# 2N2222 experiment is indicating incorrect gains

Okay, so I put a 20kΩ resistor between a 5V supply and the base of a 2N2222 with the collector connected directly to the 5V rail and emitter to ground. What I'd expect is that, roughly, 220μA (i.e. (5V-0.6V)/20kΩ) goes into the base and less than 22mA (i.e. 220μA*100) comes out of the emitter. This isn't what happens at all. I tend to get over 45mA coming out of the emitter.

First, I thought it was my resistor precision. I checked and the actual resistor value is 20.2kΩ (0.1Ω resolution).

I thought it might be the specific transistor but I tried two other 2N2222's. The first was from Mouser and the second two from China. All give roughly the same result.

I then thought maybe it was because I'm using a noisy buck converter. I switched to a linear regulator that gives 4.999V stable for several volts over the +1.6V.

I checked the circuit in CircuitLab using the exact values of my configuration and it gives me what you'd expect: 215.0μA in to the base and 23.17mA out of the emitter.

What's going on? Is this happening because I have no load? Is it because I'm doing this on a breadbord? Do I need capacative decoupling/bypass? Or do I seriously have 3 broken/fake transistors where one of them came from Mouser? If so, why are they all giving the same output?

simulate this circuit – Schematic created using CircuitLab

• Which part of "somewhere between 35 and 300" doesn't match your observations? Oct 27, 2016 at 18:08
• Where did you get the impression that a real physical 2N2222's gain is precisely and repeatably 100? Oct 27, 2016 at 18:09
• Congratulations, Anthony, you're measuring the actual gain (under these conditions) of each individual 2N2222. Note that the datasheet only specifies a minimum gain, and the conditions under which it has been tested. Oct 27, 2016 at 18:14
• @brhans : from simulation ;-)
– user16324
Oct 27, 2016 at 18:14
• @Kaz: Monte Carlo analyses like the one you're describing are pretty easy to set up in LTSpice: k6jca.blogspot.ca/2012/07/…. The default distribution is flat, but the comments on that blog post describe setting it up for Gaussian. Oct 27, 2016 at 22:14