So Robb. Can I infer here that digital is not all the way there yet when it comes to tonally adding a treble boost, as compared with best studio analog hardware? Specifically talking about treble boost here, the “air band”
Watts Up...?
- Thread starter wink
- Start date
Rob Watts
Member of the Trade: Chord Electronics
- Joined
- Apr 1, 2014
- Posts
- 3,154
- Likes
- 12,561
I am afraid I don't know the answer to that for sure - apart from knowing conventional EQ is subjectively flawed. I know a lot of recording engineers prefer analogue desks (using DACs and ADCs with analogue EQ), saying they sound better. How much is down to preferring distortion, or how much is down to digital EQ being poor, I don't know. I suspect it's a mixture of both.
Treble EQ is much easier than bass EQ, as the bi-quad coefficients are large values. The largest problems occur with bass - some of the coefficients are very small. So that small value once truncated and then fed back creates significant errors that then accumulate. With IIR filters, the signal is infinitely fed back (hence IIR meaning infinite impulse response). But if the signal is truncated away to zero, then the IIR is no longer functioning as a filter for that signal.
Treble EQ is much easier than bass EQ, as the bi-quad coefficients are large values. The largest problems occur with bass - some of the coefficients are very small. So that small value once truncated and then fed back creates significant errors that then accumulate. With IIR filters, the signal is infinitely fed back (hence IIR meaning infinite impulse response). But if the signal is truncated away to zero, then the IIR is no longer functioning as a filter for that signal.
Treble may be easier in a digital EQ implementation environment (from a programming viewpoint? I am only a tweaking audiophile, no engineer. But from my perspective on the hi fi playback end, treble EQ in its final sonic resultant SQ is HARDER to get right. I spent years to find the best audiophile listening solution to rolled off treble recordings. Your Mojo 2 is the best digital solution I’ve ever heard. Will you be implementing it in pricier desktop non portable designs? The best analog solution for treble roll off for me for the last 10 years remains a Charter Oak PEQ-1 in both my loudspeaker chain and my headphone chain. It leaves the whole frequency spectrum unmolested while creating the most beautiful air lifts that really breathe life into the appropriate recording. I think the studio engineers quite honestly are on to something. 
ecwl
500+ Head-Fier
Being able to hear conventional digital EQ or analog EQ artifacts is always challenging. This is because most of the time, there is a strong reason for the EQ in the first place. So the benefits of EQ by a few dB in a problematic frequency would frequently outweighs the harm and distortions caused by digital or analog EQ. In fact, often, I have found that the benefits of EQ would completely dwarf the more subtle issues introduced by EQ. Moreover, these benefits are usually more easily audible compared to the newly generated distortions.
Also, just as some people have trouble hearing the improvement in transient accuracy of 1 million taps vs 100 taps, some people can't hear or aren't bothered or even prefer noise floor modulation. So it is difficult to say to them that the easily audible improvement in frequency response from EQ is a bad idea because of the new distortions they have introduced.
Another interesting aspect of EQ in speaker systems is that the need for EQ usually originates from a room mode which not only creates a bass frequency peak but also a bass frequency resonance. Any EQ can only turn down the dB of the peak but cannot change the resonance. So EQ itself is a compromise because you can either turn down the dB so that even with the resonance, the frequency response is flat but when the note is struck, it won't be as dynamic because you've turned down its volume dramatically. Or you can aim for the initial note volume to be roughly equal to other notes but then with time, the resonance of the note would still dominate the room. After all, there is no replacement for addressing room acoustics other than to actually treat the room acoustics. I find many people don't pay attention to this aspect/compromise when they EQ.
Ideally I would love to be able to implement EQ in my living room using Rob Watts technology. In the mean time, as with everything in audio and in life, I try to live with the compromises that are acceptable to me. And I think for people who are blessed who find no problems with their current digital EQ or analog EQ, kudos to them.
Also, just as some people have trouble hearing the improvement in transient accuracy of 1 million taps vs 100 taps, some people can't hear or aren't bothered or even prefer noise floor modulation. So it is difficult to say to them that the easily audible improvement in frequency response from EQ is a bad idea because of the new distortions they have introduced.
Another interesting aspect of EQ in speaker systems is that the need for EQ usually originates from a room mode which not only creates a bass frequency peak but also a bass frequency resonance. Any EQ can only turn down the dB of the peak but cannot change the resonance. So EQ itself is a compromise because you can either turn down the dB so that even with the resonance, the frequency response is flat but when the note is struck, it won't be as dynamic because you've turned down its volume dramatically. Or you can aim for the initial note volume to be roughly equal to other notes but then with time, the resonance of the note would still dominate the room. After all, there is no replacement for addressing room acoustics other than to actually treat the room acoustics. I find many people don't pay attention to this aspect/compromise when they EQ.
Ideally I would love to be able to implement EQ in my living room using Rob Watts technology. In the mean time, as with everything in audio and in life, I try to live with the compromises that are acceptable to me. And I think for people who are blessed who find no problems with their current digital EQ or analog EQ, kudos to them.
I am lucky I’ve got a good listening space. No boomy bass frequencies spike up. Very even bass response
GuiltyRocker
1000+ Head-Fier
@Rob Watts what's your take on artificial ground boxes like the Nordost ones and others in the market? They claim that grounding devices to them makes a big difference.
Last edited:
Rob Watts
Member of the Trade: Chord Electronics
- Joined
- Apr 1, 2014
- Posts
- 3,154
- Likes
- 12,561
Scepticism...
But having said that, a very dear friend of mine by the name of Mike Priestland (who sadly died in the 90's), who was my first believer, mentor and supporter in the beginning of my career, and someone whom I will never forget, created a dedicated earth for his hi-fi system. And that did indeed make a big difference to performance. So of course there is something to a clean earth, I am just not sure that these magic boxes are the best solution.
And of course, it's not just ground, but the capacitive coupling to the mains, and RF noise corrupting everything that's the bigger issue. And star grounding is not effective at all for RF noise - it's completely useless for RF and will make matters worse as you ground plane for RF, not star ground (star ground is where you have tracks or wires running to one point).
Note that good ground means a warmer sound quality - if it's brighter and artificially transparent, it's worse due to more RF noise.
But having said that, a very dear friend of mine by the name of Mike Priestland (who sadly died in the 90's), who was my first believer, mentor and supporter in the beginning of my career, and someone whom I will never forget, created a dedicated earth for his hi-fi system. And that did indeed make a big difference to performance. So of course there is something to a clean earth, I am just not sure that these magic boxes are the best solution.
And of course, it's not just ground, but the capacitive coupling to the mains, and RF noise corrupting everything that's the bigger issue. And star grounding is not effective at all for RF noise - it's completely useless for RF and will make matters worse as you ground plane for RF, not star ground (star ground is where you have tracks or wires running to one point).
Note that good ground means a warmer sound quality - if it's brighter and artificially transparent, it's worse due to more RF noise.
audionewbi
Headphoneus Supremus
- Joined
- Jan 24, 2010
- Posts
- 10,663
- Likes
- 6,126
Have you tried using different housing your devices to test out the RF rejection of various metals?
Jawed
1000+ Head-Fier
- Joined
- Dec 7, 2014
- Posts
- 1,143
- Likes
- 597
I’ve been doing experiments with upsampling and thought I’d see what happens with square waves, which lead to my question.
This is a 1KHz square wave band-limited for a 44.1KHz sample rate and then upsampled 16x using 100,000 taps. The input signal is unquantized 64-bit precision and so is the output with 64-bit floating point math used for all computations:
The reference signal and the upsampled signal overlap, so that only the upsampled signal in orange can be seen.
The green line is the error, in dB, and its values are referred against the right-hand axis.
What’s interesting is that the error varies massively, with the largest cycle-to-cycle variation occurring as the signal passes through 0. Sometimes the error is similar in size to the error of nearby samples (e.g. 0.0001dB) but I’ve seen an error that’s 135dB!
Since the sample rate is not an integer multiple of the square wave’s frequency, in general the crossing through 0 is not aligned with a sample. Also the crossing through 0 has, in this case, a 21KHz component.
Overall it seems that in upsampling the worst errors occur as the signal crosses through 0, which is why I originally asked about the value of this experiment.
Now it might be that looking at the variation in error at zero-crossings for a sine signal at 22049Hz is all that needs to be investigated, instead of for a square wave.
To make this "more transient" arguably you can start with a 1Hz square wave. Then you're looking at almost 500ms of a "steady" signal followed by a crossing through zero that lasts only 32 samples at 705.6KHz. To put a 22049Hz component in there, start with a 0.9586521739Hz square wave instead.
@Rob Watts, I was wondering if your mojo2 DSP suite will be implemented in pricier full or tt models?
Deary me!
You are being far far too polite.
Do these boxes come with real reptile-oil inside?
Last edited:
If I had included a grounding box or similar woo in any of my designs for my Electronics degree I would’ve been laughed out of the room then given a fail and made to sit on the naughty step
What is even worse are the Synergistic fuses. Besides the fairy dust are they authority approved? CSA, UL, CE etc.
Users who are viewing this thread
