Dangerous Music Source

Only a few hours warm up and that's what I'm hearing. Sounds like mediocre quality opamps with high feedback output stage. Maybe even a mediocre cap or two for the analog filter. This don't sound like some Chris Muth masterpiece of an analog output circuit. Veiled and also a bit flat (microdynamics). It's mid-bassy, which seems like a Dangerous Music DAC trait.

It does need a little bit of warm up. Cold it's a disaster. @zerodeefex said to give it two days and it reaches peak performance. More in a bit.

P.S. Thanks to @Clemmaster for the loaner. He gets first dibs on one of the $$$ DACs I'm getting in. The main issue is Crane Song being able to deliver. The Convert 2 is already incoming and will arrive on Friday. Not enough in SBAF community funds (we still need to sell a few things), so I'm putting my own skin into this.

 

Here is how I am going to set things up. Comparison with Gungnir Multibit (with latest analog card upgrades).

• PC to Schiit Eitr (USB to SPDIF converter). I won't use USB because USB sucks from issues such as "USB haze", excess digititus, reduction in microdynamics and microdetail, wide variance in sound quality depending upon implementation on PC, etc. My other data sources are CD spinners, which don't suck. For this comparison, I won't be using them because it's unrealistic for people to buy CD spinners. However, it is not unrealistic for people to grab an Eitr. The Eitr outperforms most USB to SPDIF solutions. The Berkeley Alpha USB is better. The Lynx PCI cards with AES output are better or different (but cannot easily be used with a laptop). The Rednet Ethernet to AES is better, but too much fuss.
• A 75 ohm to 110 ohm coaxial to AES transformer to used to convert SPDIF SE out to AES into the Dangerous Music Source DAC. I've used this approach the Yggdrasil with excellent results. The Source being a piece of pro gear only has balanced inputs and outputs. The Eitr digital out is split and fed into both Gungnir Multibit and Source at the same time. Gungnir input is secured on the BNC jack since it does not provide digital AES inputs.
• Cinemag CMLI 15B line level transformers are used to convert the Source's outputs to SE into the EC Aficionado. I wired up another XLR to RCA ISO box so I can use the switch in the back of the amp to toggle between sources. This same approach is taken with the Gungnir, using its balanced analog outputs. In fact, the Gungnir sounds better using the balanced outs converted to SE via transformer instead of the SE outs. The SE outputs on the higher line Schiit DACs suck. They lack clarity and impart a greyishness to the sonic canvas. I elected to use these Cinemags instead of the Jensen ISO box because the Jensens warm up the sound a little bit. I wanted to maintain a neutral sound signature. The Cinemags are also 1:1 and do not step down like the Jensens. I suspect this might make them more transparent.
• There does not seem to be a speaker line level bypass on the Source, so I will adjust this to match the output levels of the Gungnir Multibit. I will assume no sonic penalty here because a pot will always be in the circuit. Level matching will be crucial.
• The EC Aficionado is configured with WE396A and vintage Hytron black plate 2A3s. The goal was a neutral sound, perhaps a pinch of warm.
• I will be using my heavily modded HD800 (Vader mod), Elex, and Verum (one back piece foam removed). The HD800 actually sounds HD650ish, but with more pronounced in a narrow spot of the upper mids / lower treble. The Elex I consider neutral up to the highs, where there is a small sharpish spike in the mid-treble, uneven lower treble, and metallic last octave. The Verum, as modified, is slightly laid back. Note that HD650 is not used. I feel that the HD650, despite whatever mods, sucks too much in the bass (issues only compounded by the mid-bass impedance bump) to evaluate bass properly since I have the non-feedback version of the Aficionado.
• Next part will be speaker rig evaluation.

 

Friends don't let friends use USB.

 

This is dumb. It's not even close. Don't get me wrong. The Source is a good DAC. It doesn't so anything wrong. It's not unlistenable like the pro DACs of old such as the Benchmark DAC-1. It offers more clarity and dynamics than the Lavry DA10. It's not an assault on the senses like the Mytek DSD. It's very solid.

At about five hours warmup for the Source. It's simply not competitive. Trust me, I really do want someone to beat the Schiit stuff because I don't think their stuff is perfect. There are a few areas where I would like to hear an improvement. But the Source just isn't doing it.

I've got Eva Cassidy, Eva by Heart playing. The Source renders the sounds very competently. The Gungnir Multibit (again, new analog boards) makes music. The Gungnir Multibit just throws so much information at you. It enthralls the senses. It sucks your soul into the music.

As far as the Source dynamics being hard-hitting, I'm not hearing it. Yes, there is a little mid-bass emphasis. (The Gungnir Multibit has more heft in the lower and sub bass) The Source is loud. Instead of crushing the dynamics downward, it crashes most of the volume levels upward, leaving all the softer range of volumes into a void. The missing information offers a level of blackness and false clarity that emphasizes the attacks. There are couple of passages, transitions, and a break in Wayfaring Stranger where the drummer hits the snare with an ff (fortismo). What we want to experience is a little jump. With the Source, those snare hits just kind of go by without much notice. With the Gungnir Multibit, they are definitely more noticed. (This is with Elex which is I feel is the most dynamic of the headphones I have on hand).

With the HD800, which I use to evaluate headstage, the Source presents a headstage which is even closer to the listener than the Gungnir Multibit's stage, which is already fairly upfront. The headstage depth totally sucks on the Source though, it's like all instruments and Eva are on a flat plate between my ears and my eyes. Imaging and localization also suck. Eva's voice seems to waver slightly up, right, left, wherever depending on the range she is singing in. Instruments are diffuse and I really don't know exactly where they are located off to the sides. I hate to say this, but the headstage almost like the worst of both OS and NOS worlds, with the exception that there sense of diffuse openness. To put things in perspective, I don't give a rats ass about headstage; but I've trained myself to evaluate this because I know that a lot of you guys do care about this.

One positive aspect is that I am finding is that the Source very listenable. No digititus nonsense. The sound is fluid. Thank goodness. So far, I honestly feel that the Source is something that is better off racked in a machine room of a sound bay or mixing room than used in the home. The balanced only inputs and outputs are not going to be compatible with most consumer gear. But if you want to get something different, it's a good option.

Will evaluate again in two days.

 

No caps is impossible. There's got to be an analog low pass filter. Opamps are worse than any of the above. If not, why did Chris Muth design a discrete circuit for the Convert-2?

Will test this and report back. This will be easy to arrange and makes things simpler.

 

@Psalmanazar: your friend status has been revoked. It's one thing to disagree or challenge my setup (which is exactly why I posted so much setup information in the first place - I am inviting challenge). It's another to abuse the dislike like an infantile teen.

 

Indeed. Now we are getting somewhere! Better highs, more cohesive throughout audio spectrum, and most importantly, I'm hearing the plankton and nuances. The downside is that everything is now a bit softer and mellow sounding. The XMOS implementation is actually pretty good here (or alternatively, the AES inputs suck). Going to do another evaluation with the USB. Headstage still sucks though. I don't like it when HD800s stage like Focals, makes HD800 pointless.

I'll just throw everything I got at it and see what sticks. Theta Data III, Sony VAIO (with the dedicated USB3), modded Marantz, etc. Best for one situation doesn't necessarily mean best for others. I just won't test with stuff I can't talk about.

 

On XMOS with the Source:

• More cohesive throughout the audio band. Better integration among bass, mids, and highs. Still that bit of mid-bass, but better integrated.
• I'd be hard-pressed to tell this was a sigma-delta design. Does not sound R2R, but doesn't have any sigma-delta hashy, strident, or mosquito nonsense. Very well done here.
• A bit different tonally and with transients: more mellow with slightly softer attacks. Still doesn't have the low and sub bass heft of Gungnir Multibit.
• Dynamics are improved from Eitr to AES.
• Headstage is still too close and imprecise. Using HD800, headstage is between my ears at worst and behind my eyeballs at best. Imaging and localization aren't any better.
• Microdynamics are also improved from Eitr to AES. The environment is more real. I don't feel like I'm so much in the Matrix anymore.
• Closer but still behind Gungnir Multibit per initial impressions using Eitr. I'd say one-third closer using XMOS USB compared to Eitr to AES. I wouldn't hesitate to recommend this to others. Just not myself. I don't change gear that often anyway. Too lazy and not curious enough and I prefer to make changes to things that make more of a difference, like speaker drivers, xover circuits, etc.

 

Not as a DAC. Just tried the RME from the built-in headouts, and ran away with my head on fire. That experience made me not want to touch it anymore, although I should have given it another chance (DAC only). I just made an assumption (possibily erroneous) that at least 40% of that glare and dead wimpy sound was from the DAC section. There is other stuff too like I never cared for the RME house sound. Life's too short. I almost died from illness a month ago, so I am more selective with what I chose to evaluate. DACs are a PIA to assess, so many variables to isolate.

The Source is worthy, I'm glad I heard it. It's not my deal, not good enough, but I am happy to award it "Not Shit" which was an honor of high regard back in the Changstar days. Very much looking forward to the Convert-2. I can only imagine the performance gains with the AD-1855 (or whatever the chip part number), a killer clock, and a well designed discrete circuit. I get the feeling I will like the Convert-2.

 

I don't use a preamp, with either my headphone or speaker setup. I can wire up a bypass on the pot of the Af or 45, but that's just going too nuts. And a pot wouldn't account for the observed differences.

A too hot signal will clip the signal or saturate the isobox transformers. This would be easily audible as horrible distortion. There is usually no middle ground - things go south very quickly. The output of the Source isn't that hot IMO, at least with the gear that I use. If no one would have told me, I wouldn't have even noticed. It's not like I don't have pro gear sitting around.

 

I'm beginning to wonder if XMOS USB is usually synonymous with soft and mellow. It seems so prolific, and I so commonly find USB to be too mellow, soft and sometimes even grey, hazy, etc.

 

In this case, I would like to think Dangerous Music focused most efforts on the XMOS USB because that's the target market these days. The AES inputs seem like afterthoughts since they suck so much. Verified this with @zerodeefex who tried his Data III CDP and Lynx card AES outs. I'll pull my Marantz out in a bit. This one has the deepest stage of my transports, so might have some complimentary effects. As an aside, the Data III is the most precise, focused, and stately. The Lynx almost has an organic liquid sound.

Until I get more experience to the contrary, because on every DAC I've evaluated so far using AES or SPDIF has always been better, I'm going to believe that the AES implementation sucks on the Source.

Yes, Cmedia USB sucks and so does XMOS. They just suck in different ways. I don't believe for one instant that DM found a magic bullet for XMOS though I'm sure they wrung out incremental improvements. Never been an XMOS fan.

 

Implementing this kind of stuff in the house is what makes wives let you die alone in the hospital. I've already pushed it too far, not pushing my luck.