New MSB Cascade DAC

For those interested in this subject, one of the major difference between space and military electronics and commercial audio electronics has to do with conduction versus convection cooling. The type of cooling impact how much you are able to accomplish with regards to EMI and RFI mitigation, but never the less similar tools and techniques often apply.
I’m fairly confident few customers would appreciate coolant lines running through their system in addition to the electrical cables :) In the end all our equipment is convectively cooled, but these days I never design for convective cooling within the case. Vented cases allow ingress of all sorts of stuff over years of service (believe me I have seen it all sorts of crazy contaminants) so we design conductive cooling solutions for all of the internal components to couple them to the casework, just for the sake of mitigating contaminant buildup, although it does assist with RFI mitigation as well (no case penetrations solely for cooling).
 
I’m fairly confident few customers would appreciate coolant lines running through their system in addition to the electrical cables :) In the end all our equipment is convectively cooled, but these days I never design for convective cooling within the case. Vented cases allow ingress of all sorts of stuff over years of service (believe me I have seen it all sorts of crazy contaminants) so we design conductive cooling solutions for all of the internal components to couple them to the casework, just for the sake of mitigating contaminant buildup, although it does assist with RFI mitigation as well (no case penetrations solely for cooling).

The problem is that unless you are machining down to mil and sub-mil level tolerances for flatness, then you might as well have slots and openings in your enclosures with regards to EMI and RFI. Hence the use of EMI gaskets.

Having cold-plates mitigates some issues but they introduce other challenges.
 
  • Like
Reactions: Mentanalyste
Thank you for posting about the genesis of the Cacade, it's very interesting.

The Cascade Link is an advanced descendant of the optical Pro ISL interface which completely shields the conversion box from any upstream noise while simultaneously producing a vanishingly low amount itself. It also has a vastly increased data capacity to feed enough uncompressed data directly to the (8) DAC modules.

Would you be open to posting more technical details about this interface? Is it synchronous or asynchronous? How are the data bits framed? Does it carry any control signals or error correction? How are the DD and Converter box kept in sync? Is there a back-channel to carry the Femto clock to the DD?
 
Would you be open to posting more technical details about this interface? Is it synchronous or asynchronous? How are the data bits framed? Does it carry any control signals or error correction? How are the DD and Converter box kept in sync? Is there a back-channel to carry the Femto clock to the DD?
The Cascade Link is bidirectional with 512x clock and link monitoring data (link error states and setup data) on the feedback channel. SFP+ phy. (must meet data rate requirements, no direct attach copper support) Synchronous with auto frame slip compensation (allows any cable latency) scrambled to approximate Gaussian thermal noise, scrambling correlation frame alignment, DC balanced, no FEC but with HD9 CRC providing data integrity check. Frames are 1 set of DAC module data + CRC + 1 bit of control data per frame about 1.5Gbps currently (8 DAC modules), depending on source data format, extensible to about 3Gbps (16 DAC modules).

It was extremely fun to design the fpga code for this one.
 
This interface discussion is fascinating, but it’s worth noting that from a pure jitter standpoint, the problem is completely solved. I enclose below a measurement of an inexpensive $800 Eversolo DAC-Streamer DMP-6 that has jitter below -140dB over the standard USB interface (source:audiosciencereview.com).

Just to ensure my post is not misinterpreted, I am in no way suggesting the Eversolo DMP-6 subjectively sounds the same to your ears as the MSB or any other high end DAC. I am just pointing out that in terms of jitter, there’s absolutely no reason to go beyond USB. It’s quite possible that such a measurement does not reveal other issues. But from a simplistic standpoint, there’s just no intrinsic reason I can see to say we need a very complex new interface to improve jitter when the distortion is below -140dB. Just so you understand, even the best loudspeakers on the planet struggle to keep distortion under -60dB over the entire frequency spectrum from 20Hz to 20 kHz. We are already -80dB below what the best loudspeakers can resolve.


IMG_6239.jpeg
 
With all due respect to the measurements:

The Sound of Eversolo as a DAC is exactly like its price.
 
  • Like
Reactions: defride and Armsan
This interface discussion is fascinating, but it’s worth noting that from a pure jitter standpoint, the problem is completely solved. I enclose below a measurement of an inexpensive $800 Eversolo DAC-Streamer DMP-6 that has jitter below -140dB over the standard USB interface (source:audiosciencereview.com).

Just to ensure my post is not misinterpreted, I am in no way suggesting the Eversolo DMP-6 subjectively sounds the same to your ears as the MSB or any other high end DAC. I am just pointing out that in terms of jitter, there’s absolutely no reason to go beyond USB. It’s quite possible that such a measurement does not reveal other issues. But from a simplistic standpoint, there’s just no intrinsic reason I can see to say we need a very complex new interface to improve jitter when the distortion is below -140dB. Just so you understand, even the best loudspeakers on the planet struggle to keep distortion under -60dB over the entire frequency spectrum from 20Hz to 20 kHz. We are already -80dB below what the best loudspeakers can resolve.


View attachment 130654
We should use measurements to validate what we hear (or don’t hear). MSB are pushing the envelope by questioning things that shouldn’t matter but sometimes do.
 
  • Like
Reactions: hopkins
We are working on a module which will provide a direct MSB PRO ISL fiber interface. For joint MSB/Taiko customers this means they can bypass/eliminate their MSB Pro USB adapter and USB cable and/or MSB Network Renderer V2 module if they so desire.
After some productive discussions in Munich with Emile, we were able to lay out plans for a collaboration to provide a Pro ISL output for Taiko. As soon as we got back to California, we sent out a development kit and DAC for them to test with. It sounds like it is all hands on deck at Taiko to get this solution up and running. We will be providing the following Pro ISL output card which they will integrate directly to their XDMI interface. As soon as they have it working and we certify it meets our bit perfect performance with clock feedback functionality we will be sure to announce it here. This will bring Taiko integration to all Discrete, Premier, Reference, Select, Cascade, and Sentinel DACs with a Pro ISL input module.

Keep us posted on your progress @Taiko Audio !
IMG_6182.jpg
 
Here, we have a DAC outperforming the Select DAC and we made the radical decision to sell it for less. Why?

We could have arbitrarily decided to sell the Cascade DAC for 165,000. This is genuinely what industry folks asked us to do when we first revealed the pricing… it’s a thing. I hate it.

So, in the end, why choose 95k vs 165k? This is where the buyer wins, not the “greedy corporation”. I am terribly concerned over industry behavior towards charging whatever will be paid.
Hi Daniel,

Your unabashed honesty is astounding and I wish this pricing philosophy was emulated industry wide. Unfortunately, this appears to be highly unlikely. If it was, I believe the hi end audio market could increase sales and become more affordable to a wider hobbyist income sector. A win - win in my view.

Thank you again for your refreshing consumer friendly, albeit rare approach to retail pricing.

Gordon
 
Last edited:
Hi Daniel,

Your unabashed honesty is astounding and I wish this pricing philosophy was emulated industry wide. Unfortunately, this appears to be highly unlikely. If it was, I believe the hi end audio market could increase sales and become more affordable to a wider hobbyist income sector. A win - win in my view.

Thank you again for your refreshing consumer friendly, albeit rare approach to retail pricing.

Gordon
Thanks for the kind words Gordon. I am optimistic things will course correct. This is an industry driven by passion for audio, something that will outlast short term schemes. I have no delusions that our products are not still very expensive, but we will try and push in favor of the end user.

Cheers,
Daniel
 
For those that are curious as to the origin of the Cascade DACs design here is an overview its origins.

Eleven years ago, when I was designing the Select DACs electronics (the seventh DAC of my design that made it to production after the MSB Platinum DAC, Link DAC 3, credit card DAC (a promotional free giveaway at CES one year),Platinum DAC 4, Power DAC, and Analog DAC) I thought that I knew everything that was necessary for a DAC platform that would last for the longest time possible and have the best possible performance.

I had worked out the role of clocking (extensive experimentation into different clock architectures and measurements on the role of sound quality), processing (digital filters and data stream formatting), noise (the importance of digital isolation and conversion noise removal), conversion hardware (a software configurable analog system that could playback any format on the fly from single/multi-bit delta-sigma to PCM) and upgradability (input formats had changed rapidly and are, still changing today).

For Daniel it was going to be his second solo chassis design after the Analog DAC, and he was getting pretty good at that as well. What I did not anticipate at that time was just how electrically noisy things would become. I remember trying to diagnose a cluster of repairs for one of our transports, going back and forth with a couple customers to figure out why they were failing, and realizing that the problem actually stemmed from the video projector they were attached to.

With tens of volts of RF flowing over the HDMI cable, failure was inevitable. Many watts of continuous RF power were being dissipated by the protection circuitry which would eventually burn out the parts (this was obviously not good for sound quality either). That is when I decided to develop the Pro ISL optical interconnect, in my opinion the first high quality optical digital audio interface (toslink is not great and “ATT/ST glass” is also very problematic compared with even lowly s-pdif). Since then noise has only gotten worse. Ambient conditions and system interconnection noise have become “THE” nearly uncontrollable problem. Why is it that everyone is talking about isolators, routers, servers, cables, transformers and even battery power as being so important?

Well they are all nibbling at the edges of the same gigantic problem. Noise is leaking from our everyday audio and non-audio appliances into the delicate analog audio signals. Proliferation of solar inverters, LED lights, dimmers, communication equipment, displays/TVs, audio servers, computers and chargers are all contributing to this problem. Even the digital audio interfaces themselves generate and conduct objectionable noise despite faithfully transmitting the data they contain. This noise can be highly circumstantial meaning some customers don’t have much of a problem, but others are constantly struggling to get good sound.

Some even become disillusioned with digital audio altogether and go back to their LPs/tapes exclusively. This noise can propagate through the air, over the power lines and over any conductive cabling. I had designed a very good (two stages of isolation, one at each of the input modules, one stage at the DAC modules themselves) isolation system into the Select, but, I have found that “very good” is no longer sufficient today. The Cascade DAC shares much of the same design methodology of the Reference/Select but it takes noise isolation to the extreme.

Only the bare minimum of circuitry necessary to convert the raw data-stream to analog is present in the converter box and that box is heavily shielded physically, magnetically and electrically from its environment. No processing occurs in the Analog Converter box at all. Even the power to its internal relays (latching) is disabled when they are not switching to eliminate the problem of noise leakage from the power supplies driving their coils (a problem I discovered when prototyping the electrostatic headphone amplifier because of its extreme 141db of dynamic range). The outboard power supplies job is to heavily shield the converter box from the noise on the power mains (via onboard power filtering, magnetically and electrically shielded transformers, ultrafast rectifiers, inductor/capacitor filtering and ultra low noise, discrete regulators).

The Cascade Link is an advanced descendant of the optical Pro ISL interface which completely shields the conversion box from any upstream noise while simultaneously producing a vanishingly low amount itself. It also has a vastly increased data capacity to feed enough uncompressed data directly to the (8) DAC modules. The first time I heard the prototype Cascade in our electrically noisy facility I was shocked at how much of a difference this “no holds barred” approach to isolation was. Milling machine working hours, office computer use, solar inverter activity, music server software updates no longer had an appreciable impact on the music quality.

From there it has been a pleasure listening to and refining every last detail of the Cascade. We didn’t rush any of the details so each component had a significant maturation process, for example just the size and exact shape of all the stainless accents had prototypes machined and evaluated daily for weeks until we were completely happy with their exact shape and form. I spent hundreds of hours programming, evaluating and refining each of the playback formats supported this far (there are 17 individual data formats at launch, each with a different playback algorithm and unique DAC configuration).

I spent hours listening to the sound of the raw bit-stream of the Cascade Link (not music, the raw data bits themselves) to match the analog gaussian noise of analog electronics as closely as my hearing could discern (“real” not “artificial” noise) so as not to pollute the optical receiver with digital noise artifacts. A high resolution FLIR thermal camera was used to evaluate and refine the exact case contact to stabilize the temperature profile of the DAC modules.

We stacked and evaluated every configuration of the separate chassis to determine the optimal placement of the magnetic shielding for every one so that our customers could place the boxes anywhere they liked in their system without worry. Daniel even designed and machined custom fasteners to optimally hold the power transformers in place. All the while we would constantly debate and recenter each design feature to keep them focused and not unnecessarily complex or expensive. Cost no object performance, but without waste, is the philosophy we aspired to follow for the Cascade. Everything that was required for world leading performance and intuitive operation but nothing more.
Still loving my "oldie" Analog DAC! :) :)
A beautiful piece of gear and a great product! Hope you'll continue to support it...
 
Last edited:
uber dac-only makers will at least need to offer a back channel wink-wink to customers about a future Taiko interface option. it's a box that will need a check mark to proceed. we can't yet know how big a deal it might be, or not be.


What is leading you to this conclusion? Is it because of Lampizator? I still don't get all the fuss around the Taiko, it's windows based for heavens sakes. How much better can it actually be than any other top-flight server (Oladra for example) and why would any company, particularly MSB build around that platform. I just can't see it happening, if anything MSB has way too much invested in their own technologies and manufacturing capabilities to partner when they likely think they can produce a better version themselves.
 
  • Like
Reactions: AndrewChen
What is leading you to this conclusion? Is it because of Lampizator? I still don't get all the fuss around the Taiko, it's windows based for heavens sakes. How much better can it actually be than any other top-flight server (Oladra for example) and why would any company, particularly MSB build around that platform. I just can't see it happening, if anything MSB has way too much invested in their own technologies and manufacturing capabilities to partner when they likely think they can produce a better version themselves.
You may want to read this:

Thread 'MSB and Taiko, let's go!'
https://www.whatsbestforum.com/threads/msb-and-taiko-lets-go.38824/
 
What is leading you to this conclusion? Is it because of Lampizator? I still don't get all the fuss around the Taiko, it's windows based for heavens sakes. How much better can it actually be than any other top-flight server (Oladra for example) and why would any company, particularly MSB build around that platform. I just can't see it happening, if anything MSB has way too much invested in their own technologies and manufacturing capabilities to partner when they likely think they can produce a better version themselves.
Our Pro ISL interface is available for any server company to implement. Antipodes is also working on an implementation(maybe this is supposed to be secret), we shared several rooms in Munich with them, great product! There are a lot of great options and we aren't here to encourage one over the other. I always want to clarify we are not developing anything for Taiko or anyone else. But, if someone gets the Pro ISL implemented with bit perfect playback, it would allow the server to be distanced electrically and physically with the fiber connection further from the analog side of the system. The Digital Director fills this roll, but every bit helps. Also, they can have a direct I2S interface instead of using standard USB drivers.

Keep in mind, none of it is real until it is. Never buy on the promise of tech, only what is in the box. Just my two cents ;)
 
Well I stand corrected. My years of owning a Discrete and then a Premier MSB was generally obstinate towards streaming in general. Then the Digital Director hit, which is why I just assumed they would want their own product. My how things have changed.
Happy to support others, but we are a long term digital company. Servers based on consumer PC's don't have the lifespan we require... Modern computers last about 4 years before they start degrading. Anyone buying an expensive server needs to manage the expectations of a short product life. The DD helps all servers do better and is a great buffer between streamers and our systems. Opens up the choice for more streamers in leveling the playing field. Something we can support long term.
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu