P-Channel Mosfet VGS

[Nikos A.]

Hi everyone learning about P-Channel Mosfets.

Looking at the datasheet below, I can see that in order to turn the Mosfet on you have to apply -1V to -3V. I a little bit confused.. It says that the max Vgs should be -3 in order to turn ON (allowing 250mA drain current) but on figure 2 there are a variety of Vgs greater than -3V.. From figure 2 I understand that I could also supply Vgs =  -10V without any problem.. Can anyone explain to me these details? What is the range of Vgs in order to turn the Mosfet ON and what is the ideal Vgs?

https://eu.mouser.com/datasheet/2/308/FDD6685-1122838.pdf

[Psi]

The mosfet is considered to be 'on' when the gate is at 3V
The mosfet will be damaged if the gate voltage exceeds 25V

[Nikos A.]

Hi, thank you for your answer.

You mean -3V isn't it?

[Siwastaja]

Hi,

I very classical trap for the young players:

VGS(th) is a completely meaningless number for switching applications, and is, in any case, varying all over the place. This number tells you where the boundary of not conducting "at all", and starting to conduct "a tiny bit" lies. For switching applications, you want to switch "fully". This information is available in the curve set: for example, the Vds vs Id curve, or the Rds vs Id curve set you posted is fine as well: look at your maximum worst case Id, and see if the resulting Rds(on) is good enough for your purpose.

Note that Rds(on) also rises significantly with the die temperature. I typically multiple the datasheet room temp values by 1.5x for the first order estimate.

Always calculate for power dissipation: P = I^2 * R, then look at the thermal resistance numbers; for example, 40 degC/W is reported on a typical PCB layout shown on page 4. This means that dissipating 1W increases the die temperature by 40 degrees C over the ambient. You need to understand that if the PCB sits in a closed box, for example, the "ambient" is the local ambient inside that box.

[Psi]

yeah, I dont like using negative voltage convention for p channel gates, but yes, -3V.

It's a bit confusing because they are trying to say the mosfet has a 3.3V logic level gate. 
But the more voltage you put into the gate the lower the on resistance because that's just how mosfets work.
They have to draw the line somewhere and say Y Vgs and X Rdson is considered 'on' so they can claim it has a logic level gate.
 

[Siwastaja]

The mosfet will be fully on when the gate is at 3V

NO! The -3V curve (Fig. 1) shows a definitely linear operation after about 3-4A of current. Clearly, at least -4.0V is required if the FET is used even near half of the current rating.

[rs20]

Hi,

The threshold in any given MOSFET will be somewhere between -1V and -3V. So if you buy a bag of 3 MOSFETs, one might have a threshold of -1.5V, one might be -2V, and the last might have a threshold of -2.5V. Just as an example.

Now, for any given P-channel MOSFET, (loosely speaking) the MOSFET is ON if Vgs is less than the threshold for that particular MOSFET. So if a particular MOSFET has a threshold of -1.5V, then it'll be on for any Vgs < -1.5V, and off for any Vgs > -1.5V.

In reality, the turn-on/turn-off is not a hard switch like a light switch as I've implied above, the threshold is actually defined as the Vgs where Id has a mere 250 microamps leaking through, so it's just the threshold where it starts to turn on.

So that's the explanation of what the datasheet means. In terms of designing a circuit to use these MOSFETs, you want your Vgs to significantly > -1V in order for the MOSFET to be off, and very significantly < -3V if you want it to turn on (because remember, some of the MOSFETs might have a threshold near -3V, which means that even with Vgs at -3V, those will only allow 250 microamps through, which hardly qualifies as being "on" in almost any application.) It looks like you'd really want Vgs to be -4V or -5V or less for it to be really properly on, based on the graph.

EDIT: Also, whatever PDF viewer you are using is screwing up the PDF. The test conditions that you outlined are supposed to be "VDS = VGS, Id = -250 uA", but the micro simple has mangled the end into a not-equals sign with an A after it. You might want to find a way to fix that, reading datasheets is hard enough without your computer mangling the symbology.

[Siwastaja]

It's a bit confusing because they are trying to say the mosfet has a 3.3V logic level gate. 

Are we reading the same datasheet? The first page clearly states the gate drive range from 4.5V to 25V.

[Psi]

The mosfet will be fully on when the gate is at 3V

NO! The -3V curve (Fig. 1) shows a definitely linear operation after about 3-4A of current. Clearly, at least -4.0V is required if the FET is used even near half of the current rating.

Not what i meant!
And you can see that i had already edited my post to remove the word 'fully' 2 minutes before you posted your reply. As i saw that it might be a bit misleading.

By 'fully' turned on i didn't mean lowest on resistance. I mean it has fully changed state from off to on.

[Psi]

It's a bit confusing because they are trying to say the mosfet has a 3.3V logic level gate. 

Are we reading the same datasheet? The first page clearly states the gate drive range from 4.5V to 25V.

The whole point of this line is to say.
 "This mosfet is ok for 3.3V logic level gate drive"
Whether or not it is a good 3.3V mosfet is another question.
All logic level mosfets can be driven at higher gate voltages for lower Rds, so it's normal for a DS to say the gate drive range is up to 25V.

[Nikos A.]

Thank you for your answers guys!! I got the point!!

[magic]

The whole point of this line is to say.
 "This mosfet is ok for 3.3V logic level gate drive"
Whether or not it is a good 3.3V mosfet is another question.

No, it doesn't say it's suitable for 3.3V drive. VGS(th) is only relevant to linear applications.

The spec you are looking for is RDS(on), and specifically, the VGS in the "test conditions" column for which RDS(on) is rated. The datasheet clearly says it's a part for 5V drive.

[Siwastaja]

All logic level mosfets can be driven at higher gate voltages for lower Rds, so it's normal for a DS to say the gate drive range is up to 25V.

Indeed, but it's worth noting that this MOSFET starts to get into the "optimize Rds(on)" region after about Vg=4V.  At Vg=3.3V, this MOSFET still cannot be modeled as a simple constant Rds(on) like we normally do for switching applications, not even approximately; the Rds(on) is a curve depending on Id.

We call the MOSFET is "fully on" after we have such high enough Vg, so that the Vds vs. Id is a straight line over the meaningful Id range, so that Rds(on) can be approximated as a "constant" independent of Id. Increasing Vg even further indeed makes sense, to reduce Rds(on) (i.e., change the slope of that straight line). That's why the front page lists Rds(on) at the recommended minimum Vgs=4.5V, and also at Vgs=10V, which gives the optimally low Rds(on).

This means, if you insist on using it as a 3.3V logic level MOSFET, yes, you can definitely do that, but you can't do the calculations based on Rds(on) like we normally do for switching applications; you need to refer to the curves with actual load current, including any peaks there might be. There is less room for error in the design. This is made especially difficult because the curve sets may be "typical", not necessarily worst case. So you need to leave a lot of margin. Ballparking from the curves, this could go up to around maybe Id=5A at Vgs=3.3V, which is only a fraction of the current this MOSFET is designed to be able to switch. Note that Fig.2 starts at Vgs=3.5V. It's very crappy at 3.3V and requires careful understanding.

No, I disagree with you. For a switching MOSFET which clearly isn't designed to be used at Vgs=3.3V, but at Vgs=4.5V and up, with only marginal specifications how it performs at Vgs=3.3V, it's irresponsible to say: "It's OK for 3.3V logic level gate drive", then later go on that it's just not good. Don't do that.

[Psi]

That's a fair point.