Noise elimination by coiling shielded cable?

[0bs3rv3r]

I had a strange problem where noise was getting into my preamp when a particular device (raspberry pi/DAC) was plugged into the preamp (via RCA interconnects). The noise was affecting the phono input channel while lisening to records with the Pi still plugged into the unused input. The device runs from an ungrounded power supply, so it should not have been a ground loop. I posted here https://www.audiosciencereview.com/...usb-dac-noise-affecting-phono-playback.10673/ about it but no replies as yet. I found the noise was reduced when I grabbed the shielded cable with my hand. I tried multiple different RCA cables. Strangely one of the quietest ones was the cheapest one. I took a good quality cable and rolled it into a coil (about 4" diameter and maybe 4 turns). This reduced the noise a LOT.

Hence my more general question here. Theory I have read, is that coiling a shielded cable does not raise the inductance of the inner conductor, but it does raise the inductance of the shield. Could it be the noise is some sort of common mode noise - hence the improvement when coiled? I wonder if a ferrite clamped on the cable might help even more?

 

[KSTR]

Looks like your preamp is having a problem when hit by large/spikey common-mode balancing currents which may actually flow *across* the PCB (which is a layout fault, in the end), producing voltage differentials on the "GND" which get picked up by the high-gain phono stage, which is what you have figured out correctly.
And yes, increasing the common-mode impedance of the interconnect coming from the D10 will reduce the ill-effect, and clamping a lot of ferrites might help big time. You could also try this on the USB feed to the D10 (together with using a long USB cable), it should have the same net effect (the balancing current must flow through the D10 as it is bus powered and the only connections are USB and line-outs). The source of the balancing current quite probably is the Pi's supply, a SMPS wall-wart? (I assume the HardDisk also is bus-powered here, so no other supply sources). Ungrounded supplies doesn't mean there can't be ground loops, 2-prong SMPS have lots of coupling capacitance to the mains.

 

[0bs3rv3r]

Looks like your preamp is having a problem when hit by large/spikey common-mode balancing currents which may actually flow *across* the PCB (which is a layout fault, in the end), producing voltage differentials on the "GND" which get picked up by the high-gain phono stage, which is what you have figured out correctly.

Thanks. Although, this is a point to point wired tube preamp. Nevertheless, the idea is valid, and I may be able to rearrange the grounding/layout to improve the preamp's rejection of this noise.

And yes, increasing the common-mode impedance of the interconnect coming from the D10 will reduce the ill-effect, and clamping a lot of ferrites might help big time.

Cheap to try. Will do.

You could also try this on the USB feed to the D10 (together with using a long USB cable), it should have the same net effect (the balancing current must flow through the D10 as it is bus powered and the only connections are USB and line-outs). The source of the balancing current quite probably is the Pi's supply, a SMPS wall-wart? (I assume the HardDisk also is bus-powered here, so no other supply sources). Ungrounded supplies doesn't mean there can't be ground loops, 2-prong SMPS have lots of coupling capacitance to the mains.

Long USB cable? Now that's interesting. Definitely worth a try, as well as a ferrite.

The Pi is the latest Pi 4 with the raspberry supplied wall wart, yes. Yes, the HD is also powered from the USB. At one point I even thought having the HD plugged in improved the noise problem.

Interesting point about capacitive coupling to the mains ground. I will keep that in mind.

 

[KSTR]

Interesting point about capacitive coupling to the mains ground. I will keep that in mind.

Another point, besides the sheer coupling capacitance values, is the points it is connected to... which happens to be between the (rectified, sometimes) mains and the output ground, bridging the SMPS's transformer. Typically we see someting like 2.2nF... required for a typical 2-prong SMPS to pass EMC. See below.

 

[Speedskater]

Thoughts:
a] Most likely it was some unnoticed, uncontrolled variable that changed the noise level.
b] Correctly chosen and placed ferrites can reduce high frequency noise. High frequency is the situation is above 100 kHz.
c] Anything that increases the end-to-end impedance of a RCA interconnect, will ingress the low frequency 'Common Impedance Coupling' noise.

 

[KSTR]

re b) Sometimes the high frequency burst noise can get demodulated and produces aritfacts in the audio range. Tube amp stages can be very prone to demodulation.
Then blocking the RF noise has a low frequency effect as well.

 

[0bs3rv3r]

Thoughts:
a] Most likely it was some unnoticed, uncontrolled variable that changed the noise level.
b] Correctly chosen and placed ferrites can reduce high frequency noise. High frequency is the situation is above 100 kHz.
c] Anything that increases the end-to-end impedance of a RCA interconnect, will ingress the low frequency 'Common Impedance Coupling' noise.

Thanks for your thoughts.

a] The starting point was the observation that placing a hand around the cable, cable length and type, positioning of the cable (and rPi/DAC) all had some effect - in that order decreasing. Then I discovered that rolling up the cable had the most effect of all. The odd thing, was that, unlike many audio noise problems, there was no noise when the preamp was switched to the rPi - i.e. while using the rPi/DAC and this cable there was no problem. The noise is only heard when switched to the the gain phono preamp section. While switched to the phono input, the noise behaved very like most noise pickup problems, as if the cable was badly shielded. Hence one of the first things I tried was a better quality cable. I cannot think of any other variables that might be involved, but of course would welcome suggestions.

b] The audible noise is buzzing sound, that sounds as though it contains quite a few frequencies, not just one, extending up into the higher end of the audio spectrum. It is not low mains frequency (50Hz here)

 

[0bs3rv3r]

re b) Sometimes the high frequency burst noise can get demodulated and produces aritfacts in the audio range. Tube amp stages can be very prone to demodulation.
Then blocking the RF noise has a low frequency effect as well.

That is the current theory as to why rolling the cable helps. I will try to get down to the shop and buy some ferrites today. The rolled cable is actually effective but looks weird and there is still a very small amount of noise - but as it is quite a bit below average vinyl surface noise, not so much of a problem.

 

[0bs3rv3r]

A follow up. Ferrites did nothing, or very little. Getting a longer cable and therefore coiling it into a coil with more turns, as well as careful placement of the coiled cable has cured, or reduced, the problem to the point I am happy with the result.

I guess computers and sensitive electronics can just be a pain at times

 

[majingotan]

I guess there’s some merit with the Wireworld Starlight USB cable where the conductors are twisted to reduce this noise effect. Plus it has been objectively measured by Amir here that their design actually causes measurable changes in the distortion products