In DAC reconstruction there are two key ways of providing the filter response function to the spikes of energy from the samples that in turn product the analogue signal.
The filter responses can be done in an infinite way - what is called IIR (Infinite Impulse Response) which uses a software recursive algorithm. Naim use this method with their AD DSP processor.
An alternate method is to use a Finite Impulse Response (FIR) approach. Essentially the impulse response is modelled into a set of samples (sometimes referred to as taps). The more samples you use the more accurate the FIR filter is. Chord use this method.
More filter response samples equals more taps equals more accurate response. Essentially Mscalar provides a very large symmetric FIR filter sample sequence applied to an oversampled stream. This can improve the performance of the combined FIR filter used in the DAC reconstruction. The down side of this method is that it is power and processor intensive and can produce electrical noise that can undermine the advantage using a larger number of taps.
Which is also it’s advantage. DACs are more efficient and can work less hard with higher resolution sampling rates. A reason why the difference between 16/44.1 and 24/192 is huge on a cheap device like a Walkman but significantly narrower on something like a high end DAC. I expect a significant benefit of the Mscalar is that it offloads a significntly large and noisy task to itself rather than the end DAC which will choose a much simpler operating mode as the resolution of the input data increases.
I think he could only qualify that with known variables like other Chord DACs. He’d have no way of knowing the impact of this processing on anything else. For a lot of gear, I suspect oversampling and processing of 16/44 is not inconsiderable.
I am not sure… i think you might be mis understanding what he was referring to, such as SPDIF noise isolation. The mscalar because of its huge kernel size and Watt’s own WTA windowing function aids more accurate reconstruction when using FIR filtering which is the method Rob Watts has chosen in his designs. (Naim do it differently)… and yes I have had this exact same discussion with Rob just prior to the launch of the mscalar… and he was proudly extolling the benefits of decoupled FIR processing of a million taps as something very special in his designs and might something I would be very interested in when launched… and he had that enthusiastic glint in his eye when he is talking about such things.
Absolutely the mscalar off loads noisy FIR processing with its huge filter kernel size. If you study DSP engineering and DACs you know that the the downside of FIR filters is the electrical power and noise produced through the intensive processing required in a period of time and can offset the advantage of improving the reconstruction accuracy by adding digital noise… there is a balance… or optimum point for a given system design. By decoupling the mscalar from the DAC you are improving this optimum point.
This has been made more possible since Xilinx and others have introduced very low power FPGAs. Even so running a filter response containing one million samples per sample is a lot of processing and computing noise !
Hi Simon, no I was clear in my question, how much was the effect of the Mscaler on Dave not using the the WTA first stage filter not being used in Dave anymore, in terms of making Dave less electrically noisy, and he said it was not a big effect at all, and the main benefit was from the 1 million taps.
Ali, you are missing the point chap, the mscalar doesn’t make anything attached less noisy from a processing noise perspective… it allows the benefits of decoupled processing (convolution) of a very large FIR filter kernel (one million samples) without, or causing minimum affect on the connected device from a processing noise perspective.
In fact noise decoupling is a key consideration, and the outputs of the mscalar are filtered, and are improved compared to the original BLU for example… but even so we know it’s not perfect as nothing ever is.
Thanks Toonartist. I’m replacing my 272/555dr with a 282/HiCapDR in a couple of weeks I may try the MScaler/HiCap on the same Fraim level MScaler on the left side of the HiCap.
Just picked up 2x Mark Grant HDX1 Pure Copper Digital Coax BNC cables (I’m 10 minutes away) lovely finish and the BNC plugs don’t wobble like the supplied cables.
First impressions a very nice lift and for only £30 per cable, bargain!!
Your change of heart is important to me because I too seem to like R2R DACs, and Audio Note make the very best of those - and to think that the Qutest can carry across the same emotive content and digital integrity as an AN 2.1 - or better in fact - is food for thought no doubt about it.
I was slightly concerned that just about everyone raved about the Chord DACs where in fact they may have been up there emotively and expressively with the likes of AN - your response perhaps proves otherwise and you are a keen listener. Perhaps the unanimous praise of Chord DACs is justified after all!
Can I ask you also if you are still using the Meridian Streamer as you choice of transport (the transport quality from spdif is highly important)?
Yes, the Meridian is still performing sterling service. A friend bought round in dCS Network Bridge just before lockdown 2 for us to compare. Unfortuneately for some reason it refused to play with Bubbleupnp and Kazoo, althoug it did before. It may be that my friend had made some config changes. Ultimately I don’t allow WiFi onto my LAN and so rather tha play with my firewall we agreed to do a replay at his place post lockdown. Give me a chance to listen to his 4.1 vs the Qutest …AND I will take over the M-Scaler (+ Krisdonia) to we can do that comparison too, should be interesting.
Unfortunately the quality of SPDIF does make a difference. I ended up buying the WAVE as once I heard the difference I didn’t want to surrender it; likewise with the ENO filter and streaming cable. That said the Krisdonia (for the M-Scaler) gives excellent VFM, followed by the SBooster & SBooster UltraII.
If you pop over to The Wam review section you can see my thoughts in excrutiating detail!
BTW I have settled on my TT2 to Etude interconnect. I have reverted to unbalanced and am using Duelund DCA20GA tinned copper multistrand wire in cotton and oil with ETI RCA connectors… parts courtesy of Hi-Fi Collective.
Absolutely gorgeous, yes the bass slam is reigned in, but seems to sound more organic and natural across a wider gamut of recordings. High end percussion tends to sound more natural along with voices and sibilance.