BJT internal emitter resistance

[EdGr]

A recent thread discussed the internal emitter resistance of BJTs and its effects on thermal stability.

Since I was curious, I fitted a BJT model with an emitter resistor to the Vbe-versus-Ic graph in the MJ15003 datasheet. A reasonable match was produced with 40 milliohms at 25C and a 0.16 milliohm/C temperature coefficient. The Vbe sensitivity to temperature was -2.1mV/C and Is was 10^-12 A.

I am curious if anyone has similar data for other power transistors.
Ed

 

[as_audio]

Thank you for this.
Using the search phrase "ring emitter transistor" you can find lots of information in this forum.

 

[EdGr]

I found a 2SC3519 datasheet with a graph of Vbe versus Ic. The internal emitter resistance is 62 milliohms at 25C with a 0.25 milliohm/C temperature coefficient. This is small compared to external resistors.
Ed

 

[as_audio]

Please explain "this is small compared to external resistors" : the internal emitter resistance
or the temperature coefficient ? This internal resistance has a temperature coefficient ?

 

[EdGr]

That's the 2SC3519 graph I found too.

Yes, the internal emitter resistance has a positive temperature coefficient.

The internal emitter resistance causes the Vbe curve to bend upwards in the semilog plot for the MJ15003. The temperature coefficient causes the 150C curve to intersect the 25C curve.
Ed

 

[wahab]

You have two devices here, the Toshiba 2SC5200 and the Sanken 2SD2560, quite different curves.

 

[EdGr]

2SD2560 is a darlington.
Ed

 

[wahab]

Right, but these two, also from Sanken, are not and the crossings are much less pronounced than the 2SC5200.

 

[EdGr]

The less-pronounced crossings are a better fit to the linear temperature coefficient.
Ed

 

[keantoken]

We generally hope that the curves are made from isothermal measurements, IE die temperature is close to the same throughout the measurement. But it is not always specified and even if an isothermal measurement is done it may not be a very good one. So I have never been confident in the validity of fitting from these curves, especially if they seem to show out of bandgap behavior.

 

[EdGr]

Some datasheets state that the measurements are made with low duty cycle pulses.
Ed