MQA update for dCS Rossini now available

Announced just now on social media.


Once all is said and done you should be at the following versions for this release:

Rossini Control Board: 1.10
Network Firmware: 368
dCS iOS app: 1.4.8

A few things to note here:

  1. This release contains a number of improvements in the Roon endpoint code which will make playback much more stable than what you’ve experienced in prior releases.

  2. MQA playback (full unfold) is supported via the Network interface using either Roon, UPnP, native Tidal, or a direct-attached USB hard drive. MQA rendering is supported by the USB1 (type B), AES, and S/PDIF inputs (meaning the first unfold needs to be done elsewhere if you’re using anything but the network input or playing files from a locally-attached USB memory stick or hard drive).

  3. You’ll notice a new filter selection (M1) which is the MQA filter and it should be selected automatically when MQA data is played. If this filter is not automatically selected then you can either do a factory reset of the Rossini (via the unit’s menu) or select the filter manually. Once the M1 filter is associated with an MQA stream on a given input it should be automatically selected in the future.

  4. Updated versions of the Rossini Player and DAC manuals are available on the dCS website:

  5. There is (will be) an update to the dCS app to version 1.4.8. When I last checked a couple of minutes ago I wasn’t seeing it on the app store yet, but it will land very soon. The dCS app has a number of improvements related to UPnP and native Tidal as well as controls for new functions.

This release was a lot of work and I know that the delays were frustrating, but it will be well worth the wait. The Roon improvements alone are going to be a big deal for all here.

If you have any questions please don’t hesitate to ask them here.



Getting ready to download NOW!!

1 Like

Just a note on the download. Occasionally where there’s a simultaneous release of network and device firmware both files won’t get properly downloaded when you tap update. Once the update is run check versions again and make sure that you got 1.10 for the Rossini Main and 368 for the network. If you didn’t get either of those then just run the update again and it will grab the one it didn’t get on the first try.

All updated and it is sounding great.

Again, thanks for providing all of the information. It made everything easier to update without issue . Waiting for the app update to come through.


The app may not drop until tomorrow. It’s in Apple’s hands now…

Be sure to check the filter setting when playing an MQA stream to ensure that the M1 filter gets selected.

The first few tracks played before I noticed that the display had not automatically switched to M1. Once I manually selected M1 it now seems to be working properly.

1 Like

I would think doing a factory reset once all updates are in place is probably a good thing.

The downside is that it will erase all customizations that you’ve done. For most things this isn’t a big deal, but a lot of people put a lot of effort into choosing filters on a per sample rate basis and these settings will be reset completely.

If I understand you correctly, #2 implies that MQA files played bit perfect from a computer over the USB interface will not be decoded, only rendering is supported via this method (ie you need a first unfold to happen on the computer). The updated docs say something different I think. On p 18 in:

where it seems to imply that untouched MQA files over the USB interface (meaning for example connecting the DAC to a computer and using Roon to play to it) will be fully decoded in the Rossini.


There are two USB interfaces. USB1 is the type B connector on the rear panel that you would normally connect a computer to. USB2 is the type A connector on the rear panel that you can connect a USB memory stick or hard drive to.

Full decoding is supported by the Network and USB2 inputs and the reason for this is that the network card is what is doing the first unfold. The USB2 interface (along with the browsing of the files on that drive) is provided by the network card.

The USB1 interface goes through a more traditional XMOS USB receiver and the only way for the DAC to perform the full unfold here would be to ship the data to the network card first then back out to the DAC.

There are some underlying reasons as to why this particular approach was chosen and it is consistent with the MQA implementation that will take place throughout the dCS product line.

1 Like

Thank you AMP for the explanation. Interesting to learn about the architecture.

It also makes more sense of how this would work with the network bridge. I presume the bridge would do the first unfold only and deliver that to the DAC to then render. I also wonder how this would work in the case of the Vivaldi DAC since this DAC does not upsample. It would probably need the Vivaldi Upsampler to do full MQA decoding. Is this right?

That is correct. The Bridge will output the results of the first unfold without any additional processing of that data. Before someone asks the next question… soon (I can verify that it’s real and it works :wink: )

The rendering process is lighter-weight and in the case of the Rossini is being done in the DAC’s FPGA along with all of the other signal processing. The Vivaldi implementation should be similar in that it will take the results of the first unfold and render it per the MQA algorithm.

The big benefit to the dCS implementation and the reason that MQA is so excited about it is that all of the MQA-related processing is folded in with all of the DSP and filtering that the dCS products do. In other implementations MQA is a bolt-on which sends rendered data out to the DAC chip. With the dCS implementation it’s completely “woven in” which simplifies some of the processing and filtering.

The upsampler will perform the first unfold much in the same way that the bridge will. The difference is that it can’t do any further processing on that stream as that would invalidate the MQA data. Per my understanding of the restrictions imposed by MQA it can’t do the full rendering as that would result in fully unfolded and rendered data being sent through its outputs (which is a no-no).

One more thing to note: Using the Bridge with the Vivaldi DAC is becoming a popular combination. The Upsampler offers better performance, but that comes at a cost which puts a streaming Vivaldi setup out of reach for those willing to buy the DAC, but can’t swing the Upsampler. Bridge + Vivaldi DAC should be fully supported for MQA streaming as well.

The rendering stage is upsampling with MQA-specified filters (which is presumably an improvement on generic upsampling). My understanding is the Vivaldi DAC does not do upsampling, period. So connecting the bridge to the Vivaldi DAC does not get you full MQA but just the first unfold (which I think is the vast majority of the effect, but that’s another discussion).

Yes, agree that MQA Ltd would not agree on the Upsampler outputting an upsampled PCM stream. One way around it would be for the upsampler to only output rendered MQA to the dual-AES encrypted output of the upsampler into the Vivaldi DAC so only this DAC would be able to use it.

Kind of, but it’s not quite as straightforward as that. Regardless, there’s some math going on.

This is incorrect. All* DACs do some form of up/oversampling in order to get the bit stream into the preferred format for the DAC itself. In the case of chip DACs this is typically DXD and in some cases DSD (or a multibit SD format).

(* there are DACs that don’t up/oversample but the vast majority of DACs on the market, and especially the very well-regarded ones, do. Makes the math so much less problematic)

The Vivaldi DAC does upsample the incoming data stream in order to convert it to Ring DAC format which in the case of the Vivaldi is a hybrid format at about 56Mbit/sec (vs approximately 18Mbit/sec for DXD). The horsepower to do rendering is there. The difference between the Rossini (which includes upsampling functions) and the Vivaldi is that you can make some selections about the intermediate formats on the Rossini whereas on the Vivaldi you can’t.

The benefit to the Vivaldi upsampler is that it allows for more sophisticated processing than the Vivaldi DAC along with selection of output format and an additional layer of digital filtering. There are also benefits to noise isolation and power supply optimization.

1 Like

I run the update. Network updated but DAC firmware still not updated. Run it multiple times, downloading, closing the app, waiting… I can still connect to the Rossini over the app (v 1.4.8 by the way) but the display shows nothing on the unit.

Ok eventually it happened and it is all good now.

Listening to the Koln Concert in MQA. Pretty nice. It does seem to be a remaster as the tape hiss is noticeably higher than other versions I have (I have them all, probably). But it is gorgeous.

Indeed good news!

I have using my brand new Rossini DAC and Clock for last 10days or so on and off. I found, like many, that the display freezes when switching to different sampling frequencies. I managed to freeze/hang the whole DAC over a handful of times and no buttons on the front panel can do anything about it, resorting to good’ol power switch at the back to reset it. I wonder if the new network board firmware will fix that.