mosfets: linear range and fully open / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

mosfets: linear range and fully open

1 2 next

Move Thread

LAN_403

IamSmooth

Wed Aug 19 2009, 04:28PM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

Can someone explain the different stress the FET undergoes if one is conducting through it in the linear range compared to turning the FET fully on. Is it more likely to blow the FET in the first case?

Marko

Wed Aug 19 2009, 05:05PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

I'm not the most knowledgeable one in the field - but I know that mosfets in parallel can suffer from thermal runaway from current mis-sharing when used in linear mode.

Temperature coefficient of resistance is positive in MOSFET's; but their treshold voltage has negative temperature coefficient.

When mosfets are driven in switch mode and fully on, they tend to share current well; but in linear mode current will tend to crowd into one device due to positive feedback of treshold voltage NTC.

But since actual mosfets are made of large number of individual cells that are effectively in parallel, this might apply for single devices too (linear mode operation could cause hotspotting) and I think that's why you get devices rated for switch mode only. So manufacturer guaranteeing a device to dissipate 200W in switch-mode might not necessarily mean 200W for linear mode, though.

Marko

MinorityCarrier

Wed Aug 19 2009, 05:56PM

Registered Member #2123 Joined: Sat May 16 2009, 03:10AM
Location: Bend, Oregon
Posts: 312

In 'linear' mode the device is dissipating more power, as heat, than it is as a switch driven into saturation. As already stated, if the device die gets too hot it will go into thermal runaway and be destroyed. Device Rds is not biased to it's lowest value, hence Ohm's law P=I X R (R = Rds)determines how much power the device is dissipating. You will need to use heat sinks much larger than those you are used to using for switch-mode applications.

Look up Safe Operating Area (SOA) on the device data sheet of a particular MOSFET to see where the device can operate in linear mode. If there isn't a SOA curve for a particular MOSFET, you may wish to not use it in a linear application.

Some power MOSFET's, like the APL501JN, are designed for linear operation, they employ an assymetrical gate layout to accomplish this.

HV Enthusiast

Wed Aug 19 2009, 06:39PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

IamSmooth wrote ...

Is it more likely to blow the FET in the first case?

Not at all. It depends on the application, the operating specifics, and thermal interface to what would be more stressing.

IamSmooth

Thu Aug 20 2009, 03:05AM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

can the output of a 555 timer run the gate of a MOSFET or IGBT?

Arcstarter

Thu Aug 20 2009, 03:23AM

Registered Member #1225 Joined: Sat Jan 12 2008, 01:24AM
Location: Beaumont, Texas, USA
Posts: 2253

IamSmooth wrote ...

can the output of a 555 timer run the gate of a MOSFET or IGBT?

Well, do you mean in astable? Monostable would work too, but i am not sure what you are actually asking. Do you just mean will it output enough current?

A 555 can drive a mosfet gate at lower frequency, and the mosfet has to have a small gate capacitance. Also, it may not turn on fully or correct or whatever.

Electroholic

Thu Aug 20 2009, 03:46AM

Registered Member #191 Joined: Fri Feb 17 2006, 02:01AM
Location: Esbjerg Denmark
Posts: 720

No, the low output is not very low, and it might not be enough to shut off the fet completely.
Use a totem pole between them.

Frosty90

Thu Aug 20 2009, 03:52AM

Registered Member #1617 Joined: Fri Aug 01 2008, 07:31AM
Location: Adelaide, South Australia
Posts: 139

It seems theres a lot of confusion over 'linerar', 'fully on/off', etc. MOSFETS like BJTs have 3 usual 'modes' of operation, Active (sometimes called saturation), Cutt-off, and Linear (sometimes called triode mode). The corresponding modes for BJTs are, confusingly; Active, Cut-off, and Saturation (analougous to Linear mode in MOSFETs).

for a MOSFET:

Cutoff is when Vgs<Vt, --> no (very little) drain current, almost completely independant of Vds.

Active (for mosfets, this is sometimes called saturation mode) is when Vgs>Vt, but less than Vds, --> drain current depends on the square of the Vgs-Vt, and is almost independant of Vds. Confusingly, sometimes in some aplications, (and very often in these forums) this mode is called 'the linear region' because this is the mode you use for linear regulators etc i.e. the region where it is dissipating lots of power (large Vds, and Id).

(the actual) Linear or Triode mode is when Vgs>Vt, but Vgs>Vds --> the mosfet acts like a closed switch, or a very low resistance, but the resistance is 'controllable' (but still small), by Vgs. In this mode, Id depends linearly on Vds, this is why its called the linear or triode (because the characteristics look similar to a triod tube). Id is directly proportional to Vgs.

For a BJT:

Cutoff is when Vbe < ~0.5v i.e. the BE junction is reversed biased. --> No (very little) collecter current, (since Ic and Ib are directly proportional) Ic almost complety independant of Vce.

Active (NOT called saturation mode for BJTs, but is analougous to the saturation mode in MOSFETs) is when the BE junction is foward biased, the BC junction is reverse biased. Ic depends exponentially on Vbe, and is almost completely independant of Vce. Confusingly, sometimes in some aplications, this mode is called 'the linear region' because this is the mode you use for linear regulators etc i.e. the region where it is dissipating lots of power (large Vce, and Ic).

Saturation Mode (NOT the same as mosfet saturation mode, but is analougous to the MOSFET linear/triode mode) is when both the BE and BC junctions are foward biased. In this case, Vce depends almost linearly on Ic, like a very low value resistor (a closed switch).

So for switching aplications, using MOSFETS, when the mosfet is Off, Vgs<Vt, in cutoff. When Vgs>(Vds-Vt), The device is in the Linear/Triode mode, and looks like a very low impedance.
For BJTs in switching aplications, the 'off' state is when the device is cutoff, the collecter looks like a high impeance (open switch). The 'On' state is when both junctions a foward biased, called the Saturation mode; the collecter looks like a very low impedance, like a closed switch.

For 'Linear' aplications (eg regulators, amplifiers etc.), MOSFETS operate in the 'saturation' mode, and BJTs in the 'active' mode. These modes are pretty similar for both devices, they are just named different. The drain of a MOSFET in saturation mode looks like a current source (high impedance), with the current dependant on Vgs, not Vds (hence high impedance). The collecter of a BJT in 'active' mode looks like a current source (high impedance), with the current dependant on Vbe, not Vce (hence high impedance).
The reason so many people call the above 'linear' mode, is because of the aplication, not the actual mode the device operates in!

The difference in naming between the two devices is weird, but it has a good reason:
The saturation mode of a MOSFET (analougous to the Active mode of a BJT, i.e. the mode you use for linear aplications), is called that, because increasing Vds doesnt change Id much, Id has saturated at a value depending on Vgs.
The saturation mode of a BJT (analougous to the linear/triode mode of a MOSFET, i.e. the mode you use for a closed switch), is called that because increasing Ib doesnt change Ic very much, Ic has 'saturated' and depends mainly on Vce.

Cheers,
Jesse

IamSmooth

Thu Aug 20 2009, 04:56AM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

Electroholic,

I was going to use the output of a comparator to trigger the igbt gate. Would the LM311 need a totem configuration? If so, would a buffer work on the output stage such as this one?

Dr. Slack

Thu Aug 20 2009, 07:12AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

For 'Linear' aplications (eg regulators, amplifiers etc.), MOSFETS operate in the 'saturation' mode, and BJTs in the 'active

This doesn't really help, even though the terminaology, which I've not bothered to plough through, may be arguably correct for the particular devices.

It may be better to invent new, very descriptive terms, which can apply accross all types of power devices. This characterises their operation by whether they have high or low dissipation, and significant AC gain or not. The three regions for all power devices like BJTs, FETs, IGBTs, Valves (but not thyristor type devices) are ...

a) High voltage and no current flowing through device, no dissipation, AC gain is zero
b) lower voltage and significant current flowing in device, high dissipation and high AC gain
c) very low voltage and the current flowing is determined by the supply and load, low dissipation, AC gain is very small

a) - we can use "cut off", most disciplines seem to understand that one, though "off" would be an excellent alternative
b) - I've always used "linear", but we could say "dissipating" or something else like "gainfull", "middle", "active"
c) - BJTs use saturated for this region, and many people including myself use this term for FETs. How about "on", or "fully on"?

How long a FET can stay in the mid voltage, mid current region depends on very subtle stuff at the device level where the current distribution across the die is unstable above a certain power dissipation threshhold. If you stay in this region for only a short time (uS), then there is not time for a thermal runaway from one part of the die to another, and you can really hammer it. If you hover here for mS or seconds, then the current distribution across the die will become non-uniform, highly stressed parts will fail, and the damage will spread across the die like a zip fastener being undone.

Linear mode operation assumes 100% dwell in the linear regime, so the power dissipation must be kept waaaaay below the data sheet headline figures for stable operation.

Switched mode operation assumes a *sufficiently* short dwell in the linear region for the dissipation experieicend there during switching. The definition of *sufficiently* is complicated by all sorts of things, and to optimise it you need to consider the gate charge, miller charge, switching losses, and the V and I of your load, as well as the current sourcing capability of the gate drive.

Can I use a 555 to drive my FET, or do I need a totem pole? The answer is Yes, depending on the details. For low switching dissipation, low rail voltage, only a low current drive is needed to charge the gate and the low Miller capacitance in a reasonable time. For high switching dissipation, or a high rail votlage which increases Miller, you need more current. When in doubt, slam it.

--- and we haven't even begun to consider multiple devices sharing the load!

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob