Strange floating base NPN behavior

I'm learning about NPN transistors and have wired up this:

I understand some of its behavior, but one is a complete mystery for me.

First, if I tie "B" (green dangling wire) to "G" (blue dangling wire) then the LED turns off, as expected.

If I let "B" float, the LED glows faintly (leakage current?). If I touch "B" the LED turns on brightly, it intensifies even if I just wave my hand near it. Again, expected, that's why I built it in the first place, to try this.

What is a complete mystery for me, is that if I touch "G" (just touch it, not hook "B" to "G"), the LED turns off. Same thing if I hook up "G" t o the ground from a power outlet. I don't understand this, I didn't expect it to affect anything. Again, not "B", I (sorta) understand that this setup is very sensitive to any signal on the base of Q1, but "G", which is the emitter of Q2 / negative of the power source.

Note, the "battery" on the schematics is a standard USB charger. I want to try this with a 9V battery, but didn't get to yet.

So, my main question is why/how touching "G" affects anything? Why does the LED turns off? More generally, I'd like to understand how to properly think about these things…

Update: I experimented more with it. Changed it to a proper Darlington configuration (which didn't change anything). And tried it with a 9V battery instead of a USB charger as a power source. This completely eliminated the mysterious behavior.

My conclusion is that the USB charger doesn't provide a pure DC power, and the 50 Hz noise gets greatly amplified by this circuit. And my body with its capacitance (and just being a big bulky antenna) can affect it in various ways...

• You should definitely try this again with a 9 volt battery because that will likely give you different results and an easier route to insight. – Andy aka Jan 8 at 13:01
• @Andyaka Thanks, I will try it tonight when I get home! – klao Jan 8 at 15:27
• Is there a reason you drive the transistor with another transistor? Its gain is more than sufficient for a 20mA led and if you drive it with common logic levels. In that case, the bjts would need a base resistor! – Sim Son Jan 8 at 16:27
• @SimSon Basically, I just wanted to experiment with the Darlington configuration as I just learned about it. I didn't quite manage to do it as you can see. :D About base resistor; these transistors are pre-biased, they do have a 4.7k resistors on their bases. – klao Jan 8 at 16:54

You have wired these two transistors in what is very close to a "Darlington" configuration. (A couple of comments correctly pointed out that this is not strictly Darlington, but it is close.)

In this mode you have made a VERY high-gain amplifier since the gain is the product of the individual transistors. In the case of the UNR4216 you have this:

hFE is the forward current gain of the transistor. So you have a gain with the pair of these in the range of:

160 * 160 -> 460 * 460 or about 25k - 200K

So any input on "B" is going to be amplified significantly, even normally minor things like leakage current.

When you ground (i.e. hook "B" to "G") you are forcing the transistors to the OFF state. They are effectively open circuits and your LED is OFF.

When you "float" the "B" pin, you basically have an antenna that picks up whatever is around it and does what it does. In this case it partially turns on Q2 and some current flows and your LED flickers as the picked-up signal on "B" fluctuates.

If you were to tie your "B" to the 5V source, through a currently limiting resistor, say 10K Ohm, then you will fully turn both Q1 and Q2 ON and the LED will turn on fully.

BTW, you should also put a current limiting resistor on the collector of Q1 to avoid damaging it or Q2. See the datasheet on the UNR4216 for maximum current ratings for Ibe (Base-Emitter Current) and other relevant parameters.

• I agree, it's not exactly a Darlington but it does come close enough to have the same behavior. – jwh20 Jan 8 at 13:16
• Thanks a lot for the detailed explanations! – klao Jan 8 at 15:28
• I'm aware that this is "almost" Darlington, I actually wanted to create a Darlington and only realized that this is not exactly that when I was drawing the schematics from the photo. :) But I tagged the question as darlington. – klao Jan 8 at 15:30
• Do you have any insight into why touching "G" (or connecting it to actual ground) changes the behavior in this way? (And sorry for spamming the comments. Is there a way to have multiple paragraphs in one comment? :D) – klao Jan 8 at 15:33
• Have you ever touched the input to an audio amplifier? You get an audible ‘buzz’ from the speakers. Just ask any electric guitarist. Your body is an antenna and the circuit is very sensitive, – Kartman Jan 8 at 15:45