Technically the ProISL-USB interface is better than the Network Renderer V2 interface in terms of specifications, but which interface sounds better or more truthful to the original sound on the recordings?
Has anyone given it a serious listening comparison?
@Daniel_Gullman and @Jonathan_Gullman, can you please share some of your experiences with the sound quality differences between the ProISL-USB and Renderer V2 interfaces using the Cascade DAC? 
I have both and havenât noticed a sound quality difference. I canât say Iâve done a critical listening test to evaluate this, but nothing has jumped out at me as needing a closer look in my normal listening.
One thing is maybe out of the ordinary in my setup. I use an Activ Audio EMI filter on the input to the Renderer and as part of a belt and suspenders approach to keeping noise out of the Cascade.
One of the advantages of the Renderer, if you use Roon, is that you can set the Roon output to fixed volume but still adjust the Cascade volume from Roon interface. In this case the volume is controlled by the Cascade, the same as using the MSB remote⌠Roon doesnât fiddle with the bits to change the level.
Dan
1 Like
I am technically curious to read the answer that the MSB owners will give you, but in my opinion the question is posed in a captious way, in the sense that we are not making an apples to apples comparison , but between two different objects.
Let me explain better:
-
The ProISL-USB interface galvanically isolates the USB signal coming from a source, but it is not an âactiveâ component in the transport, meaning that it isolates the signal coming from the outside, but takes it as it is, without any modification, and takes it to the DAC (to the DD part) for all the further elaboration, etc.
-
The Network Renderer V2 interface, on the other hand, takes a train of digital information arriving from the LAN via a Cat5/6e cable and applies its own signal processing to transform it into the same signal (hypothetically) that would arrive at the USB connector of the ProISL-USB from a hooked source.
So one transports (ProISL-USB) and one transforms (Network Renderer V2).
Maybe I missed some discussion here on the forum or on other forums online, or in interviews on various social media or YouTube, but if you wanted to compare the qualities of two interfaces that do âdifferentâ things to understand which is more respectful of the âoriginal musical signalâ, you should think that:
-
It makes absolutely no difference which network server/player you connect to the ProISL-USB because the signal coming from a Raspberry Pi 5 with proper SW or a Taiko Olympus server is somehow homogenized and brought to the DAC âat the same quality levelâ without preserving any sonic differences (if they are present and real).
-
The quality of the Network Renderer V2 is so high and specific that all the audiophile awesomeness regarding the network servers/players on the market are just an idea that we users have, but in the end the final result is identical both electrically and in terms of sound to the original message âjustâ connecting a LAN cable to this interface.
Mine is a real question, perhaps naive, and I certainly donât want to embarrass anyone or worse open Pandoraâs box, but really understand how things are because I think it is always important to learn and have a mind free from prejudices, so any answer (if possible) is very welcome.
Iâm really asking to understand better, even a PM would be fine and I apologize in advance if it can create some embarrassment o worse, it is just to talk and have a real picture from industry insiders.
I agree with @Dan on this. With the Cascade I feel it would be hard to find a difference between the two. In regards to the djm filter, weâve known about and used it for years. Itâs great and very reasonably priced for the performance gained.
I find the choice between Renderer and USB to be more of a decision of what software do you want to use. If you want to use Roon or JPlay and control the DAC volume, get the Renderer. If you want to use an Aurender and their software, get the ProUSB. The ProUSB is more flexible as it works with everything (Windows, Linux, MacOS) and almost all software all while provided great isolation.
1 Like
Hi Alberto,
Both ProUSB and Renderer are capable of bit-perfect transfers. Both go through a bunch of
transformation hoops to push the bits over the medium they are using. ProUSB can handle a higher bit-rate than Renderer, but both do bit-perfect transfers.
MSB gives us a way to test whether or not we are doing bit-perfect transfers.
If an interface fails this test, something is broken.
So, except for the possibility that one or the other interface somehow lets some noise into the Cascade I canât see a difference in audio quality. But in the end our ears are the best tool for assessing audio quality, and itâs easy to think that an electronic circuit is âperfectâ only to find out later by listening that it isnât.
Roon gives you a choice of sending bit-perfect into to the DAC or processed bits. As long as you donât use Roon Muse, or the Roon head room adjustment or volume control you should be sending bit-perfect to the DAC. The Roon interface lets you see what it is sending. Here it is telling me it is send bit-perfect via RAAT over the internet to the Renderer.
And here it is telling me it is sending bit-perfect over ProUSB.
I have both interfaces and Iâm torn between which one to use. Iike the ProUSB because it connects to the Cascade via fiber, i.e. there is no galvanic connection⌠I live in Central Florida 
I like the Renderer because I can control the volume from the Roon interface.
I guess these are first-world problems 
Dan
4 Likes
This basically comes down to the statement, âbits are bitsâ and if itâs bit perfect it should all sound the same. However, we all know that for some reason there are still differences. Why do two servers sound different with the same level of isolation? There are a bunch of theories as to why, but none of them are so strong as to be the clear answer.
We noticed some of these sound quality differences developing the digital director. The ProISL/ProUSB blocks 100% of the electrical noise because of the fiber. However, passing it through the digital director (with no processing) it sounds significantly better than direct to the DAC. The DAC is getting the same data exactly from the same source just one has the double isolation. Why the difference? One of many theories is cleaner power supplies on the fiber transmitter resulting in cleaner transmission of the data. This an many of these theories were implemented into the Cascade DAC.
The result with the Cascade or the Digital Director the differences between inputs is vastly smaller. Weâve gotten many reports that the difference between the inputs is small enough that they couldnât pick it out in a blind A-B test.
All of this is based on the assumption the playback software remains bit-perfect and doesnât do data manipulation, watermarking, etc⌠which happens more often than you would expect.
3 Likes
Annnnnnnd, before the introduction of the DD, the Pro USB often outperformed the Renderer. Adding a renderer computer and network into the DAC conversion chassis was never a good thing. We did our best, and with proper setup it was very close.
But with the DD, and proper setup, I can not hear a difference personally. This is of course not comparing different sources on either one, but the same source to both.
3 Likes
Yes, we should be comparing with the same source using the same server-streamer that has both USB digital output and Ethernet output, e.g. an Antipodes server-streamer. That way we can keep the server and player apps from the same hardware for the USB vs Renderer comparison.
5 Likes
In this way I ubderstood better what you meant, thanks.
1 Like
Ah ah ⌠you are right !
Thanks Dan 
1 Like
This got a great sense, I mean, if I invest in a super product as an MSB DAC and it plays the brain of my system, acting also as preamplifiler, I can chose a network player that suits my needs in term of software interface and capability, or âsimplyâ use Roon with the MSB Renderer and âjust enjoy musicâ rather than looking for a needle in a haystack
5 Likes
This is why we worked so hard on making all the inputs sound equally great. The entire point should be listening to the music you love using an interface you like.
2 Likes
I assume this would apply also to my humble MSB Discrete DAC, without the DD? I was advised that the ProUSB / ProISL option would be better for my recently acquired Discrete than the Network Renderer. But an issue between the interface of my Aurender and Discrete may necessitate me having to get the Network Renderer to use Roon which has a helpful function/ setting that eliminates the issue. BTW Iâm new to the MSB family. The Discrete is my entrance to this world. Thanks, and thanks for this thread.
2 Likes
Hello and welcome! Correct, the Pro USB would probably be the superior choice, but we also need to keep in mind these discussions are for eeking out every tiny bit of available performance. You wonât be disappointed with the Renderer in the Discrete DAC. We try to level the playing field of all the input modules so you can use what works best for you. 
Thank you.
I donât know if it is in your line of sight, or even if it hardware-wise feasible, but the integration of hqplayer naa in the renderer would be a great addition for those who like to have different flavors to the music.
Hqplayer signal processing is very well implemented, and the filter options do sound different, especially when doing âcd ratesâ x DXD vs DSD, with long and short filters.
Integrating in the rendering would spare having a streamer like a raspberry pi or something like that to bridge network and dac.
The main problem is the HQ player is very processor intensive. You donât want the processor in the DAC churning away on some noisy algorithm inside the chassis. Much better to get that noisy stuff in some remote computer system instead. And leave the renderer to pollute the environment inside the chassis as little as possible. The DSPs run deterministic code that is synchronized to the playback oscillator and actually does a significant amount of âuselessâ processing to significantly reduce the noise generated by evening out the power draw of the processors. This cannot be done with the renderer which has to communicate with a network for playback and all the processing is asynchronous. This means that any extra processing that the renderer has to do directly increases the RF pollution inside the DAC.
Thanks for the reply, and I completely agree.
But I guess I did not make myself clear. HQPlayer server requires heavy lifting and lots of compute power. The only company I am aware to integrate HQPlayer in its dac is Acqua. But to me it makes no sense to integrate it in a dac or in the digital director.
My suggestion was to integrate the hqplayer endpoint, which is called naa (network audio adapter). It is a very light piece of software, just like the roon bridge. And just like roon bridge, it communicates via network with the computer running hqplayer server (mine is even kept on another room, so the noisy fans may run full fledge).
The endpoint (naa) is easily run in almost any computer board, including a rasp pi2 with 1GB of ram. Again, just like the roon bridge. It would be a very light load, and work as another option to the renderer (you would have 3 options to choose from: upnp renderer, roon bridge (roon renderer) or hqplayer naa (hqplayer renderer).
Please take a look here for further reference (including manufactures who already integrated the naa on their network products):
https://signalyst.com/network-audio-adapter/
If you have any computer at hand (windows, Mac or Linux), you can easily download the naa and take a look on how lightweight it is.
Just speaking for myself.
I find the Cascade to be utterly quiet. I would not want to see any additional computation load inside the DD (or in any of the other chassis), even that which would be necessary to just receive data via NAA.
Iâm pretty sure, one way or another, there would be clock management issues if the DD implemented a NAA interface. The issues Iâm thinking of would be around the Cascade having to resync the incoming data to its clock. Just that would introduce more noise to the system.
If MSB were to implement an NAA interface, I would think, actually hope, it would be as a digital input module connected by fiber, as the ProUSB module is. That way the noise could be better managed by pushing as much as possible of the implementation on the far end of the fiber. For example, the word clock could be sent down the fiber so that the data could be resyncâd to the Cascade clock away from the Cascade, in the dongle that the USB cable plugs into. I think that is what is currently done with the ProUSB interface.
I would like to use HQPlayer myself, just to be able to try some of the features it has. If I do that, it will be with, maybe, a Raspberry Pi implementation that attaches to the Cascade via a ProUSB Cascade module. My gut feel tells my that HQPlayer would not improve on the accuracy of Cascade processing the original bits of the stream, but I would like to experiment to confirm that.
I wouldnât want an NAA interface implemented inside the DD. I wouldnât even want that logic inside the DD even if I wasnât using it.
Again, these are just my thoughts⌠I donât want to see anything done to the Cascade that could, even, possibly compromise the low noise level that it has. I think that low noise level is a key factor in its incredible performance.
Dan
I beg to differ. The noise would be the same as in the roon bridge, which is already implemented in the renderer. In fact, running htop, roon bridge demands more resources than the naa. I would not expect to have any difference in that regard.
And it would be a lot simpler just to connect the ether liner cable, than now, having to run a rasp pi, with its own power source etc.
It is not the end of the world, but it would be value added.
As for hqplayer, you really should experiment with it. You would be surprised how much the sound signature changes depending on the oversampling and digital filter elected.