SC5000 High End Roll-off


#65

Yep, like I stated 4 months ago, I hope it’s firmware repairable and not hardware related.


#66

Since it occurs both on the analog and digital outs, I would suspect it is something in the signal processing algorithms and not the hardware.


#67

Yeah but if it’s related to a processing chip and one that’s not programmable… I hope it’s an algorithm too.


#68

I’m pretty sure that @Gee_DenonDJ will say that it’s not possible to fix that, only in terms of hardware issue, it makes me sad that the sound is not as perfect as CDJ 2000 nexus 2 :frowning:


#69

The SC5000 is a computer with a CPU, memory, buses, ports, interface, display, sound card outputs, and the software OS. All these anomalies and quirks with the digital to digital sound are entirely digital signal processing code.


#70

I guess it all depends on the reasons why the filter was put in the signal chain in the first place. Only Denon can tell us about that.

The technical specification of the SC5000 gives us a frequency range of 22Hz to 22kHz, although the DAs could play up to 48kHz. I could live with that, if the frequency range was flat - which unfortunately it isn’t. The SC5000 doesn’t manage the 22kHz by any standard (-3dB, -6dB, -10dB).


#71

Agree don’t understand the point to make this, it’s strange because i already saw this (loss of frequency range) with Traktor unfortunately the picture are not available ! https://www.native-instruments.com/forum/threads/traktor-2-11-output-problem-vs-ableton-itunes-winamp-more.313933/


#72

I suspect Elastique v3 is actually within the code stream at all times, and if so it’s neither neutral tonally nor full bandwidth. The easiest short-term solution, assuming this is actually the reason, is when keylock is Off completely cutting Elastique out of the code stream pathway in the OS. With keylock Off, you’d just use simple sample rate conversion and a very gentle high-pass filter with negative pitch and a very gentle low-pass filter with positive pitch. At zero pitch with keylock Off, you’d have no sample rate conversion and no filtering, resulting in bit perfect digital out which can be easily validated.

Later on I’d recommend, assuming my original supposition is correct, them working with zplane to improve the Elastique tonality and bandwidth first, and then down the line perhaps resolving some of its other anomalies like going completely haywire.

It’s certainly true that its keylock is extremely good compared to other players when you are lower than -6% pitch. In the positive pitch this difference is less noticeable like zooming out on a very large digital image. Near or at zero, the competition is way ahead of the SC5000 currently.


#73

Actually there is even one more thing to all of that. The filter is most certainly not phase compensated. So the phases of the high frequencies will be shifted and further affect the sound of the players.


#74

The filtering is very aggressive, so it’s probably like an 8th-order filter or multiple lower-order filters, and yes, they’d have a ton of group delay issues. The only time you can get away with that extreme a filter without the phasing getting wonky is for infrasonic types to protect subs.

However, we’re talking about probably some amazing harmonics synthesis stuff going on with interpolation upward in frequency and probably content harmonics shifted downward in frequency. So some of what Elastique is doing is probably a much more aggressive version of sub-bass synthesis, tube-like harmonic generation, and interpolative upsampling… among other things. i understand that any of these, let alone all of these, are going to produce aliasing, noise, and other grunge and they’ll have to find a way to probably noise shape to move it to a less audible band and then roll it off with filtering. So I don’t want to imply I know how to do extreme negative key correction without any of this or better than them, but when keylock is off at least, the processing should be minimal, at least as a first step to improving this situation.


#75

UP guys agree with Reticuli , it must be corrected in a future firmware please @paul_denondj could we have information if it’s possible or not ?


#76

Not to beat a dead horse here, but I spent this Friday tutoring a new DJ prior to her first big show opening for a major headliner and setting up the sound system in her studio. She just recently bought a NXS2 system and already had some nice monitors. After working on the speaker arrangement, room acoustics, and ergonomics situation, I then set up the DJ front-end optimally, and this was my first time with the whole NXS2 in a controlled, studio situation outside of a venue. The NXS2 system definitely sounded dynamic, extended, smooth, and refined except when key-locked while in substantial negative pitch.

This was in subtle but noticeable contrast to the slightly-compressed, dark, rolled-off top, slightly-mushy, sort of weirdly-textured & processed-sounding quality that I’ve been experiencing on the SC5000. Does the new Denon sound outright bad and unusable? No, and negative pitch with key lock is certainly amazing and unprecedented. But it reminds me of some aspects of the lower-fi, crude DJ needles not quite set up even as well as possible that a lot of people deal with, and substantially apart from hi-fi needles. And yes, a little like Torq 2.0.

I also want to reiterate that while the Rane MP2015’s SPDIF and line inputs certainly need the ability to change their levels in relation to the trim/gain knobs like we are allowed for the phonos, the SC5000 very obviously needs to be 6dB quieter internally. I would recommend not even doing the whole optional pad thing. Just permanently cut the internal work signal DSP levels by half. Don’t worry about those past analog output noise issues with the capacitive touch display. A little boost in S/N ratio is not worth this side effect. No point in a limiter when the 1/2 sample amplitude fix is available and the SC5000 SPDIF output is this hot.

Step two on the next firmware regarding this, IMO, needs to be a simple option in the settings so that when key lock is off, Elastique is off and the processing is maximally minimal. If this means with that setting you get a slight audible tick or millisecond blanking when you switch key lock (and thereby Elastique) off, that’s fine. It’s an optional setting.

After that, a long-term goal on other future firmwares down the road: the engineers can start looking at the high frequency extension and aggressiveness of the processing’s effects on the audio’s texturing and precision. it may be possible for the low-pass filter to be gradually progressive overall as you get away from zero and/or also for possibly a distinctly less aggressive processing used when keylocked in the positive pitch, a state that sounds pretty darn acceptable on the NXS2 key lock even with their modest algorithm and hardware.

In the meantime, I am enjoying key locking while pitching deep into the negative on the SC5000s. Impressive stuff. Their deep bass is also great. And further trying to look at the sound right now in a glass half full kind of way, it doesn’t sound that out of place alongside my DJ vinyl.


#77

it’s not beating a dead horse in any way. Things like this help push towards future firmware updates or even hardware enhancements for the next run. Collectively as a whole there are so many more pros when it comes to the Prime Gear but some of the basics have yet to be addressed. But I’m sure they are listening. It’s become clear with the firmware updates we’ve seen so far. Just hoping something like this actually can be addressed via firmware. The gear feels way ahead of it’s time often.


#78

And that’s the bottom line.

Why use 2000 words when “it sounds fine” is accurate and what 99.9% of any dance floor partier is gonna say, if asked.

IS this tiny roll off of high frequency sounds, if the sound levels are above digital zero related to the Elastique pitch algorithm? Some people have made wild allegations that it is, but I haven’t read any confiming posts from Denon about that.

IF it is something related to Elastique then should Denon Put out some firmware out that switches off/out Elastique IF the pitch slider is set to zero? For God sake NO! Can you imagine your sound sounding fine, as it does already at minus 2 pitched and then suddenly getting extra crisp as you drag the pitch through zero, then losing that crispness and going back to fine as you continue up above zero pitch (this is a move I use a lot when shifting from one genre to another.

I for one, don’t want one sort of sound at zero, and a different sort of sound at plus or minus. It doesn’t need fixing as it’s not actually broken.

Some enhancement might be nice though or if it is the Elastique algorithm that is slightly pulling in the top end frequencies over digital zero db theb maybr an option in the preferences for an older version of Elastique which doesn’t offer the amazing pitching clarity, but doesn’t have the digital zero high end damping.

Let each of then choose which setting we need. That option might be something we set once and leave or something we change ten times per set as all DJs are different.


#79

It appears to be a high-order low pass filter with the corner frequency (-3dB down point) at about 14khz. It is already at half amplitude at 16khz. So that basically makes this 96khz sampling rate playback player, what, having an effective frequency range equivalent to around a 30khz sampling rate or less?

If compression/limiting is going on with already fairly dynamically-compressed popular music masters that would actually probably benefit from an expander, then you could potentially be looking at less actual effective dynamic range than even the original files most people are playing had to begin with.

Elastique is the only processing that is known to be occurring on the SC5000. Its implementation seems the most likely culprit, other than the minimum necessary processing required to change the track speed. You have a good point, though, that we do not know. Deckadance’s Elastique does not suffer from the dark, processed quality Torq 2.0 has, even though neither has the low pass filter. I don’t have anything else other than the SC5000 with version 3, though.

It’s possible this default sound of the unit could be instead caused by some data compression to reduce the throughput and memory requirements of handling the audio upon loading the track. That would mean your lossless audio files would not be handled even remotely in a lossless manner by the players, and even a lossy file would be re-compressed. That is consistent with the uniform reduction of frequency & dynamic ranges.

I suppose a third option could be there’s dynamics compression in there to make everything sound louder, normalization going on, along with an aggressive low pass filter to reduce compression artifacts when pitch is changed with keylock on. One might be motivated to implement such measures if there had been issues with noise on the analog outputs and you were trying to boost the music signal in the digital domain up higher past that occasional noise.

So to back this up in reverse… This last noise mitigation side effect thing would actually be the easiest to get rid of if that’s all it is. Data compression would unfortunately be the hardest to drop, as that would basically be a necessity of the current operating requirements. Elastique being the issue would be resolvable easily to some degree with the aforementioned optional defeat, either completely or just with key lock off, and then further improvements later when it’s on. Wow, I sure hope it’s not data compression.


#80

I would really like to get a statement from Denon on this one as well. I’ve noticed this in my gigs as well with my full prime set up, but I didn’t invest any time into why this would happen and just saw this post as I was looking for other users reporting this, and voila I found it. So @paul_denondj can we get an answer on this one? Is this a hardware limitation or can the frequency roll-off be removed software wise (however this would be implemented).

I think we all want to have a professional clear sound as most of us prepare our tracks tagged correctly, without clipping peaks and most of the time well EQ’d. I don’t want the system to somehow “minimise” the quality of my tracks.


#82

Agree with all of you !for christ’s sake @paul_denondj could it be possible to finally have an answer about this frequency roll off ? it’s been months now… we just want : “no problem guys, our engineers work to make the best sound possible it will be corrected before the end of the year”


#84

Elastique Efficient, which I would assume would be used because it’s the one most suited for realtime keylock, has a function called CElastiqueV3If::SetCutOffFreq. For a player having up to 24/96 source file compatibility, an Fs/3 value used for this to reduce processing requirements I believe might yield an effective sampling frequency of 32khz. This might explain what’s happening at least with the low pass filtering. If that’s the case, Fs/2 would give us full audible frequency. Fs/4 only gives 20-25% reduction of processing, anyway. This still doesn’t address the loud SPDIF output, sense of compression and ever-present processed quality, or the fact that frequency response is diminished even with keylock off.

So my best hypothesis at this point is that CElastiqueV3If::SetCutOffFreq is at Fs/3, there is dynamics compression and limiting occurring, then perhaps normalization, and Elastique doesn’t completely bypass when keylock is off.

If this is the case, that function needs to be changed to Fs/2, which changes the data processing savings from up to 25% to instead as low as 13% savings, you’d need a global -6dB pad to halve the digital signal amplitudes and remove the necessity of a compressor/limiter, stop normalizing if it’s happening, and make Keylock Off fully bypass Elastique’s code, at least as a preferences/settings option if there is a slight glitch when you hit the button.


#85

I’ve meant to share these analyzer read outs for quite some time. First one is white noise at -60dB over the full 96kHz sampling rate range (20Hz-48kHz). At this gain level any kind of dynamic limiting is out of the question.

Second one is the same signal from the SC5000 through the digital out. We can clearly see the sharp dip in frequency response way before 20kHz. This is whithout any pitch adjustment or harmonic processing.

The third picture is a zoom in of the frequency range where the filtering takes effect. The notches are typical for a filter response.

The SC5000 frequency response is nominally 22Hz-22kHz. In reality it is clearly not. As said before it is more like 22Hz-14kHz (-3dB) or 22Hz-18kHz (-10dB).


#86

I don’t understand why Denon as a company don’t proactively respond to issues like this. Same with the MCX8000 screen freezing. It’s very poor indeed. If there is a problem it should be identified and fixed. Communication is absolutely awful.