Thanks for the suggestion. I discovered today turning off the matrix pipeline does not produce as frequent dropouts just takes longer to start one channel at a time.
Second, it’s impossible to get the output to show DSD even though I have SDM enabled in HQP. I can get DSD repeatable with DoP checked on though.
I have tried every combo of NOS on/off and DSD Direct on/off on the R26 too.
In addition, when I do enable the pipeline again and the dropout occurs, it will sometimes be one channel, then alternate to another and finally a click and no sound. Turns out it’s the Oor/Hypsos amp I have to power cycle to hear again even though it remains on and it’s playing in Roon with DSD showing the DAC.
Is this normal behavior for the amp to loose signal to the inputs? Strange.
There are various DSD-over-I2S “standards”, it is even more non-standard than the I2S on cable in first place. So typically it needs careful checking of the documentation and configuring DDC and/or DAC so that their connector pin arrangements match for DSD.
Hi @Eminent_One, my 5cts… Please take it with a pinch of salt as I don’t onw the same rig as yours and these are only suggestions that may or may not help…
Hardware layer.
As I understood you have the following chain [Win PC] USB [DDC] IIS [DAC] XLR [Pre/Amp] - headphones
few remarks to check here:
IIS layout setting on your DDC to make sure it corresponds to the DAC - I suppose “Gustard” default should be selected on DDC? As alternative did you try USB connection between the two?
are you using “bypass” on your Oor/Hypsos preamp? You have mentioned you needed to power cycle Oor after some click, I assume it was some type of current protection that was triggered. R26 also mentions bypassing volume in PCM NOS and Native DSD. So this chain requires extra attention for sure in these modes.
Software layer:
how does Roon playback alone behaves? can you play Native DSD 512 over your chain via Roon directly ?
Jussi commented on ASIO driver to check, but I dont know what could be alternative, ASIO4ALL may be?
No, absolutely no ASIO4ALL. I meant that if one uses Roon + HQPlayer, common reason for problems is that ASIO driver, or WASAPI device has been left enabled in Roon Settings → Audio. With Thesycon drivers, symptom is that DSD is not available.
Ok, with some further testing I was able to isolate it to be an IIS issue as USB in only is not triggering any weirdness and I can play 44.1x256, albeit with @35% CPU usage.
When I incorporate the U18 DDC and IIS, after a minute or so the Hypsos shows a climb in output voltage and that’s when one channel starts cutting out in one channel followed shortly by complete cutout and the display showing double the output voltage in the mid 30’s as opposed to high teens typically.
Home audio Fidelity, but it is impossible to import them in the concerned section
I have to use Matrix or convolution, I’m lost
he made me several filters, do I have to import them all?
attached is the procedure he described to me
You must download the zip and extract the files in the directory of your choice. There are 2 mono wav files for each sample rate.
***To implement the filters in HQP, you have to use the “Matrix / Convolution set up” option. *** Just select the files for each channel, knowing that the _LL files are for the left channel and the _RR files for the right channel. You must then check the “Enabled” box to activate the convolution. If you are upsampling in DSD for example, I advise you to use the filters at 352.8 kHz.
I go back to PCM once in a while out of interest to compare. Mostly I use ASDM7ECv2/DSD256. My dac, Denafrips Pontus II, can handle PCM1536 / DSD1024 on it’s usb input. It’s fed from my HQP machine via fiber to a Cisco switch then Cat 6 to an ultraRendu which supports PCM768 / DSD512. However, the ultraRendu can be pushed harder, up to (PCM1536 / DSD1024) if your network is solid.
I read a comment by Jussi “With 1.5 MHz PCM rates, the output begins to be decently clean”. The settings below sound quite good to me, NOS active on dac.
I started wondering, what happens if I use just “shaped” dither without any filtering for redbooks. Also is that better than no dither at all (for redbook)? The only settings I would be using:
-10db fixed volume
no filter/upsampling
“shaped” or no dither
DAC bits 20
Jan Meier cross-feed
Dac is Holo May. I think I like most no processing, but I do like crossfeed and attenuation (as I have sensitive headphones and too powerful dac) so something will be done to the bitperfect sound anyway. I also noticed that “shaped” dither seemed to enhance realism without overdoing it in any way. This works great for Hi-Res material, but for redbooks and 44.1-48khz/24bit this may be problematic as the manual says that “shaped” is meant for >88.1khz. I wonder how bad the situation is? I haven’t seen any measurements studying this directly but here it looked like shaped wouldn’t be pushing anything to the audible levels:
I’m not interested in what people think sounds best but the measurements. I wouldn’t had asked opinion to such an unconventional selection of filters/dither if I wouldn’t had followed my own ears. The reason I’m asking the question is that I don’t understand these filters enough and using them in wrong way may have unwanted consequences. It’s much nicer to hear comments from someone who understands what the algorithms are doing.
For example:
“Yeah, don’t worry about it. If it sounds good to you, it’s all good. It just generates some noise to 17-20khz area, but it’s -80db anyway.”
“Personally I wouldn’t do that. If you use it with 48khz, this filter doesn’t work correctly and it creates very strong error spikes in area where you can’t hear those, but they start to irritate your ear and it may even cause hearing damage at loud levels. Stay with TBDF or gauss1 if you are not going to upscale redbook to at least 88.1khz.”
I’m also a technical person who is just interested in what happens (ie. the background for this warning/guideline to not use that dither for lower rates)
I am also a proud owner of The Holo Audio May KTE. Your comments are interesting. I’m going to have a play around with no filtering and just dither into NOS. I only use NOS anyway because what’s the point in buying a May unless you only use NOS mode🙂
Point is to run OS in HQPlayer, not in May. And have May operate as a bit-perfect DAC.
Running May or any other DAC at 44.1/48k rate in NOS just creates horribly distorted results.
But if you run May at 1.4/1.5 MHz rate with HQPlayer upsampling (OS) you get pretty good results, especially when you set DAC Bits to 20 and use LNS15, NS9 or NS5 noise-shaper.
Or alternatively, run it at DSD256 or DSD512 always. I’m running my Spring 2 and Spring 3 DACs always at DSD512 from HQPlayer.
At 1x rate outputs you get slightly upward tilted noise floor in the top octave. But the difference is largely swallowed in the NOS distortions.
For decent performance, you need to run Holo Audio DACs at least at 705.6k if you want to use the PCM mode. This also allows use of noise-shapers to linearize the ladder (you can get rid of the R2R inherent low-level non-linearities).
Mr. Jussi that’s what I’m doing my audio friend. I wouldn’t own a MAY if it wasn’t for you. I’m using Poly Sinc Long LP with LNS15 into my MAY in NOS at 1411.2 which is bloody lovely:musical_note:
I’ve never managed to try poly-sinc-long-lp as everything just freeze if I try that.
Some good combos IMO (all 20bits to dac and Bauer cross-feed 650 / 9.5):
poly-sinc-gauss + gauss1 @ 705.6/768 kHz
Perfect combination of fun, transparency and MASSIVE space when music calls for it. Try this with Thomas Bergersen - Promise. Used poly-sinc-gauss-long at first, but it started to feel like it’s overdoing space so that I lose the emotional tie to the music, poly-sinc-gauss is better balanced.
none + none @ native
It’s fascinating to listen to NOS like this. A bit harsh sometimes, but it’s interesting idea that no processing happens anywhere and you can almost see the bits in music. Works beautifully on good acoustic recordings (like Jazz).
none + shaped @ native
Same as previous, but shaped adds juuuuust enough realism that it sometimes blows my mind. Try (really, try!) Dimmu Borgir - Ritualist. Go as loud as you dare. The song starts so that the ritualist is dragging chains and starts to speak. It’s like suddenly you have that wind in your room and the ritualist is right behind you. The ultimate creeps I’ve never got from audio. Shaped also adds a nice touch on Jazz (adds realism).
poly-sinc-ext2 + LNS15 @ 705.6/768 kHz
Less fun than poly-sinc-gauss and also I feel like “more technically correct”/less colored. Focuses less on emotions and more on the contents in the recording. I’d call this as “the best 90s sound” there is. What I mean by that is that the digital ringing is still present. It makes the music a bit more digital, but this digitalism is also beautiful (I grew up liking the way pre-ringing sounds, it makes the sound recording-like which I associate as cool / not banal). This also fills the music somehow. It sounds like the sounds are interacting with each other and music is alive. This is really works well fast paced music like metal (makes the music sound even faster than it is which is nice).
that’s what I’m doing my audio friend. I wouldn’t own a MAY if it wasn’t for you. I’m using Poly Sinc Long LP with LNS15 into my MAY in NOS at 1411.2 which is bloody lovely:musical_note: