I’ve set the ref level on ADI-2 DAC to the lowest setting, -5 dBu, but EC ZDT Jr. is still too loud with some songs, and if I turn down the volume further on the amp, channel imbalance occurs. So, where should I reduce the volume, in Windows, Spotify, or ADI-2 DAC? Or is there another solution?
You can add a passive pre in front of the Eddie Current.
While a lot of people will claim that digital volume is detrimental to a signal, if the calculation is being done with more bits than the original signal the loss is at best marginal, and all your really doing is trading off that loss in quality against the sound of the crappy pot in a passive preamp.
I’d do it at the ADI-2, the odds are Windows and spotify are working with a 24 bit output signal, and the ADI is very likely 32 bits internally.
FWIW it’s sometimes better to reduce volume digitally in the chain so you can get to the best sounding part of a particular pot’s range on the amp.
My DAC has a built in analog pre using a relay based volume control, it would have the least quality loss at 0dB’s where that control is basically bypassed, but I’ll set that as low as -23dB’s for some of my headphones so I can get the volume pot on the WA33 closer to optimal.
I have the same problem with the cheap factory pot in my Bottlehead Crack. My solution is to set the BHC volume at 50% to 75% and run the input through the excellent powered preamp feature of the RebelAmp. I agree with @Polygonhell to avoid passive/unpowered preamps, as they tend to flatten dynamics and can shift the tone.
Windows volume/mixer processing is, effectively, 32-bit (convert to float, process, dither, convert to integer), where as the RME ADI-2 DAC FS is 42-bit. So doing it on the RME unit is the best course here.
Things like Roon and JRiver Media Center do their processing using 64-bit math, and are even better, but are not options for Spotify.
I’ve had gain problems in my main desktop system for years. I was OK as long as I stayed w/DACs having ~the 2.0 volts output that was nominal/normal for single-ended DACs. But then I had a series of DACs w/higher output. All were either multibit or NOS DACs, easily sonically superior to the older delta-sigma DACs–but all had hotter output, including the current DAC, the MHDT Labs Orchid (output @3.0 volts).
With my primary headphone amp/preamp, the Violectric V281, I can get around this by setting both the headphone output & line output to -12 dB using the handy dipswitch arrays for both.
But then I got a new amp/preamp with its own gain issues: the Kinki Studio THR-1. It has insane gain and no gain switching capability. So I tried a transformer-controlled passive in line between that & the DAC (it works well). I also tried a pair of 10 dB attenuators (link below) that also work well and are sonically transparent:
They also have a version for balanced systems.