Hello guys, I need some help here.

I'm using a high gain amplifier sourced with a 5V split regulated power supply. For some reason I don't know, I get a 41kHz signal of 1.69V amplitude at the negative power line.

Here is the circuit:

First stage is preamplifier, then comes two filters and the last one is another amplifier. This noise is getting amplified at the output.

 

What power supply do you have there? Do you have decoupling caps close to the opamps?

 

If that signal is measured on the negative power rail relative to the circuit ground then it suggests your -5V regulator is oscillating. Could you post the schematic of the power supply and regulator please. What's the construction method here, breadboard?

A useful check would be to disconnect your circuit from the power supply and test the power supply on its own, loading it with a resistor that approximates the current drawn by your circuit.

One final question, is it a switch-mode PSU?

 

Power supply noise is a good idea if you're using a switching supply. Extra capacitors is a good idea. Finding the current source that is polluting the ground line and using a "star" ground might be a help if the signal is being generated on the circuit board.

ps, what ARE you trying to amplify? Something in the audio range?

pss, 3 legged regulators can oscillate (from richard.cs)

 

Here is the source:

kubeek: C1 and C6 are 1cm (0.4 inches) from the amp.

Here is the picture:

And this is what happens if I disconnect the amplifier and place a FAN 12V 140mA using the 10V between positive and negative outputs. Even more noise, at 50kHz. Could it be a consequence of a defective LM7905?

 

I meant 100nF power supply decoupling caps close to the power supply pins of the opamps.
Anyway the datasheet for 7905 recommends a 1uF low esr cap on the output, as close to the regulator as possible. Does the psu oscillate even without the amplifier connected?

 

Yes. And those are rather expensive capacitors! You might try a couple of ceramic surface mount capacitors soldered right to the leads on the regulators.

 

You might have a short on the negative line (solder bridge or similar?). It is very odd to to have a power supply jumping like that without a major problem. I am guessing the lm7905 is fluttering on over-current protection.

What voltage is feeding the LM7905? Do you have any way to measure the current draw on your negative rail (on the input side of the LM7905). If not, put a low value resistor (1 ohm or less) in series with the input supply on the LM7905 and put the scope across the resistor to calculate current draw.

Also, could you explain that jumper block? How is it really connected and what does it do? (J1)?

 

SOLVED!

In the first place: thank you all.

The LM7905 was defective in fact. The fan test made clear it wasn't coming from the amplifier. And I remembered that LM7905 was recycled, I picked it from the box of used parts.

Just for the record the amplifier has no decoupling caps on the source line of the NE5532. I know it's not the best amplifier ever but it works as its only purpose is amplify a very small sine (around 200Hz to 2000Hz) in order to get the actual frequency.

Thank you again!

 

All the diagnosis in the world can't fix a defective part.

 

You're right. Thank god I suspected the regulator because it was recycled. It could have taken years to discover it if the regulator was brand-new but defective.

 

So you didn't increase the 0.1uF caps on the regulator outputs?

 

So you didn't increase the 0.1uF caps on the regulator outputs?

Nope, is it that bad? The capacitor's values were taken from the 7805 datasheet.

 

7905 is not the same thing as 7805, so you should read its datasheet too. It tells you to use 1uF output cap.

 

And 10uF (or 47uF) is better than 1uF.

Especially when you have imperfect layout, long traces, circuit built on stripboard etc.