Holo Audio Cyan 2

My desktop DAC has been a Yggdrasil A2 for the last three years or so and I have been happy with it, but recently I wanted to try a different source, perhaps something a touch more neutral yet similarly detailed and resolving. I enjoyed the Holo Spring 3 KTE that was on tour here at SBAF a couple of years ago (my review here). At the time I stated that for my preferences that DAC would (mostly) be a side grade - a trading of one set of strengths for a different set.

The Holo Cyan 2 is Holo's lowest tiered and priced DAC and costs less than half of a new Spring 3 KTE (and about $1K less than a new Yggdrasil). How does it stand up against it & the TOTL Schiit DAC?

notes:

• Chain is Roon > upsampled PCM 4x (176/192) or 8x (352/384) in either Roon or HQPlayer (DSD not tested) > Cyan 2 (USB & I2S via Pi 2 Mercury)> bal & SE output to Freya + (for bal-to-SE conversion and remote volume control) > DNA Starlett & Schiit Lyr + (mostly run in SS not tube mode) > Jar HD650, ZMF Atrium open, and Focal Utopia
• This is a NOS resistor ladder DAC, but I don't particularly like NOS 44/48 PCM generally, so I upsample/oversample (choose your term) everything in software. I have not even tried 44/48 at native rate and probably never will. I am not interested in devoting the computing resources to DSD conversion at this time either.
• The first day or two the DAC had an "edge" (not sibilance exactly) in the upper mids/lower treble that was beyond the expected R2R sound. However after pumping music through it 24 hours a day for a few days that is gone, and this is true whether using I2S or USB as input, so the DAC (input/conversion/output stages) settled down after about 72 hours.
• USB input is Holo's lower tier "galvanically isolated" version (can not be upgraded like the Spring), however in the Cyan it is quite good (more on inputs below).

Character and Sound


Pretty darn well is the answer the above question. To my pleasant surprise, the micro detail, separation, and FR neutrality I heard in the Spring is all there, or nearly. The Cyan 2 is even, or nearly (perhaps a "1/4 step") behind the Yggdrasil as far as microdetail and absolute resolution. Being more neutral in both FR and sitting a few rows farther back in headstage, as well as having a less in-your-face dynamics, the Cyan 2 is more "neutral" then the Yggdrasil A2. Indeed I think this is the most neutral DAC I have ever heard. It is neither "polite" (something tempting to say after listening to the Yggdrasil) nor is it "slamming". It is neither "forward" nor "laid back", and no aspect of the FR is either emphasized or de-emphasized. It has plenty of subass/bass/mid/treble, yet no aspect of the FR is out of balanced with any other. If this sounds boring well it might be, but I am really liking just how coherent this DAC is and how it "just gets out of the way" (perhaps an overused cliche) while at the same time bringing the expected R2R strengths (e.g. timbre).

The 'blackground' Holo is known for is there, but I don't think it equals the Spring I heard two years ago. The attack and decay speed is fast (but not 'sharp') like the Spring, but again just not as much - bringing it more into line with the Yggdrasil. Too much can be made of these differences however, the Cyan retains the Holo house sound of notes "popping" out of a blackground and decaying back into it without any emphasized decay. If you prefer a more "organic" sound then a Holo anything might not be for you.

What the Cyan 2 is missing when compared to the higher end Holo DAC is that beguiling-yet-somewhat-unnatural "wrap around" headstage (I have not heard it with speakers). The stage is very on par with the Yggdrasil as far as width an depth, yet at the same time because your sitting a few rows behind the Yggdrasil it might even have a touch better separation, at least for some recordings.

I recall saying that the Spring did not have the "inner note realism" of the Yggdrasil and for the Cyan 2 this is still true, but perhaps because the Holo "pop" is not as emphasized in the Cyan 2 it sounds more organic and closer to "neutral" (and thus the Yggdrasil) than the Spring. I would say the Cyan 2 is more neutral than both the Spring & the Yggdrasil in absolute terms, in this reminding me of the Gungnir MB A2 I had in the past (but without that DAC's particular FR emphasises) .


Input/output Agnosticism


I like how the inputs I have so far used (USB and I2S) and outputs (bal vs SE) are close enough in sound I am not confident (beyond placebo) I can tell the differences when comparing equivalent sample rate. In my chain I am converting bal to SE (to my SE only Starlett & Lyr) via Freya's buffer mode (not tubes), so not exactly apples-to-apples however but no obvious difference like the Yggdrasil's


Value and recommendation


I would say that because of its price and character, at its position in the market the Cyan 2 is an easier general recommendation than any Schiit DAC that I can think of as long as an owner can upsample a little bit (88/96 is fine). Sure, if your after a certain emphasis (say extra strong dynamics, or an "organic" warmth or a "detailed" brightness) there are other options, but darn it this is a good DAC at a good price.


Automatic input selection

No doubt in a bid to keep costs down the Cyan features no buttons, remote, or controls of any kind. Input selection is "automatic", which in practice means USB is prioritized over I2S and I suspect anything else. If I am streaming I2S from Roon and I stop, switch and stream to USB, sometimes Cyan will switch over. On the other hand it will never do the opposite and switch away from USB back to I2S - forcing me to power it down & unplug USB (disabling device in Roon and/or Windows has no effect) before it will sense an I2S stream again. If you switch inputs alot this will be an issue for you.

Update: DSD playback

So I thought I better verify that the DAC makes sound with DSD input during warranty period. I tried HQPlayer at first, but it was a no go even at PCM > 64x, possibly due to the fact that I limit my CPU to keep the fan from spinning up, but probably because HQPlayer is exhibit A under the heading "UI that should have never seen the light of day" and I never got a workable config.

So I went to Roon and 28 seconds later (7th order CLANS, 128 & 256x with very little increased computing overhead) I was listening to PCM > DSD via USB of course. Shucks, I kept listening. On the plus side, Headstage deepens a bit, as well as a touch of that Holo "wrap around" becomes evident. Microdetail becomes a tiny bit more as does separation, and small increase of warmth and "air". On the minus side, the DAC darkens a smidgeon, and on some recordings there is an artificiality to the increased "air" - almost like a reverb plugin has been turned on (on a very low setting). The overall experience is a just a bit more "ethereal" . Notice I keep using terms like "small" and the like as these differences really do take full on critical listening concentration to discern...still, they are there. I have been considering building a fanless PC to run Roon core and possibly HQPlayer so DSD experimentation is another motivation to not put this project off...

2nd Update: HQPlayer (via Roon) PCM & DSD playback

After getting a fanless/headless PC with a recent Intel CPU (running Ubuntu Linux), and successfully getting HQPlayer upsampling PCM & PCM>DSD, I get it. HQPlayer's filters when paired w/ Cyan are a step up from Roon's. My revised opinion now it is a solid (full step up) 'tweak'. I did not really want this to be so - if it wasn't I was going to put this new PC to work elsewhere, but it is true. With PCM > DSD in particular, the space, placement, and transient "rightness" is pushing the Cyan towards an very "organic" & "natural" presentation that is equaling if not exceeding the Yggdrasil in this area.


Pics or did not happen:



(I sell the audiophile 2x8 you see separating the Cyan from the Lyr for $469 shipped. Also, damn, I need to dust...)

 

thank you for the great writeup!

 

Pretty sure you just sold a few Cyan 2s. I was thinking it likely was a good Swiss army knife for people who enjoy HQPlayer and will be using oversampling regardless. This review seems to confirm that.

For people looking in this price point and with this specific use case, this seems like a super solid option. Seems like a super niche customer, but a great tool for that customer.

 

Curious how this reminds me of Soekris dac2541. This one has its own internal upsampling filters, but those seem similar (AFAIK) to what's available for PCM in Roon. Then it's R-2R for both. And the price is similar too...

 

Thanks for the review @crenca

This one's very much on my radar for various experiments with HQPlayer. I'm curious, in particular, if HQP upsampling ends up flattening the curve between various DACs.

It will be interesting, too, to see the eventual comparisons between the Spring and May at different levels of those builds.

 

I dusted off my old copy of HQP (v3) only to discover I needed v4 or higher to work with Roon, so I'm running an evaluation right now. In the end rather than purchasing a new desktop I may very well end up going the embedded route explained helpfully by @Woland here:

https://www.superbestaudiofriends.o...on-a-pi4-invigorate-your-digital-audio.11231/

My opinion about Cyan 2 and PCM 'upsample rolling' so far (I may revise with more time/experimentation): It is a tweak. I would compare it to tube rolling, but tubes are more of change/down/side/upgrade or can be. Maybe it is like going from a good tube with attributes you prefer to an even better but essentially similar tube. The DAC's essential sound and character remain the same, and 'upsample rolling' is very unlikely to make or break the DAC for you...or maybe I should say for me. If you like to go NOS at times, upsampled PCM other times, DSD still others, then maybe HQP/upsampling will be more important and impactful. So far I am gravitating to the closed form/'Sinc' filters which probably not coincidently are closed form filter's of the same type used by Schiit's combo burrito.

Question someone may be able to answer: HQP has a setting called 'DAC bits' that your supposed to match with the analog bit depth of your DAC to optimize the noise shaping 'NS' dither options. Holo claims about 21 bits of resolution (127db of dynamic range), but does not say if that is with the PCM ladder, the DSD circuit, or both. Any suggestions how I should determine this setting for Cyan 2?

 

I have both HQP 5 Desktop and Embedded, for different systems. Their documentation of "DAC bits" uses different text, but here's have I've inferred from both. This feature has two effects. For unidirectional connections (S/PDIF, AES3, I2S) where HQP cannot "see" the DAC's word length, it's important to set this appropriately for the DAC (which means no greater than what the DAC would accept). In addition, for all connections (including USB, which is bidirectional and so HQP can "see" the DAC's actual word length) it sets a correct word length for dithering and noise-shaping that seem important for linearity in R-2R DACs. One of the docs says that the correct setting for Holo DACs is 20.

 

Update - added DSD section to OG post.

 

20 bit for Holo Dacs for hqplayer settings is correct. But that is only applicable for pcm playback. If you only use dsd then no need to set the bit depth.

a raspberry pi 5 running hqplayer is sufficient for pcm 20/1536 upsampling, with oc it is capable of dsd256x48 with the 7ECUL modulator

 

I should note that so far my favorite settings is to use cyan 2 to Freya + passive preamp mode at max volume (unit gain) and use hqplayer to do digital volume control.

settings are either dsd256x48 Super 7EC or dsd512x48 Super 7EC 512fs, default filters.

I occasionally would use the tube mode but find it masked the natural, organic sound of the cyan 2, especially in the mids

 

So after listening to the cyan 2 since January.

Compared to my ifi pro idsd.

I have both going through the Freya + passive preamp cranked to max volume (unity gain) that means it’s pretty much just a passthrough.

Both are set with hqplayer at dsd256 7EC super, default filters. Both are connected to a passive usb hub into an intona. So it’s pretty easy for me to choose the dac in hqplayer settings

I love them both. To me It would be a tie for me.

The Cyan 2 sounds more neutral, organic, natural, believable.

The ifi pro sounds more musical, warmer, a bit more coloured. With a pair of great nos WE396a tubes, music can sound unbelievably beautiful, then next thing you realize it’s too beautiful it’s unreal… with or without the tubes, the ifi pro does hit a bit harder in bass and has a wider soundstage but the Cyan 2 is more precise in its imaging.

 

I’ve been lusting after a Spring 3 for a while, but it costs more than I want to spend on a DAC. Does anybody know what the technical differences between the Spring 3 and the Cyan 2 are? Is it only the obvious missing features or are there differences in the DAC part, too, in the power supply, the USB interface, you name it?

 

USB is equally excellent. Miska said based on his measurement, Spring 3 is still measured a little bit better, but with hqplayer reconstruction filter, the gap gets a bit closer.

 

The biggest difference between Cyan 2 and Spring 3 or May is the pcm performance according to Miska “Cyan 2's PCM ladder cannot quite reach what Spring and May can, it is somehow strangely noisy always compared to DSD mode. In DSD mode, Cyan 2's performance is absolutely great.”

 

Jeff Zhu once again proves why he is among the DAC designer elite. Cyan 2 is an example of knowing where to cut the costs without compromising the performance. I am impressed, comparing to both Spring 2 and 3. Some example measurements:

Distortion at 1 KHz - Holo Audio Cyan 2 spdif in Bal out

Wow! This is exceptional for an R2R DAC, at any price.

Dynamic Range - Holo Audio Cyan 2 spdif in Bal out

Very few DACs achieve this level of Dynamic Range.

Listening so far has confirmed Cyan 2 to be engaging, immersive, non-fatiguing, and thoroughly enjoyable. Spring 2 and 3 DACs offer higher technical performance, better engagement factor. But blimey, the level of enjoyment for Cyan 2 is incredible. The price is almost unfathomable considering the musical rendering it delivers.

 

Thanks for this, wonder if you might have done any measurements on usb in?

cheers

Deric

 

A complete suite of measurements takes quite a bit of time. I will do such in the future as time permits. Until then here is a link to the Spring 2 measurements for an example of what a complete suite looks like:
https://superbestaudiofriends.org/index.php?threads/kte-spring2-technical-measurements.9258/

 

The Draco SE output debacle prompted me to make a very quick measurement of the Cyan 2 SE output.
Again Jeff Zhu demonstrates the difference between his excellent designs and the typical disposable or near disposable Chi-Fi rabble.

Holo Audio Cyan 2 100 Hz sine at -90 dBFS spdif in SE out (analyzer input DC coupled)


Holo Audio Cyan 2 100 Hz square at -90 dBFS spdif in SE out (analyzer input DC coupled)


No DC offset, no gimped performance from Balanced output.

 

Hi, Cyan 2 owner and enjoyer here with some input on why this is.

Look at the pictures of the Cyan 2 internals and count the rows of resistors in the ladder, now compare to Spring and May. (I'll save you the effort, the Cyan 2's PCM section is cut in half but the DSD section is same size as Spring and May)

EDIT: I had this backwards, it's the DSD part that's cut in half but they use a new topology that lets it perform better.

This DAC was built with DSD + HQPlayer in mind for sure.
Oh and DSD256 measures best on it btw, DSD512 starts to fall off (very slightly)

 

Does all of this mean that the Cyan2 is not a solid recommendation over say the Soekris2541, if the concepts of DSD and PCM are completely new to me, and my entire library is (PCM, probably)? Like I just have thousands of random FLACs of varying bitrates and some 320kbps MP3s.

If it matters, my main headphones are HD800S and ZMF Atrium, while my Amps are Singxer SA-1 V2 and Eufonika H5DS.