How could they play CDs straight (unprocessed) via Rossini DAC's SPDIF input?

The quotes below are from a review titled “Has Rossini reached its peak?” published in HFC issue 67 released in July-August-September 2022 by Martin Colloms.

I wonder how they could play CDs ‘unprocessed’ via the Rossini APEX DAC’s SPDIF input. As far as I know everything gets upsampled to either DXD or DSD including signals fed to any inputs including SPDIF. Am I missing something?

Another related question. How does upsampling by a DCS transport affect/interfere with upsampling by the DAC? I would think upsampling as early in the chain as possible makes sense but doesn’t double upsampling (both by the transport and the DAC as it can’t be disabled in the DAC) has any unwanted/detrimental effects?

“I listened in DSD at many sample rates, and in PCM at rates up to 96kHz, the filter settings creating greater differences than expected for these options, and some clear favourites emerged (see Filter Settings), while for a straight CD input (via wired SPDIF) with no further processing, the sound quality was considered excellent right away, up with some of the best DACs auditioned. Minimal jitter artefacts were audible, and the Apex DAC sounded ‘fast’, upbeat, and musically engaging, being superbly revealing of detail, but also dynamic and expressive. This firm foundation made the later comparison of those various upsampling and filter options so easy, while the digital volume control was audibly transparent.”

" 1. Straight CD 16/44.1: ‘word clock’ set to ‘auto’
as standard, and which reference baseline replay actually sounded really good. We then progressed through the many options for upsampling and filtering, and the various filter maps. Some of these setting combinations and options had quite strong audible effects and were not always for the better.

I suspect that some industry critics have favoured some options just because they could ‘hear’ a difference. For example, with ‘100%’ as reference for plain CD, invoking Map 1 was now marked down at 90%, Map 2 at 65 % with noticeably poorer rhythm, and Map 3 at 70 % but little else of consequence to remark."

That is correct. Vivaldi , however, has what is called a Clone Mode that bypasses any upsampling. One of the differences between the two models.

AFAIK you can’t double upsample but maybe someone has further information. So it is the same data in and out.

Sorry of this has moved slightly from you question but it indicates that dCS have thought of this for their top model. I wish it was available in others.

1 Like

Thanks. I wonder what that review is talking about then, it is even comparing playing CDs unprocessed vs upsampled, using different filters and maps. I always take reviews with a pinch of salt but this seems a huge mistake if it’s not possible. The review is about the Rossini APEX DAC.

By double upsampling I didn’t mean like an upsampled 44.1kHz to 352.8k would be upsampled again by 8 times resulting in a 2,822.4M but like if the transport upsampling is set to a different setting than the DAC’s. For example the transport upsamples to DSD while the DAC is set to upsample to DXD or to different sample rates, etc. I wonder what exactly woukd happen than because it sounds like it can result in unnecessary conversions if they don’t know abouit each other.

I don’t mind upsampling. I’m actually a big fan of it and I couldn’t stand the CD sound before upsampling was invented. I don’t think the fact that we can’t hear sine waves above 15-20 kHz justifies chopping off everything above 22k using a steep filter messing up the phase and the impulse response. I’m pretty sure the roll off curve of our hearing above the cut off frequency is not as sharp as the filter and it doesn’t mess up the phase either so we can hear the difference between a clear and fast unmolested impulse and a badly rounded off and ringing impulse in the time domain like the sound of hitting the rim with the drumstick, etc. even though this depends just as much or more on the recording format and the ADC than the DAC used to convert the music back to analogue.

I previously used a steamer transport set to upsampling everything to DSD128 as far as I remember and once, when I was testing a different DAC that didn’t support that format I turned upsampling off but forgot to turn it back on and forgot about it. For the next about half a year I avoided listening to my digital gear and always ended up listening to vinyl instead. It sounded flat and 2D. I knew something was wrong with it but I couldn’t figure out what, until I went into the settings again to change something else and realised that upsampling was turned off…

I queried Martin on the CD playback issue back on the Hifi Critic forum, when it existed. His reply seemed to suggest that he just used the factory default settings, so the mapper and filter would still have been in play.

Thanks, so at least it’s not only me who spotted this, the sentence below still doesn’t make sense. Whichever map is considered default there are only three maps and he compares them all to straight CD playback.

I thought Martin was a reasonably technically savvy reviewer but apparently not. The mapper is essential to the function of the Ring DAC, without a mapper no voltage gets applied to the resistors and no output is generated. Same with the filter, no filter no analog output. These are not “optional extras” like upsampling, they are core functionality. Martin’s comments don’t seem to make any sense.

I agree. Even if we let alone technical savviness I wonder what they compared to what, what was the reference they refer to as straight unprocessed CD playback which they compared all the filters and map settings to.

Am I right to assume that apart from applying maps and filters which as you put are essential, is an SPDIF 16/44.1kHz CD signal always upsampled to DXD or DSD?

After reading a bit more about the Ring DAC, the signal ultimately has to be oversampled to 706.8 KHz to feed the Ring DAC so I’m curious at what point the DXD, DSD signal is present?

Would be good to see a block diagram including an upsampling DCS transport connected to an upsampling DCS DAC to see the whole process.

There is no such thing as “straight unprocessed CD playback” in a system comprising a Rossini Transport & DAC when playing CDs. Upsampling will always occur somewhere and one or other Mapper and Filter will always be active. So what one chooses as a baseline for any comparison is essentially an arbitrary decision.

The output format/resolution from the Transport (for PCM, i.e. when playing a CD) is determined by which output is used: AES 1+2 will always be upsampled - to DXD, DSD or DSDx2 per the Upsampling Mode setting on the transport. The AES 3 and S/PDIF outputs always pass non-upsampled 16/44.1 data.

If the data entering the DAC are already upsampled to DXD (AES 1+2) and the Upsampling setting on the DAC is set to DXD then no further upsampling is performed in the DAC. This is the intended architecture as it offloads processing from the DAC which provides a “quieter” environment for digital to analog conversion. If the Upsampling setting on the DAC is set to DSD or DSDx2 then the corresponding conversion will be performed.

If the data entering the DAC are not upsampled (AES 3 or S/PDIF) then they will be upsampled to DXD, DSD or DSDx2 per the Upsampling setting on the DAC. This is done in the final stage before DAC, i.e. after the oversampling to 706.8kHz.


Regarding the Hi-Fi Critic review: there is relevant discussion and info on the listening tests carried out by Martin Colloms - from @jonhoneyball who was there - in the thread “Bought a Rossini today”. I contacted MC myself to ask about the review, as I found it confusing, and concluded from his answer that it had all been done in rather a hurry.

Thank you struts for your detailed response, all clear and much appreciated as I’m relatively new to DCS and I usually want to understand everything.

Your response also confirms that upsampling early in the chain or separately from the DAC is the desired way and that’s why the Vivaldi uses a different architecture.

My last question in the topic. Is there any way to see what format the current track played is upsampled to? I know I can check if DXD, DSD or DSD2 is selected and I can see the source format on the display and in Mosaic and could figure it out based on these two pieces of information but it would be easier to just see it somewhere. It would also come in handy when playing unsupported formats which get ‘downscaled’ like 32 bit or DSD256 tracks, etc.

Thanks Simon, further confirms my view on magazines. I used to buy a lot of car and hifi mags but then I realised they all just repeat each other including and adding mistakes and their only goal is to fill 30++ pages and have a catchy front cover to sell rather than to provide useful and reliable information.
Sometimes I still buy some occasionally but only to catch up on new products and trends.

Also thank you for that thread it discusses many very interesting topics, I will go through it.

Just check the display on the DAC. For PCM the final result is always determined by the Upsampling setting on the DAC regardless of what has or hasn’t been done upstream.

No trouble, Tom. I have a somewhat similar opinion of the magazines (I started reading them in the 1970s!). There are some individual reviewers I trust, and some publications are generally better than others, but mostly I’m sceptical.

Same here, but Colloms was one of the reviewers I came to respect. He seemed technically savvy and thorough. I am dismayed that he put his name to such a poor review.

Yes, the product is complex and yes, the documentation while fairly comprehensive is not exactly written for easy consumption. But I am sure dCS would have explained everything to him clearly and patiently had he asked. This review undermines his credibility totally IMO.