I’m puzzled, so perhaps someone can enlighten me.
Per below, a screenshot from my Mosaic, I have the option to set digital filters in the Vivaldi UPS and the DAC. How do the two sets of filters inter-react? Should the same filter be selected in both boxes or does one over-ride the other, or are they completely independent and yet another complicating factor in filter selection?
Thanks in advance.
The filters are completely separate and do not inter-react ( except in your subjective judgement of the end result of the combination).
Although there are 6 filter numbers this is just to identify them. The filter types may vary depending on the sample frequency as can the number of filters offered in UPS. Please consult your user manuals for each unit.
You do not need to select identical numbers for UPS or DAC. In fact you may find that you cannot ( again depending on sample rates).
So, yes, another complication. Personally, I just use the defaults provided in UPS.
Perhaps my 5 years of wondering has just been answered!
Are you saying, based on the sample rate of the playing track, certain filters may not be able to be engaged?
When playing certain tracks, and not having taken detailed notes my suspicions are these were always hi res tracks, when I engaged certain filter numbers (5 or 6) they would not “hold” and would always jump back to the previous filter of 1 or 3.
This happened with my Rossini, and now my Vivaldi.
Quite correct Gregg. Here are the number of valid combinations from the manual (i.e. in many cases there is only one):
If you think this is making things clearer then I’m afraid you are mistaken.
Going back to my original post, how do these filter combinations work in both the UPS and the DAC.
Is it the case that filters that won’t work with some sampling rates on the UPS (per the table) will work on the DAC with the Upsampled data?
Sorry! Yes, the FPGAs in the UPS and the DAC both see the input they are presented with. So the output of the UPS becomes the input of the DAC meaning that upsampling to DXD or DSDx2 in the UPS leaves you (by design) with at most one option in the DAC.
Thank you Andrew!
Now I know why I spent the last 5 years in the wilderness. This chart only appears in the Vivaldi documentation, and not in the Rossini manual!
Am I reading the chart correctly - does the number indicate the available filter number? If so I do not see any filter 3 or filter 6?
More likely the numbers above represent the number (quantity) of different filter “combinations” available for each input and output rate? If so is there a chart showing the makeup of each combinaion ie number, for example 5, one could see the 5 combinations available? Using UPS/DAC, for example, filter combinations 3/3, 3/5, 1/5, 1/6, or 3/3 would be available
good question
The upsampler changes the incoming sample rate to the one chosen by you. That updampling process has to use the filter in UPS.
The upsampled result is passed to DAC where it is treated just like any other incoming file and it is processed using the DAC filter as selected by you.
Sorry, seems I am increasing and not decreasing confusion here.
The platform was designed such that the Upsampler offers a way to offload D2D workload from the DAC thereby reducing EMI and associated noise in the DAC chassis**. However this is subject to the practical 24/384 limitation of AES 3 (we can’t reliably send higher bandwidth payloads over AES 3). The idea was not to enable different arbitrary divisions of upsampling workload between the UPS and the DAC.
And again, Upsampling is all about influencing the cutoff frequency at which anti-alias filtering is applied and not about ”increasing the resolution”. The audio data are the same, upsampling can’t ”create” more audio.
So although they may amount to essentially the same thing I would urge you all to think of it not in terms of ”valid combinations” but more ”for each given input sample rate, what is the valid subset of transformations that can be applied to it?”.
That is what the table above is showing. For each input frequency how many valid upsampling options are there? I.e. the table is the same regardless of whether you are looking at the UPS or the DAC (but of course in the case you have both, the output of the former defines the input seen by the latter).
** Oversampling is always happening in the DAC as it is part of the Ring DAC design but that is not dispensable.
Andrew, thanks for clearing that up and the operation of the filters when using UPS and DAC now makes a lot more sense.
I don’t think the manuals are very clear TBH.
CL
TBH I tend to agree. I’ll discuss with Mat.
@AndrewS - any chart showing the above - the actual combinations available for each I/O rate?
Not that I know of but I think you could make one pretty easily. I am on the road at the moment but let me get back to you when I can grab some minutes alone with a Vivaldi stack and can have a play.
I think parts of the manuals are written in such a way that you need to be confident in what you are reading to have any chance of understanding it.
Obviously that’s not 100% great and will leave many unaware on whether you are doing it right or not, i have myself struggled and not even now would i be 100% confident that i have thing’s all correctly set.
I have and do change filter settings and i certainly understand them better than i first did, but it’s a confusing arrangement, but at the same time great to have.
Probably getting bit’s reworded in more of a normal consumers way, would be beneficial to all concerned going forward.
Agreed. The filter section of the manuals are a perfect example of an expert writing what he thinks is obvious without regard for an audience less familiar with the subject, thereby leaving them baffled.
+1 to the comments above
You are right. I have been a dCS user for over a quarter of a century and am still finding things that I didn’t know. It is also clear from so many topics posted here that users have either misunderstood the manual or have been put off from reading it.One issue is that there is usually a ( long) start up section which is as far as many seem to get yet the following detailed advice contains caveats that they need to be aware of.
Believe me when I say that the current manuals are a big improvement over the initial ones for their first domestic range ( Elgar). The products themselves have also become easier to use. For example if I remember correctly Elgar ( or Purcell) required the user to select a waveform for dither compatible with the sample rate.
Nevertheless these are still complex items posing hurdles to be overcome for explanation to those lacking engineering degrees ( like me).
From what I have read Varese has features that assist setup . Maybe future products will inherit something along these lines.
Addressing the most recent comments here, filtering is a really tricky topic to talk about in a concise way. It’s quite technical, and becomes very difficult to explain the trade-offs of each filter choice without getting into theory of things like transient response and Nyquist imaging. I have made some posts previously on filtering (see here and here) to explain in more detail than in the manuals exactly what happens with the filters in a dCS DAC or an Upsampler, and what happens when for example you increase the filter length (so move from say PCM filter 4 to filter 1).
Aside from that, if we are discussing what filters would be recommended for a particular user, it is a topic with no right answer - there are simply too many factors at play for us to be able to say “this is the filter you should be using” - listening and using your ear to decide which filter you subjectively prefer is always going to be the best option. None of the filters in a dCS DAC (or the Upsampler for that matter) are ‘wrong’, so you won’t be missing anything major by not having the PCM filter set to anything other than the default F1 across the board, but it can help personalise the sound to your preferences.
In terms of listening to filters, I very rarely find that folks will hear a difference large enough to be able to A/B test filters on the go. It typically takes the form of choosing a filter option, living with it for a while (a few weeks perhaps) and making a change. After the change, do you find yourself listening for longer, with less fatigue? Are you more emotionally engaged with the music? If so, you’re on the right lines, and if not, changing back would be a good idea.
One of the areas we are working on in terms of usability, currently for Varèse, is to simplify audio settings (such as PCM filter choice, Mappers etc.) into two configurations recommended by dCS. These will hopefully either act as starting points to experiment with settings from, or can be where you leave your system.
I should note, and this is often met with a bit of scepticism from users, that the default settings of PCM Filter 1, DSD Filter 1, Mapper 1, DXD Upsampling is a perfectly valid settings configuration and will provide excellent results. There is nothing wrong with this configuration, and there is absolutely no necessity for you to change any of these settings if you are happy with the sound.
If there is anything I can clarify about filtering, I’m more than happy to go through questions here.
