Intrigued by
this post I wanted to double check on my JM20, and i can confirm it.
My tests are in Linux -> audio properties panel -> hardware -> select the device you want to test -> speakers test, clicking the 2 buttons you hear respectively "front left" and "front right"
Jcally JM20 (CS43131 - Savitech bridge - ASYNC usb)
high gain
no ramp-up\fade-in effect, you hear the entire phrase but with short
pink noise in background for the first half second played when the device is idle (not playing) for at least 5 seconds.
low gain
no issue, you hear the entire phrase with clean background, no matters how much time you leave it idle
Then I wanted to re-check my other dongles, changing gain when it is possible:
Tempotec sonata BHD (dual CS43131 - Savitech bridge - ASYNC usb)
low gain - mid gain
ramp-up issue: clicking on "front left" you can only hear "left", then after 5 second clicking on "front right" you can hear "t right", so right channel come up a fraction earlier than left one (panning effect), both truncating the very first samples.
high gain
no issue, you can always hear the entire "front" word on both channels, no matters how much time you leave it idle
Kuang Pai Player 3 (dual CS43131 - Savitech bridge - ASYNC usb)
low gain - high gain
ramp-up issue: clicking on "front left" you can only hear "ont left", then after 5 second clicking on "front right" you can hear "ront right", so right channel come up a fraction earlier than left one (panning effect), both truncating the very first samples.
So it's a bit faster than Sonata (truncate less samples) but has the issue in any mode you use it
Ugreen HiFi Audio Pro (CS43131 - CS46L41 bridge - SYNC usb)
low gain - high gain
no issue, you can always hear the entire "front" word on both channels, no matters how much time you leave it idle
CX-Pro (Grave Audio DA06 CX31993 - TTGK brdge - SYNC usb)
only 1V mode
no issue, you can always hear the entire "front" word on both channels, no matters how much time you leave it idle
Hi-Max (CB1200AU chipset - ADAPTIVE usb)
only 1V mode
no issue, you can always hear the entire "front" word on both channels, no matters how much time you leave it idle
So, different behaviors, some devices are always ok, others change with gain mode, others are always affected. In my case the only common factors between the device that are affected by some extent is the Savitech bridge and the corresponding ASYNC usb transfer.
Maybe the JM20 pink noise is one of the artifacts that Fosi Audio referred to saying that the ramp-up was a necessary setting to avoid them, but it happens only in high gain, while Sonata is perfect in high gain and has the ramp-up in lower gain modes, so high variability and nothing can be concluded with confidence.
Did not run tests in other applications, since it is proven that depending on the application and how it keeps the device asleep the issue will not appear even if the device is affected, for example if i leave an audio editor like Ocenaudio with the dongle as output card, it is kept always asleep even if nothing playing.
From my point of view, if the issue appears in the OS speaker test, a device is to be considered affected.
Since my tests are under Linux they are not compliant to
@jkim method, so I'm not going to update the spreadsheet and will leave the choice to him.
in the following days, but I just gave JM20 a quick run to try it with an amp, since that what I bought it for and I sure am confused. 
www.hwinfo.com
