RCA to balanced speakers - background noise problem

[incans]

WiiM Device: WiiM Ultra (Firmware 5.2.730566)
Network: Ethernet (hard-wired)
Speakers: KRK VXT8 (active)
Subwoofer: B&K P12-300SB-PR
Sources: Samsung S95B (HDMI ARC), DLNA Streaming (hopefully)

My new WiiM Ultra arrived today. Having connected it up I notice an immediate problem with background noise from the speakers. The noise is semi-random tics and pops, but it's clearly not actually random because for example changing the volume level on the WiiM causes a stream of audible pops in time with the stepping of the volume level.

The speakers are active studio monitors with XLR/TRS balanced inputs. They are connected to the Ultra via moderately long (~6m) single pole (mono) RCA->TS (jack plug) cables. The whole system is connected to the same mains supply circuit (ring main).

The speaker documentation says they are ok with an unbalanced source, as long as the "cold" (negative) side is linked to the ground. When using a mono jack plug this happens automatically as the single sleeve contacts both the Ring (cold) and Sleeve (ground) contacts in the TRS receptacle of the monitor.

This connection arrangement worked fine with my old preamp, a Rotel 1070 stereo unit, but the WiiM definitely has a noise problem.

The monitors have a "ground lift" option that is supposed to help with noise caused by ground loops, but this has no appreciable effect.

It was the combination of features and price that drew me to the Ultra, so it will be really disappointing if it turns out that it's just not compatible with balanced power amps (even in unbalanced mode), as the nearest comparable devices with balanced outputs are substantially more expensive.

If anyone has experience of mitigating with this problem I would appreciate any ideas.

 

[ralphs99]

Hi Nanook,

Thanks for your explanation of the situation where the unbalanced output device is a Class II design. I agree that connecting the shield at both ends is the better choice as it avoids the effects of leakage currents

What happens in the case of a device with a balanced input being Class II? Where is pin1 usually connected? Directly to 0V?

 

[Nanook]

Hi @ralphs99,
I think the figure copied in post #14 answers your question. It's for the ncore NC400 module.
The connection to chassis on one of the 4 pins of the input connector shown in this figure must be the active-low enable signal not gnd. See section 5.2 in the datasheet.

https://www.diyclassd.com/media/85/b5/1c/1646058855/NC400_04xx.pdf

Figure 2 in section 9.4 of the NC400 datasheet shows the setup for an amplifier module that is not grounded (2-prong inlet; the thick black "wire" is the connection between the NC400 and the DC power supply.

Thus the shield of the external signal cable (pin 1 of the XLR) is connected to the chassis directly at the input connector and the internal gnd of the amplifier (via the shield of the internal input cable) is as well connected to the chassis in close proximity to the XLR connector.
Thus the differential input stage of the NC400 only has to deal with the CM signal due to the I*R-drop along the short (and good) internal input wiring (I=leakage current of amplifier power supply ; R=resistance of the shield of the internal input cable).
When connected to a source, the amplifier's leakage current will cause an I*R drop along the shield of that connection as well and this adds to the I*R drop at the internal cable, so using an extremely cheap connection cable with a high shield resistant is not recommended.

However: be the shield resistance a few Ohms - that's vastly less than the assumed input Impedance in the example given in my earlier post. The assumed 0.1mA would cause an I*R drop of less than 1 mV - compared to 330mV.

Hypex specifies a CMRR of typ. 55dB across the audio range (this is quite good, Bruno seems to use 0.1% resistors in the differential stage) so his suggested wiring sounds "bullet-proof" to me.

 

[Nanook]

... the internal gnd of the amplifier (via the shield of the internal input cable) is as well connected to the chassis in close proximity to the XLR connector.

I have not really a clear idea why Bruno did draw the connection of the shield of the internal signal cable to a close but different point near the XLR connector.
Maybe he wanted to suggest that we should not solder this shield to pin1 of the XLR because this would allow some RF interference to enter the chassis due to the inductance of the short wire connecting pin1 to the chassis. This RF would be present on the input cable.
Attaching this shield to a different screw of the XLR panel jack does restrict the entering of the RF to this short piece of wire acting as an antenna. The actual input cable would be "clean" except for the mutual inductance in case these are routed in parallel - drawing these to connections to the chassis to different sides of the XLR connector may suggest to minimize their mutual inductance.
I bet this detail was not coincidentally.