2
\$\begingroup\$

I've been studying MOSFET switching applications recently, and have been playing with the Circuit Simulator Applet here. In my diagram below, I have Vds usually set to either 12 or 14, as I am trying to design a decent buck converter for automotive applications, in which a constant DC output that has minimal oscillations is necessary.

I've noticed that using a Vgs of 12V, while also have Vds being 12V, results in a lot of heat being dissipated over my MOSFET. I've attempted to minimize switching losses by playing with the timing and inductance values, but have hit a wall.

However, I noticed that upon increasing Vgs higher than Vds, I minimize my switching losses almost entirely, if Vgs is set high enough. In this case, I have it set to 18V, which I cannot really hope to achieve in my application without some kind of voltage doubler/charge pump, of which I have not yet played with.

I'm wanting to make sure the behavior I am seeing here is consistent with reality and not just something happening in the simulator. Is it better to have Vgs higher than Vds? If so, am I wrong in my understanding that MOSFETs are designed to be low-voltage triggered switches, designed to control high voltage, high current loads?

My circuit can be found here that I have set up. Along with some of the sliders I was playing with to try to make it efficient. To anyone that knows whether this is true or not, please explain as much as you can! Very interested in this subject. Higher Vgs than Vds results in less power loss?

\$\endgroup\$
12
  • \$\begingroup\$ Your "problem" is that you are using a High Side N CHANNEL FET. An N Channel FET requires a Vgs that is positive relative to source. In this case, to drive the source to V+ (as you want to do) you must dtive Vg above V+ to get positive Vgs. | If yuou use a high side P Channel FEt then to turn on it requires Vgs negative to source with source connected to V+. This allows a drive voltage less than V+ \$\endgroup\$
    – Russell McMahon
    Apr 27 '20 at 12:46
  • \$\begingroup\$ @RussellMcMahon appreciate your response Russel. I am going to look into either a charge pump/high-side driver IC in order to drive this mosfet. I wanted to avoid P-channel, mostly for the reason that I've read they have a higher rds(on). I will however definitely explore the idea youve set out. \$\endgroup\$ Apr 27 '20 at 12:50
  • \$\begingroup\$ You can get some very good modern P Channel FEts. Unless your application is extremely extreme you can get excellent PFETS at a very modest cost premium. You don't say what current you require (or switching frequency)(you should) but odds are at 12V your need will be easily met. I use www.digikey.com as a selector guide in most cases. Good availability, OK prices, vast range. Choose any 3. \$\endgroup\$
    – Russell McMahon
    Apr 27 '20 at 13:16
  • \$\begingroup\$ @RussellMcMahon I will definitely check those out. I was under the impression that pfets suffered higher rds on, but the material I'm reading may be dated, or may be unreliable material. My maximum load would be around 5A, assuming nearly 100% duty cycle of the fet. Switching frequency can be arbitrarily set as I just wanted to create an efficient design, and am not too concerned with frequency so long as I'm not violating any kind of radio frequency laws. Ideally, I'd probably go to the lowest frequency possible while still remaining efficient. \$\endgroup\$ Apr 27 '20 at 13:18
  • 1
    \$\begingroup\$ Also, do you plan for this to be regulated in any way? \$\endgroup\$
    – DKNguyen
    Apr 27 '20 at 18:17
1
\$\begingroup\$

The configuration you have is a high side switch. If MOSFET is fully on then VS=VD (assuming very low RDSon and thus low voltage drop). But for this to happen, VG needs to be sufficiently high enough, for example 10 to 12V higher than VS. This is why VG =18 V works better than VG = 12 or even VG=14V. Since for these lower voltages the VGS is only 0V and 2V. Not enough to properly turn on, and thus causing power dissipation. In this configuration a charge pump is required.

AS a side note, if you use N Channel MOSFET in low side configuration then VDS can be more than VGS with no problem since it’s VGS that determines how much MOSFET is on. For example VDS of 20V with a VGS of 10V.

\$\endgroup\$
3
  • \$\begingroup\$ Thank you Leo, this was more along the lines of what I was looking for. I am attempting to design my own buck converter and what you describe is what I was experiencing. I was seeing a Vds even at 100% duty cycle around 3-4v, way too high for a fully-on mosfet. \$\endgroup\$ Apr 27 '20 at 3:14
  • \$\begingroup\$ Yeah this problem of needing a VGS higher than supply is a reason why the high side gate driver was designed. Check out part LT1910 or similar for an idea of how it’s used. \$\endgroup\$
    – Leoman12
    Apr 27 '20 at 3:30
  • \$\begingroup\$ Better yet check this out analog.com/media/en/technical-documentation/data-sheets/… \$\endgroup\$
    – Leoman12
    Apr 27 '20 at 3:37
1
\$\begingroup\$

Yes, that is the whole point if you want to use a MOSFET as a switch. Read up on operating regions of a MOSFET.

It's like a valve. If you want to use it as as switch and not to modulate or throttle the flow (like an amplifier) you open the valve as much as possible (drive the MOSFET as hard as you can) so what can flow through the pipe will flow through the pipe. Of course, there is a limit on the diameter of the pipe so if enough water is flowing to fill the diameter of the pipe, opening the valve more won't increase the flow. There is a similar limit on MOSFETs (it might just overheat first before it ever gets to the point where it cannot conduct any additional current through it though).

Just to be clear, you do not "set" Vds. Vds is the voltage across the source-drain terminals of the MOSFET. It is not your supply voltage.

You then apply a voltage across the gate-source terminals of the MOSFET. This is important. You are not applying a voltage to the gate terminal relative to ground. The MOSFET does not care about what ground is. In this case, it can't even if it wanted to since it has no terminals connected to ground. It cannot see ground. The only thing it cares about is the voltage across its gate and source terminals.

Vds then responds accordingly and as the MOSFET turns on more and more, Vds decreases. At some point, Vds decreases below the Vgs you are applying. If you continue to drive the Vgs higher and higher the MOSFET continues to turn on more and more and Vds continues to decrease to the point where the MOSFET cannot conduct any more current through it.

Most MOSFETs require a Vgs of at least 10V, preferalby 15V to fully turn on. Logic level MOSFETs only require 5V, 3.3V, or even 1.8V. Easiest way is to check the Vgs used to obtain the Rdson in the datasheet. Ignore Vgs_threshold. That is not useful for use as a switch. Of course, you can also use the IV curves.


Now, remember when I said the MOSFET only cares about the voltage between gate and source terminal? Not GND? In your circuit, you are applying a voltage to the gate relative to ground. That's where problems lay.

As you apply a voltage to the gate relative to ground, the MOSFET turns on, but as it turns on the voltage across the inductor and load rise, which pushes the source voltage up. The result is that the gate-source voltage is decreased since the voltage you are applying to the gate is ground referenced so does not rise even as the source voltage rises.

The way to get around this is to use a driver which applies a gate drive voltage relative to the source terminal (not ground). A so-called high-side gate driver.

The most common high-side gate driver is a bootstrap diode and capacitor. These are not capable of 100% duty cycle since the capacitor, which is charged relative to GND and then is floated up to be between the gate and source terminals, needs to be periodically refreshed. This refresh typically happens by having the a transistor on the low-side (which is not present in your circuit) turn on periodically in a half-bridge configuration. This low-side transistor replaces the diode in your circuit.


Alternatively, you could use a PMOS. On the low-side, the source-terminal of an NMOS is conencted to the ground rail which is fixed. That makes it easy to work with. On the high-side using a PMOS has the source-terminal fixed to the positive rail. This makes it simpler than trying to chase a source-terminal whose voltage is floating.

schematic

simulate this circuit – Schematic created using CircuitLab

  • If your gate drive signal source is logic level, then your NMOS will have to be a compatible logic level to turn on.
  • The NMOS max Vds must be able to survive 12V, obviously.
  • max |Vgs| is alwayws less than |Vds|. For a PMOS in that position with its gate being pulled all the way to ground, the limiting factor is max Vgs not max Vds. The Vgs must be able to the survive the gate-source voltage difference when the gate is pulled all the way to ground. If it cannot, the circuit has to be modified so that the gate is pulled below the +12V rail enough to turn the MOSFET on, but not all the way to ground where it will blow (using zener clamps is simplest).
  • R1 was used for simplicity but will slow down things which will probably be unacceptable when you are switching at high frequencies like 120kHz. A more complicated "gate drive circuit" as it is called will be required to get around this. Such as a push-pull gate or totem pull gate drive stage that uses a pull-up and pull-down transistor instead of a pull-up resistor and pull-down transistor as I have shown here.
\$\endgroup\$
14
  • \$\begingroup\$ Is there a relation between Vgs and Vds in this case? Does Vgs need to be higher than Vds in order to drive it properly? I am confused how loads like 60v are controlled with an absolute maximum Vgs of usually 20v, without heavy losses in the MOSFET. Currently, I am only expecting to drive about 4-5A through this circuit. I also would like to avoid heat sinks as much as possible due to space concerns. \$\endgroup\$ Apr 27 '20 at 2:57
  • \$\begingroup\$ This is the confusion I was wondering if you were having when you said you "set" Vds. The supply voltage is NOT Vds. Vds is the voltage across the DRAIN-SOURCE TERMINALS of the MOSFET. It is NOT the supply voltage which is the voltage across the source-drain-load. In your schematic 12v is NOT Vds. That is just the supply voltage. \$\endgroup\$
    – DKNguyen
    Apr 27 '20 at 3:04
  • \$\begingroup\$ You don't set Vds. You connect the MOSFET source-drain in series with a load and connect all that across a voltage source. You then apply Vgs. Vds is responds accordingly and as the MOSFET turns on more and more, Vds decreases. At some point, Vds decreases below the Vgs you are applying. If you continue to drive the Vgs higher and higher the MOSFET continues to turn on more and more and Vds continues to decrease to the point where the MOSFET cannot conduct any more current through it. \$\endgroup\$
    – DKNguyen
    Apr 27 '20 at 3:04
  • \$\begingroup\$ Basically, you don't set Vgs to be higher than Vds. You can't. What you do is you drive Vgs high enough so that the Vds decreases and decreases to the point where it is far below Vgs which is characteristic of the MOSFET having low losses across drain-source. \$\endgroup\$
    – DKNguyen
    Apr 27 '20 at 3:04
  • \$\begingroup\$ Sorry, I don't mean to overstate it. Just trying to fully understand this. In the case of an LED this is different as Vgs can be expected to be higher than the led voltage. In my case, 14.4v would usually be Vds and creating a higher potential would result in more circuitry. I just want to be sure that I do want the higher voltage to properly drive the fet before I go off trying to design in more to accomplish that, or if it's a limitation/bad setup in the simulator I was using. Trying to find a way to drop the 14.4v down to say 8v minimum efficiently while controlling a 4-5a load. \$\endgroup\$ Apr 27 '20 at 3:05
0
\$\begingroup\$

To use an N-FET on the high side as you’ve shown, the gate voltage needs to be brought above the supply to turn on the FET fully. This is because the swing at the inductor is also going all the way to the supply rail.

So the gate voltage needs to be above the rail voltage (that is, the inductor input), enough so that the FET is fully ‘on’ and so doesn’t have any internal IR drop.

That higher gate swing usually done with a gate driver running on a ‘bootstrap’ supply that is derived from the inductor flyback.

Let’s say your FET has a threshold of 4V, and your supply is 12V. You need to bring the gate to at least above 16V, and preferably as high as 18~20V to achieve the lowest Rds(on) and minimize losses in the FET.

How to do that? Use a high-side gate driver IC that can do that bootstrapping for you.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.