R.G.'s reverse polarity protection (MOSFET)

[markusw]

Just tested it on breadboard: used a BS250 with 1 M from G to GND, 1k pot from S to gnd (via a DMM), 9V battery to D. Adjusted 1k trim to give ~20 mA on DMM. Then measured voltage drop via DS with a second DMM: 150 mV OK it's far better than a series 1N400x but it's definitely easy to measure with a DMM.

After looking at the data sheet this voltage drop seems reasonable since it states a DS resistance of "typically" 9 ohm. Thought "Ok R.G.says use a BS250P". Looked at the data sheet of the BS250P    it states a DS resistance of 14 ohm.

Now my question is: probably it's not a BS250P but a BSP250 which has a DS resistance of 0.25 ohm? On the other hand this one is only available in a SOT223 package.

Any help would be highly appreciated since I think this trick is really "sexy" and I would like to use it for my future projects.

Thanks in advance,

Markus

[R.G.]

Since you're seeing turn on of the MOSFET, I'm not sure there's anything wrong with the circuit as you have it set up, except possibly getting the device backwards. Check that - make sure it's properly backwards.

I remember being mildly surprised that the voltage drop was that low. I checked it a couple of times to be sure. If I recorded as little as 150mv, I'd still have been quite happy. It could be that I got an especially good batch of MOSFETs. I'll have to go see if there are any of those left and check again.

Just playing Devil's Advocate - if it works and you only lose 150mv to good polarity protection, is that possibly good enough?

[markusw]

Since you're seeing turn on of the MOSFET, I'm not sure there's anything wrong with the circuit as you have it set up, except possibly getting the device backwards. Check that - make sure it's properly backwards.

I remember being mildly surprised that the voltage drop was that low. I checked it a couple of times to be sure. If I recorded as little as 150mv, I'd still have been quite happy. It could be that I got an especially good batch of MOSFETs. I'll have to go see if there are any of those left and check again.

Just playing Devil's Advocate - if it works and you only lose 150mv to good polarity protection, is that possibly good enough?

Thanks a lot for your reply. I tested it with the MOSFET (BS250) in both orientations (G to GND via 1M with both) and actually got roughly the same value independent of the orientation.  And, yes for sure 150 mV is fine. But why not get lower if the price is the same ? (I think it's simply my noobeeish tweaking-spleen)
Therefore another Q: if I use e.g. this MOSFET http://www.irf.com/product-info/datasheets/data/irfr9024n.pdf the voltage drop should be much lower (DS resistance of 0.175 ohm) than with the 9 ohm of the BS250? Or is the "Static Drain-to-Source On-Resistance" the wrong parameter to look at 

Markus

[R.G.]

I tested it with the MOSFET (BS250) in both orientations (G to GND via 1M with both) and actually got roughly the same value independent of the orientation.

I guess that makes sense - the channel resistance should be the same. I was wondering if the substrate diode had any effect, and it seems that it doesn't.

Q: if I use e.g. this MOSFET http://www.irf.com/product-info/datasheets/data/irfr9024n.pdf the voltage drop should be much lower (DS resistance of 0.175 ohm) than with the 9 ohm of the BS250?

I think it will be.Their typical Id versus Vgs curves show that with Vgs=-8V, Id of 600ma pulls the Vds up to 0.1V, giving an rds of .1667, close to what they spec for rds. Sure - that one should get your drop lower. It looks like rds is linear with that one, at least at higher currents, so extrapolating we'd think we gets Vds of 3.5mV at 20ma.

Actually, taking a stroll through the digikey catalog, IRFD9024 in the four pin dip pack may be the winner; its spec'ed at rds of 0.28 ohms and costs $0.88.

It's probably worth your time and money to toss in a 12V zener from gate to source for protection.
==============update....==================
The more I thought about this, the curiousier I got. I went out to the garage and dug out my old stocks.

I found my old BS250Ps, and when I ran the test again, I measured about 1mV of drop at 50ma. Which seemed decidely odd, given your results and the rds you're finding on the data sheets. While puzzling over that, I put on my magnifiers and looked at the packages really closely. They say "BSP250", not BS250P.

I got onto the net and looked around. I can't find any "BSP250" in to-92 packages. Phillips makes a BSP250 in the sot 223 SMD package that is specified at 0.25 ohm rds.  All of the other chips I can find that match the "BS250P" and are in TO-92 packages have rds specified at about 7-12 ohms.

I think I got a batch of TO-92 packaged BSP250s, not BS250Ps, then hosed up writing down the part number. So the results seemed artificially good.

TO-92 Pchannel MOSFETs that work as I wrote the article:
Mouser Fairchild FQPF11P06, TO-92  rds =0.175   $0.86
Mouser Supertex VP3203N3, TO=92 rds = 0.6 $0.94

Interestingly, Mouser offers the Vishay Semiconductor BS250/E6 (no P's involved!) for $1.14, and it's max rds is spec'ed at 5 ohms, leading to a 0.1V drop in your setup. This one is dead in the middle!

Another winner might be the SupertexVP2106N3. It's rds is 12 ohms, so you'll see a loss of about 0.2V, but it's only $0.31 each in units!!
Mouser Supertex VP2106N3, TO-92 rds = 12.0, $0.31

So the bottom line is - I had an unusually good batch that doesn't seem to be available any more in that package under that part number. There are a lot of P-channel MOSFETs that work just fine as I drew it if you want to use SMD parts. There are at least a couple of parts in the TO-92 package that work well. There are lots of cheap, easily available parts that work OK, but you lose a small fraction of a volt. I like that IRF part in the hexdip package. That's easier mounting than a to-92.

It's worth noting that N-channel MOSFETs in to92 packages that have rds under a couple of ohms are widely available. If you protect on the negative side, they'll work, so positive ground pedals are easy to do this for.

Bottom bottom line - thanks! You pointed out some things I need to update. I hope the info I dug out helps.

[markusw]

Thanks for your clarifications and for the tip with the zener diode. It's always highly informative to browse through your posts.

I think I will give the IRFU9024 with it's "straight lead package" a try. Curiously it's about half the price compared to the IRFD9024 here in Austria. Although it's roughly 7 mm in height it should'nt consume much more space on a PCB than a 1N400x diode.

One more newbie Q if you don't mind 

the zener diode is for protecting the MOSFET from electrostatic discharges (i.e. keep the gate-source voltage below the maximum ratings) ?  higher voltages per se won't hurt the MOSFET, just if the gate-source voltage gets too high (>20 V in case of the IRFU9024 and IRFD9024)? so actually the zener could also be from gate to drain?

Markus

[R.G.]

the zener diode is for protecting the MOSFET from electrostatic discharges (i.e. keep the gate-source voltage below the maximum ratings) ?

Yes.

higher voltages per se won't hurt the MOSFET, just if the gate-source voltage gets too high (>20 V in case of the IRFU9024 and IRFD9024)?

Yes.

so actually the zener could also be from gate to drain?

Hmmmm... let me think for a moment.

No. You should put it from gate to source, but with the cathode to the source, anode to gate. The P-channel drain and gate normally resist high voltages when the polarity is in the correct direction for the device. in this case, that's the reversed-polarity direction we're trying to protect against. As long as the gate is less than a gate-threshold-voltage negative from the channel, the eice is non-conducting and undamaged. A zener with anode to gate, cathode to source clamps incoming voltages to a diode drop (less than the threshold) in the reversed direction, and less than the zener in the other. The zener will conduct for reversed-polarity connections, but it conducts through that 1M resistor in series with the gate, so the current is limited by the 1M and by the MOSFET being off. The voltage can't turn the MOSFET on because it's clamped to below the threshold, so no current flows. The body diode prevents damage in the other direction.

So the zener goes anode to gate, cathode to source. 

[markusw]

 :) 

As far as I understand electrostatic dicharges may burn tiny holes into the gate/source insolation (oxide or whatever). Thus the insolation layer is destroyed. Hm, but why is the insolation layer between gate and drain more stable to electrostatic discharges?? Or how is the body diode able to prevent the damage of this insolation layer?

Thanks for your patience.

Markus

[R.G.]

I'm pretty solid on bipolar and JFET semiconductor physics, less so on MOSFET. I'll have to go look it up. My comments are based on keeping things working similar to normal operation of the device. May take me a day or so to dig it out, as the reference books are still in boxes.

[markusw]

I'll wait patiently 

Also tried to find an explanation by googling but without success. Actually,  it may well be that I found an explanation but did'nt realise because of my lack of knowledge.

[KORGULL]

When selecting the gate protection zener diode, could I use any zener that has a reverse breakdown voltage rating which is higher than the power supply, but lower than the MOSFET's gate-source maximum voltage rating?
For example, if my circuit is running from a 12V battery and the MOSFET's maximum gate-source voltage is 40V, would I be able to use any zener between about 15-30V?

I'm thinking that the zener diode is just there for electrostatic protection, so it only needs to attenuate or redirect any voltage that is higher than what the MOSFET gate can withstand; so the main thing is that the zener needs to (reverse) conduct at a voltage that is lower than the max V the MOSFET can take without damage.
And you don't want the zener to do anything as long as no voltages higher than its reverse breakdown rating are present; which is why you want it to have a reverse rating higher than the power supply.

Can someone either confirm what I'm thinking, or help me understand this better?

Between using a PNP MOSFET, hooking it up backwards, adding in a zener, and considering the MOSFET's body diode, I get confused and have some trouble picturing all at once how this whole thing works.
I don't have much experience at all with using MOSFETs yet.

[KORGULL]

After re-reading (yet again) the other thread where I asked about this, I see my answer was there the whole time.

R.G. wrote:

Actually, you want a zener that never conducts, except in conditions where the gate would be damaged. The gate is damaged at 20V, so you want one lower than that, and you'd like the actual power supply not to turn it on, so it needs to be more than 9V plus any variation above 9V. A fresh battery is 9.3-9.4V, and some power supplies get to 10V. I personally would use a 12-15V zener. 12V is kind of standard for MOSFET gate zeners in many apps.

http://www.diystompboxes.com/smfforum/index.php?topic=44966.0

I don't know what's wrong with my brain lately -spent way too much time trying to sort out a cap switching mod on a perf layout diagram last night ...a diagram/mod that I drew and implemented less than six months ago. I think I found a mistake in it though; which might (partially) explain why it wasn't making sense.