4
\$\begingroup\$

It's common sense that the current gain β (hFE) in a bipolar junction transistor is somewhat unpredictable, and one must design a discrete transistor circuit to be insensitive to the variation of β.

In most textbooks and tutorials, the common explanation is something similar to the following: The transistors are not coming from the same batch, so there will be variations in the manufacturing process. Some variations also exist within the same batch. If one wants matched parts, one should purchase hFE-grouped transistors. Better matched parts are selected from the same wafer, and the best possible matching is possible when transistors are on the same silicon die, i.e. an IC.

But none explained what physical mechanism is responsible for the manufacturing variations, possibly because the question is out-of-scope of an applied electronics book. Question: What precisely is the physical mechanism for β/hFE variations in manufactured bipolar transistors? Is β/hFE an parameter that is intrinsically difficult to control?

\$\endgroup\$
5
  • 2
    \$\begingroup\$ base-emitter doping is the biggest factor for hFE which must be at least 10x more than collector-base doping. There are many process variables that affect this. \$\endgroup\$ Commented Feb 28, 2020 at 6:41
  • \$\begingroup\$ @TonyStewartEE75 -- Would Laser-trimming be a possible way to produce transistors having a particular beta with greater accuracy? For learning electronics, and building our own audio amplifiers, for instance, tighter beta control would allow more experimentation that might even produce more ic designs, as many circuits would be prototypable in discrete technology. I have an ic that I want to produce. \$\endgroup\$ Commented Aug 19, 2021 at 23:03
  • \$\begingroup\$ No that’s not doable. @MicroservicesOnDDD better designs use current mirrors and resistor ratios with feedback \$\endgroup\$ Commented Aug 20, 2021 at 0:11
  • \$\begingroup\$ @TonyStewartEE75 -- The biggest reason in my eyes that "No that's not doable" is because the chip companies had to sell chips. Unless you provide hard evidence, I will assume you are providing a mantra handed down from one high priest to another. You're a physics guy, and I would like some physics reasons, please... \$\endgroup\$ Commented Aug 20, 2021 at 15:30
  • 1
    \$\begingroup\$ If Rohm makes the tightest hFE ranges by bins of all suppliers and they can’t do it, assume the explanation is beyond our experience to understand. @MicroservicesOnDDD The same is true for Vt, Rdson, Vf, Rce which all have wide tolerances, unless they binned from a single wafer. It’s a 3D surface process and you are proposing a 1D solution that would destroy the substrate. \$\endgroup\$ Commented Aug 20, 2021 at 16:00

1 Answer 1

3
\$\begingroup\$

beta depends on width of the base region which, along with doping levels within the base, determines the fraction of re-combination within that base region.

thus you have 3 doping levels and temperature as effects.

additionally the Vcb affects the width of collector depletion region which affect the effective base region; thus current mirrors can seriously mismatch, what with one transistor biased as a diode and the other in normal common-emitter mode

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.