Fluctuating high-pitch noise in op-amp circuit

If you look at the picture there's a thick white line on the board; it looks like it was meant to accept some kind of shield, there are even pads...

Spot on, agreed. Wonder if the original designer expected such a shield plate to be needed -- perhaps even tested to confirm it. Then production had some piddly problem with it, submitted an ECO, and got it taken out of the design.:mad:

You might try to scrounge up a mild steel plate to fit in there -- maybe 16 gauge or so, or whatever you have. Cut it tall to fit close to the cover, mounted using those 4 holes if they're grounded, taking care not to short any top-side traces.

May not solve your problem, but it might eliminate one more possibility.

By any chance, do you have some light industrial zoning within a block or two? There are plenty of assembly and processing machines that are not at all polite in their treatment of the A/C.

Best Regards, R
 
Last edited:
Ooops, forgot 'til just a second ago -- they aren't the right sound character or timing, but:

- fluorescent light ballasts have made some nasty line hash -- esp if they're failing

- ordinary incandescent dimmers -- esp if they're 'controlling' an LED bulb

If the voluminous reading is to be believed, the engineers have some real nightmares trying to pack SMPS into the tiny space, with no power dissipation, and still respond proportionately to the triac-based dimmer.
 
AX tech editor
Joined 2002
Paid Member
Did you try to diagnose where the whistling originates? For instance, when you short the amp input, is it still there? If you use a preamp instead of the xover, is it still there? If you plug the various components in different mains outlets, is it still there. Does it change in character with the different setups?

It's no use operating the patient if you don't know what's wrong with him.

Jan
 
Spot on, agreed. Wonder if the original designer expected such a shield plate to be needed -- perhaps even tested to confirm it. Then production had some piddly problem with it, submitted an ECO, and got it taken out of the design.:mad:

You might try to scrounge up a mild steel plate to fit in there -- maybe 16 gauge or so, or whatever you have. Cut it tall to fit close to the cover, mounted using those 4 holes if they're grounded, taking care not to short any top-side traces.

May not solve your problem, but it might eliminate one more possibility.

By any chance, do you have some light industrial zoning within a block or two? There are plenty of assembly and processing machines that are not at all polite in their treatment of the A/C.

Best Regards, R

This is some cheap Chinese device sold under three different names on places such as Ebay and Aliexpress. There are some variances in price and they are all slightly different if you look at the pictures (main difference is wich capacitors are used), none has a shield, most of the pictures have seen online have ungrounded enclosures - well they don't have a heart connection at the IEC socket connected to the enclosure like mine has
There are empty spots on the PCB, there's a big pad with nothing connected marked "ground"... obviously corners have been cut and they know a lot of buyers will tweak it themselves. Despite that noise problem - inbetween two "whistle attacks" it sounds very, very good (better than my modded TDM 24-CX4) and I'd really like to make this work!

If I put a shield plate on the white line there it will have to double as heatsink for the voltage regs, because they are so very close, I better make use of it rather than risk an accidental contact...

There's no industrial zoning nearby but the level above my apartment is an office full of computers and most of those stay on 24/7...
 
Did you try to diagnose where the whistling originates? For instance, when you short the amp input, is it still there? If you use a preamp instead of the xover, is it still there? If you plug the various components in different mains outlets, is it still there. Does it change in character with the different setups?

It's no use operating the patient if you don't know what's wrong with him.

Jan

I always use a preamp; I haven't tried without crossover but with my TDM crossover I never got any noise issue.
I've tried putting 0,1uF mLCC caps between hot and cold on the RCA sockets just to see if the noise would go away; all it did was reduce high-end response (wich was to be expected as 0,1 is too big a value) but it didn't stop the noise so I removed them; I will put some smaller value because it won't hurt anyways but I doubt it will cure anything...
I've tried plugging the power strip (aluminum casing, shielded Ollflex cable) in different wall sockets it didn't help

I could still try plugging a source directly and use the x-over as a preamp but I'm almost sure it won't help
 
Last edited:
Yes if it is not self-oscillating. Plate has to be bonded so it is connected electrically to the safety grounded case. Transformer can stop RF but the wires to it from line cord have to be short so they don't act as transmitters. There are IEC sockets with RF trap in them if this has unprotected one of those to the line cord.
I tack the 33 pf caps right on top of the feedback resistor with short leads. I don't see them. 22 or 47 could help too.

I will put the caps on the feedback resistors

would a schaffner filter doubling as IEC receptacle help?
 
The brand of IEC socket that has RF filters I'm familiar with is corcol.
You need to look at the specs of schaffner and see if it is an RF trap.
But kill the gain in the RF band first with the disk caps across feedback resistors. Realize there will probably be a parts order & 3 day wait. I happened to have a dozen 33 pf disks from the Radio Shack grab bag I bought about 1974. Voltage rating 50 or greater if your op amp supplies are +-15 v.
If you have some junk equipment with a high end IEC socket, you could measure from input to output on each pin with LC meter and see if it contains 5 to 15 microhenries of inductance. Or make a test jig of a 8' power cord and use a 200 mhz scope to see what waveforms are on a short wire on the inlet side. I don't have either device. I used junk toroid chokes from AC inlet of trash TV's to kill RF coming into my disco mixer. Put them right next to the barrel connector bringing the DC in through the steel wall from the wall transformer. No fuse even.
 
Last edited:
AX tech editor
Joined 2002
Paid Member
I've tried putting 0,1uF mLCC caps between hot and cold on the RCA sockets just to see if the noise would go away; all it did was reduce high-end response (wich was to be expected as 0,1 is too big a value) but it didn't stop the noise so I removed them;

Were this the input or output RCA, on which unit?

I could still try plugging a source directly and use the x-over as a preamp but I'm almost sure it won't help

We're not looking for a solution yet, we're looking where the issue is. It's an advanced concept called diagnosing ;-)

You have a chain of stuff and at the end it whistles. Now, where in the chain does it start? That's crucial for a solution. No use 'fixing' something in the preamp if it originates in the power amp.

First run the poweramp on its own. Does it whistle? Does it whistle when you short it's inputs? Next, add a preamp. Does it whistle? Does it whistle with shorted preamp inputs?
Now replace the preamp by the xover. Does it ....etc. You get the point.
Doing it this way you home in on the problem very fast. With trial and error fixing you could easily spend a week without going anywhere.

Jan
 
Were this the input or output RCA, on which unit?



We're not looking for a solution yet, we're looking where the issue is. It's an advanced concept called diagnosing ;-)

You have a chain of stuff and at the end it whistles. Now, where in the chain does it start? That's crucial for a solution. No use 'fixing' something in the preamp if it originates in the power amp.

First run the poweramp on its own. Does it whistle? Does it whistle when you short it's inputs? Next, add a preamp. Does it whistle? Does it whistle with shorted preamp inputs?
Now replace the preamp by the xover. Does it ....etc. You get the point.
Doing it this way you home in on the problem very fast. With trial and error fixing you could easily spend a week without going anywhere.

Jan

I understand of course, the point is, I get the whistle only when this crossover is in the chain. if I substitute it with my other crossover, leaving all the rest exactly the same, there's no whistling :) so it's fair to assume the problem is in the crossover... I'll add that lowering the output potentiometers of the crossover lowers the whistle in the same proportions
 
Were this the input or output RCA, on which unit?



We're not looking for a solution yet, we're looking where the issue is. It's an advanced concept called diagnosing ;-)

You have a chain of stuff and at the end it whistles. Now, where in the chain does it start? That's crucial for a solution. No use 'fixing' something in the preamp if it originates in the power amp.

First run the poweramp on its own. Does it whistle? Does it whistle when you short it's inputs? Next, add a preamp. Does it whistle? Does it whistle with shorted preamp inputs?
Now replace the preamp by the xover. Does it ....etc. You get the point.
Doing it this way you home in on the problem very fast. With trial and error fixing you could easily spend a week without going anywhere.

Jan

when I added those ceramic caps on the RCA it was both at the input and at the outputs
 
Yes if it is not self-oscillating. Plate has to be bonded so it is connected electrically to the safety grounded case. Transformer can stop RF but the wires to it from line cord have to be short so they don't act as transmitters. There are IEC sockets with RF trap in them if this has unprotected one of those to the line cord.
I tack the 33 pf caps right on top of the feedback resistor with short leads. I don't see them. 22 or 47 could help too.

I assume the feedback resistors are in the signal path, is it ok sound wise to use 33pF ceramic there? Would 33pF polystyrene work as well, or is ceramic the only type to use?
 
No. Hope they are ceramic, wound plastic has more inductance. Inductance resists high frequencies, you want inductance of RF short capacitor at a minimum. Impedance of inductance is directly proportional to frequency. Capacitance of X7R diaelectric varies with voltage, which is between +- 15 at this point, negligible IMHO. 1 uf ceramic caps in between RCA input & op amp input, you might be able to hear X7R versus tantalums, but the X7R will last longer. 1 uf caps are 30000 times bigger than a 33 pf cap.
I'm listening to 4.7 uf COG ceramic caps in the sT120 right now, sound fine to me. Speaker distortion is way above the 3rd digit of HD where John Curl made some money on all those books.
 
Last edited:
39pF will be just fine. Inductance and dielectric won't matter as much since this is only a 'does THIS make any difference' test.

How do you get along with the people upstairs? If it's an office, they have printers/fax machines. Ask them to let you know when they're about to run a multi-page print job. (So you would at least have a stable, consistent symptom.) Or at least trade phone numbers -- maybe they wouldn't mind an occasional 'Hey, what do you have running Right Now?' call.

My bet is still on line-conducted/transmitted, vintage CPU-sourced -- but I've been wrong before.;)

Rick
 
Okay so I tried to locate the feedback resistors around the op-amps; the thing is, I don't have a schematics, and I can't recognize what is what :-( I'm not good enough

So I started with putting those caps on the RCA plugs; turns out those caps must be pretty bad because I immediately got some hum (or maybe I shorted one by spreading its legs too far apart, they seem to crumble easily - **** quality, I don't wanna use them), so I removed them.

So I'm back where I used to be. I think I will buy a Schaffner Filter and go from there...
 
AX tech editor
Joined 2002
Paid Member
I understand of course, the point is, I get the whistle only when this crossover is in the chain. if I substitute it with my other crossover, leaving all the rest exactly the same, there's no whistling :) so it's fair to assume the problem is in the crossover... I'll add that lowering the output potentiometers of the crossover lowers the whistle in the same proportions

OK, clear. Now, what happens if you short the input to the crossover, disconnect the input cables?
If it disappears, what happens if you only connect ONE input cable, L or R? Does it then whistle?

Jan
 
OK, clear. Now, what happens if you short the input to the crossover, disconnect the input cables?
If it disappears, what happens if you only connect ONE input cable, L or R? Does it then whistle?

Jan

well that is interesting: disconnecting input cables yields a total absence of noise (bare the faint hiss that's always present, a normal level of hiss). Made me realize there's also some hum when the cables are connected.

connecting the preamp straight to the high pass amplifier using the same cable yields a very slightly higher level of hiss, barely noticeable difference; that's the noise floor of the preamp coming trough probably

then why does connecting the cables to the crossover input causes noise?

I've noticed: there are two ground planes as far as I can tell: one for the PSU section of the PCB, that one is grounded to the aluminum enclosure; and one for the signal part of the PCB wich, as far as I can tell, isn't connected to the enclosure... should the signal ground plane also be connected to the enclosure?...