I've had similar problems with my Hugo TT and DAVE over the years. Jitterbug and ferrites make no difference. I found the same as you, that by increasing the sample rate in the Windows audio properties (i.e. because I'm using Windows mixer), the soft clicks would stop.
WASAPI from Foobar is a complete solution.
Optical is also a solution in my system, working at any sample rate (upto 192KHz) with Windows mixer.
On my TT with the SD input, which does not use the Chord driver, there was no problem. The SD input also has no galvanic isolation and only supports 44.1KHz/48KHz sample rates, so it's really there for direct connection of phones/tablets.
With my DAVE, if I change track sometimes I get a few clicks within the first few seconds. On YouTube videos this is very common. It seems to me that DAVE (and TT before it, which did the same) is trying to get the PC to do something shortly after the music starts and this interferes with the transmission of music.
So, I have to conclude that the fault lies entirely with Chord. My old PC might be relevant, but this is a common complaint with Chord's USB.
I very much doubt a USB device between PC and DAVE will solve this problem, as the PC itself is being "interrupted" as it tries to send the music.
Jawed, it appears you are 99% correct. Thanks for your insights and in confirming that my issue is reproducible on other (your) units. Here's what I discovered:
My 2 USB regen devices came in today and I tested them. No difference. What is even more enlightening is something else I thought of trying -- it *DID* correct the problem, but has some tradeoffs.
On a whim, I uninstalled the Chord drivers and used the Windows 10 Creators edition built-in class 2 driver (derived from Thesycon code base, which almost every other non-Chord DAC seems to use)...
No more soft clicks when using Windows mixer in 44.1kHz!!! I played though 2 hrs of Mahler 3... which was more than enough to present clicks in the past. (Usually within about 10 minutes, I would hear at least one click).
The trade-off is -- for some reason, the DAVE is recognized as a 32 bit only device. If one uses Foobar or the like, no problem -- set the WASAPI output bit depth to 32 and it will pad the data.
But the trade-off lies with Tidal... Tidal refuses to enter WASAPI mode. I suspect it gets an error (bec if I set Foobar to 24 bit, it reports an unsupported error) and falls back to non-WASAPI.
My current theory is that it is not a "sample too late" due to timing problem (which USB regen should correct unless there's a big gap).... I am a bit baffled by why the upsampling with Windows mixer works around the clicks. (Maybe it's a missing sample issue and missing 1 sample out of 352k instead of 1 out of 44k becomes masked?)
I am less baffled by why WASAPI mode works -- I am guessing that's the focus of Chord testing and driver development as that's the main mode used for audiophile playback. This is definitely a bug in the Chord drivers. (Which I suppose is a good thing, bec it isn't a bug in inadequate buffering int he DAVE itself... Drivers are upgradable.... DAVE firmware [today] is not.)
Undecided whether to use the Windows included Class 2 driver or go back to the Chord driver. I am leaning towards Windows included driver bec the mixer upsampling is known to have some bad characteristics.... and de facto, most Tidal content is in 44.1kHz, so if I keep Force Volume on, even if it goes through mixer, it will be close to original PCM bits (if not in fact bit perfect).
I don't play DSD content and have not tested DSD with Windows included driver.