the "source" is the
>output of the MOSFET while drain is the input.


That's way too low for consistently driving a standard power 
MOSFET.  That's low even for a lot of logic level MOSFETs, many have 
maximum thresholds of 2V (some are 2.5 or 3 IIRC).  Some of the newer logic 
level FETs might get away with that but I wouldn't want to try it on any of 
the logic level power FETs I've used.

>Yes, drive the gate harder and "on" resistance will be lower. But the
>most bang-for-bucks is just over 2 volts.


Don't drive it hard enough and you are either in the linear region or it 
doesn't turn on at all.

>One of my favorite parts is the IRF7307:
>http://www.irf.com/product-info/datasheets/data/irf7307.pdf


Well if you want to use a small logic level part :)  One of the devices I 
use at the moment is an ST 
part  http://www.st.com/stonline/books/pdf/docs/9231.pdf  They control a 
little more power ;)


>Notice in figure 3 how they have about 9 amps flowing thru a 5A part,
>pulsing, with 2 volts GS. Figure 6 shows gate charge and you can see
>how the gate takes quite a charge before it rises above 2 volts.


That's the normal turn-on knee.  The device isn't usually considered fully 
on until you rise above that point.

>I don't recommend the IRF7307 for this application as its SOIC-8 and
>therefore hard to prototype with. Also its only 20V.


Not to mention it won't handle 10A continuously.


>Digikey labels the IRF7307 "logic level" but not the IRF7317:
>http://www.irf.com/product-info/datasheets/data/irf7317.pdf


Regardless of what Digikey calls it, with a 0.7V threshold it's a logic 
level FET.  The biggest issue I'd have with these is getting them off and 
keeping them off.  I've seen problems even with higher threshold 
devices.  I notice the spec appears incomplete, I don't see the max 
threshold.  I may just have missed it I didn't spend much time looking.

>In Figure 5 of the IRF7317 datasheet we see a knee in the plot of "on"
>resistance vs. gate voltage at 3 volts.
>
>One other point is that maximum Vgs is +-12 volts on these parts while
>the maximum Vds is 20. If one were to drive the gate +15 then it would
>be outside of its ratings.


That's not unusual for logic level parts.

>The real reason for speaking up is to suggest the combination of P- and
>N-channel MOSFETs for the purpose of switching a higher voltage with a
>lower voltage.


That would work well for the high side FETs.  It won't work for the low 
side FETs in the H-bridge though.  They will see 0 - 18V on the source as 
the K2232 PWMs.  Hmm, you might actually get away with that since no 
current flows as the 2232 turns off anyway.  Funny things might happen at 
the gate though.  You would end up turning one of the low size BUZ11 on and 
off with the K2232 (that could get interesting), the other BUZ11 could get 
a negative gate voltage.  Ugly.  It would really make more sense to PWM the 
BUZ11's and remove the K2232.

Robert

" 'Freedom' has no meaning of itself.  There are always restrictions,
be they legal, genetic, or physical.  If you don't believe me, try to
chew a radio signal. "

                         Kelvin Throop, III