Transistor Curve Tracing Shows Weird Scope Display

Thread Starter

RonFredericks

Joined Nov 13, 2009
11
I recently crafted a transistor curve tracer with my Oscilloscope and Arbitrary Waveform Generator. Most BJT Si transistors show typical/normal traces but one device undertest produces abnormal trace when collector-emitter voltage (Vce) goes higher than about 6 Volts.

Does anyone know what is the cause of this abnormal trace (transistor or test setup for examples)?

I present a few images below showing my test setup, a good trace, and the weird trace.
  • image 1: LTSpice showing the circuit I am using
  • image 2: breadboard showing the circuit I am using with scope probes and AWG waveform inputs
  • image 3: AWG setup
  • image 4: a 2N3906 transistor showing typical results
  • image 5: the problem result - a 2N3906 under test where Vce seems to break down after 6 volts DC

Link to my report on this subject:
http://www.biophysicslab.com/2021/0...lloscope-and-arbitrary-waveform-generator-awg
 

Attachments

Audioguru again

Joined Oct 21, 2019
3,543
The transistor is not upside down. Since the supply and input signals are positive then the 2N3906 PNP transistor should be replaced by an NPN 2N3904 transistor.
 

Thread Starter

RonFredericks

Joined Nov 13, 2009
11
Thank you folks for your replies and for looking at my circuit and issue. But maybe the Occam's razor approach is not too helpful in this case.

The transistor under test is not upside down because the second 2N3906 plugs into the breadboard the same way and works fine and is from same manufacturer, and the weird trace transistor still shows a normal hFE value. The signals are correct (negative supplies coming from terminals marked positive as shown in AWG image and in LTSpice circuit signals).

From three different forums where I have posted this issue for help I got this summary that others have additionally suggested seems correct to me. The transistor is likely just bad.

Looks like the CE junction got zapped and acts as a SCR with a negative incremental resistance . Might be good for a tunnel diode oscillator at RF if biased at -7V. ;) This is a breakdown arc failure with say 40K/mm might be 40V/um so 7V is about a 175 nm gap that the E-Field bridged This PNP should be good to -40V and act like a Zener near -50V.

Definition of Occam's razor

: a scientific and philosophical rule that entities should not be multiplied unnecessarily which is interpreted as requiring that the simplest of competing theories be preferred to the more complex or that explanations of unknown phenomena be sought first in terms of known quantities
 

DickCappels

Joined Aug 21, 2008
7,741
Or to elaborate on Audioguru Again's comment - why is the emitter of a PNP transistor more negative than the collector and the base more positive than the emitter. Not sure which curve is which but the one with a "hook" in it looks like negative resistance.
 

Thread Starter

RonFredericks

Joined Nov 13, 2009
11
Or to elaborate on Audioguru Again's comment - why is the emitter of a PNP transistor more negative than the collector and the base more positive than the emitter. Not sure which curve is which but the one with a "hook" in it looks like negative resistance.
Because LTSpice and Rigol AWG function generators allow the user to reverse polarity of the signal.
In LTSpice this command for example does the "trick"

PWL REPEAT FOREVER (file=pwl_triangle_minus.txt) ENDREPEAT

In actual function generator (as shown in diagram) High value set to 0 and Low value set to -5. So ground is actually the plus side of the circuit.

Why do this? Because then NPN and PNP devices can be tested with the same circuit breadboard just by reversing polarity of the input wave forms.

Not sure why this is so hard to see? The oscilloscope probes are all showing negative voltages as well with respect to "ground". I guess those folks at Rigol, Tektronix, and Analog's LTSpice gave us this option for a reason. And now you know.

The ground symbol is the circuit's reference node which is "almost" always at 0V. In this case it is not - boom.
 
Last edited:

Thread Starter

RonFredericks

Joined Nov 13, 2009
11
Opps - Last line above. Meant to say

The ground symbol is the circuit's reference node which is "almost" always at 0V. In this case it is at 0 too. But the symbols marked as "+" in LTSpice voltages are actually now more negative than ground.
 

Thread Starter

RonFredericks

Joined Nov 13, 2009
11
As an fyi: When building this simulation then breadboard I discovered that I can't swap the voltage sources so as to read "-" to ground. I found that I could swap the power rails in LTSpice. But when I tried that with my AWG and Oscilloscope the earth ground would clobber the results as there are two voltage sources and only one ground. Neither of my lab instruments have a floating ground option to my knowledge. So I went with what now seems to be an certain amount of confusion seeing "+" on the simulated power supplies and "Red" function generator probes in the physical circuit.

Sorry if it looked like I was fighting with you earlier. Your feedback is very helpful, thank you.
 

Thread Starter

RonFredericks

Joined Nov 13, 2009
11
I have never designed, built or tested anything that has a negative power supply. I use maybe 20 times more NPN transistors than PNP transistors.
Old cars in the UK had the battery "backwards" (and also drive on the wrong side of the road).
I think many Op Amp designs use a "+ to ground" and a "- to ground" set of dual power supplies. I feel it is very common. But non-the-less I feel your pain so I have tried to improve my schematic documentation.
 

Attachments

Thread Starter

RonFredericks

Joined Nov 13, 2009
11
The schematic in the simulation still shows positive signal generators, not negative ones.
OK Audioguru Again. One more try at leveraging your experience. I made a new voltage source for LTSpice that I think may be more clear. Screenshot of the working simulation with this new symbol attached.
 

Attachments

Thread Starter

RonFredericks

Joined Nov 13, 2009
11
Ron, instead of making-up symbols, you should use the standard way of showing a negative signal or supply:
Since a signal generator is not a balanced power supply with + Gnd - terminals maybe I will never please all the folks. A signal generator can be configured to go from 0 to negative output without any positive output by nature of the signal as shown in my AWG photo at the top of this thread. I did run this issue by the LTSpice user group and came up with this these options... (Note: I started with option 4 below). I added a new photo showing options (3) and (2) combined. But my signal generator can't flip its ground so I still prefer the second photo below with"+" output actually being a negative voltage as in real life for this signal generator's actual configuration. See second photo.

If your design reviewers object to the LTspice voltage source symbol, there are at least three (maybe five) ways to handle that:

(1) Make your own voltage source symbol.

(2) Use the "signal" voltage source symbol. It also has (smaller) "+" and "-" next to the symbol's pins, but also has a bigger squiggle line, and perhaps it might suggest to them that it can be a changing voltage and not a static DC voltage with the polarity they want to think it is.

(3) Invert the voltage source symbol. Rotate it twice. Now its pin labeled "-" is up. Of course you'll need to change the parameters (PWL files) too.

(4) Show your reviewers that the voltage at the output of the voltage source is negative. Plot it and point to it.

(5) Add a Comment on your schematic next to the negative voltage wires that says, "This is a negative voltage".

I don't understand what you said about using "0" in the voltage source symbol for ground. The symbol doesn't have a ground and doesn't care if it's connected to ground or not.
 

Attachments

Last edited:

Thread Starter

RonFredericks

Joined Nov 13, 2009
11
To conclude this discussion: On this support site I asked for comment on why my new curve tracer test board produced a very weird output for one 2N3906 part - making clear that other 2N3906 transistors worked fine in this same curve tracer test jig.

What I got in return was a lot of false suggestions about transistor pins being placed wrong in my breadboard, positive voltage going to the wrong pins, and other arm chair half thought out quick responses. Other forums I participate in, include some call to action by a senior participant saying something like "User asked for help on why one part out of many giving unexpected results" or some such. That never happened.

Instead I reshared here what 2 other EE forum support boards concluded on why one transistor might produce these odd results - ie bad transistor. Yes, those discussion boards also had their fair share of wrong way answers about pins on a transistor probably not being configured correctly. But to be clear, there was always enough moderation by others to steer the conversation onto the users question. That of course never happened here.

On the other hand, I think my LTSpice simulation schematic has improved with a better symbol and thanks to AudioGuru Again for that.
 
The transistor is not upside down. Since the supply and input signals are positive then the 2N3906 PNP transistor should be replaced by an NPN 2N3904 transistor.
The simulator schematic shows polarity marks on the generator symbol which indicate true polarity when the assigned voltage is positive but in this case the assigned voltage seems to be negative so in this situation the + Marks on the generators are actually negative, consistent with testing pnp transistors. While this is consistent it is confusing and demands careful study beyond a casual glance; also it's a trap for the uninitiated. My preference is to allow the marks to be true polarity and turn the symbol upside down as required. Then there's no hidden context, a favorite software gotcha.
 
Last edited:

atferrari

Joined Jan 6, 2004
4,433
To conclude this discussion: On this support site I asked for comment on why my new curve tracer test board produced a very weird output for one 2N3906 part - making clear that other 2N3906 transistors worked fine in this same curve tracer test jig.

What I got in return was a lot of false suggestions about transistor pins being placed wrong in my breadboard, positive voltage going to the wrong pins, and other arm chair half thought out quick responses. Other forums I participate in, include some call to action by a senior participant saying something like "User asked for help on why one part out of many giving unexpected results" or some such. That never happened.

Instead I reshared here what 2 other EE forum support boards concluded on why one transistor might produce these odd results - ie bad transistor. Yes, those discussion boards also had their fair share of wrong way answers about pins on a transistor probably not being configured correctly. But to be clear, there was always enough moderation by others to steer the conversation onto the users question. That of course never happened here.

On the other hand, I think my LTSpice simulation schematic has improved with a better symbol and thanks to AudioGuru Again for that.
Missing the troopers in Stack Exchange?
 
Top