Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › Metrum Acoustics Octave
New Posts  All Forums:Forum Nav:

Metrum Acoustics Octave - Page 45

post #661 of 666

I will give the Hex that is has balanced outputs and other functionality the other DACs in the family do not, but the incremental improvements and refinements to sound (relative to Quad, not Octave, which I haven't heard yet - but Mk2 does use a Hex board) still make me wary of the base $2.5K or so price tag. If one has money to blow, though, then by all means, the Hex is great. I loved mine and only had to sell it for moving money.

 

I have some, not all, of the DACs I've owned or tested in my profile, most of which I directly compared to the Hex.

post #662 of 666
Quote:
Originally Posted by zilch0md View Post
 

Thanks again, Amine.  

 

For the record, I should make it clear that Cees Ruijtenberg (via an email reply) did not condemn the use of upsampling, altogether - he just warned me of the possibility of undesirable artifacts (that I should listen carefully as I experiment with upsampling.) But yes, OS DACs can more readily benefit from upsampling- especially when starting with 44.1kHz files.

 

I think one of the greatest benefits of DSD, with sampling rates of 2.8 MHz and higher, is that it might actually be easier (I'm conjecturing, here) to design a DA converter to handle those rates - certainly easier than doing so for 44.1kHz - and perhaps that's why we are seeing a proliferation of inexpensive DSD-capable DACs that sound better than the same or similarly priced DACs sound when decoding 44.1kHz files.  DSD could actually transform the industry by bringing "clean" DA conversion to the masses, training their ears to find fault with the inferior 44.1k DA conversion we've come to accept.

 

Will any DSD DAC ever be as good as what I'm hearing with the Octave MkII?  I don't know - I'm just trying to emphasize the possibility that the higher the sampling rate of the recording, the easier it is to design a DAC to handle it.

 

Mike

 

Thanks for clarifying the upsampling point. 

 

With regards to DSD vs. PCM, I think that it might be easier overall to construct a DSD

 

Below is supposed to be a comparison of impulse responses of various formats: 

 

 

It is clear, from comparing different impulse responses, that DSD is the closest one to analog, followed by the higher sampling rates of the PCM format.

 

I do remember that at some point, on Metrum's website, that they published comparisons of impulse response for NOS vs. oversampling DACs. From memory, the impulse response of the Metrum DAC looked more like analog (or DSD) than the 48K (or 96K, 192K) PCM shown in the picture above.

 

I believe that part of the Metrum DAC signature (clean, fast and devoid of digital edginess) is due to that clean impulse response. The best SD DACs might be able to reduce the pre and post ringing, but it is probably much easier to get a clean sound with either an R2R DAC playing or restricting SD to playing DSD. But that is just pure speculation.

 

 

As for the Hex vs. Octave comparison, I am very interested in hearing more about it for those who have had the chance to compare the two. On paper, for single ended use, and in comparison with the Octave MKII, it doesn't seem as if the Hex has much to go for it besides the better power supply. As for the Lundhal transformers, I don't know what to make of it.

Given that my main (tube amplifier) is single ended, I feel that if I upgraded to the Metrum Hex, half of the chips will go to "waste". 

I hope that Metrum can come up with DACs equivalent that is 100% dedicated to single ended (btw, that was the main reason I didn't get an Audio-gd Reference series DAC which are also built for balanced use).

post #663 of 666

I really enjoy reading your stuff, slim.a!

 

Thanks for that follow-up (above)  ^

 

I think it's the traits which we've not yet learned how to measure, but which our ears can appreciate, that most excite me with the Octave MkII.  I'm so content with its performance that I'm really not even inspired to hear the Hex.  Perhaps that's just my frugal nature kicking in, but really, every combination of amp and headphone I put downstream from the Octave MkII is made better than I've heard the same gear when sourced with other DACs, admittedly a fairly small sample of nothing but oversampling DACs. 

 

Mike

post #664 of 666
Quote:
Originally Posted by slim.a View Post
 

I do remember that at some point, on Metrum's website, that they published comparisons of impulse response for NOS vs. oversampling DACs. From memory, the impulse response of the Metrum DAC looked more like analog (or DSD) than the 48K (or 96K, 192K) PCM shown in the picture above.

 

As for the Lundhal transformers, I don't know what to make of it.

 

That's true, but there's much more to the picture than that. While you may get a clean looking impulse response or square wave on NOS DACs, you'll also get stair-stepped sine waves that look like this: 

 

 

An analog sine wave would look smooth, and you could make the argument that music is composed more of sine waves, not impulses or square waves. When you get to high treble frequencies, especially at or over 10KHz, the NOS sample-and-hold characteristic makes the measured signal look so stair-stepped that you can barely recognize the sine wave. One upside is that going to higher sampling rates greatly improves this. Even a 96KHz-sampled music file does wonders for this stair-stepped pattern on NOS DACs. (If you run NOS DACs, you owe it to yourself to get music at higher sampling rates or try to find software upsamplers without filter ringing.)

 

However, there's one idea floating around that by the time the wave goes through an amp, a driver, and hits your ears and brain, it'll be smoothed out. I think there may be some truth to this, but I'm not sure. On one hand, I've heard a NOS DAC that was able to play 0dB sine tests cleanly, subjectively, all the way up to 20KHz. This was a PCM1704-based NOS DAC. On the other hand, I've heard another NOS DAC (Metrums, TBH), that as you moved up to higher sine wave frequencies at 0dB, it would start to fall apart and sound more like a 56K dial-tone and would measure with a ton of distortion. A counterpoint to this is that music probably isn't going to have treble signals anywhere near the 0dB level, and high treble sine waves did sound subjectively cleaner and less distorted to me on the Metrums the more you lowered their level.

 

I know some believe upsampling/oversampling filters with pre and/or post-ringing don't have audible effects, but my (potential flawed or totally placebo) subjective tests make me think otherwise. The most noticeable, audible difference will be the slight treble droop on NOS DACs. But I do think there is something to be said about human hearing and what upsampling/oversampling filters with lots of ringing might do to the time domain in an audible manner.

 

There are potential trade-offs either way, in theory. I say just go with whatever sounds best to you, and I did love how smooth and musical the Metrum DACs could be. It's good to try to know what's going on with your sound on a technical level, but it's also best to trust your ears in the end and go with what works for your tastes. I know the Schiit Yggy is supposedly meant to optimize the frequency and time domains, so perhaps it'll actually be the best of both worlds. (Clean impulse and square wave on NOS being focused on time domain, smooth sine wave and no treble-droop on OS DACs being focused on frequency domain...sorta like that, if that makes sense.)

 

Now, as for those transformers in the Hex, I've heard some say the Quad and Octave sound more "direct" due to not having output transformers. Whatever that means. Some prefer it without, some prefer it with (Hex).


Edited by hans030390 - 10/20/14 at 12:22pm
post #665 of 666
Quote:
Originally Posted by hans030390 View Post
 

 

That's true, but there's much more to the picture than that. While you may get a clean looking impulse response or square wave on NOS DACs, you'll also get stair-stepped sine waves that look like this: 

 

 

An analog sine wave would look smooth, and you could make the argument that music is composed more of sine waves, not impulses or square waves. When you get to high treble frequencies, especially at or over 10KHz, the NOS sample-and-hold characteristic makes the measured signal look so stair-stepped that you can barely recognize the sine wave. One upside is that going to higher sampling rates greatly improves this. Even a 96KHz-sampled music file does wonders for this stair-stepped pattern on NOS DACs. (If you run NOS DACs, you owe it to yourself to get music at higher sampling rates or try to find software upsamplers without filter ringing.)

 

However, there's one idea floating around that by the time the wave goes through an amp, a driver, and hits your ears and brain, it'll be smoothed out. I think there may be some truth to this, but I'm not sure. On one hand, I've heard a NOS DAC that was able to play 0dB sine tests cleanly, subjectively, all the way up to 20KHz. This was a PCM1704-based NOS DAC. On the other hand, I've heard another NOS DAC (Metrums, TBH), that as you moved up to higher sine wave frequencies at 0dB, it would start to fall apart and sound more like a 56K dial-tone and would measure with a ton of distortion. A counterpoint to this is that music probably isn't going to have treble signals anywhere near the 0dB level, and high treble sine waves did sound subjectively cleaner and less distorted to me on the Metrums the more you lowered their level.

 

I know some believe upsampling/oversampling filters with pre and/or post-ringing don't have audible effects, but my (potential flawed or totally placebo) subjective tests make me think otherwise. The most noticeable, audible difference will be the slight treble droop on NOS DACs. But I do think there is something to be said about human hearing and what upsampling/oversampling filters with lots of ringing might do to the time domain in an audible manner.

 

There are potential trade-offs either way, in theory. I say just go with whatever sounds best to you, and I did love how smooth and musical the Metrum DACs could be. It's good to try to know what's going on with your sound on a technical level, but it's also best to trust your ears in the end and go with what works for your tastes. I know the Schiit Yggy is supposedly meant to optimize the frequency and time domains, so perhaps it'll actually be the best of both worlds. (Clean impulse and square wave on NOS being focused on time domain, smooth sine wave and no treble-droop on OS DACs being focused on frequency domain...sorta like that, if that makes sense.)

 

Now, as for those transformers in the Hex, I've heard some say the Quad and Octave sound more "direct" due to not having output transformers. Whatever that means. Some prefer it without, some prefer it with (Hex).

 

I forgot about stair-stepped issue... However, it seems that (some of) our ears are able to smooth out those sine waves very efficiently :)

 

As for the PCM1704uk, I tried 0x oversampling (NOS) vs. 8x oversampling on the audio-gd dac19DSP and I preferred the sound with oversampling. At least that is what seemed "right" to my ears.

post #666 of 666
 

 

Below is supposed to be a comparison of impulse responses of various formats: 

 

Looks like some Sony propaganda. There is no 'analog' recording system that reproduces a 3uS pulse like they've shown. It would need bandwidth upto a MHz or so to get edges as steep as that. But since 3uS wide pulses never occur in performances, the whole comparison is rather moot. The DSD result is faked because there's still a low-pass filter recommended on the output of a DSD converter, From memory I think its around 80kHz corner frequency so the pulse would get smeared out but probably without the pre-ringing shown for the 192kHz PCM. The amplitude of the pre- and post-ringing looks rather exaggerated for effect in the 192kHz case.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Dedicated Source Components
Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › Metrum Acoustics Octave