Author Topic: P-FET gate drivers  (Read 13639 times)

0 Members and 1 Guest are viewing this topic.

Offline jayTopic starter

  • Regular Contributor
  • *
  • Posts: 62
  • Country: ee
  • Coder
P-FET gate drivers
« on: October 07, 2014, 06:27:41 am »
I have a P-channel FET switching up to 30V supply (and I would like to allow the circuit to tolerate a bit more than that). The max Vgs for the FET is -20V so i cannot simply pull the gate to ground without exceeding the limit. I figured out couple of circuits to solve the problem and found few more circuit options by googling it. I think using a 12V zener diode to limit the voltage is the nicest solution so far but there must be ICs for the purpose.. I've found few but those don't match the specs.. For example, LT1693-5 supply range is limited to 14V. Do designers usually go for N channel component with gate driver that has built in charge pump or other way of bring up the voltage for N channel driving? I realize that N channel FETs are cheaper and better but P channel components don't seem too bad so I would expect there are applications for a P-FET driver ICs.. Or did I miss something or misunderstand something again? Even if there are no drivers suitable I think I'll use P-FET anyway because it's interesting to try out a level shifting circuit made of discrete components. Maybe I'll run in some problems driving the high gate capacitance fast and learn something useful :-DD
SW engineer trying to design HW because it's more fun.
 

Boltar

  • Guest
Re: P-FET gate drivers
« Reply #1 on: October 07, 2014, 10:44:45 am »
I've used a P-FET in a high speed switching application and used a gate driver, but that was only 12V not 30V. You can use a n-channel driver (low side driver) if it has an inverting input/output. I used a TC4426, which is a low side inverting driver and it worked fine. All it essentially is is a fast response logic shifting not gate with a seriously low output impedance. But, it has an upper supply limit of 18V so that's no good for you. A compatator can theoretically be used, there are 44V comparators out there, but to drive a fet gate you'll need quite a high current capability from the comparator which you'll struggle to find. I think your best best is to use an n-channel with a boosting gate driver. That's what I've switched to anyway.

EDIT: I'm only a very green beginner, so you'll need one of the experts here to tell you for sure. But this seems to at least simulate correctly.

IC1 would have to be a high voltage, high speed, rail to rail comparator and I'm not sure if there is such a thing. The trannies need to be at least 1.5A capable as well I would think. But I've probably overlooked something or made a mistake somewhere. Like I said, I'm no even close to being an expert.

EDIT2: Indeed I've missed stuff, I think a small resistor is needed between the fet gate and tranny junction and a large pull-up (or pull down depending on your point of view) resistor from the gate to VCC. In fact ignore me, lol. Wait for someone qualified to help properly. I'll just make matters worse.
« Last Edit: October 07, 2014, 11:48:34 am by Boltar »
 

Offline bktemp

  • Super Contributor
  • ***
  • Posts: 1616
  • Country: de
Re: P-FET gate drivers
« Reply #2 on: October 07, 2014, 12:28:56 pm »
The main problem with this circuit is the cross conduction in both transistors:
The NPN transistor will be on for 0.7-30V, the PNP will be on for 0V-29.3V. There is a large overlap where both transistors will be conducting a heavy current even if the time is short. Worst case they will short the power supply even before it has reached the minimum operating voltage for the opamp/comparator to get a stable output.
Better use emitter followers.
You can use a simple circuit like this:
Q1 is a current sink. With 5V at the input it draws about 13mA wich results in a voltage drop of 13V across R1. This voltage is buffered by Q2/3.
« Last Edit: October 07, 2014, 12:36:47 pm by bktemp »
 

Offline jayTopic starter

  • Regular Contributor
  • *
  • Posts: 62
  • Country: ee
  • Coder
Re: P-FET gate drivers
« Reply #3 on: October 07, 2014, 12:57:06 pm »
I wasn't asking about discrete solutions.. As I said in the original post I figured it out and found plenty of PDFs showing ways of solving the problem with discrete components. Two FET vendors have nice documents about it. I'll try to adjust the question.. Do professional engineers actually implement this using discrete components or do they simply use N-FETs with drivers when the driving voltage is significantly below the switched voltage? It would be nice to know what the pros prefer and why. Maybe using a P-FET doesn't make sense when the driving gets complicated?
SW engineer trying to design HW because it's more fun.
 

Boltar

  • Guest
Re: P-FET gate drivers
« Reply #4 on: October 07, 2014, 01:14:39 pm »
I've looked at and considered a lot of commercially available buck converters and they all use a synchronous buck with 2 n-channels one high side and one low side with the high side one's gate being driven from a charge pump. I guess it's probably cheaper, high current power p-fets of similar capability to n-fets tend to be several times more expensive. I have heard that n-fets tend to fail open and p-fets tend to fail shorted, so maybe that's also a consideration, although I don't know how true that is. I've been researching various power converters and in almost every example schematic I've seen, it's the same way, s gate kicked n channel. So from a purely observer viewpoint, I'd say the tendancy is to use n-channels.
 

Offline jayTopic starter

  • Regular Contributor
  • *
  • Posts: 62
  • Country: ee
  • Coder
Re: P-FET gate drivers
« Reply #5 on: October 07, 2014, 01:56:02 pm »
Thanks, Boltar. The N channel drivers seem quite complex beasts so I didn't realize that in the end N-FET + fancy driver might be still cheaper than a P-FET. Googling for failure modes lead me to http://www.richieburnett.co.uk/mosfail.html (so many of them!). So far I didn't find a confirmation if it's actually more common for P-FETs to fail shorted than N-FETs but it kind of makes sense - intuitively.
SW engineer trying to design HW because it's more fun.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
Re: P-FET gate drivers
« Reply #6 on: October 07, 2014, 03:01:12 pm »
N-channel MOSFETs have an advantage in die size and therefore gate charge for a given voltage and Rds(on) which often makes up for the extra complexity of the required high side drive circuit.  This advantage becomes more pronounced at high voltages.  For simple designs though, a P-channel MOSFET with zener diode and appropriate level shift circuit is competitive.
 

Offline Dave

  • Super Contributor
  • ***
  • Posts: 1352
  • Country: si
  • I like to measure things.
Re: P-FET gate drivers
« Reply #7 on: October 07, 2014, 04:08:15 pm »
Guys, don't forget most MOSFETs won't tolerate having 30V+ between their gate and source. Most MOSFETs are only good up to +-20V.

Edit: Whoops, nevermind, I didn't even bother reading the posts, just looked at the two diagrams and typed up a comment. :palm:
« Last Edit: October 07, 2014, 04:28:20 pm by Dave »
<fellbuendel> it's arduino, you're not supposed to know anything about what you're doing
<fellbuendel> if you knew, you wouldn't be using it
 

Boltar

  • Guest
Re: P-FET gate drivers
« Reply #8 on: October 07, 2014, 04:22:32 pm »
Assuming VCC is 30V. If the source is at 30V (as it would be in a high side configuration, well just below depending on the fet's resistance and the load) and the gate is at 40V it's fine isn't it? That's a 10V G->S. Just have to be sure to drive the low side fet at regular logic.

@Dave, hehe. I moderate on a C# forum and I can't tell you how many times I've done exactly that. It happens.
« Last Edit: October 07, 2014, 04:36:54 pm by Boltar »
 

Online Kjelt

  • Super Contributor
  • ***
  • Posts: 6459
  • Country: nl
Re: P-FET gate drivers
« Reply #9 on: October 07, 2014, 08:31:21 pm »
It looks like there are much more N fets available esp. in high currents than P fets.
To check this I went to Farnell selected "MOSFET" it resulted in 7322 hits.
I then set the filter at minimum current 25 Amps and it resulted in 2272 N fets. (31%)
I then reset the filter and set it at maximum current of -25 Amps and it resulted in 107 P fets. (1,5%)
I agree not a very scientific method but it still gives a clear picture.

My guess is that the N process is much cheaper or easier to manufacture so that is the direction were the new developments are made.

 

Offline nuno

  • Frequent Contributor
  • **
  • Posts: 606
  • Country: pt
Re: P-FET gate drivers
« Reply #10 on: October 07, 2014, 10:43:20 pm »
Thanks, Boltar. The N channel drivers seem quite complex beasts so I didn't realize that in the end N-FET + fancy driver might be still cheaper than a P-FET. Googling for failure modes lead me to http://www.richieburnett.co.uk/mosfail.html (so many of them!). So far I didn't find a confirmation if it's actually more common for P-FETs to fail shorted than N-FETs but it kind of makes sense - intuitively.

While developing a "power" h-bridge, I burnt a few P and N fets by shootthrough and too much heat caused by long periods of high power dissipation on their diodes, and they all failed shorted except one P which would never get quite OFF. Some would get D or S shorted to G (and had an event where this short took an output transistor on my discreet gate driver), some would have all 3 legs shorted.
 

Offline DutchGert

  • Frequent Contributor
  • **
  • Posts: 257
  • Country: nl
Re: P-FET gate drivers
« Reply #11 on: October 10, 2014, 07:34:03 am »
It looks like there are much more N fets available esp. in high currents than P fets.
To check this I went to Farnell selected "MOSFET" it resulted in 7322 hits.
I then set the filter at minimum current 25 Amps and it resulted in 2272 N fets. (31%)
I then reset the filter and set it at maximum current of -25 Amps and it resulted in 107 P fets. (1,5%)
I agree not a very scientific method but it still gives a clear picture.

My guess is that the N process is much cheaper or easier to manufacture so that is the direction were the new developments are made.

Kinda good guess :). Due to the physical difference between P and N MOSFETs its harder to make a P version with a low RDSon than a N version. Therefore in general P fets with the same spec as a NFET is more expensive and/or bigger than its N equivalent.
Downside of a NFET is that u have to boost up the gate voltage above Vin to set it 'on'
 

Offline DutchGert

  • Frequent Contributor
  • **
  • Posts: 257
  • Country: nl
Re: P-FET gate drivers
« Reply #12 on: October 10, 2014, 08:31:24 am »
The main problem with this circuit is the cross conduction in both transistors:
The NPN transistor will be on for 0.7-30V, the PNP will be on for 0V-29.3V. There is a large overlap where both transistors will be conducting a heavy current even if the time is short. Worst case they will short the power supply even before it has reached the minimum operating voltage for the opamp/comparator to get a stable output.
Better use emitter followers.
You can use a simple circuit like this:
Q1 is a current sink. With 5V at the input it draws about 13mA wich results in a voltage drop of 13V across R1. This voltage is buffered by Q2/3.

That doesn't work, the Output never gets switched off
« Last Edit: October 10, 2014, 11:40:32 am by DutchGert »
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
Re: P-FET gate drivers
« Reply #13 on: October 10, 2014, 02:02:37 pm »
That doesn't work, the Output never gets switched off

That is odd.  If anything it should have a problem turning on but that will not happen with the supply voltages shown.  This driver circuit should work great.

Oh, you have the P-channel MOSFET configured with its source and drain reversed.
 

Offline DutchGert

  • Frequent Contributor
  • **
  • Posts: 257
  • Country: nl
Re: P-FET gate drivers
« Reply #14 on: October 10, 2014, 02:48:13 pm »
That doesn't work, the Output never gets switched off

That is odd.  If anything it should have a problem turning on but that will not happen with the supply voltages shown.  This driver circuit should work great.

Oh, you have the P-channel MOSFET configured with its source and drain reversed.

That doenst make a difference in this case.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
Re: P-FET gate drivers
« Reply #15 on: October 10, 2014, 03:14:30 pm »
That doesn't work, the Output never gets switched off

That is odd.  If anything it should have a problem turning on but that will not happen with the supply voltages shown.  This driver circuit should work great.

Oh, you have the P-channel MOSFET configured with its source and drain reversed.

That doenst make a difference in this case.

It sure does if you expect the high side P-channel MOSFET to be able to switch the output off.  If the source and drain are reversed, then the body diode between them will conduct when the output is about 0.6 volts below the input no matter what the gate to source voltage is.
 

Offline DutchGert

  • Frequent Contributor
  • **
  • Posts: 257
  • Country: nl
Re: P-FET gate drivers
« Reply #16 on: October 10, 2014, 05:02:27 pm »
Sry, stupid of me, ofc it does. See attachment.
« Last Edit: October 10, 2014, 05:09:53 pm by DutchGert »
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
Re: P-FET gate drivers
« Reply #17 on: October 10, 2014, 06:28:31 pm »
Woohoo!  \o/  That looks right to me if I am reading your simulation results correctly.

I would make R3 a little larger to limit fault current through the zener diode.
 

Online Kjelt

  • Super Contributor
  • ***
  • Posts: 6459
  • Country: nl
Re: P-FET gate drivers
« Reply #18 on: October 10, 2014, 07:49:19 pm »
I have seen such mistakes many times. Why do a lot of schematic symbols of fet's miss that body/freewheeling diode? :-//
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
Re: P-FET gate drivers
« Reply #19 on: October 11, 2014, 12:43:15 am »
I have seen such mistakes many times. Why do a lot of schematic symbols of fet's miss that body/freewheeling diode? :-//

That is a great question but I would not know since I make my own symbols and include the body diode. :)

To be fair, not all MOSFETs have the body diode but if I was using both types, I would include use different symbols to distinguish them.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: P-FET gate drivers
« Reply #20 on: October 11, 2014, 02:47:06 am »
The body diode is the little triangle inside the symbol.  It's always there, whether you notice it or not.  Or, if you were always wondering... :)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline c4757p

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: P-FET gate drivers
« Reply #21 on: October 11, 2014, 04:22:55 am »
The body diode is the little triangle inside the symbol.  It's always there, whether you notice it or not.

:-+

The extra one only belongs in schematics from the Department of Redundancy Department.

Huge pet peeve. :P
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
Re: P-FET gate drivers
« Reply #22 on: October 11, 2014, 05:48:53 am »
The body diode is the little triangle inside the symbol.  It's always there, whether you notice it or not.  Or, if you were always wondering... :)

How does that make sense?

There are MOSFETs with and without the body diode and there are both N and P channel MOSFETs.  The standard symbols do not show the body diode whether it is there or not and the only way to determine the MOSFET polarity is the inside triangle.

Even worse, the alternative MOSFET symbol which looks more like a bipolar transistor reverses the direction of the triangle so it backwards from the direction of the body diode.  I have never used that one but it does have a certain appeal since it looks consistent with bipolar transistors.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: P-FET gate drivers
« Reply #23 on: October 11, 2014, 06:11:58 am »
There are MOSFETs with and without the body diode

Yes, albeit very rarely; and also, there are those with the substrate ("back gate") taken out to a separate pin (or.. there were).  In that case, of course, the diode behavior is related to the substrate pin, and the "diode-ness" of the triangle remains relevant, just not automatically strapped to the source like most.

One example of current technology that doesn't have a substrate connection is EPC's eGaN FETs.  From what I've seen in their appnotes, you can indeed run them in reverse: except that, because Vgs(max) limits how far you go, and Vgs(th) is low, they still kind of act like they have body diodes, in recommended operation.

The correct symbol in this case, could go a few ways: the "channel" could be a solid or broken line (I think the intended difference on that one is enhancement vs. depletion mode), with no triangle to indicate the substrate junction (because it doesn't exist), but, some means is still needed to indicate the polarity of the channel (P or N).  The triangle nomenclature could still be used, but merely with no connection made to it (which is true).

Quote
and there are both N and P channel MOSFETs.  The standard symbols do not show the body diode whether it is there or not and the only way to determine the MOSFET polarity is the inside triangle.

Like commonly used in CMOS IC schematics?  I think they only do that as shorthand.  The substrate connection is implied (to the respective rail) and therefore ignored when possible; some still draw it, e.g., the original CD4066 datasheet is a good illustration of CMOS, with proper symbols, and showing different substrate connections as well (necessary for the bilateral switch action).

Quote
Even worse, the alternative MOSFET symbol which looks more like a bipolar transistor reverses the direction of the triangle so it backwards from the direction of the body diode.  I have never used that one but it does have a certain appeal since it looks consistent with bipolar transistors.

Yeah.  This, I'm pretty sure, is a corruption, and best left forgotten, or preferably, burned in a firey pit. >:D

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Online Kjelt

  • Super Contributor
  • ***
  • Posts: 6459
  • Country: nl
Re: P-FET gate drivers
« Reply #24 on: October 11, 2014, 09:54:59 am »
The body diode is the little triangle inside the symbol.  It's always there, whether you notice it or not.  Or, if you were always wondering... :)
Tim
Yeah I am still wondering what you are saying, are you saying that ALL fets have a freewheeling diode?
Is there a difference between a triangle or arrow?  ;)
Here are just three examples of the Eagle fet library to me the last symbol is the most clear, but I know for a fact that the IRF510 also has a freewheeling diode (datasheet).
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf