using mosfets as rectifiers?

[Ken Smith]

No, because the MOSFET has an intrinsic reverse diode built into it.

See elsewhere in this thread.

Besides, where would you get the 10 volts you'd need to turn the
MOSFET fully on?

If you pick your MOSFET carefully, 3V is enough to drive the gate.

 

[Albert]

Thanks Asimov, that was exactly the info I needed!

So the zero dropmosfet can only be used as a rectifier IF teh bias
voltage is switched on and off when the input goes in the desired
direction!

Awesome, I think you again.

A

 

[Albert]

Jim, Win and Ken,

Yes, you have uncovered the dilema and understand my question. Thanks
to you all!

I've already found (and ordered) a 2.2 volt turn on mosfet for my
switching transistor. This should give me the the head room I need to
drive the gate with a lower voltage.

With regard to the mosfet as a diode issue, I understand what you all
are saying.....without a means to tell the mosfet when to conduct, it
will conduct in either direction. Since I don't alot of extra power to
generate the signals to control the mosfets conduction periods, this
option appears closed to me.

My shottky diodes are already dropping just under 4 tenths of a volt,
so switching to a ger diode gets me very little and might not help at
all as some of them have very high leakage currents.

I also note that Maxim makes some single supply op amps that operate
on very low voltage and draw microamps of supply current. Perhaps a
small battery can be used to power a 3X amplifier since the load
impedance of my mosfet switch is high, it really doesn't need alot of
power or a high fidelity audio amp!

Again, thanks to you all. I've learned alot from this discussion.

A

 

[Winfield Hill]

Ken Smith wrote...

Your understanding of the OP's situation must be different than mine.
I though that the only way his circuit got power was via the rectifier.
When power is first applied to the system, there is no voltage to run the
comparitor from and hence it can't be controlling the FET and hence
the body diode must do the rectifying until enough voltage is produced
to run the comparitor.

If that's true, the FET should be paralleled with a Schottky diode.
For that matter, did anyone suggest he simply use Schottky diodes?

But to address your point about diode reverse-recovery time, the
intrinsic diodes in power MOSFETs have characteristics much like
ordinary rectifier diodes. In some high-speed SMPS applications
this can be an issue, where one wants fast-recovery (10s of ns)
instead. It wouldn't likely be an issue here. Once again, using
FETs isn't a very good way to attack the problem, but if they were
to be used, the most painful issue might be their high capacitance.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Ken Smith]

But to address your point about diode reverse-recovery time, the
intrinsic diodes in power MOSFETs have characteristics much like
ordinary rectifier diodes.

See if you can find a VN10KM (Siliconix) laying around in your junk box
and measure its recovery time.

Many years ago, I worked on the design of a system that used a VN10KM to
drive a long line. When the big stuff switched the MOSFET was fed a
reverse current spike. The resulting recovery time was many mS long. I
ended up adding a diode to the design to keep the current out of the
device.

IIRC, the data sheet for the part in question didn't show a body diode and
there was no spec for it.

 

[Winfield Hill]

Ken Smith wrote...

See if you can find a VN10KM (Siliconix) laying around in your junk
box and measure its recovery time.

I've got a few. I'll do it, someday. :<)

Many years ago, I worked on the design of a system that used a VN10KM
to drive a long line. When the big stuff switched the MOSFET was fed
a reverse current spike. The resulting recovery time was many mS long.
I ended up adding a diode to the design to keep the current out of the
device.

It's charge that has to be removed. You must've had almost no current,
so t = q / i was very long.

IIRC, the data sheet for the part in question didn't show a body diode
and there was no spec for it.>--

Yep, *all* VMOS parts have body diodes.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Ken Smith]

Ken Smith wrote... [.. VN10KM recovery ..]

It's charge that has to be removed. You must've had almost no current,
so t = q / i was very long.

IIRC: 6mA give or take a bit. The MOSFET pulled to the -15V rail. There
was, I think, a 4.7K to the +15V. I think I may still have the schematic
at work. I'll have to take a look this has me curious about it.

Yep, *all* VMOS parts have body diodes.

I think that Siliconix didn't think it was important enough to mension or
perhaps they measured the recovery and decided just not to bring the
subject up.

 

[Tony Williams]

I think that Siliconix didn't think it was important enough to
mension or perhaps they measured the recovery and decided just
not to bring the subject up.

I got bitten by Siliconix when trying to use the
VN10KM as a direct replacement for a JFET bipolar
analogue switch. The presence of a shunt body diode
is not acknowleged on any Siliconix VMOS data sheet,
but there is a mention in one of their app notes.

A similar situation exists with Zetex DMOS Fets.
There is no explicit mention of a body diode in the
process.... you have to notice that some data sheets
carry an added a Source-Drain Diode spec.

 

[Winfield Hill]

Ken Smith wrote...

Ken Smith wrote... [.. VN10KM recovery ..]

It's charge that has to be removed. You must've had almost no current,
so t = q / i was very long.

IIRC: 6mA give or take a bit. The MOSFET pulled to the -15V rail.
There was, I think, a 4.7K to the +15V. I think I may still have the
schematic at work. I'll have to take a look this has me curious about it.

Please do. The setup and source of diode charge and discharge currents
isn't clear from your description.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Winfield Hill]

Tony Williams wrote...

I got bitten by Siliconix when trying to use the VN10KM as
a direct replacement for a JFET bipolar analogue switch. The
presence of a shunt body diode is not acknowleged on any
Siliconix VMOS data sheet, but there is a mention in one of
their app notes.

A similar situation exists with Zetex DMOS Fets. There is
no explicit mention of a body diode in the process.... you
have to notice that some data sheets carry an added a
Source-Drain Diode spec.

The rule for all these parts is there's an intrinsic diode,
even if it's not mentioned. We all know that now, or should,
but in the early days, with exotic-looking names, numbers and
descriptions, this reality wasn't so obvious.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Ken Smith]

I got bitten by Siliconix when trying to use the
VN10KM as a direct replacement for a JFET bipolar
analogue switch. The presence of a shunt body diode
is not acknowleged on any Siliconix VMOS data sheet,
but there is a mention in one of their app notes.

You also had to watch out for the gate protection zener. It had a *noisy*
leakage current that went to the channel when it was on.

 

[Jim Adney]

So the zero dropmosfet can only be used as a rectifier IF teh bias
voltage is switched on and off when the input goes in the desired
direction!

Not quite true. Asimov's response was for FETs which, as he described,
are fairly symmetric, but your question was for MOSFETS, which are not
at all symmetric.

MOSFETs won't work for a completely different reason. See my other
post.

-

 

[Jim Adney]

starfire wrote...

This isn't an issue because a "turned-on" MOSFET happily conducts
in both directions,

That still keeps it from being much of a rectifier, however.

-

 

[Jim Adney]

If that's true, the FET should be paralleled with a Schottky diode.
For that matter, did anyone suggest he simply use Schottky diodes?

I may be wrong, but I think both you guys have missed the point of his
question. He stated that he is already using Schottkys. He wants
something with less voltage drop. He asked about MOSFETS.

I believe he was hoping that he could use some kind of active
rectifier circuit using driven MOSFETs so that there would be NO
forward voltage drop. I suspect that he has no knowledge of the
reverse intrinsic diode (which makes this approach a complete looser)
and hadn't yet thought about how he might come up with the voltage
necessary to turn the MOSFETs on.

The short answer is that MOSFETs are not the solution he was hoping
for.

-

 

[Jim Adney]

I've already found (and ordered) a 2.2 volt turn on mosfet for my
switching transistor. This should give me the the head room I need to
drive the gate with a lower voltage.

I don't follow this stuff much, so I didn't know that such things
existed. I still have to ask: Is 2.2V where the turn-on starts, or
where it is really fully on?

Thinking about this, I gather that you intend to run these MOSFETs
backwards, so that the "rectifier" is on when it is parallel to the
body diode and off when in the usual forward direction. Does that
work? I never considered the possibility....

My shottky diodes are already dropping just under 4 tenths of a volt,
so switching to a ger diode gets me very little and might not help at
all as some of them have very high leakage currents.

In my limited experience, Schottkys have a lower forward drop than
even Ge.

I don't know your application, but I wonder if you could transformer
couple your signal and boost it a bit, just enough to give you the
extra voltage you need to keep this job simple.

-

 

[Ken Smith]

look up

Please do. The setup and source of diode charge and discharge currents
isn't clear from your description.[/QUOTE]

I have someone else checking their records. I didn't turn up a copy of
the schematic. I think I, remember where the charge current came
from.


Remote preamp Long cable Main unit
.................. +15 .....................
+15V .--------------------------.
! . GND .
/ .--------------------------.
\ . .
/ . .
\ . Signal in question .
!----------------------------------------- Reciever
!- . .
!- Q1 . .
!- . -15 .
! .--------------------------.
-15V . .
.................. .....................


With Q1 on, the signal is at -15V.
With Q1 off the signal is at +15V
This works fine until we add some more info.

There is some big currents switching in the pre-amp section. Q1 turns on
just as the other thing switches off and remains on for a few mS. The
other circuit switching causes the preamp's GND to jump vs the ground in
the main unit. The combination of the cable capacitance and the receiver
circuit caused Q1's current to flow backwards. Either the bias on Q1
wasn't enough to keep the voltage below 0.7V or the current was backwards
just when Q1 was supposed to go off.

The cable could be as much as a mile long. The large signalling voltage
swings were needed to deal with the ground differences.

 

[Ken Smith]

I may be wrong, but I think both you guys have missed the point of his
question.

You may be assuming incorrectly that we are aiming at his point. We are
talking about using MOSFETs as recifiers and one strange case I had some
years back.

He stated that he is already using Schottkys. He wants
something with less voltage drop.

If you use a MOSFET as a rectifier, you get less drop than a Schottky.
The sticking point is that until you've rectified some power you don't
have the voltage needed to make the gate drive for the MOSFET. We solve
this by putting the Schottky in parallel with the MOSFET. Initially, the
Schottky does the work and gets the voltage part way up. Once enough
voltage has been made to run it, the MOSFET takes over the work.

[....]

forward voltage drop. I suspect that he has no knowledge of the
reverse intrinsic diode (which makes this approach a complete looser)

No, all you have to do is flip the Source and Drain legs of the MOSFET and
the idea works just fine, assuming the gate drive is there as needed.

 

[Ken Smith]

That still keeps it from being much of a rectifier, however.

10V Push button
0.1Hz AC /
--------------/ O-------------- DC load

If I push the button only while the AC input is positive, I've made a
rectifier. The switch would conduct in both directions if I just held the
button.

 

[Ken Smith]

Thinking about this, I gather that you intend to run these MOSFETs
backwards, so that the "rectifier" is on when it is parallel to the
body diode and off when in the usual forward direction. Does that
work? I never considered the possibility....

Works good. Many DC-DC converters that run 1.8V logic use this method to
make a lower drop than a Schottky.

At 1.8V, 0.4V means you lose 22% of the power in the diode.

 

[Spehro Pefhany]

On Thu, 10 Jun 2004 03:26:48 +0000 (UTC), the renowned

Works good. Many DC-DC converters that run 1.8V logic use this method to
make a lower drop than a Schottky.

At 1.8V, 0.4V means you lose 22% of the power in the diode.

Just Google on "synchronous rectification" for *lots* of info.

Best regards,
Spehro Pefhany