Transistor HFE - IC curve tracer?

Ok, you said: Bs = Vy = (Vo - Vi) * (-1) = Vi - Vo

Vo = Bn * Vi / Bs
Vi = Vc We look at Bs, that is sampled/sensed beta and is the same for all transistors, so Vi is a constant voltage = Vc

Vy = Vc - Bn * Vc / Bs
Vy = (1 - Bn / Bs) * Vc equal to your equation: Vy / Vx = ( 1 - Bnom / ß )

Your paper example calculations for beta 80, 100, 125 left out Vc.

Your beta curve is normalized (or something) to Bn. Bs becomes zero.

But Vy is not beta! Vy only depends on beta somehow.

So your beta curve follows beta but the actual numeric beta Bs is unknown because Vc becomes Vi which rides on the triangle, and also Bn is unknown.

IMHO from Vy = (Vo - Vi) * (-1) you can visualize the curve but not calculate actual numeric beta.
Vbeta-compensation is not beta but something like 1/beta.
Inverting Vbeta-compensation mirrors the curve visually only.
I did not know what it was, so I guessed "something like 1/beta", but it is normalized.
 
Last edited:
Vi = opamp Vin- = opamp Vin+
It is not fruitful to start a closed loop equation with this statement: the output of the opamp (U3D) contains the correction (what is sought after), because there is no openloop gain. -> Vout = Ao * [ Vin+ - Vin-]. Sure U3D wants to minimize the error, it differs Ao / [Ao + 1] , which is almost 1, but not exactly.
Your Vout is my Vz -> Vz = Ao * (Vg - Vx) .
Vi = Ic * R * gain...
There is a distiction between Vin+ and Vin- (my Vgenerator and Vx-output). Without the openloop gain Ao, it is flattened to equal (and no correction part anymore).
-> Vx = Aina * Rmeas * Ic .
... & Vo = Ib * R * gain
The gain of the controlled current source is 1, the devider in front is 1/100 ("k4"), R55 specifies the CS-current and is 100ohm (in the eq's 'R5').
-> Vz = Ao * ( Vg - (k4 * ß * (Rmeas/R5) * Aina) * Vz (this the closed loop equation, yields to...)
-> Vz / Vg = ( 1/Ao + k4 * ß * (Rm/R5) * Aina ) , the 1/Ao part nullifies.
Now one could extract ß out of this, but this is not usefull: the terms Vg and Vz (opamp output) are still in the eq.
We want the opamp output Vz, containing the original Vgenerator and the correction added to it.

Here kicks in various restrictions and revelations:
Know the expected ß (say: 100x). The Vg is 1V~ but to large for the BJT and the CS, so a devider first (k4), 1/100 (=1/ß).
The CS makes 10mV/100ohm = 100uA base current, the BJT multiplies with ß ('100') to 10mA.
This Ic multiplied with Rmeas (1.00 ohm) is 10mV (again!), the INA multiplies this 100x, Vx output is 1V~ (Vx = almost Vg)

Needed now is the development of the eq towards a Vx - Vz eq.
k4 * Aina = 1 ( 1/ß*ß=1 !!!), R5/Rm = 100 (the expected ß !!!)
Let we take R5/Rm to be the nominal (ideal) ß, naming it 'Bn' (this 'Bnominal' is constant).
-> ß * Vz = (R5/Rm) * Vg (the rest is unity)
-> Vz = (Bn / ß) *Vg
Both sides minus Vg (to remove the Vg part), but we use the Vx because Vx = Vg (!!!).
Vz - Vx = ( (Bn/ß)-1) * Vx
Then the inverter.
Vx - Vz = (1 - (Bn/ß)) * Vx = Vy output (!!!!!)
Vy related to Vx, and not to Vz anymore...
Vy / Vx = ( 1 - Bn/ß ) (Vy = funtion (Vx) )
The X-Y scope or X-Y plotter can show this (#79 picture).

If the datasheet beta is to way off from the above values (k4, Ai, R5/Rm), those need adaptations to suit the actual beta involved.
This is of course an elaborate hardware development, let alone to adapt the circuit to PNP BJT's too!
 
So I see Vi+ = Vi- as a valid assumption.
Vi+ Vi- is valid. It cannot be equal or there is no output at all.
Bn is not defined, it is unknown.
The circuit has to be adapted to an expected beta, hence the introduced Bn.
Vx is variable with time, it is unknown.
Yes and no. Any signal generator is a timed device. Beta as a function of Ic is wanted, time is irrelevant here. There is no time in the ß = f (Ic) specification.
ß is a denominator in the equation
To extract ß, it is made relative to Bn.
The correct equation would be:

Vy = ß
That would be a certain static value, not a characteristic where ß is related to Ic.
Ic is represented as a voltage Vx, ß is represented as a relative voltage Vy. The oscilloscope is used in the X-Y mode, not in the X-t & Y-t mode.
 
It seems to me that you could verify the circuit concept in simulation. Does it, or does it not, produce the correct Beta-Versus-Ice plot? Simply use a (simulated) transistor whose Beta-versus-Ice is known beforehand, then connect this transistor model to your measurement circuit. Measure how much your circuit's answer deviates from the known correct answer. Aha, validation & certification.

To name one example, the LTSPICE built-in model for the 2N2222 transistor
Code:
.model 2N2222 NPN(IS=1E-14 VAF=100 BF=200 IKF=0.3 XTB=1.5 BR=3  
+      CJC=8E-12 CJE=25E-12 TR=100E-9 TF=400E-12 ITF=1 VTF=2  
+      XTF=3 RB=10 RC=.3 RE=.2 Vceo=30 Icrating=800m mfg=NXP)
gives a beta-versus-Ic curve that decreases when Ic increases. LTSPICE simulation plot below.

Incidentally, it's a fun little five minute brain teaser, to discover ways to make LTSPICE create a plot like this. How do you get Ice on the horizontal axis? Woo hoo, plenty of enjoyment.

_
 

Attachments

  • Beta_2N2222_is_NOT_constant.png
    Beta_2N2222_is_NOT_constant.png
    13.8 KB · Views: 81
"Using_TINA.png" is an image file. The file extension .png signifies that it is of type Portable Network Graphics ... here is Wikipedia's explanation of the details of PNG files, their organization and compression.

This particular image named "Using_TINA.png" happens to be a screen capture of my laptop, when running Texas Instruments' free circuit simulator TINA. It is a plot of a (TINA simulated) NPN transistor's Beta (Ic/Ib), versus collector current (Ic). The image demonstrates that it is, in fact, possible to make the TINA program plot log(Icollector) on the horizontal axis, and (Ic/Ib) on the vertical axis. All that the circuit designer has to supply, is a little ingenuity. Solve a little brain teaser. Figure out how to get it done, rather than merely Googling "How do I get it done?", then whimpering and muttering when you can't find the answer by trivial searches.
 
It is a plot of a (TINA simulated) NPN transistor's Beta (Ic/Ib), versus collector current (Ic). The image demonstrates that it is, in fact, possible to make the TINA program plot log(Icollector) on the horizontal axis, and (Ic/Ib) on the vertical axis. All that the circuit designer has to supply, is a little ingenuity. Solve a little brain teaser.
Spend some hours and shared the effort... ehmm...
 
Here's a typical example of the difference between datasheet and circuit simulation (in this case: TINA), using .MODEL parameters.

When you get your tester working and accuracy verified, you can show examples of the difference between datasheet and measurements on actual real-life devices.

_
 

Attachments

  • datasheet_4401.png
    datasheet_4401.png
    37.8 KB · Views: 67
  • simulated.png
    simulated.png
    9.2 KB · Views: 68
Another example. Again TINA simulation is not a wonderful match to the datasheet.

However, as proposed in post #89 above, you could drop this TINA-simulated 2N5210 transistor into a TINA-simulation of your Beta-versus-Icollector hardware device. As long as your TINA-simulated results from the simulated hardware device, match the raw Beta-versus-Ic from a TINA simulation of the transistor without the hardware device . . . . you know the hardware device got the right answer. It extracted a Beta-versus-Ic curve which approx. matches how the simulator says the raw transistor behaves.

_
 

Attachments

  • datasheet.png
    datasheet.png
    38 KB · Views: 68
  • simulator.png
    simulator.png
    9.5 KB · Views: 75