transistor testing question

[p.larner]

I have one of the chinese atmegga328 lcd based component testers,it works ok but if i try to measure darlingtons it reads wrong,for example measuring a tip-121 it gives a hfe of 48 and a tip 122 gives 50,it should read nearer 1000 beta/gain,any ideas,ie is that normal?.

[golden_labels]

Without knowing, what exactly the tester does: no.

But my guess would be: it’s not capable of providing voltage high enough to cross threshold of two⁽¹⁾ base-emitter junctions.

⁽¹⁾ See floobydust response below and my response. The situation is a bit more complex.

[aliarifat794]

The tester may not be properly calibrated for measuring high-gain transistors, leading to erroneous results.

[Ian.M]

Notice anything?

[madires]

The original OSHW firmwares (k and m) would show the base-emitter resistor (the sum in this case). The modified firmware versions of the Chinese clones lack some (a lot?) of functionality in favour of a more fancy display output.

[floobydust]

I grabbed some from the junkbox:
TIP122/TIP110/TIP102: E-B Fluke 87 reads 0.76V (forward) or 2.3V (reverse), resistance 6kΩ.
I think the last E-B junction is effectively shorted by the 70-120Ω internal resistor and cannot be seen.

GM328 reads like a normal NPN w/C-E diode, not a Darlington h_FE 35-44. V_BE 1.17V
Markus Frejek and Karl-Heinz Kübbeler I made reads NPN U_f 1.3V but h_FE=0.

[golden_labels]

Good observation. However…

The base resistor would draw 1–10 mA, but in normal operation that is only a fraction of the output transistor’s base current. The purpose of a Darlington pair isn’t heating.

With voltage only across the base-emitter junctions and small current, there isn’t enough juice for the output transistor’s BE junction to ever start conducting. This is normally supplied by the input transistor and for that all three inputs of the Darlington pair must be powered. So the diode test alone isn’t going to help much here.

I still stand by the suspicion, that the tester can’t provide enough voltage to the base.

Also note that you may be unable to see the full advertised gain on a tester. TIP121 is optimised to work in the 4–5 A collector current range. Below that the gain drops quickly. Not quickly enough to explain your readings, but you’re unlikely to see the rated 1000–2500 gain if running at 100 mA.

[Ian.M]

'328 component testers generally drive the test pins via 680 ohm and 470K resistors. That means, with a 5V supply the maximum Ic that can be provided and measured is about 7mA, with the emitter pin directly driven low. However the 470K resistor cant provide enough base drive for a power transistor in common emitter, so I suspect the test firmware configures power transistors as emitter followers for h_FE measurement.  The max. Ie will be under 7mA due to Vbe drop reducing the voltage across the 680 ohm resistor, and assuming Vbe drop of around 1.2V for a Darlington, Ie will be around 5.5mA.  The ST TIP120-TIP127 datasheet fig.3 shows h_FE of a bit over 60 @10mA Ic & 25°C.  10mA Ic is the left edge of the plot but if you extrapolate the straightish line downwards to 5mA the tester's low h_FE results look reasonable.

[madires]

With a current m-firmware:
TIP130: NPN, h_FE 19, freewheeling diode, R_BE 7.723 kOhms, V_BE 2.299 V
TIP135: PNP, h_FE 26, freewheeling diode, R_BE 4.789 kOhms, V_BE 904 mV

Is it sufficiently obvious that h_FE and V_BE need to be taken with a grain of salt when there's an R_BE? Or would it better to suppress the output of h_FE and V_BE in case of an R_BE?

BTW, the firmware measures h_FE in common emitter and common collector setup and keeps the higher h_FE value.

[Terry Bites]

Internal Resistors in the package!

[floobydust]

I breadboarded a TIP120 using the component tester 680Ω and 470kΩ values, off 5V. Because things can go different than speculation.
With 470kΩ to the base, of course nothing happens due to the internal E-B resistors, I got V_BE=60mV and the transistor is off.
With 680Ω to the base, 5.4mA supplied V_BE 1.30V and the Darlington goes saturated V_CE=0.61V I_C=6.5mA with R_C=680Ω
I'm not sure how hFE is calculated by the testers but lack of an intermediate resistor value seems to make the tester not work in this case.

[Ian.M]

Configure it as an emitter follower with 680 ohm base and emitter resistors.  The ADC measures the voltage drop across each to calculate the ratio of base to emitter currents.  A small fudge factor should be added to the resistances in the calculation to compensate for the non-linear and level dependent GPIO output resistance.

[floobydust]

OK on the breadboard as an EF, for the science, base 680Ω to +5V, gives 0.1V drop and 4.90V on the base, for I_B=0.147mA. V_BE=1.17V
Emitter is 3.73V w/680Ω for I_C=5.5mA giving h_FE=36.3 38.3 edit: corrected h_FE calc forgot I_E=I_C+I_B for EF.

(I use 27Ω for the '328 outputs when sourcing)

[Ian.M]

... and now your results are 'in the ballpark', which is as good as can be expected, considering its being operated outside the range of the datasheet graphs, and the limitations of the component tester's ADC, which cant measure that 0.1V base resistor drop very accurately, as 0.1V is only 20 counts. Note the ATmega328P's ADC is specced as 0.5 LSB Integral Non-Linearity and +/- 2 LSB Absolute Accuracy, so its measured h_FE could be off by a bit over 12% (2 counts + half a count uncertainty)