Aries Cerat Kassandra II Ref DAC: Wow!

Believe High Fidelity said:

Thank you Astro for taking the time to write up such a detailed and accurate review of the Kassandra Ref DAC. I wish you the most enjoyment possible out of the equipment

Click to expand...

Thanks for the positives. I have posted on head-fi and hifi-wigwam as well, as I go one those quite a lot. Yeah. one AC component has transformed my system, so it shows you can start with that, enjoy it anyway. I would love to have more AC bits in there as well, I can imagine the synergy and the effect of that. The simplest way I can describe the effect on me personally, is it has 'returned' to real music. Hard to describe, but it is less hi-fi somehow, less in the way, more pure musical enjoyment.

Julian, not enough audiophiles either find the components that talk to them.

Or they do, but then try and learn a new language.

It is VERY good to have the realisation dawn on you that you have reached nirvana, and that it’s time to kick back and relax.

Brad, maybe in another life, but I’m settled.

My lightbulb moment on digital has already come with the Eera Tentation cdp.

I’m v happy with my NAT 211s, and the Zus suit my room and general sound preferences.

No, just acknowledging Astrostar’s good fortune.

Has anyone tried comparing NOS vs upsampling and applying a 24b/384k or 24b/352.8k input through USB from a capable source like HQPlayer? I was wondering if a better (or worse) result is achieved through upsampling on this DAC considering the 18-bit DAC IC used, thanks.

kma said:

Has anyone tried comparing NOS vs upsampling and applying a 24b/384k or 24b/352.8k input through USB from a capable source like HQPlayer? I was wondering if a better (or worse) result is achieved through upsampling on this DAC considering the 18-bit DAC IC used, thanks.

Click to expand...

I have extensively. While it can make a difference in other DACs, the Kassandra does not benefit in anyway from the filters and upsampling through HQPlayer. Upsampling as a feature (not improvement) tends to collapse the soundstage in favor of better or fuller imaging. You mileage may vary, but I do not subscribe to these "filters" or "tweaks" as anything but exactly that. Not needed with the Kassandra.

Believe High Fidelity said:

I have extensively. While it can make a difference in other DACs, the Kassandra does not benefit in anyway from the filters and upsampling through HQPlayer. Upsampling as a feature (not improvement) tends to collapse the soundstage in favor of better or fuller imaging. You mileage may vary, but I do not subscribe to these "filters" or "tweaks" as anything but exactly that. Not needed with the Kassandra.

Click to expand...

My findings as well. Feed it resident rate, no messing with the signal, keep it as pure as is possible. Upsampling in DS was created for other reasons than sounding better, rather to fix problems created by the filtering. Huge subject, but as far as the Kassandra sounds, leave the rates alone. It loves 96 and higher rate data as well, again, leave the rate as the file playing.

Thanks guys. Interesting observations on the upsampling for this DAC. I guess it does make sense that every DAC will respond differently. I have an Aqua La Scala (the PCM1704 version), and I actually notice the opposite when I upsample with HQPlayer (soundstage grows and image density softens). I tend to like upsampling better for my DAC.

In looking at some of the pictures posted of the internals of the Kassandra, I see a Pulse Clock OCXO is being used (which is very low phase noise, resulting in superior jitter per the DAC manufacturer's claims). However, I only see a single OCXO in the box, which probably means the DAC is optimized for jitter for multiples of 44.1kHz rather than 48kHz Fs. Therefore, upsampling a redbook format to 192kHz or 384kHz may not be as optimum as 176.4kHz or 352.8kHz from a jitter perspective. Maybe this has something to do with the magic being lost with the upsampling. However, given you guys probably tried all combinations of upsample ratios from native rates, then maybe the overall jitter difference is not the reason for the change in sound. Or maybe I can't see everything in the box just from the pictures I see on the net

I've read from various message boards that Astrostar has gone through several different versions of kit and UK versions of AN DACs. I'm actually slowly working on building a heavily modified 4.1 kit. In addition to the usual parts changes/upgrades, I'm actually doing a little DIY on it. The plan is to design my own DAC board which mimics the Kassandra to some extent.. the goal is 16 parallel AD1865N-K chips, balanced, powered with ultra low noise shunt regulators, driving transformer based I-V. The input will be from an XMOS USB, which is then re-clocked with an ultra low phase noise XO (probably the Pulse Clock OCXO). Its a side project. The challenge will be to get all of it to fit in the 4.1 chassis (including the extra toroidal power transformer).

I noticed the Kassandra doesn't appear to be using the MSB trimming option of the AD1865 to eek out even better element matching within the DAC IC (better THD+N). Sure, the chips are K grade, but if I'm not mistaken, each chip could be dialed in manually for best low level linearity. I'll try to do this and see how it goes too. It would take a lot of time to calibrate each DAC chip though, so its not something you would find in a commercial design I guess.

If all this comes to fruition, and I have a final product one day, I may stop by Believe High Fidelity's storefront for a listen and comparison, since I live in the Austin area. But, no way to purchase considering the cost

Sorry, meant Pulsar Clock, not Pulse.

kma said:

Thanks guys. Interesting observations on the upsampling for this DAC. I guess it does make sense that every DAC will respond differently. I have an Aqua La Scala (the PCM1704 version), and I actually notice the opposite when I upsample with HQPlayer (soundstage grows and image density softens). I tend to like upsampling better for my DAC.

In looking at some of the pictures posted of the internals of the Kassandra, I see a Pulse Clock OCXO is being used (which is very low phase noise, resulting in superior jitter per the DAC manufacturer's claims). However, I only see a single OCXO in the box, which probably means the DAC is optimized for jitter for multiples of 44.1kHz rather than 48kHz Fs. Therefore, upsampling a redbook format to 192kHz or 384kHz may not be as optimum as 176.4kHz or 352.8kHz from a jitter perspective. Maybe this has something to do with the magic being lost with the upsampling. However, given you guys probably tried all combinations of upsample ratios from native rates, then maybe the overall jitter difference is not the reason for the change in sound. Or maybe I can't see everything in the box just from the pictures I see on the net

I've read from various message boards that Astrostar has gone through several different versions of kit and UK versions of AN DACs. I'm actually slowly working on building a heavily modified 4.1 kit. In addition to the usual parts changes/upgrades, I'm actually doing a little DIY on it. The plan is to design my own DAC board which mimics the Kassandra to some extent.. the goal is 16 parallel AD1865N-K chips, balanced, powered with ultra low noise shunt regulators, driving transformer based I-V. The input will be from an XMOS USB, which is then re-clocked with an ultra low phase noise XO (probably the Pulse Clock OCXO). Its a side project. The challenge will be to get all of it to fit in the 4.1 chassis (including the extra toroidal power transformer).

I noticed the Kassandra doesn't appear to be using the MSB trimming option of the AD1865 to eek out even better element matching within the DAC IC (better THD+N). Sure, the chips are K grade, but if I'm not mistaken, each chip could be dialed in manually for best low level linearity. I'll try to do this and see how it goes too. It would take a lot of time to calibrate each DAC chip though, so its not something you would find in a commercial design I guess.

If all this comes to fruition, and I have a final product one day, I may stop by Believe High Fidelity's storefront for a listen and comparison, since I live in the Austin area. But, no way to purchase considering the cost

Click to expand...

If this is ever completed, I'd love to hear it as well. I'm also in the Austin area.

Hi Lordcloud, sure. Don't hold your breath though. The difficult part is to try to fit everything in the same 4.1 chassis. Without this constraint, it would be very much possible. I may have to offload the existing power supply and regulator to another chassis, or figure something else out if space becomes an issue. I'm still in the early planning stages.

kma said:

Hi Lordcloud, sure. Don't hold your breath though. The difficult part is to try to fit everything in the same 4.1 chassis. Without this constraint, it would be very much possible. I may have to offload the existing power supply and regulator to another chassis, or figure something else out if space becomes an issue. I'm still in the early planning stages.

Click to expand...

Interesting, and good luck with your build! To build a multi chip array you will need some high end measuring equipment to get it to be linear I would think.

My DAC 4.1 kit was modded with Audio Note UK parts and Duelund capacitors, bigger blackgates, bigger I/V transformer stage, best tubes etc. But even then it is miles behind the Kassandra, as was the Audio Note UK DAC 5.

The digital section in the Kassandra is part of the mix, the other part is the fantastic gain stage, basically a true SET pre-amplifier in it's own right in the same chassis. And this combined with NOS gives it it's natural sound signature.

The next big deal is the over specified power supply, huge capacitor banks. This aspect is normally a big compromise in most DAC designs, along with weak op amps. It is the whole package.

The USB input has double clocking and can accept 384 data lengths.

astrostar59 said:

Interesting, and good luck with your build! To build a multi chip array you will need some high end measuring equipment to get it to be linear I would think.

My DAC 4.1 kit was modded with Audio Note UK parts and Duelund capacitors, bigger blackgates, bigger I/V transformer stage, best tubes etc. But even then it is miles behind the Kassandra, as was the Audio Note UK DAC 5.

The digital section in the Kassandra is part of the mix, the other part is the fantastic gain stage, basically a true SET pre-amplifier in it's own right in the same chassis. And this combined with NOS gives it it's natural sound signature.

The next big deal is the over specified power supply, huge capacitor banks. This aspect is normally a big compromise in most DAC designs, along with weak op amps. It is the whole package.

The USB input has double clocking and can accept 384 data lengths.

Click to expand...

Thanks for your input. My personal opinion is that, the weakest link of the AN kits, and for that matter, even the high end UK AN DACs (like the 5), is the digital implementation overall. I've heard an AN DAC 5 in demo, and from what I recall, it doesn't come across as a "modern" design, in that modern designs try to keep all the attributes of analog, while preserving the detail and dynamic ability that could be achieved with modern digital. I know the UK DAC 5 uses really nice parts, but the digital board is still using the decades old Cirrus PCM chipset (or something equivalent.. can't tell for sure what exactly is there because I never owned one or opened the hood). These digital implementations based upon older tech are missing substantial improvements in jitter reduction. In addition, there are much better, low noise, lower impedance shunt regulators available to be used with the DAC IC, which I think is also far out of date on the stock kit or UK implementation.

Any current source (like the output of the DAC IC) wants to see a ground, or virtual ground (no voltage rise, full current shunt), hence, the popularity of opamp based I-V conversion. Considering the relatively low output impedance of the DAC chip with its R2R implementation, passive loading via proper resistor value becomes even more critical. With a single DAC chip, a relatively high load resistor value needs to be used to generate acceptable voltage output. Parallel DAC chips, although not really improving the current source situation from an individual DAC chip point of view, does allow you to closer approach a short circuit load for the current source output (ie, lower load resistance to generate the same output voltage because the total current has been increased by N number of chips). Dropping the value of the load resistor increases linearity at the high signal level side, and the increased current improves noise immunity at the lower signal level side. In addition, my guess is the slew rate of the output decreases during fast and large transients (high dV/dt) due to less drive capability as the output terminal voltage increases, which is also partially rectified with parallel chips. Using a balanced approach at the I-V conversion stage, in addition to subtracting out even order harmonics, does help reduce overall dither like noise from the DAC output and also some common mode power supply noise. Attempting to address all these issues translates into "better", more realistic sound (quieter, more dynamic, more linear, and more resolving). I'm assuming these are the areas where the Kassandra excels relative to an AN DAC. In short, trying to address all these aspects should yield the biggest returns on the ANK platform, which is my goal

I'm a believer in stout power supply design, however, there is such a thing as "too much" reserve capacitance. Too much doesn't necessarily produce any negative side effects, but it also doesn't produce increasingly positive effects either. A DAC just doesn't draw as much dynamic power as a power amp would. A power supply should be designed for ample reserve capacity from the wall, through the power transformer, and into the power supply reserves (overall bulk capacitance). Its not too difficult to attain plenty of total reserve capacity and get it all charged up, but the "quality" of the reserve is important (ie, low ESR). Charging the reserve bank of capacitance is one thing, but the ability for a demanding load to draw/discharge upon these reserves quickly and without lag (at the frequencies of current demand, which will vary based upon the load circuit) is more important to good power supply design. Low power supply impedance across frequencies of interest for local decoupling capacitors also plays into this. No doubt, the Kassandra has already hit on all of these subjects. But, my point is, comparing overall total of bulk capacitance does not automatically imply a better power supply.

As far as the Kassandra, if all that you guys are using is NOS (44.1kHz), if your Pulsar reference clock is not the lowest possible common denominator frequency to support the related MCLK rate, then I would inquire with the designer on if the Pulsar OCXO could be swapped to a lower frequency (depending upon his overall implementation). There is a huge improvement in phase noise for lower frequency crystals (just physics). For example, if the current reference clock is 45.1584MHz, you may be able to use a /2 or /4 crystal (for only 44.1kHz), and get a huge improvement in your sound. That is, at this already low jitter level overall, if you could perceive further jitter level improvements. Also, this assumes the DAC chip itself sounds the same at different rates, which maybe it doesn't. It would be something worth asking about though.

kma said:

Thanks for your input. My personal opinion is that, the weakest link of the AN kits, and for that matter, even the high end UK AN DACs (like the 5), is the digital implementation overall. I've heard an AN DAC 5 in demo, and from what I recall, it doesn't come across as a "modern" design, in that modern designs try to keep all the attributes of analog, while preserving the detail and dynamic ability that could be achieved with modern digital. I know the UK DAC 5 uses really nice parts, but the digital board is still using the decades old Cirrus PCM chipset (or something equivalent.. can't tell for sure what exactly is there because I never owned one or opened the hood). These digital implementations based upon older tech are missing substantial improvements in jitter reduction. In addition, there are much better, low noise, lower impedance shunt regulators available to be used with the DAC IC, which I think is also far out of date on the stock kit or UK implementation.

Any current source (like the output of the DAC IC) wants to see a ground, or virtual ground (no voltage rise, full current shunt), hence, the popularity of opamp based I-V conversion. Considering the relatively low output impedance of the DAC chip with its R2R implementation, passive loading via proper resistor value becomes even more critical. With a single DAC chip, a relatively high load resistor value needs to be used to generate acceptable voltage output. Parallel DAC chips, although not really improving the current source situation from an individual DAC chip point of view, does allow you to closer approach a short circuit load for the current source output (ie, lower load resistance to generate the same output voltage because the total current has been increased by N number of chips). Dropping the value of the load resistor increases linearity at the high signal level side, and the increased current improves noise immunity at the lower signal level side. In addition, my guess is the slew rate of the output decreases during fast and large transients (high dV/dt) due to less drive capability as the output terminal voltage increases, which is also partially rectified with parallel chips. Using a balanced approach at the I-V conversion stage, in addition to subtracting out even order harmonics, does help reduce overall dither like noise from the DAC output and also some common mode power supply noise. Attempting to address all these issues translates into "better", more realistic sound (quieter, more dynamic, more linear, and more resolving). I'm assuming these are the areas where the Kassandra excels relative to an AN DAC. In short, trying to address all these aspects should yield the biggest returns on the ANK platform, which is my goal

I'm a believer in stout power supply design, however, there is such a thing as "too much" reserve capacitance. Too much doesn't necessarily produce any negative side effects, but it also doesn't produce increasingly positive effects either. A DAC just doesn't draw as much dynamic power as a power amp would. A power supply should be designed for ample reserve capacity from the wall, through the power transformer, and into the power supply reserves (overall bulk capacitance). Its not too difficult to attain plenty of total reserve capacity and get it all charged up, but the "quality" of the reserve is important (ie, low ESR). Charging the reserve bank of capacitance is one thing, but the ability for a demanding load to draw/discharge upon these reserves quickly and without lag (at the frequencies of current demand, which will vary based upon the load circuit) is more important to good power supply design. Low power supply impedance across frequencies of interest for local decoupling capacitors also plays into this. No doubt, the Kassandra has already hit on all of these subjects. But, my point is, comparing overall total of bulk capacitance does not automatically imply a better power supply.

As far as the Kassandra, if all that you guys are using is NOS (44.1kHz), if your Pulsar reference clock is not the lowest possible common denominator frequency to support the related MCLK rate, then I would inquire with the designer on if the Pulsar OCXO could be swapped to a lower frequency (depending upon his overall implementation). There is a huge improvement in phase noise for lower frequency crystals (just physics). For example, if the current reference clock is 45.1584MHz, you may be able to use a /2 or /4 crystal (for only 44.1kHz), and get a huge improvement in your sound. That is, at this already low jitter level overall, if you could perceive further jitter level improvements. Also, this assumes the DAC chip itself sounds the same at different rates, which maybe it doesn't. It would be something worth asking about though.

Click to expand...

Few points. Have you tried an over specified power supply to know it makes no difference? The Kassandra has a huge power reservoir to create a 'pseudo' battery supply or an ideal noise free linear supply. That is how I understand it.

On current to voltage, the AN DACs use an I/V transformer that is specially wound, as does the Kassandra and the Ypsilon DAC 100. It is a more expensive way to do it but has it's sonic benefits.

The AN DAC boards are indeed very old, about 2005 vintage for the same DAC 5 digital board. They have little jitter reduction or clocking. They tend to get sold with the AN transports which have clocking at their end.

My Kassandra has a 44.1 clock on the SPDIF as I asked for it. So good for 44.1, 88.2, 176.4 etc. It has double clocking on the USB input and can accept up too 384. I have played data up to 192 so far and it does sound great, better than 44.1, but that is tracks at that resolution. I am not convinced by up sampling in HQ Player. That would be better possibly in a DAC with a filter, to enable that DAC to avoid less of the problems of a filter. In my case, there is no filter.

I think we (me and you) probably know a tiny amount of what is required in a good digital board design. It is not an easy task. And probably why there are very few really good DIY kits around. TotalDAC used to sell the boards before they went only finished products. Soekris sell a board, but it is probably not that great.

I would hazard a guess and say the power supply and gain stage is easier to design / apply than the digital section, as there are lots of good DIY pre-ampilfer kits around.

Going back to the Audio Note DAC 5, it was held back to a great extent by the coupling caps in my example. Once I changed them to Duelund Copper Cast things got a lot better. But still way behind the Kassandra....

Hi, I will try to respond to your comments, thanks.

Few points. Have you tried an over specified power supply to know it makes no difference? The Kassandra has a huge power reservoir to create a 'pseudo' battery supply or an ideal noise free linear supply. That is how I understand it.
I think the implication is throwing more capacitance into a power supply automatically makes it better. My point is, although I agree to some extent, there is definitely diminishing returns, and in an extreme case, once a certain threshold is crossed, that there will be no change at all. Its not inconceivable that there is an upper limit. A "noise free" (or I would say, very low noise) supply could be made with a linear supply or switching power supply with finite bulk capacitance. Asking me if I've tried this is not really a simple question to answer, because it always depends upon the power supply circuit, the type of load such power supply is connected to, the overall noise immunity of the load, as well as the target noise level in the system. To be honest, bulk capacitance is cheap. If all a power supply needed to be as low noise as a battery was more capacitance, then everyone would do it. I'm not making a claim that the Kassandra designer overbuilt his power supply. Unless I know his design, I cannot comment. I could imagine that he is using ultra clean regulators to supply power to the subsystems in his design, and usually ultra low noise comes with a power/heat penalty. If the regulators are drawing a lot of power and generally taxing a power supply, the bulk capacitance should be up sized accordingly. The Kassandra has done this. To be fair, the regulators used in most DACs, probably including the AN UK ones, are less power hungry (and probably have higher output noise). Such systems do not need as much bulk capacitance, but are probably further short changed anyways for other reasons. For the changes I plan to make to my DAC, it will require a lot more additional bulk capacitance to adequately feed all the digital subsystems.

On current to voltage, the AN DACs use an I/V transformer that is specially wound, as does the Kassandra and the Ypsilon DAC 100. It is a more expensive way to do it but has it's sonic benefits.
The resistor load I was talking about is in conjunction with the transformer. The typical transformer used for this application is too high of an impedance at audio frequencies to be used alone. The load resistor (either on the primary or secondary side of the transformer) sets the average load the DAC sees, which is required to be low impedance. In essence, the resistor shunts the current, and sets the signal voltage that gets amplified by the turns ratio of the transformer. So even with transformer I-V, this load resistor can be made quite a bit smaller with parallel DAC chips, which was the point I was trying to make.

My Kassandra has a 44.1 clock on the SPDIF as I asked for it. So good for 44.1, 88.2, 176.4 etc. It has double clocking on the USB input and can accept up too 384. I have played data up to 192 so far and it does sound great, better than 44.1, but that is tracks at that resolution. I am not convinced by up sampling in HQ Player. That would be better possibly in a DAC with a filter, to enable that DAC to avoid less of the problems of a filter. In my case, there is no filter.
The clock I was referring to is the master clock in the system, which is probably being used to reclock for jitter reduction. If you review the Pulsar OCXO datasheet, you will see it is only offered as low as ~11MHz. My point was to ask the designer which of the available crystal rates he is using as his master clock in the Kassandra. If it is higher (like lets say ~45MHz), the phase noise would be much better if a lower frequency crystal was used. If your main use case is 44.1, then this could be an option to further improve the jitter performance of the DAC. The possible tradeoff may be lack of support for the highest sample rates. If such compromise exists, or if any modification is possible with the current Kassandra, I have no idea. Depends upon the design architecture of the Kassandra. Seeing that you actually use higher native sample rates in your system, then there is no point to inquire however.

My current DAC (Aqua La Scala) has no analog filter and does no digital filtering internally either. But, I personally find it more engaging using HQPlayer. I'm actually disabling the noise shaper in HQPlayer too. It sounds better this way for me. I guess all DACs respond differently (and our ears respond differently too).

I think we (me and you) probably know a tiny amount of what is required in a good digital board design. It is not an easy task. And probably why there are very few really good DIY kits around. TotalDAC used to sell the boards before they went only finished products. Soekris sell a board, but it is probably not that great.
I've not used a Soekris board, but there are a lot of people who like it. Soekris chose to make a discrete R2R with FPGA, including digital filtering. The board does a lot, but maybe doesn't do any particular thing "best in class". So, its a decent platform to build a decent DAC around, but is probably not going to be in the big leagues. But, I don't think it was ever targeted as such. Implementing a DAC IC on a board is much simpler than implementing your own DAC IC in discrete components (FPGA + discrete resistors) on a board. I've done enough board designs at work at much higher frequencies than audio (where it actually does get challenging). Every design has challenges. And sure, we all make mistakes and learn. But of course, the fun is in the trying! Making boards is relatively cheap nowadays with all the small Chinese shops around. Oh, and don't be surprised, but most domestic companies use overseas PCB fabs and assembly houses as well. The challenge with a good DAC board is in the power supply schemes, implementation of reclocking schemes for ultra low jitter, and preventing unwanted coupling paths on board to name a few. DIY communities seem to focus on 1 aspect, rather than putting together the whole package. I could imagine there are some well executed DIY designs out there. In those cases, the designer does things for himself, without advertising his work online. There are a lot of smart people out there!

I would hazard a guess and say the power supply and gain stage is easier to design / apply than the digital section, as there are lots of good DIY pre-ampilfer kits around.
But where's the fun in that? Yeah, a simple amplifier stage would be a little easier to deal with. As a continuing project, I will probably start playing with the gain stage later on down the road (if I'm still feeling ambitious). Power supply design is actually a lot more challenging. Power hungry shunt regulators can be made at near battery noise levels, but they come with a price (power/heat). Using an off the shelf linear regulator is easy, but most are more noisy comparatively speaking. I have little knowledge of tube rectified supplies too.

Going back to the Audio Note DAC 5, it was held back to a great extent by the coupling caps in my example. Once I changed them to Duelund Copper Cast things got a lot better. But still way behind the Kassandra....
Good call on that. I'm planning to put Duelund CAST hybrid Cu/Ag in my kit as a coupling cap in the gain stage. There is a lot more sophistication in a Kassandra, so I fully expect it to wipe the floor with a stock, or boutique parts modified, AN UK design.

Kma, it is an interesting project you have set yourself. Good luck! I enjoyed my foray with the Audio Note kit 4.1 maxed out. The biggest sound quality hike was upgrading the I/V transformers from the tiny ones in the kit to bigger AN UK versions (496). You don't have to buy AN UK transformers, other manufacturers make them, and cheaper. Though the implementation and spec is quite technical to get them right.

The Kassandra as far as I understand it, has 2 pathways to the chip array, and SPDIF in my example goes through the extra 44.1 clock. I can bypass that, or enter via USB which has 2 x clocking and takes up to 384 data. As I said, files at higher (true higher rate originals) sound great, and in the cases of the same master, do sound better than Redbook 44.1.

I haven't heard an improvement with upsampling and HQ player yet, in my test it was inside Roon.

IMO the general AN DACs and the AN Kits are good, but held back by a coloured gain stage and the ok-ish power supplies. There are big increases to be had by going big and better in those areas IMO. The digital section in the DAC 5 is better than the Kit DAC 4.1, but I guess you knew that. The 4.1 kit doesn't have shunt regulated power supply and other extra smoothing. I have heard both boards in the same DAC and the DAC 5 board is better.

Finally, the output transformers in the DAC 4.1 kit, and even the AN UK DAC 5 are so so, nothing special. I believe the superior (and bigger) units in the Kassandra again distance the DAC from the AN sound, which is dynamically compressed, coloured and fairly muddy / asleep by comparison.

The Aqua La Scala is a nice DAC for the money, but I question if it surpasses the AN DAC 5. I sold my DAC 5 back to the guy I bought it off, and he had the 2017 La Scala new from the factory for 4 months. He told me it wasn't as good as the DAC 5 in his system. As it shouldn't be really, the DAC 5 is list 30K but no idea who buys then at that price. It is not competitive today on price v performance, so maybe only by the hard core AN crowd possibly?

Hi Astrostar, thanks for all the useful tips. I was planning to upgrade the I-V transformer to either a Sowter or Lundahl. Do you have any experience with either of those?

You should try HQplayer rather than Roon as the upsampler. I have both together, but I don't use Roon for any upsampling duties. HQPlayer is very powerful and changing settings there can amount to differences on the order of changing out a whole DAC. There are hundreds of combinations that could be utilized in HQPlayer, if you factor in all the filter types, noise shaper choices, and sample rates. It seems people find the filters that work best for their ears and for their DACs. Well, maybe there is no reason to try this if you get the perfect sound through SPDIF at 44.1.

Just curious. How did you determine the depth of coloration of the output stage and output transformers on the AN designs when there are so many other variables that are all inter-related (as you said, power supplies, DAC board, etc)? Definitely a lot more than 1 variable is different between a Kassandra and an AN DAC. My personal belief is, with a little bit of power supply help, an I-V transformer upgrade, and a whole new digital board approach (supplies, clocking, and multiple D/A converter chips), the majority of the difference can be accounted for. Sure, when you are playing at an extremely high level, every piece/section can make a difference. But from a "get you 90% there" approach, I was thinking you can get there with a mostly stock output stage and output transformer. Well, I have nothing to prove this, but its just a hunch. The triple C-core output transformer on the ANK DACs are not shabby. Or do you subscribe to using an audio power amplifier output transformer in its place? Do you have any pictures of the insides of the Kassandra power supply box (I assume the output transformers are in there?). I'd be curious to see how big these units are..

On the digital board of the DAC 5, yeah, I never saw one, or would never have the luxury to do an A/B swap of DAC boards for comparison like that, so your comparison is useful. But, as you know, there is still a whole lot more left on the table with the DAC 5 digital board relative to a Kassandra..

I don't have the flagship Aqua DAC. I have the La Scala, which is a tube version, and mine is the first revision which uses PCM1704 rather than discrete resistor/FPGA. I heard the first revision of the Formula, which is the flagship, and although it was nicer than the La Scala, it was missing something without the tube stage. What they need to make is the Formula with a tube stage. Oh well. My recollection of hearing all AN systems in demo rooms tells me that the Aqua and AN house sounds are totally different. Both are nice, but both have their own audience. If it were me, I'd probably want a hybrid between the two. I guess thats why I've gone down the ANK path and trying to DIY it a bit. I'd like to see if I could accomplish a good compromise between both house sounds. Then again, the ANK sound is more akin to the Aqua sound than it is to the AN Japan or UK sounds I'm sure, now that they've diverged quite a bit.

I also wonder who would shell out that kind of money on a new DAC 5. I'd personally like to see the UK guys modernize their unit a bit. It could do with some updated design work. Then, possibly it might be worth its sticker price. I don't know the exact sticker price of a Kassandra, but I think its on order of the DAC 5? In that league, why would anyone give the DAC 5 a chance against the Kassandra.

Hi Kma

I don't subscribe to up sampling if the DAC has no filter or extra logic. It is not required. The only advantage I see in up sampling it to reduce the hack job a digital filter does to the sound. The Kassandra is filterless, and logic less i.e. data straight through. I can feed it 44.1 or 384 resident rate and it just plays it without any messing with the data.

I don't use Roon to upsample.

On the Audio Note DAC 5 current model, which is so close to the one I had circa 2008 (it is not worth trying to part them), yes the deign of the DAC 5 is old and needs updating. But the fundamentals were there. It didn't sound digital, not edgy or flat soundstage as many DS DACs, but unfortunately it wasn't the most dynamic DAC around. It lost too much in the circuit design IMO, too many filter caps and caps in the signal path. AC uses no coupling capacitor in the signal path.

Also, as you may know, the Audio Note gear has a 'sound' as they like to call it. That is ok, but I could call it coloured. It is warm and cuddly but is not terribly fast or the most transparent sound around. So the AN DACs are leaning very hard into the warm camp, calming things down, going away from the digital sound and detail freak DACs. But IMO they loose in other ways, and also drift away from an 'ideal' analogue sound we all crave IMO.

The output transformers from AN are ok, but very over priced. And the upper levels are just pain crazy prices, they don't make an ok DAC into a top DAC, certainly not good value for money. Probably the Hi-B copper is as far as I would suggest you go on that.

The Sowter I/Vs are very good, they have been making them for years, and in fact made them for Audio Note way back. You can ask them for the correct I/V unit for your 4.1, they will know, and the resistor loading you will need at the chip and the line stage ends.

To really build a better DAC you may need more space. The power supply on the stock 4.1 kit is quite small. It will be better using 2 x IE trans, one for the digital, one for the tubes. And a separate filter PCB for the digital power feed. Big caps for the power supply will take up space mighty quickly.

How much are you investing? I ask as if you go beyond a certain point, you may as well source a used DAC 5, or better save for the Kassandra?

The Kassandra is 18K, the DAC 5 is 35k, and the Kassandra is well beyond it in terms of sound quality. Actually in build quality as well, and size of parts, weight etc. 60 kilos v 19 kilos. It is also huge.

Where are you based? Maybe you should visit my place for a demo?