Holo Cyan2 user manual is up.
I actually asked them to include output impedance and to mention USB galv isolation since their original launch website did not mention (although USB isolation was obvious from photos).
Very low output impedance which is nice
When is the 1st batch of Cyan2 expected to ship? Really looking forward to some initial impressions and measurements!
According to the website 5-6 weeks after orderā¦ they are made to order.
My friend ordered couple weeks ago so Iām still approx 4 weeks away from borrowing it (assuming he lets me borrow immediately! 
)
Yes it isā¦ Or you shouldnāt say ānoiseā here, but instead ādistortion and spurious tonesā or something like that.
The apparent level of random noise depends on FFT and averaging parameters. With longer FFT and more averaging it keeps getting lower and lower, since random noise is equally distributed over more and more FFT bins. So the per-bin level drops.
Exactly the kind of noise/resonances/sounds not belongen to music I have spent years now to try to kill. My friends, private hifi geeks and a professional actor in the industry say my rig is the cleanest they have heard, including their own studios, and this DAC would be an interesting contribution.
Here one can really tell the beautiful contribution of HQP 
The SMSL D-6 is alsoā¦ in my plot it normalized to 0dB, in Jussiās it isnāt
Thatās why I prefer to talking about the amount of clean range (120dB for example)
All the crap with D-6 is 120dB below the signalā¦
My plot is not averaged either, it is live RTA
Yeah, my plot is normalized to 0 dBFS signal level. So it uses same level normalization as THD, IMD and other measurements. So the test signal is 0 dBFS signal, IOW, sum of those test tone harmonics is 0 dB.
I personally consider it wrong to normalize the plot so that each harmonic is at 0 dB, because that means the actual test signal level is about +20 dB in the scale.
But doesnāt change the result, just needs to be taken into account while reading the plots.
Yep likewise
I can produce same looking plot , all distortion below -140dB
just different vertical axis unit used
I only do this for quick look at the amount of clean range
Less mental maths 
The clean range is from the signal max, which is at 0 dB in my plots!
But anyway, Iām not sure if your and mine have same bandwidth and FFT parameters either. So for the random noise levels it is question mark.
Anyway, hereās the D-6 again for comparison measured same way as the others:
SINAD-wise, the T+A DAC 200 is cleanest still. And maybe the Gustard A26 as well.
I wonder if its worth picking up a S.M.S.L D-6, as they look to be around Ā£160 in the UK. I can then save up a bit for a Holo. It would sit in my main rig instead of the Topping D90, I would not use any pre-amp functionality as I have Topping pre90.
Assuming I am using HQP, feeding upscaled DSD content, would there by any disadvantages from swapping the Topping D90 for a S.M.S.L D-6?
Any DAC should be fed the max speed compliant as their onboard DSP will more or less route through and not convert, given you do not set any features requiring DSP, such as e.g. volumeā¦ But DACās offering direct DSD should set accordingly as the DSP in all DACās I have so far tried, and they are some and expensive oneās, are inferior to HQP. By far, in my opinion. I try not to out any brands, as I am in the industry, but I must question the value of some very expensive oneās, even tried with taylored upsamplers but it will not cure the decease of the system ārattlingā I clearly detect already at a short audition. That opinion also include the overall noise. But the ālawā in my book states, if the source is the cleanest possible, the lesser the system resonances and other noise become active. The more strain put on any device, especially supersonic noise, the more audible noise it will create. A common width is some 10 - 100.000 Hz, which the electronics manage to modulate and filter quite good, but as soon as the very high frequency noise is introduced, there will be consequential audible artifacts, and those are not possible to filter out as it is glued to the signal and should never been introduced by any pipeline in the devices. Especially cable shields for computer based interfaces are experts of landing sh*t in the hifi gear as it connects every device ground to eachother. Grounding earthplanes and filtering all power sources, sometimes also in the device itself, is of essence to maintain the beautiful -140 dB noise rejection, otherwise it will remain impossible to detect.
Here is my SMSL D-6 measured in Sep-2022 , DSD256
Different scale to my earlier plot, noise/distortion floor down at -160dBFS
FFT 2M, no averaging
NID multitone peaks are at -1.6 dBFS
In the D90 user manual there is no mention of enabling DSD Direct mode, so it doesnāt even belong in this thread? It is like all ESS based DACs , canāt bypass AKM SDM modulator
SMSL D-6 works very well up to DSD512 (after I told SMSL about a bug at DSD512 showing them my measurements, and they fixed it with a firmware update)
I have the non-mqa version of the DAC, and I think you can run it in DSD64-DSD512 Native and DSD64-DSD256 DoP. I remember there being a note that the performance sweetspot was DSD 256.
I think my version of the DAC is AKM rather than ESS. Or am I mis-understanding something?? (which I may well be!).
Check the user manual I linked above - there is no DSD Direct mode mentioned? Some Toppings do mention it but this does not.
I donāt think MQA vs non-MQA version makes a difference in terms of DSD Direct.
Unless you can show me something mentioning it? But I couldnāt find in searching
Thanks, yes I did have a look. I also did a side by side with my saved manual, and could not see āDSD Direct modeā anywhere. Just to check, what is the significance of not mentioning DSD Direct mode in the manual but referencing the ability to play DSD?
I will have a look around. I basically remember something in terms of the DAC chip itself, but have not saved a bookmark or similar. My understanding was with the pre-amp mode off, DSD Direct was available, insofar as the signal would be fed to the conversion stage.
Update - I have found the AKM AK4499 datasheet, and the block diagram seems to suggest DSD content goes straight to conversion, bypassing the modulator, but I am very much not an expert! Just an idiot with a computer!! Lots of forum posts seem to reference the āDSD Directā being possible, but I have not found anything official.
Only if Topping enabled the user to select
As I mentioned, some Topping manuals mention how to enable and disable. D90 manual does not
The chip alone is not enough. The implementation is key
Singxer SDA-6 Pro 2
Dual AKM 4499EX chips - donāt see the user manual anywhere about DSD Direct.
And who knows if it is buggy AKM implementation or not. If I can home trial one later, I will get it to test.
But if not, not that interested. More interested in discrete Holo Cyan 2.
