amplifying white noise?

[dentaku]

This is a weird situation. I built a crude white noise generator with two NPNs and it does what it's supposed to do. I get a quiet SHHHHHH out of it at around 1Vpeak-peak.
http://www.experimentalistsanonymous.com/ve3wwg/doku.php?id=noise_generator   I used 12V instead of 15V

I tried amplifying it a bit using a TL072 (and even an old LM358) and I get nothing out of the opamp using either an inverting or non-inverting amp. I plugged my function generator into it and it's amplifying as it should but when I try with the white noise I get nothing.
I can use the OpAmp as a buffer and it works but the instant I try and amplify the signal it just goes quiet.
I've even tried buffering the signal with one half of the opamp them amplifying it with the other half but it still doesn't work

Why would the opamp work as a buffer but not as an amplifier when you input white noise?

[Andy Watson]

Schematic?
Taking a flying guess - have you removed the D.C. component of the signal before amplifying? (i.e. have you put a capacitor between the noise source and your amplifier).

[Zero999]

Interesting circuit.

I've tried a similar thing before but found it difficult to get the maximum noise output which is reached, just above the minimum current required to turn the reverse biased transistor on,  so it operates in the avalanche region. This circuit achieves this by putting Q2 in the feedback loop of Q1 and AC coupling its emitter to 0V. As soon as Q2 starts to breakdown, Q1 turns on a bit, diverting the current round it, so it balances itself with a tiny current through Q2.

Yes, you need to AC couple the circuit to the amplifier, in order for it to work. The DC voltage on Q1's collector will be around  6.6V (Q2's breakdown, plus V_BE of Q1). Say you need a gain of 5. Without AC coupling, the amplifier will try to drive its output to 6.6*5 = 33V, plus the noise signal, which will be out side the power supply voltage range. If the DC is blocked by a capacitor, then it will just amplify the AC noise signal, not the 6.6VDC bias.

[dentaku]

The DC offset was the first thing I thought of yesterday but I never was able to get it amplified.
This morning I tried various capacitors (again) on the output of the noise generator and I eventually got it working but only after placing my scope probe in different places on the circuit and watching the trace move as the capacitors either charge or discharge then I can finally amplify the noise. If I unplug my breadboard it stops working again until I mess with it some more.

Oh well, it's not important. It was just an experiment. I can always just output much more controllable noise signals from my other computer instead of doing it with transistors

original on top amplified on the bottom

[Andy Watson]

You need to provide DC paths to correctly bias the opamp. This is probably what your scope probe was achieving.

A schematic of your circuit would help pin-point the problem.

[Zero999]

Is it running of a dual of single power supply?

If it's dual, check you've not confused 0V with -V.

If it's single, make sure there's a DC feedback around the amplifier and it's DC biased properly.

Of course post a schematic. It's difficult to help you without one.

[dentaku]

I've tried running the OpAmp using both 0v/12V and +-12V

This is what I have at the moment.
It starts working only after I connect my scope probe to the non-inverting pin #3 of the opamp for a few seconds and I see the trace come down the screen then settle. After this I can probe the output of C3 and it looks fine.
I used a 10K pot instead of the 5K resistor shown in the image so I can adjust it but otherwise I think this is exactly what I have on my breadboard.

This all started because I wanted a noise source to test out some analog synth style resonant lowpass filters I wanted to build and for some reason I got stumped on something as basic as this

[rob77]

place 1meg resistor between ground and the inverting input - that way you'll have the "zero" defined and the noise will "wigle" the input through the cap. you'r scope's 10Meg input imedance between the non-invert input and ground was the reason why it brought your opamp back from "bumping it's head". and definitely use teh symmetric supply - otherwise you'll have to create an artificial ground (voltage divider followed by a opamp buffer).

[dentaku]

place 1meg resistor between ground and the inverting input - that way you'll have the "zero" defined and the noise will "wigle" the input through the cap. you'r scope's 10Meg input imedance between the non-invert input and ground was the reason why it brought your opamp back from "bumping it's head". and definitely use teh symmetric supply - otherwise you'll have to create an artificial ground (voltage divider followed by a opamp buffer).

OK. I just put a 1.5Meg resistor from non-inverting to ground and it sprang to life 

[Zero999]

place 1meg resistor between ground and the inverting input - that way you'll have the "zero" defined and the noise will "wigle" the input through the cap. you'r scope's 10Meg input imedance between the non-invert input and ground was the reason why it brought your opamp back from "bumping it's head". and definitely use teh symmetric supply - otherwise you'll have to create an artificial ground (voltage divider followed by a opamp buffer).

OK. I just put a 1.5Meg resistor from non-inverting to ground and it sprang to life 

Good. This is one of the golden rules of using op-amps: the inputs mustn't be left floating at DC. Both inputs must be DC biased to a voltage within the common mode range.

[Dave Turner]

A photo of the actual circuit would help the gurus to comment practically.

[KJDS]

Op-amps need to have a DC path for current to flow out of their input pins to ground. Ideally it should be the same resistance on both pins otherwise the input bias current will generate an offset. It will only be a small offset, but in precision circuits that can make a critical difference.