Schiit Gungnir Multibit impressions

@atomicbob,

The measurement @skem posted was Yggdrasil A1 trying to driving 600 ohms. A1 didn't drive 600 ohms with any consistency. This was a well known issue, discussed here with EC Studio owners that used studio grade 600 ohm ISO transformers, before JA even posted his measurements.

 

LOL, that's what Jason was thinking. It was a mystery at first and I did some troubleshooting and going back and forth between Jason and Craig. What made things even more interesting is that some Yggdrasil units did better than others. It was the luck of the draw with the transistor. Craig went to the best ISO transformer that he knew of, a 600 ohm load, for this flagship amp. He worked at JBL/UREI making pro-gear back in the day, so that was his thinking.

In the end, Schiit tweaked the circuit and Craig vowed to stop using 600 ohm ISO transformers in his future amps.

 

You are making it too complex. Just subtract one from the other using the db scale and get the result. db is inherently relative - not absolute.

 

Well in the end it doesn't matter, unless your preamp inputs are actually 600 ohms, e.g. you are running an old transformer coupled mixing console.

 

It's not your fault really.

JA needs to change his procedures to test 10k ohm loads instead of 600 ohms. The 600 ohm test isn't very applicable today. JA is very well-spoken, a great presenter, cracks great jokes, but he's kind of a dipshit with his analysis. I mean, not total dipshit. Just kind of a dipshit. He measurements are good in the sense that he has great gear and has good practice. Kind of like the the old farts in one of them new 911s cars that almost drive themselves around the track.

 

I think you are indeed missing something. But I need to go out eat with my family.

Part of it is that the measurement is done at -60 dBFS and such a measurement is somewhat special. The fundamental is fairly low. This measurement, if I remember correctly, is done to get a feel for noise. Because at such a low fundamental, the distortion may not dominate THD+N (SINAD), it's noise power.

The more standard measurement is done at 0 dBFS (or close).

If you are looking at -60 dBFS, and the distortion is 65 dB below that, then you are looking at issues around -125 dBFS. That's like right at the 20-bit LSB!

EDIT: Note also that I don't feel the crossover distortion is unnatural or whatever. But the fact that the newer iterations of this product improve on this issue the way they do is pretty nice.

 

This is not a matter of "maybe so". It is a matter "it is not so".

We need to compare apples to apples. You cannot compare a full 0dbFS 1kHz measurement into a 600-ohm load vs a -60dbFS 1kHz measurement into a 100,000-ohm load. Attention to detail, the signal level, and the load, are crucial to understanding the plots.

However much you may want to have these measurement characteristics explain the relative harsh and edgy sound that you hear, they do not. Otherwise this Metrum DAC would have sounded this way, but it does not. It's sounds much more laid back and forgiving compared to than any of the Schiit DACs: https://www.superbestaudiofriends.o...s-amethyst-dac-measurements.4589/#post-152572

 

1. I've heard the the non-glitch and glitch DACs. No difference, at least with casual listening. They sound the same. This was on DACs with the glitch fix only, not full A2 upgrade.
2. What ameliorated the grainy sound was the full A2 upgrade, which was more than the glitch firmware fix.
3. The Schiit DACs are still incisive and not a relaxed sound. These are not easygoing DACs.
4. The Gungnir A2 is less warm, The Yggdrasil A2 is more V-shaped now.

TRANSLATION: I doubt you will like the new A2 Schiit DACs either.

You already explained yourself. You don't need to go to measurements to explain what you heard because the measurements do not correlate to what you heard. There are also other DAC with odd-order dominant distortion which are much more polite and forgiving than these Schiit DACs.

TRANSLATION: The bit-flip glitch and odd-distortion distortion of the Schiit DACs don't explain what you are hearing.

No. The distortion pattern is a result of multi-bit architectures, R2R ladder, 2R string, hybrid of such, etc. There is no such thing as noise at the LSB.

 

The moral of the story is don't look deeply or rush to judgement on DAC measurements unless they are really bad, on the scale of -80db or higher. This is assuming the measurements were not borked, which is possible even with experienced people.

Finally, system synergy and personal preferences are going to play a HUGE role.

 

I actually meant absolute scale of any signal (not distortion product below fundamental) - depends how loud you listen.

Let's say I listen to music with peaks at 105db SPL equating to 0dbFS, which is fricking loud. An quiet living room at night is around 35db SPL. That means anything below -70dbFS is likely going to be masked by environmental sounds and noise. Let's pad this by 10db more just to be conservative. That -80dbFS. Anything below -80dbFS / 35dbSPL won't be heard - not to mention that -60dbFS is about the limit for human beings to detect distortion.

 

I think the "glitch" would indeed be harmonic given the test signal is a periodic sinusoidal. The "glitch" you are seeing toggling in the LSB range is periodic. But like @purr1n said, this is not noise but distortion.

It's distortion because it's correlated in time (periodic). The difference becomes more clear when you use an analyzer and depending on background. But we are drifting away from the spirit of this thread.

Regardless, like @purr1n said, I don't think you will like these MB DACs. I don't think it's the "glitch". You should search for what you like. Maybe try the DS @schiit DACs.

Currently I use the 2i2 (DS) DAC output on the back of the unit which is pretty good even single ended.

Maybe I'll write an essay about why I like the 2i2, but maybe you don't... But I have a feeling nobody is going to give a shit.

 

A shit. That's right. The -60 dBFS measurement is the good 'ol Dynamic Range measurement. Indeed this is not a THD or THD+N measurement which are done at closer to 0 or -3 dBFS (which I would argue JA purposely fucked up by using a fairly low 600 ohm load for a DAC - JA can be like that sometimes).

Indeed the above shows that the "glitch" is basically a non-issue relative to other sources of distortion at normal to high listening levels (-60 dBFS is pretty low).