NE5532 picking up HF noise?

[TransistorMan]

Recently, I started playing arround with some Sallen-Key active filters for my upcoming audio crossover project.

I breadboarded 4th order low pass and high pass filters, with X-over frequency arround 160Hz (100nF caps and 10k resistors), and I used NE5532 opamp.



Power supply is just regular transformer + rectifier + 7815/7915 supply.

Resistors used are Vishay 1% metal film resistors, so they aren't causing issues either.

When connected to my "old" amplifier (Class AB output stages, mains frequency transformer and rectifier, so absolutely no switching circuits) everything seems to work fine and quiet, just as it should be.

However, when I connect that same circuit to my new amp (IRS2092 class D output stages, IR2153 unregulated SMPS), I hear some hissing from my speaker. Not much, maybe 30-ish dB SPL (slightly quieter than standard fan). 

Now here is where things get weird - just replacing NE5532 with TL082 more or less completly eliminates that hissing,or at least "pushes" it below hearing level!

I would like to know why is this happening - I understand that TL082 has JFET inputs, and i'm assuming that is what makes this difference.

P.S.

I got same issues when using NE5532 as both non inverting amplifier, and inverting amplifier. For some reason, when I used NE5532 in differential (XLR line receiver) mode, things got quieter.

Since I designed my PCB for SMPS and Class D output stages, it is possible that it's emitting quite a bit of unwanted noise. But still, it's weird that NE5532 is more susceptible to picking up that noise than TL082.

[Audioguru]

You show a second-order filter but you say fourth-order.
The NE5532 is a dual opamp. What did you do with the second opamp? Did you properly disable it or did you leave it disconnected and oscillating?
A solderless breadboard also usually causes opamps to oscillate because of the capacitance between the rows of contacts and the wires all over the place.
The wires all over the place pickup interference.

[TransistorMan]

You show a second-order filter but you say fourth-order.
The NE5532 is a dual opamp. What did you do with the second opamp? Did you properly disable it or did you leave it disconnected and oscillating?
A solderless breadboard also usually causes opamps to oscillate because of the capacitance between the rows of contacts and the wires all over the place.
The wires all over the place pickup interference.

It's true that 2nd order filter is drawn. I stacked 2 of those in series (that's what I did with second opamp).
Could it be that for some reason NE5532 is more prone to oscillation on solderless breadboard than TL082?
I litteraly pulled NE5532 from my breadboard and put TL082 in his place. When TL082 is removed and NE5532 returned, issues return, so it's not bad joint or something.

I tried even using NE5532 as voltage follower, with non-inverting input connected to ground via 10k resistor(without any other input signal). It's enought to generate some audible noise.
When non-inverting input is left floating, noise increases.

[ruairi]

The NE5532 is a dual opamp. What did you do with the second opamp? Did you properly disable it or did you leave it disconnected and oscillating?

Listen to the Guru, this is important.

Also you don't show bypass caps on your power rails.  You do have them?

Have you verified that your "new" IRS amp actually works with a known good pre amp? 

[JS]

What values are you using? High impedsnce circuits will have less noise on jfet input opamps as current noise is much lower.

Also, as you said something about it, NE5532 is a faster opamp, might be more prone to oscillations than the TL082, try adding decoupling right between Vcc and Vee of the IC. They should go to ground on the final design but quite hard to get that right on the breadboard using short paths. If that doesn't work but changes something (less or more noise) you could try using some way of creating a low impedance path to ground for using two decoupling caps. Use a close connection point to plug the decoupling caps, connect it to ground with a short wire and ake your output from this point, not the general ground.

Other source of noise in BJT input amplifiers is the abuse of the input devices, if you happened to reverse bias them they could become noisier, so try a fresh IC after you check everithing is propperly connected.

JS

[mzzj]

Thats expected. BJT inputs are generally more sensitive to RFI than FET inputs. BJT transistor junctions rectify the low-level RFI way above op-amps expected bandwidth, resulting offset or noise in the output.  100Mhz RFI from SMPS can easily modulate the op-amp output.

http://www.analog.com/media/en/training-seminars/tutorials/MT-096.pdf
https://books.google.fi/books?id=dunqt1rt4sAC&pg=PA720&lpg=PA720&dq=bjt+rectifying+rfi&source=bl&ots=oXtOqo7IW5&sig=3p9g3ffCbfma4ORjv40AQb7Pz9k&hl=fi&sa=X&ved=2ahUKEwjUpcGzwP3cAhXMbZoKHbYQBVcQ6AEwAXoECAkQAQ#v=onepage&q=bjt%20rectifying%20rfi&f=false

[Zero999]

Thats expected. BJT inputs are generally more sensitive to RFI than FET inputs. BJT transistor junctions rectify the low-level RFI way above op-amps expected bandwidth, resulting offset or noise in the output.  100Mhz RFI from SMPS can easily modulate the op-amp output.

http://www.analog.com/media/en/training-seminars/tutorials/MT-096.pdf
https://books.google.fi/books?id=dunqt1rt4sAC&pg=PA720&lpg=PA720&dq=bjt+rectifying+rfi&source=bl&ots=oXtOqo7IW5&sig=3p9g3ffCbfma4ORjv40AQb7Pz9k&hl=fi&sa=X&ved=2ahUKEwjUpcGzwP3cAhXMbZoKHbYQBVcQ6AEwAXoECAkQAQ#v=onepage&q=bjt%20rectifying%20rfi&f=false

I agree.

Try connecting an RFI filter between the op-amp's output and the amplifier. A simple RC T filter should do the job. First try 1nF for C and 100Ohms for R.

[schmitt trigger]

Many good suggestions here, but definitively start with Audiguru's first.

[TransistorMan]

Thanks for all advices

Many good suggestions here, but definitively start with Audiguru's first.

As I said, second "gate" is used (one op-amp IC is used as 2 2nd order filters,connected in series to create a 4th order filter)

The NE5532 is a dual opamp. What did you do with the second opamp? Did you properly disable it or did you leave it disconnected and oscillating?

Listen to the Guru, this is important.

Also you don't show bypass caps on your power rails.  You do have them?

Have you verified that your "new" IRS amp actually works with a known good pre amp? 

"new" amplifier is dead silent without preamp connected, or when TL082 is used in circuit instead of NE5532.
Bypass capacitors, sadly, aren't helpfull in this case.
I tried 10uF, 1uF and 0,1uF capacitors across power rails, and  I haven't noticed any noise reduction.

What values are you using? High impedsnce circuits will have less noise on jfet input opamps as current noise is much lower.

Also, as you said something about it, NE5532 is a faster opamp, might be more prone to oscillations than the TL082, try adding decoupling right between Vcc and Vee of the IC. They should go to ground on the final design but quite hard to get that right on the breadboard using short paths. If that doesn't work but changes something (less or more noise) you could try using some way of creating a low impedance path to ground for using two decoupling caps. Use a close connection point to plug the decoupling caps, connect it to ground with a short wire and ake your output from this point, not the general ground.

Other source of noise in BJT input amplifiers is the abuse of the input devices, if you happened to reverse bias them they could become noisier, so try a fresh IC after you check everithing is propperly connected.

JS

R is 10k, and C is 100nF in my filter.
I tested my "old stock" NE5532, as well as my "new" NE5532 (I bought them few weeks ago, from reputable store), and they are equaly noisy.

Thats expected. BJT inputs are generally more sensitive to RFI than FET inputs. BJT transistor junctions rectify the low-level RFI way above op-amps expected bandwidth, resulting offset or noise in the output.  100Mhz RFI from SMPS can easily modulate the op-amp output.

http://www.analog.com/media/en/training-seminars/tutorials/MT-096.pdf
https://books.google.fi/books?id=dunqt1rt4sAC&pg=PA720&lpg=PA720&dq=bjt+rectifying+rfi&source=bl&ots=oXtOqo7IW5&sig=3p9g3ffCbfma4ORjv40AQb7Pz9k&hl=fi&sa=X&ved=2ahUKEwjUpcGzwP3cAhXMbZoKHbYQBVcQ6AEwAXoECAkQAQ#v=onepage&q=bjt%20rectifying%20rfi&f=false

So far, this seems to be the main issue. My new aplifier is spreading too much RFI, and NE5532 being  BJT is just more prone to picking them up than TL082.



Now here is something unexpected - when I connected T filter (R=100 ohms, C=1nF) on the output, noise got reduced, but not at the filter output, instead it got reduced on the filter's input? (modified schematic is attached below)

UPDATE:

When I short RCA jacks grounds together on my amplifier, noise is eliminated, or at least attenuated below hearing treshold!
I guess that for some reason lack of that connection helps some RF noise leak from amplifier, and NE5532 picks it up

[David Hess]

Proper layout is important to avoid ground loops and weirdness no matter how much decoupling and filtering is used but most schematics are not drawn to show this.  Once the class-D switching noise gets into a ground loop, it will be everywhere.

Rectification can not only occur at the operational amplifier inputs but also through the power supply and output connections.  They all need to be protected.

You may need to use *real* decoupling with RLC networks instead of just capacitors including at the class-D amplifier but it would be better to first ameliorate any loop problems in the layout.  Check out Analog Devices application notes 202, 345, and 347 for some ideas.

[bson]

Keep in mind some transformers can produce significant fields.

[Wolfgang]

Hi,

can you exclude that the noise comes from your power supply. NE5532 is quite fast, so this could be a good reason. Maybe some filtering and deblocking can help ?