I designed this circuit to be an analog white noise generator, but when breadboarded, it exhibits strange behavior. Reverse-biased emitter-base junction of Q2, whose breakdown voltage is 5 volts, is the source of broadband noise, and its noise current is the signal input to the base of Q1, a common-emitter amplifier, and the noise output is taken from the top of R2.

The problem is, the voltage measured at the base of Q1 is 5 volts, at the emitter of Q1 is 10 volts, and at the collector of Q1 is 0 volts. It appears that 5 volts across the base-emitter junction of Q1, in the forward direction, is not enough to make it conduct! I thought I might be burning out the base-emitter junction of Q1 with the charging current for C1, so I added R3 to limit the inrush current to a few milliamperes, replaced both transistors, but the odd behavior persists. I tried increasing the value of R3 to 10k, again with new transistors, but the odd behavior still persists. Why does this circuit not work as expected?

 

Welcome to AAC!

Q1 collector at 0V and emitter at 10V means the PNP transistor is not conducting.
5V across the base-emitter junction means something is wrong. It only takes 0.7V to turn on the transistor.
Either you have the leads wrong or the transistors are bad.
What are the voltages reported by your simulator?
I think R3 is too low. Try 470kΩ with fresh transistors.

 

Welcome to AAC!

Q1 collector at 0V and emitter at 10V means the PNP transistor is not conducting.
5V across the base-emitter junction means something is wrong. It only takes 0.7V to turn on the transistor.
Either you have the leads wrong or the transistors are bad.
What are the voltages reported by your simulator?
I think R3 is too low. Try 470kΩ with fresh transistors.

Of course Q1 is not conducting. I know it takes about 0.7 volts to forward bias a transistor. That is why the circuit behavior is strange. I checked all the connections carefully, and they are correct. I tried 10k for R3, and that should limit the base current to 500 microamperes, far below what it would take to burn out the junction.

As for simulation, LTSpice does not model the reverse breakdown of the emitter-base junction of the 2N3906, so for Q2 I substituted a voltage source of 5V DC with a 10mV AC signal riding on it, and that simulation worked perfectly, and Q1 provided good amplification. The strange behavior occurs when I breadboard the circuit. On the breadboard, I tried substituting a 100k resistor for Q2, and I got the expected 0.7 volt drop across the base-emitter junction of Q1, and the expected voltage drops across R1 and R2. That verified that I had it wired correctly. But then when I put Q2 back into the circuit, the strange behavior returned. Anyone who doubts this should breadboard the circuit and see what happens.

 

I can look at it tomorrow. Meanwhile, here is your basic white noise generator. This one uses NPN transistors.
Note that the signal is AC coupled to the amplifier stage.

 

The schematic I uploaded should show 10k as the value of R3, which I changed it to from the original value.

 

You burnt the base-emitter of Q1 before you added R3.
The moment you turned on the 10V, C1 was discharged causing the emitter of Q1 to be at +10V but its base was held at +5V or +6V by the zenering emitter-base of Q2. Then the base-emitter of Q1 had unlimited current. Also, Q2 might be burnt.

 

You burnt the base-emitter of Q1 before you added R3.
The moment you turned on the 10V, C1 was discharged causing the emitter of Q1 to be at +10V but its base was held at +5V or +6V by the zenering emitter-base of Q2. Then the base-emitter of Q1 had unlimited current. Also, Q2 might be burnt.

Please go back and re-read my original post. I wrote "I thought I might be burning out the base-emitter junction of Q1 with the charging current for C1, so I added R3 to limit the inrush current to a few milliamperes, replaced both transistors, but the odd behavior persists."

 

I discovered that the actual emitter-base reverse breakdown voltage of the 2N3906 PNP transistor is not the expected 5 volts but greater than the 10 volts of the supply. I also discovered that the emitter-base reverse breakdown voltage of the 2N3904 NPN transistor is about 8.2 volts, making it usable in this new circuit which utilizes that transitor. Noise output of about 600 mV p-p is taken from the collector of Q2. Gain could be increased by bypassing R1 with a 22 uF capacitor, and that would necessitate a resistor of several kilohms connecting the bases of Q1 and Q2, to limit the inrush current charging the capactor. However, the 600 mV noise signal is adequate for my application.

 

Although the above schematic shows a 10 volt supply, I used a 9 volt battery to power the breadboarded circuit. The DC operating point at the collector of Q2 is 2.45 volts. The values of R1 and/or R2 could be modified to work with a 10 volt or other supply.

 

The problem is, the voltage measured at the base of Q1 is 5 volts, at the emitter of Q1 is 10 volts, and at the collector of Q1 is 0 volts.

I tested that circuit and read 8.2 volts on the base of Q1 and 0 on the collector with 12 volts.
All the transistor noise generators I've seen use NPN transistors and require at least 12 volts.
I tested this design using the 2N3904 with excellent results at 12 volts.
I'll try your circuit in post #8.

 

I'll try your circuit in post #8.

I didn't get the same results. No noise until the supply was at least 11.4 volts. Tried several 2N3904 with the same results

 

I didn't get the same results. No noise until the supply was at least 11.4 volts. Tried several 2N3904 with the same results

ok, my circuit in post #8 is not a universal circuit, just something that works for my application with a specific set of parts. It requires the emitter-base breakdown voltage of Q1 to be at least 1.5 volts less than the supply voltage even to work. And, the DC operating point is a function of both breakdown voltage and supply voltage.

 

Transistors are not made to be zener diodes, the avalanche breakdown is accidental then its voltage is not controlled.
Since the emitter-base is tiny and fragile, heating it with avalanche-breakdown permanently reduces its hFE when it is used later as an ordinary transistor.

 

Transistors are not made to be zener diodes, the avalanche breakdown is accidental then its voltage is not controlled.
Since the emitter-base is tiny and fragile, heating it with avalanche-breakdown permanently reduces its hFE when it is used later as an ordinary transistor.

That's good to know.

 

I didn't get the same results. No noise until the supply was at least 11.4 volts. Tried several 2N3904 with the same results

Check on the source of your transistors. There are hundreds of thousands if not millions of popular transistors sold under popular part numbers, that in reality are not even similar to the advertised transistors past their TO-92 packages. I know, I have a couple hundred on my desk.