In the last few posts there are several hints that the "pops" are related to the interaction of D-6s and the amplifier connected.
The D-6s has no coupling capacitors in the output (just like the application circuit for the ES9039 suggests).
These caps would need to be film caps (electrolytic caps are not well accepted in the audiophile community) and thus large and expensive (and still there would be people that feel that these caps need to be polypropylene and thus even larger and more expensive).
I just measured the output offset voltage and output resistance on my D-6s:
------- RCA output ----------
RCA, left, active: +0.37mV , 101 Ohms
RCA, left, muted: +0.00mV , 110 Ohms
RCA, right, active: +0.32mV , 101 Ohms
RCA, right, muted: +0.00mV , 110 Ohms
The DC output voltage (offset voltage) is fairly small. In the muted state the outputs are connected to Gnd with a similar source resistance.
(I guess they use the SGM3710 muting chip and this chips has either 1 Ohm (active path) or 11 Ohms (muted path), so SMSL should have implemented a 91 Ohm resistor in the muted path and not 100 Ohms like in the active output path).
These ca. 0.36mV should show up on the amplifier output as ca. 7mV assuming an amplifier Gain of 26dB. On my setup this click/pop is audible, but this is an old integrated amplifier with almost 43dB gain with the volume knob fully open.
There is no significant change in the output resistance in the muted state, so even a DC-coupled amplifier (with some input offset voltage or bias-current) should not produce a significant "pop" noise. The turn-on, turn-off and break-before-make times for the SGM3710 are in the order of 100 ... 200ns, so the output of the D-6s will be left "open circuit" for just 100ns (break-before-make-delay).
------- XLR output ----------
XLR, left, "+", active: -0.68mV, 110 Ohms
XLR, left, "-", active: -1.77mV, 110 Ohms
XLR, left, "+", muted: +0.00mV, 21 Ohms
XLR, left, "-", muted: +0.00mV, 21 Ohms
XLR, right, "+", active: -1.26mV, 110 Ohms
XLR, right, "-", active: -1.91mV, 110 Ohms
XLR, right, "+", muted: +0.00mV, 21 Ohms
XLR, right, "-", muted: +0.00mV, 21 Ohms
The DC output voltage is in the order of -1.5mV with the difference for the individual pair being ca. 1mV. This is nothing to worry about, not even when feeding a transformer based input since the source resistance is 200 Ohms and 1mV translates into 5uA assuming the transformer has a DC resistance of 0 Ohms.
The output resistance changes significantly when changing to the muted state. This may produce a DC- step (pop) when connected to a DC coupled input stage with relatively large offset voltage or unequal bias currents, but equipment with XLR inputs usually is AC- coupled.
Looking at the PCB of a Topping E50 you find 4x 100uF electrolytic caps near the output/filter stage. I guess they do AC-couple the XLR outputs and the RCA output is still DC-coupled. Since the RCA output subtracts the XLR "+" and "-", the offset will be fairly small like for the SMSL D-6s.
Edit: Looking at the layout carefully these caps in the Topping E50 are power-supply bypassing.
Remark: The measured value for the output resistance in the active state is influenced by the offset voltage; reversing the input terminals of the DVM and taking the mean value of both readings gives an accurate result.
My personal conclusion:
There's nothing wrong with the implementation of the filter / output stage. The small DC- offset voltage will be easily tolerated in almost any application. Omitting AC-coupling capacitors in the output is a tribute to the audiophile community (cheap caps in the signal path are not welcome) and the ESS application circuit does not suggest capacitors in the outputs either.
I thus have no real clue what's the reason for the audible "pops" that
@Feesh observes. Still it would be interesting to see if an "old school" amplifier with AC-coupled input does produce a "pop" too.