Mosfet design question

[Cside]

Hi guys any help would be appreciated.

I want to build a circuit equivalent to a P-chan mosfet without a body diode.
I know I could just add a diode in series but efficiency is too important and I cannot afford the voltage drop.
My theory is to put two mosfets in series sharing a common drain. I haven't tested, but based on the Semicon technology It looks like the P Channel mosfet with Gate pulled low, should have minimal resistance for current in either direction - not just from source to drain. Is this correct?

If so, then such a design should theoretically work, but what about the Gotchas? any experience?

To qualify further, this is purely for DC power charging not an AC or otherwise switching system.

Again, any experience with this would go a long way before I breadboard it up.

[c4757p]

Like this?

Yeah, FETs can be used backwards. It's somewhat common in reverse polarity protection circuits, like I've attached: in forward polarization, the substrate diode conducts first, supplying initial power - raising the source above the gate and switching the FET into full conduction. In reverse polarization, there's nothing to pull the gate and source apart (as the FET is controlled by V_GS) and the diode can't conduct, so nothing goes through.

Just remember that V_GS controls the transistor, so make sure that no matter what polarizations you apply, this voltage is known.

[Cside]

Hmm
except I was using common drain. Would this not work? Is common Source the standard way to do this?
attached my idea

[c4757p]

You can put the drains together - but that puts the source of the second FET at the unknown load, making it hard to control V_GS. In your circuit, if the load went above PV+, V_GS would become negative and the FET would switch on. The other one's diode would conduct. In mine, if the load goes above PV+ the second FET's diode will conduct, but the current will push the gates up with the load, keeping both FETs switched off - and the load can't control V_GS.

[Cside]

You can put the drains together - but that puts the source of the second FET at the unknown load, making it hard to control V_GS. In your circuit, if the load went above PV+, V_GS would become negative and the FET would switch on. The other one's diode would conduct. In mine, if the load goes above PV+ the second FET's diode will conduct, but the current will push the gates up with the load, keeping both FETs switched off - and the load can't control V_GS.

yeah noticed this after I posted.

Excellent idea to have the sources common my mind has finally wrapped around it. Thanks a lot that was fantastic advice.

[CoilKid]

Is there any reason you can't use an FET? Why do you have to build one? Is this to better understand how an FET works?

[Cside]

Is there any reason you can't use an FET? Why do you have to build one? Is this to better understand how an FET works?

the MOSFET is a type of FET? are there other types of FETS that do not have a body diode and have low DC impedance when Vgs is saturated?

[David Hess]

Not all MOSFETs have the body diode but in the past, a JFET (junction FET) would have been used for this.  They are still used in small signal circuits as analog switches.

[c4757p]

Is there any reason you can't use an FET? Why do you have to build one? Is this to better understand how an FET works?

No... it's to get a decent switch. A "FET" (assuming you missed the J) isn't a very good switch.

[T3sl4co1l]

If you give us a precise description of what you're trying to accomplish, we can probably show you the way.  It's very likely been done already.  Attempting to withhold information just frustrates those who are trying to help you...

On a more theoretical level, MOSFETs make very good constant-voltage switches, which, for a certain range of applications, naturally require an antiparallel diode action.  The nearest inverse would be a vacuum tube triode, which does not conduct in the reverse direction, and which exhibits a somewhat constant-voltage output characteristic (not very well, which belies its true nature, but much more so than any other device).  Considering vacuum tubes aren't so common (or easy to use), the most practical analogous device would be an SCR, which does not conduct in reverse, however because it latches in the forward direction, it cannot be used for linear amplification purposes.  IGBTs do not conduct in reverse, in and of themselves, but do not tolerate much reverse voltage, and are frequently packed with diodes for protection.

To be fair, a MOSFET does not necessarily conduct in reverse, but because substrate and source are generally hard-wired internally, you don't have any choice.  Very old signal MOSFETs had the substrate taken out to a separate pin.  CMOS devices use the substrate implicitly (making analog switch ICs possible).  Some devices are available which have no conducting substrate, such as GaN-on-Si transistors, though within recommended operating limits, they also don't support much reverse voltage, and end up acting like diodes as well.

Tim

[T3sl4co1l]

On a completely unrelated topic...

^^ Does that mean you're a HOT-headed type, Mr. Toshiba transistor?   (Or maybe that's from an audio amp, dunno.)

Tim

[Cside]

If you give us a precise description of what you're trying to accomplish, we can probably show you the way.  It's very likely been done already.  Attempting to withhold information just frustrates those who are trying to help you...

That isn't fair, I've posted the entire schematic above. But if it isn't clear, I am charging a battery with a PV cell, tracking the maximum power point and maximum charge current.

I think the currrent configuration (as in pdf) but switch to common source as suggested by c4757p is the best method for this purpose.

[David Hess]

No... it's to get a decent switch. A "FET" (assuming you missed the J) isn't a very good switch.

This depends on the application.  JFETs are not power devices although they used to make power and even high voltage JFETs and in small signal applications where charge injection is a major problem, MOSFETs do not particularly perform any better.

In some cases bipolar transistors with emitters and bases tied together can be used as high performance analog switches.  Operation in reverse is low gain but also low saturation.  I suspect the advantage over JFETs and MOSFETs is low charge injection.

Diodes in a current driven bridge configuration can be used to make very high performance analog switches.

But the king of high power bidirectional switches now is the thyristor or back to back power MOSFET.

[c4757p]

^^ Does that mean you're a HOT-headed type, Mr. Toshiba transistor?   (Or maybe that's from an audio amp, dunno.)

Somewhat, I suppose

(No idea what the transistor's from, my nick's completely unrelated to the transistor )

This depends on the application.  JFETs are not power devices although they used to make power and even high voltage JFETs and in small signal applications where charge injection is a major problem, MOSFETs do not particularly perform any better.

True, I meant specifically in this application. Even when power JFETs were made, the on-resistance was never all that good, compared to the modern MOSFETs you can get now measured in milliohms.

[David Hess]

True, I meant specifically in this application. Even when power JFETs were made, the on-resistance was never all that good, compared to the modern MOSFETs you can get now measured in milliohms.

At higher voltages, bipolar transistors have a lower forward voltage drop than MOSFETs for a given area because minority carrier conduction supports higher current densities.  This is why you do not find any low voltage IGBTs.  Before power MOSFETs were available, I occasionally ran across solid state relays which used bipolar transistors.

Rds(on) for a MOSFET increases at something like the square of the breakdown voltage.

[T3sl4co1l]

At higher voltages, bipolar transistors have a lower forward voltage drop than MOSFETs for a given area because minority carrier conduction supports higher current densities.  This is why you do not find any low voltage IGBTs.  Before power MOSFETs were available, I occasionally ran across solid state relays which used bipolar transistors.

Rds(on) for a MOSFET increases at something like the square of the breakdown voltage.

I don't think I've seen any HV BJTs that I would consider that attractive; not this day and age.  Your average HOT boasts a 5V drop in saturation, and that at hFE = 2.5 or whatever.  Of course, IGBTs work even better, and are easier to drive.  Back in the day, you had your choice of HV BJTs, or really freakin' old toobs -- which considering the 5kV+ peak handling, are actually pretty good (though at last eclipsed by MOSFETs these days), so that 5V drop wasn't a big deal.

The quadratic scaling is over, now that superjunction technology has been perfected -- high voltage devices are as good as any, relatively speaking.  SiC was hyped to fill this gap, but then SJ came along and made it merely competitive.

Tim

[T3sl4co1l]

If you give us a precise description of what you're trying to accomplish, we can probably show you the way.  It's very likely been done already.  Attempting to withhold information just frustrates those who are trying to help you...

That isn't fair, I've posted the entire schematic above. But if it isn't clear, I am charging a battery with a PV cell, tracking the maximum power point and maximum charge current.

I think the currrent configuration (as in pdf) but switch to common source as suggested by c4757p is the best method for this purpose.

Ok, that's helpful, because it's not what the schematic is showing -- namely, how can you have MPPT if you aren't doing any conversion?  Between "PV" and "Vbatt" there's a shunt sense resistor and a switch.  And... why would you ever want current to flow back into the PV from Vbatt?

Since you don't have a converter, you're literally as good as using a diode -- well, an "ideal diode" anyway.

Tim