Mosfet question -- substrate connected to source

This question is geared toward the enhancement mode mosfet devices that have the substrate connected to the source.

Since the device is not symmetrical, what is the downfall of driving current from the source to the drain and effectively switching their roles when the substrate is connected to the old source?

Thanks in advance.

Reply to
Tom McAndrews
Loading thread data ...

If you try to do this, the intrinsic substrate diode starts to conduct. As I understand it, it's not a terribly great diode in terms of either switching speed or forward voltage drop, but its presence pretty much eliminates the ability to use the MOSFET in its usual way. Remember, when you switch the connection roles around you do not magically change the internals of the MOSFET - the substrate is still connected to the "source" pin.

I have seen one interesting application for a MOSFET wired up in this "backwards" fashion, in "Troubleshooting Analog Circuits" by Bob Pease (page 164). It makes a nifty polarity-reversal protector for power-supply-to-load hookups. For example, you install an NMOSFET on the ground side of the load, with the drain grounded and the source towards the load, and drive the gate via a resistor or voltage divider from the high side of the power supply.

If the power supply is hooked up correctly, the positive voltage biases on the MOSFET, _and_ forward-biases the substrate diode, so the MOSFET switches on and allows power through to the load. If you use a MOSFET with a low Rds[on], you end up with less voltage drop than you'd have from a simple silicon-diode protector.

If the power is accidentally hooked up backwards, the substrate diode is reverse biased (doesn't conduct) and the gate is pulled several volts negative with respect to the source and drain and switches the MOSFET off quite hard. The reverse-voltage standoff will be that of the MOSFET - probably a lot more than you could get with a low-forward-voltage-drop Schottky diode.

To quote Pease, "Yes, this circuit looks funny. Yes, it is correct, and yes, it works well."

--
Dave Platt                                    AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
 Click to see the full signature
Reply to
Dave Platt

Since the drain to source path is paralleled with a diode, using the fet as a conductive path only makes sense if it will always be turned on when the drain and source are reversed or the applied voltage will be less than about 300 millivolts. They are used this way as synchronous rectifiers.

--
John Popelish
Reply to
John Popelish

Another notable example of an instance where we make good use of the integral diode is in totem-pole MOSFET drives for inductive loads.

In these applications, there has to be a delay between switching OFF the conducting device and switching ON the device that was previously OFF, in order to prevent shoot-through. What then happens is that the inductor current commutates into the body diode of the device that was previously OFF, thereby providing a continued path for the current, and the diode is then shunted by the low Rd-s on of the MOSFET when it is later turned ON. (Yes, it does conduct in both directions).

You kids have it easy these days though. Back in the eighties, you had to be really careful using power MOSFETs in totem-pole configuration. If one of the devices was OFF, but the current flowing through its body diode, you couldn't simply turn ON the other MOSFET to take over the current as this would 'pop' either the body diode of the OFF MOSFET, or the d-s channel of the MOSFET that you had just turned ON, due to the reverse recovery limitations of the in MOSFET's body diode.

Regards

Ted Wilson

Reply to
Ted Wilson

Why is everybody talking about diodes? Power? Goodness.. there is only one thing that this will do. If the voltage on the drain of the FET is more negative then the source, then it's topologically the SAME as inverting the device (exchanging the drain and source) and tying the substrate to the drain. What that does is that it requires less voltage on the gate to turn the device on. In your question, you forgot to mention something.. it this pFET, or nFET?

Unless I'm wrong, the device is symmetrical (It's a FET afterall). The only thing that changes is that you're effectively increasing the potential on the substrate, thereby lowering the Vt of the device.. the rest depends on whether you have a pFET or an nFET.

Veronique

Reply to
Veronique

[snip]

It's called a "body" diode for a reason. Unless the "body" is a separate connection, weird things can happen in reverse.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

I don't think that a MOSFET _has_ to be symmetric (you could build a symmetric bipolar transistor, after all, and even "regular" ones have some action when run in reverse). I just think that MOSFET designs tend to come out more symmetric than bipolars do.

If you dig up a drawing of a MOSFET in a textbook you'll see that the interfaces between the substrate and the source and drain are PN junctions (a depletion-mode MOSFET will have one region that makes up the source, drain and channel). This junction will have some bias on it, and will conduct if it's a forward bias. So for a three-pin device you connect the substrate to the source, and all of a sudden your device has a built-in diode.

When they first started using power MOSFETs in switching supplies this was a pain, because the diode is right where you put your snubber diode, only it has first crack at turning on. Since the device is optimized for MOSFETness, the intrinsic diode is usually a pretty bad snubber diode and can have an extremely long reverse-recovery time. Everyone has learned how to deal with this now, but it's still an issue to be ignored at risk.

the

only

Reply to
Tim Wescott

a

how

Can you talk some more about that? I noticed recently (while looking for replacements for a blown MOSFET) that some of them claim "ultrafast intrinsic diode" or the equivalent, so I assume that's one solution to the problem. You make it sound like there are other solutions - what are they?

Thanks!

Reply to
Walter Harley

No problem when the device is used as a switch and is ON. But when it is off, you have to contend with the the diode.

It is common to see two devices in series with their sources tied together. That way, when they are off, only one of them will have a forward biased substrate diode.

--
local optimization seldom leads to global optimization

my e-mail address is:   AT mmm DOT com
Reply to
Roy McCammon

I read in sci.electronics.design that Tim Wescott wrote (in ) about 'Mosfet question -- substrate connected to source', on Thu,

22 Jan 2004:

Philips did; the OC140 was a silicon alloyed device. Used in video processors, and designers were very annoyed when Philips suddenly discontinued it.

--
Regards, John Woodgate, OOO - Own Opinions Only. 
The good news is that nothing is compulsory.
 Click to see the full signature
Reply to
John Woodgate

The only other way that can remember is to use a Shottkey in parallel with the MOSFET, pointing the same way as the intrinsic diode. Very fast transients will want to go through the MOSFET anyway, so you have to be careful that the diode sees a nice low-inductance path.

was

it

and

they?

Reply to
Tim Wescott

com>) about 'Mosfet question -- substrate connected to source', on Thu,

I recall there were some others, done like that for use as analogue switches.

Regards Ian

Reply to
Ian Buckner

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.