Fluctuating high-pitch noise in op-amp circuit

Hi all,

So I'm currently playing with this cheap but neatly built active crossover from Breeze Audio (China cheapie).

Ever since I've started using it I started hearing fluctuating high pitch noises in the tweeters, like a whistling sound that comes and goes; it is fluctuating in level and in position, going from one channel to the other like a UFO or just in both; then disappearing. sometimes a lower frequency buzz comes along.

My speakers are 104dB horns and compression drivers, I think it would be much fainter on regular cone/dome speakers, probably barely audible.

I've been experimenting with cheap Chinese discrete op-amps wich went quickly out as they had very bad DC offset; I now use OPA2132 but with all op-amps tried I was getting these interferences.

At first I've been using this device with ultra cheap, unshielded interconnects and I thought they were the reason, but now with shielded interconnects the noise still comes and goes, although it is maybe fainter? (not sure here).

I have a wi-fi modem just a meter away from my stereo equipment; not an ideal situation but I didn't have any issue so far. There's also an office with about 6 computers just above my head. Parts of the electric circuit are shared as this is an old house.

The crossover itself is enclosed in a nice, all aluminum, thick box that should provide some protection against RFI...
I've received yesterday some bag full of cables and tweaks from a friend; there are some ferrite beads to clip on cables, various sizes, I've tried few locations but while it seemingly improved on some stuff (probably placebo!) it didn't help with the noise so far.

Is what I'm hearing really RFI or are my OPA2132 oscillating (but I've never ever had an OPA2132 oscillating...)? I could add some 0,1uF MLCC cap under each chip and see if the noise goes away... but its fluctuating nature makes me think more of interferences...
 

Attachments

  • psu2.jpg
    psu2.jpg
    520.7 KB · Views: 377
At first I've been using this device with ultra cheap, unshielded interconnects and I thought they were the reason, but now with shielded interconnects the noise still comes and goes, although it is maybe fainter? (not sure here).

I have a wi-fi modem just a meter away from my stereo equipment; not an ideal situation but I didn't have any issue so far. There's also an office with about 6 computers just above my head. Parts of the electric circuit are shared as this is an old house.

The crossover itself is enclosed in a nice, all aluminum, thick box that should provide some protection against RFI...

It does sound like you are describing RFI from the wireless.

Simple test... move the wifi somewhere else, a few meters away and see if the problem persists.

The aluminum cases on these gadgets are nice and sturdy and they look nice, but they are seldom grounded. If not grounded, they tend to act like antennas and can actually make an RFI problem worse. If you can add a ground lead from the case to the outer ring of the RCA connectors, you might solve the problem.

Also, adding low value capacitors in the 10pf to 50pf range across the audio hot lead and ground right at the RCA connectors might knock it down. At least it would stop your cables from acting as antennas.

Also make sure your input and output cables are well shielded and the shield is connected to RCA ground at both ends ... You should note that a lot of the really expensive "audiophile" cables are not.
 
It does sound like you are describing RFI from the wireless.

Simple test... move the wifi somewhere else, a few meters away and see if the problem persists.

The aluminum cases on these gadgets are nice and sturdy and they look nice, but they are seldom grounded. If not grounded, they tend to act like antennas and can actually make an RFI problem worse. If you can add a ground lead from the case to the outer ring of the RCA connectors, you might solve the problem.

Also, adding low value capacitors in the 10pf to 50pf range across the audio hot lead and ground right at the RCA connectors might knock it down. At least it would stop your cables from acting as antennas.

Also make sure your input and output cables are well shielded and the shield is connected to RCA ground at both ends ... You should note that a lot of the really expensive "audiophile" cables are not.

Actually I looked closer at the circuit and couldn't find any local ceramic decoupling capacitor close to the op-amps (the circuit came with NE5532). There is 22uF silmic and a small plastic (?) cap of unknown value on each supply rail to ground, near each op-amp, but not nearly near enough for a faster device.
So I added 0,1uF MLCC under each chip between pin 4 and pin 8 and the noise is gone, all I hear now when I put my ear very close to the tweeter is the very very faint hiss inherent to all active circuit, but nothing weird.
Looks like the OPA2132 were oscillating like crazy!

It also sounds better, more focused and quiet, less fuzzy.

The enclosure is grounded! :)
 
Actually I looked closer at the circuit and couldn't find any local ceramic decoupling capacitor close to the op-amps (the circuit came with NE5532). There is 22uF silmic and a small plastic (?) cap of unknown value on each supply rail to ground, near each op-amp, but not nearly near enough for a faster device.
So I added 0,1uF MLCC under each chip between pin 4 and pin 8 and the noise is gone, all I hear now when I put my ear very close to the tweeter is the very very faint hiss inherent to all active circuit, but nothing weird.
Looks like the OPA2132 were oscillating like crazy!

It also sounds better, more focused and quiet, less fuzzy.

The enclosure is grounded! :)


Excellent. I would still ad the low value caps on the input jacks just to play it safe. They should not affect the sound, just take RF off the input cables to ground.
 
So I added 0,1uF MLCC under each chip between pin 4 and pin 8 and the noise is gone, all I hear now when I put my ear very close to the tweeter is the very very faint hiss inherent to all active circuit, but nothing weird.
Looks like the OPA2132 were oscillating like crazy!

It also sounds better, more focused and quiet, less fuzzy.

The enclosure is grounded! :)


Its not widely appreciated enough that modern and even not so modern high performance opamps need proper decoupling with ceramic low-inductance capacitors between the two rails to behave. Lack of decoupling can lead to very subtle effects sometimes (measurably greater distortion than the datasheet without audible oscillations - hard to spot).
 
Its not widely appreciated enough that modern and even not so modern high performance opamps need proper decoupling with ceramic low-inductance capacitors between the two rails to behave. Lack of decoupling can lead to very subtle effects sometimes (measurably greater distortion than the datasheet without audible oscillations - hard to spot).

It's not all pink yet: listening to some vocal music this morning shows a subtle metallic edge that I really don't like, and it's not as silent as I'm used to from my TDM crossover.

Regarding the "metallic edge", maybe it would have been better to decouple from each supply pin to ground, instead of between supply pins? What is the best practice? I need to reopen my TDM once, I don't remember what I used for OPA2132 in there; I just know that for LM4562 I always get best results with one 0,1uF MLCC between V+ and V-, and one 0,1uF MLCC between V- and ground, with some 100uF Silmic on the rails nearby. No matter the circuit that always works best with LM4562 for me.

Yesterday I replaced two 22uF Silmic with 100uF Muse KZ, near the input buffer op-amp (the one that's the farthest from the PSU); maybe I need to let those break-in a little before judging on the sound.

Ultimately I would like to replace the main PSU caps with Kendeil or Mundorf, and all the 22uF Silmics with their 100uF version.
 
Last edited:
It does sound like you are describing RFI from the wireless.

Simple test... move the wifi somewhere else, a few meters away and see if the problem persists.

The aluminum cases on these gadgets are nice and sturdy and they look nice, but they are seldom grounded. If not grounded, they tend to act like antennas and can actually make an RFI problem worse. If you can add a ground lead from the case to the outer ring of the RCA connectors, you might solve the problem.

Also, adding low value capacitors in the 10pf to 50pf range across the audio hot lead and ground right at the RCA connectors might knock it down. At least it would stop your cables from acting as antennas.

Also make sure your input and output cables are well shielded and the shield is connected to RCA ground at both ends ... You should note that a lot of the really expensive "audiophile" cables are not.

All I have at hand is those 0,1uF MLCC caps, could I use those on the input RCA?
 
Having played quite a lot with active crossovers based on opamps the 'subtle metallic edge' could well be down to layout. When using single-ended connections the shield carries common-mode currents and unless those largely ultrasonic noise currents are kept from imposing noise voltages across 0V references, they tend to generate sibilance in bipolar opamps. LM4562 is particularly susceptible to RF noise ingress.
 
All I have at hand is those 0,1uF MLCC caps, could I use those on the input RCA?

No. The value is high enough to cut into the treble range of the audio.

You're going to want to keep those values under 100pf since all you want to do is shunt off any RF that might find it's way into your interconnects. Small ceramics are not expensive, probably about 10 cents each and they're common as rice. Almost any electronic repair shop should be able to sell you a few.
 
Okay so I've changed my decoupling scheme.
Now I've put 0,1uF MLCC from each power pin to ground - instead of between V- and V+.

But now they are paralleled with what was originally in the circuit: some 22uF (those will stay or be replaced with 100uF), but also some 0,1 (apparently) plastic (?) caps that are between each supply and ground but just a few mm farther.

I wonder if it's ok to leave those plastic 0,1uF there or if I should take them out now that I put the 0,1 MLCC closer to the power pins?
 
In any case nothing is doing anything; I still get this fluctuating high pitched noise coming and going. Noise floor between those peaks is quite hearable as well - but I could live with that.
Probably this cheap crossover is just badly conceived and there's nothing that will fix it... :( without redesigning everything wich is above my pay grade.
 
Regarding the "metallic edge", maybe it would have been better to decouple from each supply pin to ground, instead of between supply pins?


Maybe, maybe not - that can inject distortion currents into a local ground trace. HF Decoupling is between supply rails to stop oscillation modes involving unintended internal feedback via the power rails. The internals of the chip are ignorant of ground potential anyway.
 
Put 33 pf disk caps parallel the op amp feedback resistors. Pins 1 & 2 and 7 & 6 usually.
I had to do that to my 33078 when I replaced the 4558 the mixer came with. In addition to a .1 uf between supply rails for every 2 packages.
When you replace a 6 v per microsec slew rate amp with a 20 v per microsec amp, make sure the high freq gain is killed by shorting it out with the feedback cap.
I couldn't hear the oscillation, but my cs800s amp fan was running on high at 1 w/ch. Had a working scope then for 2 weeks, saw ~ 1 mhz.
BTW the 5532 is a pretty quiet part. Just a bit of a hog on power supply current. There weren't any leaded DIP packages in stock the day I was buying "low noise op amp" from newark. Paid all of $.38 ea for ST33078. NJM2068 is another forgotten low noise low cost part. If you need low noise + high drive current for long lines or high capacitance filters, Peavey used NJM4580 in the 90's.
 
Last edited:
Have you added the RF filtering ceramic caps on the inputs yet? Worth a shot

Well I did try with what I have at hand - the 0,1 mlcc - it did nothing for the noise problem

I’m gonna take them out since value is too big anyways - I wanted to see if it would cure my noise problem

It’s very frustrating and never experienced something similar... maybe something is broken? It’s a really strange noise. It’s just like something frying then it modulates higher then lower then louder then it’s just the fried noise again. I tried disconnecting nearby appliances (WiFi, tv box,...) but it stays the same. I put ferrite beads, I put a filtered IEC cable... noise is always there. Enclosure open or closed won’t change a thing. It must come from within the circuit. But how can a simple filter circuit with 5 op-amps be so difficult to deal with?..

I’m about to give up but it’s very frustrating as, if that noise wouldn’t be there, it would be more musical than my TDM crossover.
 
Last edited: