Any Electronics Engineers here? Need semiconductor help

Wanting to hook up with someone that might take pity on my electronics ignorance and allow me to email a few questions to better understand some basics in designs.

Example, I am looking at a pair of MOSFETs in a PS. Obviously 1 is certainly shorted but I would like to work more on a test circuit or learn if the DMM diode testing is a satisfactory test method.

Also looking at a couple NPN transistors where I am testing across emitter to collector with the meter and get OL one way and 15M the other. BUT, on the next identical transistor, I get 15K both ways...

I also made a test circuit with a a 5V source and an LED to test and seems to work decent for the MOSFETs but not so much on the NPN transistors.

Next example is where I screwed up and installed an MOV in place of an ICL because I could not ID the part. Obviously no harm was done, just did not conduct any power... DUH! I am coming around that MOVs are usually in parallel to absorb transients and ICLs are always in series. The one in question had a pin hole and was NOT shorted or lets just say it is 15 ohms which I think is about right at 75F.


Anyway, you can see by my questions that I have been around the stuff but I am still a ways from EE status.... My goals are just to be able to ID common problems in PS and amplifier problems. A huge problem I have is finding data on components that are poorly marked. Example is this ICL that is marked DSP 104...

You can't test emitter to collector because those are back to back diodes. You can check the base emitter diode or the base collector diode. The base is often biased high or low through a resistor, so you need to have the device out of the circuit to test it. In the results you presented I don't think you are testing the device but rather the surrounding circuitry plus the device. MOSFET is a little different. Drain to source should be open, provided there is no bias at the gate. You can't test a MOSFET with a diode tester because there's no diodes. The gate is isolated by dielectric. I don't mean to sound harsh but your measurements are meaningless for the most part. You need to isolate the transistor and get it on a curve tracer to know what is going on.

Alan

All components are out of circuit. I know better.

I was using diode testing because it passes a voltage that can be detected. Again, this is why I am asking the questions. There are certainly documents out there that suggest to test from collector to emitter. I am trying to find some concrete documents to work by.

It's not super complicated, just don't measure with the base floating. Tie the base and emitter together and connect the positive lead. Connect the negative lead to the collector. You should conduct through the diode. Same for the other way, tie the base and collector together and you should see another diode.

You're not "passing voltage", you're conducting current. You need some way to plot I vs V. That will give you the resistance of the device in saturation and the turn on voltage. This will tell you if its a working transistor.

Alan

viper said:

Example, I am looking at a pair of MOSFETs in a PS. Obviously 1 is certainly shorted but I would like to work more on a test circuit or learn if the DMM diode testing is a satisfactory test method.

Also looking at a couple NPN transistors where I am testing across emitter to collector with the meter and get OL one way and 15M the other. BUT, on the next identical transistor, I get 15K both ways...

Click to expand...

You mention the DMM diode testing, and then go on to show measurements you've made that look like resistance measurements. So, you need to clear that up.

That said, a DMM with a diode test mode is a great way to get started checking semiconductors. It often works in-circuit, because of the simple fact that a lot of time semiconductors (both bipolar and MOSFET) fail by shorting.
A little Google searching turned this up, which may be of some help:

Basic Testing of Semiconductor Devices
Testing semiconductors with analog and digital multimeters

In diode test mode, the DMM attempts to force a small current through the device under test, but the DMM displays the voltage developed across the device, not the current and not the resistance. A good silicon diode will thus test at about 0.7V in one direction, and OL in the other direction... A shorted device will typically show less than 0.2V in both directions.

Certainly there are more precise ways to check semiconductors, but a DMM with diode test mode (and knowing how to use it) is step one....

For a bipolar transistor, there IS a quick and dirty test for "gain", i.e. "working".....

Using a meter WITH A DIAL AND NEEDLE, choose a resistance checking range that gets you near midscale when you check the base to emitter diode. That is often the "R x 1" scale.

While measuring the base to emitter diode, slide the probe on the base over and contact the collector as well.

if the transistor is good, and has gain, the reading will go down. How MUCH it goes down is a rough measure of the "beta" or current gain.

Mosfets

Mosfets DO have a diode, it is "intrinsic" to the device, and is poled to conduct opposite to the normal conduction direction.

A mosfet which, when gate is shorted to the source, measures open one way, and like a diode the other way, is "probably" good.

If the gate also measures open to both drain and source, the mosfet is almost certainly good. (watch out, some have a zener gate to source, which measures like a diode one way)

Most typical failures will cause a gate to drain short. That usually damages the gate driver as well, so any bad mosfet (or IGBT) should probably be assumed to have cooked at least the gate driver as well.

IGBT

These are a bipolar transistor, with a mosfet type gate. If they have a diode, it is separate, and tells you nothing about the goodness of the part itself.

And, you don't get to check the base, it is internal, spoiling many useful tests.

Gate should test open to collector and emitter, if it does, it is probably good. if that checks, and the collector to emitter check is "open", it is probably good.

if you really need to know, two things for mosfets and IGBTs...... but they are fussier tests. They may help find "damaged but not dead" parts.

1) check the leakage......(Idss) current from drain/collector to source/emitter with gate connected to source, at circuit voltage. Should be within spec for the device.

2) for mosfets, checking the gate threshold voltage (Vth) is often good. Connect gate to drain, and connect drain to voltage through a resistor that will give the spec current, (usually between 250 uA and 1 mA). source to common.

Check voltage from drain to source, which should be within the spec in the datasheet.

Whenever working with mosfets or IGBT, use a properly grounded static strap and grounded iron. In 15 years + of working with those parts, I have yet to see anyone get away with not using a static strap and grounded soldering iron. Not for long, anyhow.

To amplify (!) what jst said, bipolar transistors are two diodes connected
back to back, with the base the connection between the two. The arrow will
tell you which direction both should conduct.

A couple of caveats: 1) not all ohm meters have the + battery connection to the
red probe. So check this before starting. 2) a simpson meter on the Rx1 scale
puts out over 100 mA of current and you CAN blow out small signal devices this
way!

If you test 1N914 signals diodes with a simpson that way, they will ALL show
defective!

Buy "the art of electronics", the student manual. Read it. Keep a pencil and some paper at hand and do the exercises. It´s really well written, hands on and plain words and it even has a dictionary for electronics slang.
It covers passive components, filters, transistors, power electronics, FETs, operational amplifiers, power supplies, digital circuitry and microcontrollers. It would still be a steal at double the price.

Well, lets start with the NPN transistors. By looking at the internal diagram, they should NEVER conduct from collector to emitter without the base triggered. However, on one of them, I am able to get 15K ohms in both directions which I think is odd.

Also, in connecting my test circuit with 5V and an LED, when connected to the collector and emitter, I get a nice light in one direction and a dim light in the other. The dim light makes me think there is some leakage of the transistor.

I realize above that testing other functions will be needed but what I am trying to do is build a test procedure that is pass/fail, before moving to more intricate testing. As I understand it, most transistors either work or they don't.



Mosfet gate driver - I have read where this can be an issue with a failing mosfet. What is typically used to drive them? Does the damage usually stop at the driver? I thought some circuits used a current limiting diode to the gate to reduce shorting damage. Maybe I am wrong there.

Johann Ohnesorg said:

Buy "the art of electronics", the student manual. Read it. Keep a pencil and some paper at hand and do the exercises. It´s really well written, hands on and plain words and it even has a dictionary for electronics slang.
It covers passive components, filters, transistors, power electronics, FETs, operational amplifiers, power supplies, digital circuitry and microcontrollers. It would still be a steal at double the price.

Click to expand...

Does it cover all the testing procedures along with internal diagrams and such? Testing in a test circuit as well as DMM? Optocouplers??

Hidden gotcha's.....

1) YOUR conductivity..... not only will you provide conductivity that shows on the meter if you bridge the test leads with a finger, as good as 10K ohms, or even lower, your conductivity can also provide base current if you inadvertently touch the base and collector.

2) A dirty surface the part is lying on, contamination of the part's case, or a flat out bad part, all can show leakage even when you are not touching things you shouldn't.

Gate drivers.....

Just assume the gate driver is bad unless proven otherwise. Even IF proven otherwise, replace it anyway! Easy insurance, cheaper than another go-around on replacing parts.

If possible, find a way to check the gate drivers output before committing the outputs to the high voltage.... A resistor or light bulb(s) inserted in series with the power source may allow checking with relatively little (but not zero) risk of damage.

it is not unknown for the damage to extend even farther back. At some point, you just count the cracked cases, and if, as is often found, there are more than 2 or 3 OTHER than outputs, it may be cheaper in the long run to replace the unit..... You probably won't find them all, and you;ll spend time and money trying to.

gate resistors are bad unless replaced, if the device failed..... just treat them as bad, and an intermittent one won't fool you.

jim rozen said:

If you test 1N914 signals diodes with a simpson that way, they will ALL show
defective!

Click to expand...

Not so fast........ The center scale value is 15 ohms, which *suggests* the short circuit current is 0.1 amp, given the 1.5V battery used in my backup 260 (the only one handy). (1.5/15 = 0.1)

BUT, the 1N914 also has a forward drop between 0,7V and 1V at any significant current. call it just 0.8V.......

(1.5 -0.8)/15 = 0.046 or 46 mA, well within the capability of the 914 or 4148. Explains why I have been happily testing them that way for 40 years or so.......

Now, if you try it with a 1N60, 1N34A, 1N21, 1N23, etc, THEN you will indeed have a bad time of it.

viper said:

Also, in connecting my test circuit with 5V and an LED, when connected to the collector and emitter, I get a nice light in one direction and a dim light in the other. The dim light makes me think there is some leakage of the transistor.

Click to expand...

The LED should not light in either direction as long as the base is floating (not connected). Yours does light so the transistor is bad.

Why don't you use the diode test function in your DMM?

I get the sense that you really want to know how a transistor works? I agree, that's worth learning... (A base doesn't get 'triggered', it gets forward biased relative to the emitter.) Feel free to PM me - I made my living as an EE for 27 years...

JST said:

Now, if you try it with a 1N60, 1N34A, 1N21, 1N23, etc, THEN you will indeed have a bad time of it.

Click to expand...

You can't be serious. Why would he be using parts like WWII X band mixers and germanium diodes?

Bill

9100 said:

You can't be serious. Why would he be using WWII X band and germanium diodes?

Bill

Click to expand...

he wouldn't...... probably.

Point is, the 260 won't hurt a 914, those others are just some that it WOULD hurt..... merely an example, my friend from Marshall Ave... merely an example.

Go try it. A simpson meter *will* burn out a 1N914.

They will flow over 100 mA through that device.

BTW the authors of that book in question are Horowitz and Hill,
and it is hands-down one of the best electronics texts available.

Amazon.com: The Art of Electronics: Paul; Hill, Winfield Horowitz: Books

Thanks for all the help guys!! Certainly will grab a book or two to study up on this stuff.

Got a quick question. We are in need of a small surface mount (SOT-223 ish) switcher than can switch AC and DC power. up to 40V and 1A. Nominal is 14V and 300mA but will be latching a power relay coil so will have to consider the inrush (still unknown). Constant draw on the coil is 200mA.

I have looked at a triac but they are a little pricy for this and not sure if there are other solutions out there. I am assuming a transistor will NOT switch full wave AC??

A triac, or SCR ( thyristor) device other than a "gate turn off" (GTO) device will not switch DC. They latch "on", and must have the current reduced below a certain minimum before they will switch "off". Ac does this naturally, DC does not.

A regular transistor or mosfet, IGBT, etc will not switch AC.

To do both/either, you would need to use a more complicated circuit employing

1) a full-wave bridge rectifier to allow the transistor to switch AC but only "see" DC.

2) A thyristor with a second parallel one that will "steal current" through a capacitor to shut off the first thyristor.

Jim:

I forgot to check the #1 Simpson at work.... I have been using that one to check all manner of diodes including 914/4148 for years..... never fried one yet, and don't expect to, as they are typically rated for 200 mA, even assuming you are correct.

The backup one is actually defective, the needle is sticky and readings don't make sense, so I need to fix it before anything useful will come from it. Got 86 mA.....

jim rozen said:

A simpson meter *will* burn out a 1N914.

Click to expand...

This Simpson schematic:

http://www.simpson260.com/downloads/simpson_260-2_schematic.pdf

Has an 11.5 ohm series resistor and a 1.5 V battery on the RX1 scale. When connected to this 1N914:

http://www.oup.com/us/pdf/microcircuits/students/diode/1N914B_philip.pdf

Which is rated at 75 mA continuous, my analysis shows less than 75 mA would flow. It would dissipate under 100 mW while being rated for 250 mW. And the 1N914 wouldn't be burned out.

Germanium diode? Different story...

Fairchild seem to be 200 mA rated, others may be lower. Old time 1N914 were around 75 mA, and 1N4148 were 100 mA, IIRC.

I'll hae to check to see what the simpson at work is like.....

There are a number of DIFFERENT versions of the 260...... Some have a single 1.5V battery, others may have more. They will vary in their ohmmeter current capability.

One version may be as Jim R suggests............ depends on the center scale resistance. And battery.

But no Simpson I have ever used has burned out a 914 or 4148.

OK, lets move on to another problem. For simplicity, I will try to only include certain relevant data.

PWM controller for small universal motor. System is run from a PIC16F876. I already know there are certain system checks performed checking for voltages, etc. PIC thinks all is just fine. When I command output, I get nothing. Output circuit is pretty simple. PIC drives an MOC3052 optoisolator. the MOC3052 then modulates power to a pair of S6025L SCRs.

I just replaced and verified function of the MOC3052. Further more, when I power up the board, the output terminals read 120V with a DMM even when output is NOT commanded. However, as soon as a load is hooked to the outputs, voltage drops to zero and the motor certainly does not try to move. I am just curious if this reading of 120V could be indication of something leaking? IE, on the output side of SCRs, should I see any voltage (static or otherwise) without being commanded?