Transistor HFE - IC curve tracer?

schematic

beta11.jpg
 
Some remarks:

The divider R35-R34 (20Ω/10020Ω = 1/501x which is approx the beta of the bjt (Q3) involved (that was my initial 'Bnominal', alike the cosmic constant).
So this sets the expected beta to be measured and should be adjusted in advance to keep the results visible on the scope (or whatever).

V4 is the fixed Vce voltage source (5Vdc in this circuit), to be adjusted also to suit the measurements.

R5 is the current source control resistor, you add another measurement resistor (R22 1k0). Why not use R5? Change R23 for higher amplification.

With your use of the AD633 in this configuration, the output of the OP27 (with R20/R21 on the minus node) is -10 * Vic / Vib (Vic being the 'X'-output, Vib being the INA111 (with R23) output. The original generator sawtooth is zeroed out by the divider. The output of the AD633 is the 'X'-output inverted.
The output inverter (R24/R25) changes the minus to a positive value: Beta = {V-} Ic / {V-} Ib !!!

Now the last challenge: a log generator!
I think of a diode-resistor ladder circuit consisting of a few diodes and low value resistors, close to the dynamic resistance (impedance) of the diodes to have the anode-cathode voltage drop less steep and better curved according to the log scale. When linear driven, its output current should also become somewhat logarithmic. This current can be measured with another INA (again!) and be used as the generator voltage to drive the beta-tracer circuit.
With two to four diode-resistor sections, a log-span of at least two decades (0.1 - 10.0 mA) should be possible, maybe three (0.1 - 100 mA) or four.
 
The divider R35-R34 (20Ω/10020Ω = 1/501x which is approx the beta of the bjt (Q3) involved (that was my initial 'Bnominal', alike the cosmic constant).
So this sets the expected beta to be measured and should be adjusted in advance to keep the results visible on the scope (or whatever).
I chose 20 ohms so that the driving opamp output is around 5V for a 100 beta transistor. 10 ohms would be 10V and low beta would cause the output to hit the V+ rail.
 
With your use of the AD633 in this configuration, the output of the OP27 (with R20/R21 on the minus node) is -10 * Vic / Vib (Vic being the 'X'-output, Vib being the INA111 (with R23) output. The original generator sawtooth is zeroed out by the divider. The output of the AD633 is the 'X'-output inverted.
The output inverter (R24/R25) changes the minus to a positive value: Beta = {V-} Ic / {V-} Ib !!!
Correct
 
Yes, LOG100 ceramic gold package.

Today I added the onboard sawtooth generator.
Still artefacts around zero mA.
Need adjust hf compensation.
But it looks much better now.
It is becoming an instrument.

For looking at the curve it is best to ac couple Y, so sensitivity can be switched without readjusting the beta trace.
With DC coupling Y and high sensitivity, transistors can be tightly matched because a small difference in beta will shift the curve.

New samples with high sensitivity:
The LS318 is a supermatched dual transistor for very low currents., gain falling after 1mA.
BC237 is wrong, it is BC238.
Span should be 2mA-20mA
The blue pot is for sawtooth DC offset.

beta13.jpg



beta12.jpg
board.jpg
 
Last edited:
All looks great though I would prefer a 'DC-view'. Maybe when a log-saw can be produced. I'll try to draw my simple log-gen (#125).
How accurate should the x-axis be? If it spans more decades it counts for all measurements equally, even if not perfect log scaled.
Is "EX 3 C 37' a BC337?
 
Good idea to increase the graticule illumination, that will help when you're comparing (Ib, Ic) data from your instrument, against (Ib, Ic) data from other test equipment. Testing the exact same transistor on (a) your instrument, and also (b) other gear. Thus helps validate that your new instrument gives the correct answer, namely the same answer as other, proven and trusted, test gear.
 
All looks great though I would prefer a 'DC-view'
Practical there is no such thing as a DC view. If you use a fixed setting such that both low and high betas are displayed on the screen, the difference between a flat and a non flat curve is hard to see.
You must zoom in, then Y offset needs be adjusted on the scope.
With ac you can just zoom in on the curve until it is clearly shown for example where is the top or when it starts falling.