Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[PrinterPrinter]

Hello Chaps,
I'm using MC on RPI and love it. I assume it works the same as the windows version.

I have set up some EQ filter (used REW to get them) to tame my sub. My set up is pre-amp-less - so the RPI goes directly to the Yggy and then to a power-amp and I use a pair of inline attenuators as well as MC's internal volume to control the volume. Because the attenuators are pretty aggressive and I also use volume levelling - which usually reduces the volume by around 10db I often find myself playing at 100 volume and could use some flexibility.

So I added another PEQ filter after the 'bass PEQ' that just boost volume by 3dB and gives me more flexibility. My question is does this impact or degrade quality in any way and is it best to boost the volume before the PEQ does it's bass things? I suspect hat because it's all done in software it's basically math and doesn't matter to SQ but I want to do it right.

Thank you!

[mattkhan]

If you are at 100% internal volume and you add more volume via the PEQ block then you will digitally clip the signal (if the signal you put through is within x dB of full scale where x is the amount of volume you added)

To avoid this, you have to turn it down (attenuate) digitally. One source of attenuation is volume levelling. This means that if volume levelling has reduced the level by more than you want to turn it up, you're ok. If volume levelling reduces it by less than your PEQ then you're not. Why are you using inline attenuators btw?

[PrinterPrinter]

If you are at 100% internal volume and you add more volume via the PEQ block then you will digitally clip the signal (if the signal you put through is within x dB of full scale where x is the amount of volume you added)

To avoid this, you have to turn it down (attenuate) digitally. One source of attenuation is volume levelling. This means that if volume levelling has reduced the level by more than you want to turn it up, you're ok. If volume levelling reduces it by less than your PEQ then you're not. Why are you using inline attenuators btw?

Hey Mate,
Thank you!

Yes I use volume levelling which sometimes gets to -15dB. So I hope adding 3dB doesn't clip the signal. Is there a way for me to see in MC if the signal is clipped?

I've been playing with EQ and some of the filters REW suggest boost certain frequencies by 10dB - is it possible to clip the signal that way? (I suspect it's trying to boost room nulls...)

I'm using inline attenuators because I don't have a pre-amp, just a DAC (Schiit Yggy) going straight into the power amp (ATI 6002 (300W) - so I got -30dB attenuators (Rothwell attenuators) which worked fine and gave me headroom. I then changed speakers to slightly less efficient ones (ML Electromotion) - so with some recordings I'm running out of volume The real solution of course is to get new attenuators, but I'm lazy ;-)

[mattkhan]

The analyser window shows you the level at the end of the chain however if the boost added is less than volume leveling then you are OK.

Any boost requires digital headroom to avoid clipping yes. If REW is suggesting 10dB of boost then that is because you have configured it to allow that. You can reduce this in its eq window.

[blgentry]

As an aside that might be important to you:

A huge part of the "magic" of the Schiit Yggdrasil is the digital filter that they call the MegaCombo Burrito Filter.  It's job is to upsample the incoming 44.1, 48, 96, etc signals to Yggdrasil's internal sampling rates.  It does this while keeping ALL of the original samples that are sent to it.  This is a "big deal" in that no other resampling algorithm is able to do this!

If you assume that the samples that are on the recording are worth preserving, then you should send *those* samples to the DAC.  This means NO DIGITAL PROCESSING before the DAC.  Any EQ, volume changes, convolution, etc destroy the original samples completely and make the Yggdrasil not all that special because it's MCBF isn't being used on the original samples any more.  It's working with newly manufactured samples, as a result of the digital processing.

Certainly Yggdrasil is a great DAC even without the MCBF.  But MCBF is what makes it special.  If you pre-process your digital signals, you're not really getting that special part of Yggdrasil.

If you believe in the Yggdrasil, you really want an analog volume control *after* Yggdrasil.  Food for thought.

Brian.

[PrinterPrinter]

As an aside that might be important to you:

A huge part of the "magic" of the Schiit Yggdrasil is the digital filter that they call the MegaCombo Burrito Filter.  It's job is to upsample the incoming 44.1, 48, 96, etc signals to Yggdrasil's internal sampling rates.  It does this while keeping ALL of the original samples that are sent to it.  This is a "big deal" in that no other resampling algorithm is able to do this!

If you assume that the samples that are on the recording are worth preserving, then you should send *those* samples to the DAC.  This means NO DIGITAL PROCESSING before the DAC.  Any EQ, volume changes, convolution, etc destroy the original samples completely and make the Yggdrasil not all that special because it's MCBF isn't being used on the original samples any more.  It's working with newly manufactured samples, as a result of the digital processing.

Certainly Yggdrasil is a great DAC even without the MCBF.  But MCBF is what makes it special.  If you pre-process your digital signals, you're not really getting that special part of Yggdrasil.

If you believe in the Yggdrasil, you really want an analog volume control *after* Yggdrasil.  Food for thought.

Brian.

Thanks Mate,
I'm not sure about this, I can see what up-sampling before the Yggy would make a difference but I was under the impression that as long as we are in the MC 64bit - things like volume levelling don't matter much?
Also my EQ is very light just to bring the sub under control (I'm running 2 mains + Sub) directly from the Yggy...

I actually asked Mike MOffat (on Headfi) if applying EQ somehow messes with the Yggy and got a typically cryptic response

In other news, I am interested to know if MC applies dither (if it's turned on) after applying EQ or volume levelling internally?
I noticed the Yggy sounds different with dither on and off in MC. Can't decide which sounds better.
My audio path on MC says - output: direct connection 32bit into the Yggy so presumably there's no word length reduction?

What do you guys think?

[blgentry]

I'm not sure about this, I can see what up-sampling before the Yggy would make a difference but I was under the impression that as long as we are in the MC 64bit - things like volume levelling don't matter much?

As soon as you apply any kind of DSP, including volume changes, all of the original samples are destroyed and replaced with new ones.  The question then becomes, is this effect audible?  Mike Moffat thinks so; he's spent more than 20 years working on the solution to this by building the MCBF that's in the Yggdrasil.  I honestly don't know because I've never heard one.

You're in a really cool position.  You own one and you own JRiver MC.  If you have any kind of analog preamp, you can turn off DSP and Volume in MC and see if you hear a difference or not.

Brian.

[PrinterPrinter]

As soon as you apply any kind of DSP, including volume changes, all of the original samples are destroyed and replaced with new ones.  The question then becomes, is this effect audible?  Mike Moffat thinks so; he's spent more than 20 years working on the solution to this by building the MCBF that's in the Yggdrasil.  I honestly don't know because I've never heard one.

You're in a really cool position.  You own one and you own JRiver MC.  If you have any kind of analog preamp, you can turn off DSP and Volume in MC and see if you hear a difference or not.

Brian.

Thanks Mate,
The problem is that I don't own an analogue pre-amp or the extra cables... ;-(

My impression is that when mike talks about retaining the original samples he's really talking about R2R vs Delta Sigma - which results in an approximation of the original samples who are thrown away... 

Lots of people use MC with the Yggy... Surly if I apply volume or minimal EQ in MC and then feed it to the yggy - that's considered 'original samples' as far as its concerned? I imagine that significant DSP or convolution may introduce phase errors etc but minimal stuff? MC has an internal engine of 64bit... changing volume just adds or removes a few zeros, no?

[mattkhan]

any music has been through innumerable processing stages to get to the end product so I find it hard to see how any concept of an "original" sample has any meaning other than in terms of translating digital to analogue, i.e. do any processing you like while you're in the digital domain.

NB: I'm not a "computer audiophile" so feel free to ignore if you operate under a different system

[PrinterPrinter]

any music has been through innumerable processing stages to get to the end product so I find it hard to see how any concept of an "original" sample has any meaning other than in terms of translating digital to analogue, i.e. do any processing you like while you're in the digital domain.

NB: I'm not a "computer audiophile" so feel free to ignore if you operate under a different system

Thanks Mate,
It sound really great the way I use it and I lack the setup to try it any other way so I'm not especially worried.

One last question about the EQ, is it considered bad form to boost room nulls? REW keep suggesting it (if I allow it) - I'm talking specifically in the bass - 70hz and 140hz

[mojave]

In other news, I am interested to know if MC applies dither (if it's turned on) after applying EQ or volume levelling internally?
I noticed the Yggy sounds different with dither on and off in MC. Can't decide which sounds better.
My audio path on MC says - output: direct connection 32bit into the Yggy so presumably there's no word length reduction?

Most ASIO devices show themselves to be 32-bit. In JRiver, if you have less than a 32-bit DAC, then you need to go to Tools > Options > Audio > Audio Device > Device Settings > Output and check "Device uses only most significant 24-bits". JRiver will now dither to 24-bits while padding to 32-bit for output.

You can also go Options > Audio > Advanced > Dither Mode and change between two types of dither or turning it off. Anytime you do any processing, including volume control, you should dither.

[mojave]

The Schiit Yggdrasil implements a SHARC DSP inside it to do its processing.  ?

"The SHARC DSP processor implements Schiit's proprietary 18,000+ tap digital filter algorithm at 352.8 or 384kHz sampling rate and 20-bit depth"

[blgentry]

Surly if I apply volume or minimal EQ in MC and then feed it to the yggy - that's considered 'original samples' as far as its concerned?

No, not at all.  The original samples get changed any time you do any volume changing or EQ with DSP.

I imagine that significant DSP or convolution may introduce phase errors etc but minimal stuff? MC has an internal engine of 64bit... changing volume just adds or removes a few zeros, no?

Sorry, but no.  Yes, MC is 64 bit internally, and (as far as I know) has a world class implementation.  But any DSP processing changes all of the samples.  Even changing the volume.

Can you hear the difference?  I honestly don't know.  I suspect that the only recordings you can really hear a difference on were made in real acoustic spaces (not recording studios) and had minimal processing done, and only got turned into digital at the very last stage.  But I'm guessing at all of that.  Again, I've never heard a Yggdrasil.  I *do* envy you that you get to hear one all the time. 

Brian.

[blgentry]

The Schiit Yggdrasil implements a SHARC DSP inside it to do its processing.  ?

Yeah, the MCBF apparently requires a lot of DSP math.  Why the confused smiley?  Is the MegaCombo Burrito Filter new to you?  If so, I could understand your puzzlement.  It's totally unique as far as I know and took Mike and the boys a pretty incredible amount of development to produce.  The math being the big part.

Brian.

[PrinterPrinter]

Ok,
So maybe I was confused by 'bit perfect' - I understand that with MC, as long as you don't apply EQ and only change volume while using their dither then it still stays bit perfect - with the blue icon lit?

Where as you say that even changing the volume messes up the 'original samples'?

I'll see if I could EQ my sub 'au natural' without an EQ just using placement and the sub controls - never tried that.

[blgentry]

So maybe I was confused by 'bit perfect' - I understand that with MC, as long as you don't apply EQ and only change volume while using their dither then it still stays bit perfect - with the blue icon lit?

Where as you say that even changing the volume messes up the 'original samples'?

The blue light is rather misleading.  It lights up blue even when volume changes are being made.  This in NO WAY implies that the bits are not being altered.  They absolutely are being altered when volume changes are being made.  I asked about this a while back and was told (and read threads) that said essentially this:  "We feel that volume changes don't make a big difference in the signal, so we keep the blue light on even when volume changes are being made".

The blue light is more of a "no EQ, channel swapping, or 'big changes' light".  It does NOT show "bit perfect delivery".

This is kind of an academic point though:  I know what I'm saying is correct.  The real question is:  does it make a sonic difference?  Only you can answer that question on your system.

Brian.

[mojave]

Yeah, the MCBF apparently requires a lot of DSP math.  Why the confused smiley?  Is the MegaCombo Burrito Filter new to you?  If so, I could understand your puzzlement.  It's totally unique as far as I know and took Mike and the boys a pretty incredible amount of development to produce.  The math being the big part.

Brian.

1.  The DAC requests 32-bit data from software via ASIO.
2.  The DSP in the DAC outputs at 20-bit.
3.  The DSP will never keep the "original samples" if the source files were at least 24-bit.

any music has been through innumerable processing stages to get to the end product so I find it hard to see how any concept of an "original" sample has any meaning other than in terms of translating digital to analogue, i.e. do any processing you like while you're in the digital domain.

I agree. There really is no such thing as an "original" sample. I do live sound several times a month at one venue and on Monday evenings at another venue. At both venues, all sound goes through four A/D and D/A conversions and three different DSP's. Volume levels are digital. What the audience hears is the original as we intended it.

[blgentry]

1.  The DAC requests 32-bit data from software via ASIO.
2.  The DSP in the DAC outputs at 20-bit.
3.  The DSP will never keep the "original samples" if the source files were at least 24-bit.

Well, yes that's technically true, but it misses the point.  No audio DAC on the planet has more than ~21 - 22 bits of effective resolution.  Their noise floors aren't low enough to discriminate the lowest bits.  The 24 bit DAC on my desk probably has 17 or less effective bits.

The Yggdrasil uses TWO 20 bit R2R DACs in tandem, per channel, to achieve a true, measurable 21 bits of resolution.  See the FAQ here:

http://schiit.com/products/yggdrasil

But the "big deal" about the MCBF isn't how many bits it can resolve.  It's about not destroying every sample that comes in and replacing it with another.  Why is that important?  Frankly, I have to go on what I've read here, as I've never heard one and I don't know the math behind it.  What I've *read* is that keeping the original samples preserves the phase information more fully.  Which leads to spacial cues being more evident.  I.E., you can hear more positional information when you listen to playback that preserves the original samples.

Again, I have NOT heard one.  This is theory for me.  Interesting theory. 

As to your point about live playback and modern recording techniques, you're correct for sure.  Which is why I said that the MCBF probably only matters with recordings that were made in a way that only digitized them once, *and* had minimal processing done.  If the acoustic cues that tell our ears where things are in a live acoustic space aren't on the recording to begin with, then this filter probably won't help.  If those samples have already been destroyed by digital DSP, then the filter won't help.

I hope this is taken in the way I'm intending it:  I'm presenting the technology as I understand it.  Technology that I think is rather fascinating actually.  I'm not arguing.  Hopefully I've given you some information you didn't already have.

Brian.

[Guybrush]

If you are at 100% internal volume and you add more volume via the PEQ block then you will digitally clip the signal (if the signal you put through is within x dB of full scale where x is the amount of volume you added)

To avoid this, you have to turn it down (attenuate) digitally. One source of attenuation is volume levelling. This means that if volume levelling has reduced the level by more than you want to turn it up, you're ok. If volume levelling reduces it by less than your PEQ then you're not. Why are you using inline attenuators btw?

I also use inline attenuators going directly into a power amp as they are necessary for reducing PC noise transmitting through my speakers. I'm adding channel-specific volume via room correction. Is the same true in that case that I am in danger of digital clipping if setting the volume (internal) too high?

[blgentry]

I also use inline attenuators going directly into a power amp as they are necessary for reducing PC noise transmitting through my speakers.

A good DAC shouldn't make "PC noise".  You should figure out why you are getting this noise and eliminate it through standard techniques of ground isolation, RF isolation, etc.  Feel free to ignore me of course.  Just my perspective. 

Brian.

[mwillems]

A good DAC shouldn't make "PC noise".  You should figure out why you are getting this noise and eliminate it through standard techniques of ground isolation, RF isolation, etc.  Feel free to ignore me of course.  Just my perspective.  

Brian.

If you have a high-enough-gain power amp hooked up to sensitive speakers, the DAC output of even very nice dacs can cause PC noise out of the speakers.  Low noise helps obviously, but the amps will amplify the noise-floor of the DAC as readily as they amplify the signal.

I have a DAC with a noise floor at -119dBFS, which is pretty excellent (about like the Yggy). If I hook it up to a block power amp with 20dB of gain and feed it into my 113dB 1W/1M sensitivity horns, I get audible computer noise (although it's fairly quiet).  In-line attenuators and/or a lower amp gain solve the problem for me entirely.

There is not a DAC that's so low noise that it can't lead to audible PC noise with the right downstream chain, especially as a lot of power amps will have amplification factors of 30dB or above.  Getting the lowest noise DAC you can afford is a good start, but not a guarantee.

[Guybrush]

A good DAC shouldn't make "PC noise".  You should figure out why you are getting this noise and eliminate it through standard techniques of ground isolation, RF isolation, etc.  Feel free to ignore me of course.  Just my perspective. 

Brian.

My DAC is an internal PC sound card, and I believe it picks up a lot of noise through the PCIe bus. Worst is when my 980Ti really ramps up to speed while gaming, or even when the ODD spins watching a movie.

But my question stands, do I risk digital clipping by staggering the channel gain in room correction and playing at high volume?

[mattkhan]

I also use inline attenuators going directly into a power amp as they are necessary for reducing PC noise transmitting through my speakers. I'm adding channel-specific volume via room correction. Is the same true in that case that I am in danger of digital clipping if setting the volume (internal) too high?

yes, if you add volume using DSP and use 100% internal volume then you can clip

[mattkhan]

But the "big deal" about the MCBF isn't how many bits it can resolve.  It's about not destroying every sample that comes in and replacing it with another. 

as far as I can see that's about making it a better DAC, i.e. so that the A more faithfully reproduces the D

I don't see how this has any relationship at all to what those bits represent or contain other than to make nice marketing blurb

[Guybrush]

yes, if you add volume using DSP and use 100% internal volume then you can clip

Thanks. What's the best way to detect if clipping is occuring?

[PrinterPrinter]

Ok Guys,
I made an experiment, I tried to use as less MC as possible - disables the sub EQ and volume levelling and set dither to 'bit perfect' mode so I had the blue icon lit...

I think I have a very resolving system: The Yggy goes diretly to an ATI 6002 poweramp that drive a pair of Martin logan Electromotions.

I didn't feel any noticeable difference in sound quality with the Yggy. Obviously EQing the sub brings it under control...
I did notice theres audible difference between the the dither modes (none, bit-exact and TPDF) with the latter sounding noticeably better, with much more air around the instruments. No EQ sounds very hash and lacking space in the soundstage.

I think I'll keep using MC with EQ and volume levelling, dither etc... It sound great and very convenient for me.
Assuming I'll disable all processing in MC, then I'll need to get a passive pre-amp for volume, another pair of balanced Townshend fractal IC and will need to get up every time to change the volume - my listening position is about 3m from the Yggy... and I;ll still need to find a way to EQ the sub...

Lastly, I'm not sure that adding this extra stage will actually improve sound, a the saying goes -
the most transparent pre-amp is no pre-amp...