MOSFET Testing

[alpha_uma]

Is an oscilloscope the only way to test a MOSFET in circuit to see if it has
gone south?

Al-U

 

[Walter Harley]

Is an oscilloscope the only way to test a MOSFET in circuit to see if it has
gone south?

Depends on the circuit. You can sometimes use an ohmmeter. For power
(enhancement) MOSFETs, if the circuit is off, you should see something
reasonably large from gate to source/drain, and something similarly large
from source to drain. If you see only a couple of ohms it's probably fried.
In my limited experience, MOSFETs usually fail shorted, not open.

 

[alpha_uma]

Depends on the circuit. You can sometimes use an ohmmeter. For power
(enhancement) MOSFETs, if the circuit is off, you should see something
reasonably large from gate to source/drain, and something similarly large
from source to drain. If you see only a couple of ohms it's probably fried.
In my limited experience, MOSFETs usually fail shorted, not open.

Thanks for your advice, Walter. The MOSFETs that I want to test are in
circuit on my dead motherboard (socket 478, P4). BTW, I have already
replaced my system with a new motherboard. I just want to troubleshoot what
has gone south, and learn something along the way.

I did the ohmmeter test on the six voltage regulation MOSFETs near the ATX
power connector. I don't know if it makes any difference where the positive
lead of the meter should be placed. But I make sure that I did them
consistently by placing the positive lead on the first of each of the pairs
(G-S, G-D, D-S) listed below. For the sake of comparison, I performed
similar tests on a spare motherboard (socket 370, P3) that is known to be in
good functioning condition.

This is what I found on my GOOD socket 370 motherboard:

For one MOSFET (B3015L):
G-S: 21.3 mega-ohm
G-D: 20.9 mega-ohm
D-S: 238.6 ohm

For second MOSFET (B3015L):
G-S: 20.9 mega-ohm
G-D: 21.3 kilo-ohm
D-S: 14 kilo-ohm

This is what I found on my DEAD socket 478 motherboard:

For the three MOSFETs NEC K3296, all gave similar (incredibly low)
readings as follows:
G-S: 18.7 ohm
G-D: 21.8 ohm
D-S: 3.2 ohm

For the three MOSFETs JS214A, all gave similar (incredibly low) readings
as follows:
G-S: 0 ohm (this is bad, right?)
G-D: 29.7 ohm
D-S: 27.6 ohm

Do you think these six MOSFETs on my socket 478 motherboard are all dead?
BTW, the three electrolytic caps around these six MOSFETs are already
confirmed bad out-of-circuit, and I replaced them with good ones.

Thanks
Al-U

 

[Walter Harley]

[...]
I did the ohmmeter test on the six voltage regulation MOSFETs near the ATX
power connector. I don't know if it makes any difference where the positive
lead of the meter should be placed.

A power MOSFET has a "body diode" (a diode that just happens to be there as
a result of the semiconductor process) from S to D. The polarity depends on
whether the MOSFET is n-channel or p-channel. Often in the manufacturer
datasheet it shows the diode explicitly. For instance, if you look at the
NEC datasheet for your 2SK3296, at
http://www.necel.com/nesdis/image/D14063EJ2V0DS00.pdf, down on page 6 you'll
see a diagram with the diodes shown. (There's also a protection diode for
the gate, on that MOSFET.) If you get your leads backwards, you'll see
conductivity through the diode, even though the MOSFET isn't turned on.

For the three MOSFETs NEC K3296, all gave similar (incredibly low)
readings as follows:
G-S: 18.7 ohm
G-D: 21.8 ohm
D-S: 3.2 ohm

In your case, when testing the D-S pair you put the positive lead on D,
which turns out to be correct for the 2SK3296, because it's an N-channel
MOSFET. So, yes, at least one of them is blown: the G-S and G-D readings
are way too low to be real (you should expect more than 100 ohms in any
circuit I've seen), and the D-S reading is also very low.

However, you probably can't tell which one is blown, because they may all be
in parallel with each other, meaning the blown one is shorting out the
others. You'll need to unsolder them till you find the one (or more) that
is blown.

For the three MOSFETs JS214A, all gave similar (incredibly low) readings
as follows:
G-S: 0 ohm (this is bad, right?)
G-D: 29.7 ohm
D-S: 27.6 ohm

In this case, putting the positive lead on the D when you measured the D-S
pair was wrong; as you know, the JS214A (aka 2SJ214A, from previous
discussion) is a P-channel device.

But regardless, seeing a dead short from G to S indicates a fault. There's
a big smoking hole in that silicon. Again, it might only be one or two of
the three; they're probably all in parallel with each other.

By the way, the group sci.electronics.repair is also a good group for this
kind of discussion.

 

[alpha_uma]

[...]
I did the ohmmeter test on the six voltage regulation MOSFETs near the ATX
power connector. I don't know if it makes any difference where the positive
lead of the meter should be placed.

A power MOSFET has a "body diode" (a diode that just happens to be there as
a result of the semiconductor process) from S to D. The polarity depends on
whether the MOSFET is n-channel or p-channel. Often in the manufacturer
datasheet it shows the diode explicitly. For instance, if you look at the
NEC datasheet for your 2SK3296, at
http://www.necel.com/nesdis/image/D14063EJ2V0DS00.pdf, down on page 6 you'll
see a diagram with the diodes shown. (There's also a protection diode for
the gate, on that MOSFET.) If you get your leads backwards, you'll see
conductivity through the diode, even though the MOSFET isn't turned on.

For the three MOSFETs NEC K3296, all gave similar (incredibly low)
readings as follows:
G-S: 18.7 ohm
G-D: 21.8 ohm
D-S: 3.2 ohm

In your case, when testing the D-S pair you put the positive lead on D,
which turns out to be correct for the 2SK3296, because it's an N-channel
MOSFET. So, yes, at least one of them is blown: the G-S and G-D readings
are way too low to be real (you should expect more than 100 ohms in any
circuit I've seen), and the D-S reading is also very low.

However, you probably can't tell which one is blown, because they may all be
in parallel with each other, meaning the blown one is shorting out the
others. You'll need to unsolder them till you find the one (or more) that
is blown.

For the three MOSFETs JS214A, all gave similar (incredibly low) readings
as follows:
G-S: 0 ohm (this is bad, right?)
G-D: 29.7 ohm
D-S: 27.6 ohm

In this case, putting the positive lead on the D when you measured the D-S
pair was wrong; as you know, the JS214A (aka 2SJ214A, from previous
discussion) is a P-channel device.

But regardless, seeing a dead short from G to S indicates a fault. There's
a big smoking hole in that silicon. Again, it might only be one or two of
the three; they're probably all in parallel with each other.

By the way, the group sci.electronics.repair is also a good group for this
kind of discussion.

Thanks for your detailed explanation, Walter. It's very helpful. I saw the
documentation for the 2SK3296 at NEC's site as well, which makes more sense
to me now after you've explained the function of those static protection
diodes.

I've tried the "electronics.repair" newsgroup already, but not too many
people responded to an earlier post there. Anyway, the information I got
there was to test with a 'scope which I don't have (so I've been browsing
E-bay for the past couple of days). Because the traffic at the ".repair"
group is busy, my message soon got buried among its numerous daily postings.
Now, thanks to you, at least I think I have some suspect chips on my
motherboard to pursue.

I will probably head for the ".repair" newsgroup again and ask for tips for
removing MOSFETs. When it comes to replacing SMDs, I'm a just a rookie. Do
you have any advice on that?

Also, could you point me to a few good sites where I can look up cross
references for replacing these two MOSFETs? I googled many times on 2SJ214A,
and came up fairly empty. I don't even know who makes that chip. I found
Fairchild's online cross reference site, and found some listing of chips
compatible to K3296, but not on 2SJ214A. In any case, I can't afford to
order thousands of K3296s from them. I tried searching for MOSFETs at
DigiKey. I didn't see any online cross references and I'm not sure if it is
enough to just specify the max A, max V and max T. What are the most
important parameters (besides RDSon) that I should be looking at when
searching for compatible MOSFETs?

Thanks

 

[Walter Harley]

[...] the information I got
there was to test with a 'scope which I don't have (so I've been browsing
E-bay for the past couple of days).

There's a scope up there right now:

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=58282&item=3830457659&rd=1

This is a Hitachi V-212, which I would highly recommend as a simple,
reliable, entry-level scope. I have one and still use it almost daily,
although it is no longer my main scope (my main scope is a Tek 2465). It
has served very well. For a buy-it-now price of $80 from an established
seller, this is a great deal. (I have no affiliation with the seller.)

I will probably head for the ".repair" newsgroup again and ask for tips for
removing MOSFETs. When it comes to replacing SMDs, I'm a just a rookie. Do
you have any advice on that?

I'm a rookie at SMT also. I haven't tried the low-temp alloy stuff, but I
want to. Barring that, I have just used a small tip on the soldering gun,
solder wick, tweezers, and swear words. Unsoldering is much harder than
soldering.

Also, could you point me to a few good sites where I can look up cross
references for replacing these two MOSFETs?

What I do is watch on sci.electronics.components and sci.electronics.repair,
and when people come up with answers to questions like that, I try to keep
track of what sites they were on; or ask them, if they didn't say in their
post. Not everything shows up on Google! If you can figure out the
manufacturer, sometimes going to their site and searching is more
successful.

Also, make sure you search for variants - J214, J214A, 2SJ214, 2SJS214,
whatever. You can't assume they're the same, but they might be.

I didn't see any online cross references and I'm not sure if it is
enough to just specify the max A, max V and max T. What are the most
important parameters (besides RDSon) that I should be looking at when
searching for compatible MOSFETs?

You have to know something about the circuit, unfortunately. There are a
lot of parameters and which ones matter depends on how it is being used.
Make sure the pinout (which pin is S/D/G) is the same! Another parameter
that matters is Qg, or gate charge - it is how much charge needs to be on
the gate in order to turn on the channel. Often you can find replacements
that have lower Rds(on), but at the expense of needing more charge; the
driver circuitry may not be able to supply the extra current.

This, and the difficulty of buying the parts in small quantities even if you
can find an appropriate replacement, is one reason why it is so hard to
repair many modern electrical devices such as motherboards.

By the way, you should be aware that when a MOSFET fails short-circuit, it
often couples whatever is driving it to the supply rails, which often
results in a failure of the driver as well. So you should expect to see
some failed components in the driver circuitry. You may as well check for
them now, because if you don't replace them at the same time, then you'll
just end up re-frying the MOSFET due to the now-defective driver circuitry.
Not to mention that when the MOSFET failed short-circuit it also probably
resulted in whatever the MOSFET was driving being killed by being connected
directly to the supply rails. That is, for instance, if the MOSFET was
stepping 5V down to 3.3V, then when it failed short-circuit it might have
resulted in a bunch of 3.3V devices seeing 5V. Or not - just depends on the
circuit.

 

[alpha_uma]

You have to know something about the circuit, unfortunately. There are a
lot of parameters and which ones matter depends on how it is being used.
Make sure the pinout (which pin is S/D/G) is the same!

Silly me. I thought they were all in the order of GDS (from left to right).
Well, at least the NEC K3296 seems to follow that order according to the
documentation.

This, and the difficulty of buying the parts in small quantities even if you
can find an appropriate replacement, is one reason why it is so hard to
repair many modern electrical devices such as motherboards.

Yes, it's just a long shot to begin with. I was hoping that it was just the
fat caps, but it's beginning to look much worse.

By the way, you should be aware that when a MOSFET fails short-circuit, it
often couples whatever is driving it to the supply rails, which often
results in a failure of the driver as well. So you should expect to see
some failed components in the driver circuitry. You may as well check for
them now, because if you don't replace them at the same time, then you'll
just end up re-frying the MOSFET due to the now-defective driver circuitry.
Not to mention that when the MOSFET failed short-circuit it also probably
resulted in whatever the MOSFET was driving being killed by being connected
directly to the supply rails. That is, for instance, if the MOSFET was
stepping 5V down to 3.3V, then when it failed short-circuit it might have
resulted in a bunch of 3.3V devices seeing 5V. Or not - just depends on the
circuit.

Thanks for this reminder. The bad news is that I'm not sure if I can
identify exactly which components on the motherboard belong to the driver
circuit. Near those MOSFETs, I can see three circular magnet-looking coils,
an IC (labeled F E151A KA7500B), a tiny tiny 10uF electrolytic, and two
transistors (whose labels I could not make out until I get a bigger
magnifying glass). These seem to be for driving the MOSFETs. And even if
these are the only items in the driver circuit, they will be testing the
limits of my soldering skill. At the very least, I need a finer tip. And it
would have helped if they had directly soldered that IC on to the board,

So, I think I will just try to find some replacement MOSFETs from, say,
DigiKey, and try my luck. If the driver circuit is shot, it would fry the
six replacement MOSFETs, but I think I can afford to burn six MOSFETs.

Al-U

 

[Walter Harley]

Thanks for this reminder. The bad news is that I'm not sure if I can
identify exactly which components on the motherboard belong to the driver
circuit. Near those MOSFETs, I can see three circular magnet-looking coils,
an IC (labeled F E151A KA7500B), a tiny tiny 10uF electrolytic, and two
transistors (whose labels I could not make out until I get a bigger
magnifying glass). These seem to be for driving the MOSFETs. And even if
these are the only items in the driver circuit, they will be testing the
limits of my soldering skill. At the very least, I need a finer tip. And it
would have helped if they had directly soldered that IC on to the board,

Here's your KA7500B: http://www.fairchildsemi.com/ds/KA/KA7500B.pdf

 

[Franc Zabkar]

Here's your KA7500B: http://www.fairchildsemi.com/ds/KA/KA7500B.pdf

FYI, the KA7500B is equivalent to the TL494. Many AT/ATX PSUs use
these.


- Franc Zabkar

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

FYI, the KA7500B is equivalent to the TL494. Many AT/ATX PSUs use
these.

Thanks, that's good to know. I've found that Sharp pulls the shenanigan
of marking their equivalent chips with some oddball number, but I didn't
know that Fairchild did so, too. Usually they just change the TL (which
stands for TI linear, IIRC) to LM.

 

[Franc Zabkar]

Thanks, that's good to know. I've found that Sharp pulls the shenanigan
of marking their equivalent chips with some oddball number, but I didn't
know that Fairchild did so, too. Usually they just change the TL (which
stands for TI linear, IIRC) to LM.

It was originally a Samsung part, not Fairchild.


- Franc Zabkar

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

It was originally a Samsung part, not Fairchild.

Ah-hah, I see. Fairchild's is a clone of the Samsung, which is a clone
of the original TI part. )

 

[Franc Zabkar]

Ah-hah, I see. Fairchild's is a clone of the Samsung, which is a clone
of the original TI part. )

It's more like Fairchild *is* Samsung.

See "Fairchild Semiconductor Completes the Acquisition of Samsung
Electronics' Power Device Division (Apr 14, 1999)":
http://www.fairchildsemi.com/news/1999/9904/ssclose.html


- Franc Zabkar

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

On Fri, 6 Aug 2004 08:34:37 -0700, "Watson A.Name - \"Watt Sun, the
Dark Remover\"" <[email protected]> put finger to keyboard and
composed:

It's more like Fairchild *is* Samsung.

See "Fairchild Semiconductor Completes the Acquisition of Samsung
Electronics' Power Device Division (Apr 14, 1999)":
http://www.fairchildsemi.com/news/1999/9904/ssclose.html

Oh. I thought National acquired Fairchile, and then later spun it off
as independent. Cornfusing. Whatever makes a buck, I guess. As long
as I can get datasheets, that's mainly what i'm concerned about.

<<
A guy goes into a restaurant/lounge wearing a shirt open
at the collar and is met by a bouncer who tells him he must
wear a necktie to gain admission. So the guy goes out to
his car and he looks around for a necktie and discovers
that he just doesn't have one. He sees a set of jumper
cables in his trunk. In desperation he ties these around
his neck, manages to fashion a fairly acceptable looking
knot and lets the ends dangle free. He goes back to the
restaurant and the bouncer carefully looks him over for
a few seconds and then says, "Well, OK, I guess you can
come in - just don't start anything."

 

[alpha_uma]

Here's your KA7500B: http://www.fairchildsemi.com/ds/KA/KA7500B.pdf

Thanks. I saw it too.

Does anyone have the documentation or a cross reference for the MOSFET
2SJ214A? Who makes the chip? I can't find it on the web, and without it, I
don't know how to find a replacement MOSFET. Any help would be much
appreciated.

 

[alpha_uma]

Does anyone have the documentation or a cross reference for the MOSFET
2SJ214A? Who makes the chip? I can't find it on the web, and without it, I
don't know how to find a replacement MOSFET. Any help would be much
appreciated.

OK, I'm going to start a new thread on the search for JS214A and hope that
more people would be able to help.

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

OK, I'm going to start a new thread on the search for JS214A and hope that
more people would be able to help.

Make up your mind. Do you want 2SJ214A? Or JS214A?

Have you tried www.datasheetarchive.com? How about a substitute at
www.nteinc.com?
Some info here (hard to read) http://tech.usaeg.com/download/00.pdf
And here's a schem that uses them
http://www.renesas.com/avs/resource/japan/jpn/pdf/linear/j204025_ha16116
fp.pdf
this is on Renesas, which might be the maker. Used to be Hitachi and
Mitsubishi? I think..

 

[rob001]

A few questions about mosfets...

I have a 100 mhz two channel oscilloscope with an auto-adjust setting. How could I use this scope to test an IC mosfet?

I'm also trying to figure out what is wrong with a DV6000 No Power - Blue DC Block lights..

Low ohms

181 ohms - 7636 (By max 8724)

51 ohms - 7636 (by ENE)

274 ohms - 6679 (opposite side of CPU socket)


Replace all? This version of the board uses an intersil SMPS controller. How would I test it? ISL6236

I also bought a logic probe. I wonder how I could use that to test motherboards.