Jump to content

Ye Olde Filter Stepping Issue - Revisited


CairnsFella

Recommended Posts

Hi all.

 

Long time since I last contributed here (though I do pop by for a read now and then.. where sadly, I see, it is much quieter nowadays).

 

I have mainly been absent because my musical efforts have been focused on my bass playing, but I continue to dabble with the G1 and other bits and pieces. But anyway, to the subject at hand. I do, of course, realise this has been done to death in the past, so I apologise if this has been covered before, but from my searches I cannot see my observations being explicitly confirmed.

 

Basically, I recently obtained a Behringer BCR2000 midi controller. I purchased this primarily to revive an old Yamaha sampler with the traditionally failed front panel encoders, though I have also been integrating it with my DAW and other midi instruments. Enter the G1. 

 

With the help of the user manual, midi implementation guide, Allenk's excellent (p1) companion, and other posts here and elsewhere I set about trying to understand the Midi machinations of the G1 - which it would be fair to say I am still trying to fully grasp. However, as part of the initial stages I was trying to get my head around the seemingly multiple ways of controlling the same parameters. Simply because it was the first parameter that presented itself, I was using filter cutoff as my guinea pig parameter.

 

It had been reported elsewhere that using a modulator to control the cutoff potentially eliminated the much maligned stepping, and I had read comment that an external controller 'may' help, though I do not recall reading that this was confirmed. Well...... I guess the first point is that I have only just realised that I did not try the controller mapped to the filter "control change" (CC) I am not at home at the moment so I cant try this for a few hours. I have my suspicions that this will produce the same results as the on board controller, but I suppose its pointless guessing when I can check for sure later.

 

It is also possible to control the cutoff, not only with sysex, but also with NRPN's. (The former I learnt a little about whilst setting up my sampler, the latter is new to me (beyond the acronym). Sorry, it has taken a while to get to this part, but it - for me - is the interesting bit. 

 

Using Sysex, the stepping usually noted with on board controller is massively reduced. I would throw a figure of around 90% less artefacts.  Of course, even a 90% reduction means that it still cannot be called smooth. I probably need to do some more tests, but I actually felt there was an almost 'new' artefact with unnecessarily HUGE filter movements, like almost one big step, albeit not quite instant. Anyway, this latter part is rather subjective at this point and should rechecked. The reduction in the original stepping though, is blatant and clear.

 

As for NRPN's. Well for some reason that I cannot get my head around - although my limited understanding of much beyond basic midi may be reason enough - produces another distinctly different result. A filter sweep with NRPN's is ..... very very very nearly ... artefact free. At least in my limited tests so far. Yes, things can still get messy with the G1's resonance screaming, but with reasonable settings, and even with pretty fast sweeps, there really isnt any stepping. I tried this with a number of solo synth settings. The Sysex and NRPN settings on the Behringer and the G1's own control are obviously all available at the same time, so could make immediate and completely equal comparisons. There really is no question. Three different control methods for the same parameter, three DIFFERENT results. Bizarre..no?

 

In one of the posts I have read in the past, it was suggested that a controller may have access to finer adjustments (i.e. a greater range than 0-127). If this were the case it would explain the CC/Sysex/NRPN variations. But no. They are all operating over the same range.

 

Even if this is an unreported finding, I guess I am preaching to the few nowadays, but if you are reading, and have a controller that you can assign NRPN's to. Please try it. 

 

If anyone is interested, I could record an audio clip to demonstrate the differences too I guess, but I'l hold off on that unless it is requested. However I will report back with the CC's controlled via the BCR later for completeness.

 

Till Then.

 

EDIT - SEE IMPORTANT NOTE IN POST 4

 

Link to comment
Share on other sites

Well, _I'm_ still listening. And still posting stuff.

 

This is a very interesting observation. I have not done side-by-side comparisons of this nature before. But I have noticed a massive slowdown in the operation of the step sequencer when a control track addresses an NRPN and smoothing is enabled. I wonder if that is related in some way? Your results would seem to suggest that Casio has already implemented some kind of smoothing for NRPNs so perhaps asking for smoothing on top of that is too much for the processor.

 

I suspect that the obvious stepping you hear when you rotate a knob is because the XW-P1 doesn't check the position of the knobs as often as it probably should. When I investigated memory limits for controller movements in phrases I discovered that I could record 50% longer before reaching the limit when moving a knob than when moving a wheel, indicating that the former was being sampled much less frequently than the latter. This also accounts for why I have always felt that adjusting the cutoff of the solo-synth filter is smoother with the modulation wheel (mapped through a virtual controller) than with a knob.

 

I suspect that when you get around to trying the CC#74 command it will behave just like sysex in terms of stepping although, of course, it will not behave the same wrt the filter cutofff frequency itself because it specifies an offset relative to the parameter's current value rather than an absolute value (or it least it is supposed to).      

Link to comment
Share on other sites

Hi Alen,

 

Appreciate you taking a little interest in this. I rather wish I had stumbled upon this a long time ago, given how the stepping filter issue was such a hot potato in it's day. Obviously using an external controller does not really nullify the original criticism, but I do feel it does show some greater potential (or perhaps less limitation of potential) than originally reported. And lets be honest, the XW was so cheap, that even if you did factor in the cost of an additional controller - that obviously be used for many many more things - you are still in the realms of budget synthesis.

 

Anyway, back to the testing.

 

So I have tried with the controller mapped to CC#74 (and for the sake of due diligence, mapped to an alternate CC, in turn mapped to cutoff via the virtual controllers.. though the results were the same).

 

Just to Recap then.

  • Onboard knob (Knob 1) - Stepping, as widely reported
  • External Sysex Control - Much reduced stepping in general, but singular large, slightly less than instant, step with fast movements.
  • External NRPN Control - To all intents and purposes, no stepping at all.

And

 

  • External CC Control - (Drum roll) .....................Well blow me if there is once again, NO stepping (looks like we both backed the wrong pony on that one). I am honestly flabbergasted that this hasn't been reported before (or if it has, that there has not been more reference to it in the past). Of course, as you intimated, externally controlled CC's do result in a relative change, not absolute, although I obviously took this into account, setting the cutoff centrally before testing the CC control.

So why the hell does the XW's own controller produce the worst results of all. I cant fathom it. (And I did also conduct the rather pointless test of reassigning the knob to an alternate CC to control cutoff, but it was just the same). Your theory seems quite plausible (I dont have one of my own) that the knobs position is not being registered literally (i.e. the values are being intermittently sampled). Really though, if this were the problem you'd have to think this could easily have been fixed...... wouldn't you. It's really madness, and especially so when , as you noted, the mod wheel also produces smother filter sweeps. (Edit. I have added another post below, just to add a little more to this point).

 

Actually on that note, I used your term "smoother" re: the mod wheel, but in actual fact, to be consistent with my other findings, I would say this was the same as my external CC and NRPN sweeps. In other words I am calling this smooth. This isn't to be picking up on your grammar, but more to understand what you mean. "Smoother" could imply 'more smooth' than before, but still with scope for improvement. Your experience is likely greater than mine, so if this is your view it probably would be a more appropriate one for others to take on board, and if this is the case then my NRPN and external CC results should be interpreted the same way.

 

I would like to try to 'justify' my own view though if that is ok. In many respects I feel the XW's filter is quite 'rough' in nature. Certainly more obviously digital than many in a lot of applications (though this may be as much or more a function of the sound sources as the filter itself). Equally, it is able to be driven more extremely than quite a few other instruments allow. With only 127 resolution (though accepting 'smoothing' can still come into play within this limitation) I dont really feel the results from the Mod Wheel, External CC or External NRPN exhibit any real "stepping" as would be reasonable discernible within a complete piece of music. At least for purposes then, 'smooth'. BUT, had this been the default state of the filter when originally reviewed, I would have still felt comments such as "the filter isn't the smoothest" would have been reasonably fair, though I honestly dont feel it would have been highlighted in this event. Without any doubt though, the comments in some forums that I have read that classified the filter as 'laughable' would not have existed.

 

Not much more I can add. I am a little reluctant to extend the testing as I too easily find 'technical interests' detract from any musical pursuits, and I am sure enough other anomalies will arise that pique my curiosity. Though given that I only tried a few sound sources, i cant help but wonder if a greater selection may have started to show up some glitches in my NRPN and External CC results. (To be fair, I did choose the default presets that appeared to perform the worst with the onboard controller). My offer remains if you do want me to record the results, though given your experience with the mod wheel, the only peculiarity you may not have experienced is the odd 'jump' with fast SYSEX sweeps. (Actually today I found that with v fast sysex sweeps, the note would cut completely. This may be an issue with my midi setup though, with a possible feedback loop).

 

On the other points you raise, I am once again quite surprised.. Especially the comment re: massive slowdown. I am somewhat curious how this may translate when hooked together with another Midi device (Something which I plan to do with my G1 and Roland MV at some point in the future). Still, I dont suppose there is much to gain by my trying to reproduce the issue now. Perhaps it's best I just thank you for the heads up on that one, so I can be aware of why this may be occurring in the event I encounter it.

 

Sometimes I wish they could just get a technician on a slow day, and get them to spend half a day reading the forum, and half giving us some frank facts that close off these points once and for all (good or bad).

 

Yes.. I still dream of the impossible

 

EDIT - SEE IMPORTANT NOTE IN POST 4

 

Link to comment
Share on other sites

OK. Just to add to the point about how the XW may be 'sampling' the midi data for controllers.

 

At this stage I should 'confess' to a variation between the tests that I hadnt really considered. On the XW, the knobs turn approximately 270 degrees for min to max values. On the default settings, my external controller turns about 450 degrees. In other words, at a like for like speed, the onboard knobs are changing values  - v v v roughly - twice as fast. However, when I did the original tests I partially took this into account, albeit a little subconsciously - because the controller displays the value being sent, and to that end I was always trying to go from min to max at the same rate. Nonetheless, given my greater awareness of this today, I redid the tests with the controller set up to a similar (if not slightly faster) resolution to the XW. The results were the same, even though I could see via midi-ox that perhaps a few more values were being skipped.

 

This lead me to a 'possibly final' test. Given that I can see the values on Midi-ox from my controller, and in Performance mode, I can see - what I assume - are the values being used by the XW when turning the onboard knob, I recorded the values returned by both when being turned as fast as possible (with the external controller on the revised fast setting).

 

So the XW :-

 

 TIMESTAMP IN PORT STATUS DATA1 DATA2 CHAN NOTE EVENT               
 001B6A28   6  --     B0    4A    03    1  ---  CC: Brightness        
 001B6A29   6  --     B1    4A    03    2  ---  CC: Brightness        
 001B6A2A   6  --     B2    4A    03    3  ---  CC: Brightness        
 001B6A47   6  --     B0    4A    31    1  ---  CC: Brightness        
 001B6A49   6  --     B1    4A    31    2  ---  CC: Brightness        
 001B6A4A   6  --     B2    4A    31    3  ---  CC: Brightness        
 001B6A67   6  --     B0    4A    56    1  ---  CC: Brightness        
 001B6A68   6  --     B1    4A    56    2  ---  CC: Brightness        
 001B6A69   6  --     B2    4A    56    3  ---  CC: Brightness        
 001B6A87   6  --     B0    4A    7F    1  ---  CC: Brightness        
 001B6A88   6  --     B1    4A    7F    2  ---  CC: Brightness        
 001B6A8A   6  --     B2    4A    7F    3  ---  CC: Brightness

 

And the External Controller :-

 

TIMESTAMP IN PORT STATUS DATA1 DATA2 CHAN NOTE EVENT               
 001F384B   7  --     B0    4A    01    1  ---  CC: Brightness        
 001F385A   7  --     B0    4A    03    1  ---  CC: Brightness        
 001F3866   7  --     B0    4A    04    1  ---  CC: Brightness        
 001F386C   7  --     B0    4A    06    1  ---  CC: Brightness        
 001F3875   7  --     B0    4A    08    1  ---  CC: Brightness        
 001F387A   7  --     B0    4A    0B    1  ---  CC: Brightness        
 001F387D   7  --     B0    4A    0D    1  ---  CC: Brightness        
 001F3880   7  --     B0    4A    0E    1  ---  CC: Brightness        
 001F3883   7  --     B0    4A    10    1  ---  CC: Brightness        
 001F3886   7  --     B0    4A    13    1  ---  CC: Brightness        
 001F388A   7  --     B0    4A    15    1  ---  CC: Brightness        
 001F3890   7  --     B0    4A    1A    1  ---  CC: Brightness        
 001F3893   7  --     B0    4A    1C    1  ---  CC: Brightness        
 001F3896   7  --     B0    4A    1D    1  ---  CC: Brightness        
 001F389C   7  --     B0    4A    1F    1  ---  CC: Brightness        
 001F389F   7  --     B0    4A    21    1  ---  CC: Brightness        
 001F38A3   7  --     B0    4A    22    1  ---  CC: Brightness        
 001F38A6   7  --     B0    4A    27    1  ---  CC: Brightness        
 001F38A9   7  --     B0    4A    2B    1  ---  CC: Brightness        
 001F38AC   7  --     B0    4A    2E    1  ---  CC: Brightness        
 001F38B0   7  --     B0    4A    31    1  ---  CC: Brightness        
 001F38B3   7  --     B0    4A    35    1  ---  CC: Brightness        
 001F38B7   7  --     B0    4A    38    1  ---  CC: Brightness        
 001F38BA   7  --     B0    4A    39    1  ---  CC: Brightness        
 001F38BE   7  --     B0    4A    3D    1  ---  CC: Brightness        
 001F38C0   7  --     B0    4A    40    1  ---  CC: Brightness        
 001F38C3   7  --     B0    4A    42    1  ---  CC: Brightness        
 001F38C6   7  --     B0    4A    45    1  ---  CC: Brightness        
 001F38C9   7  --     B0    4A    47    1  ---  CC: Brightness        
 001F38CD   7  --     B0    4A    48    1  ---  CC: Brightness        
 001F38CF   7  --     B0    4A    4C    1  ---  CC: Brightness        
 001F38D2   7  --     B0    4A    4D    1  ---  CC: Brightness        
 001F38D5   7  --     B0    4A    4F    1  ---  CC: Brightness        
 001F38D8   7  --     B0    4A    51    1  ---  CC: Brightness        
 001F38DC   7  --     B0    4A    54    1  ---  CC: Brightness        
 001F38DE   7  --     B0    4A    56    1  ---  CC: Brightness        
 001F38E2   7  --     B0    4A    57    1  ---  CC: Brightness        
 001F38E5   7  --     B0    4A    5B    1  ---  CC: Brightness        
 001F38E8   7  --     B0    4A    5C    1  ---  CC: Brightness        
 001F38EB   7  --     B0    4A    60    1  ---  CC: Brightness        
 001F38EE   7  --     B0    4A    61    1  ---  CC: Brightness        
 001F38F1   7  --     B0    4A    63    1  ---  CC: Brightness        
 001F38F4   7  --     B0    4A    65    1  ---  CC: Brightness        
 001F38F7   7  --     B0    4A    68    1  ---  CC: Brightness        
 001F38FA   7  --     B0    4A    6A    1  ---  CC: Brightness        
 001F38FD   7  --     B0    4A    6D    1  ---  CC: Brightness        
 001F3900   7  --     B0    4A    6E    1  ---  CC: Brightness        
 001F3904   7  --     B0    4A    73    1  ---  CC: Brightness        
 001F3908   7  --     B0    4A    77    1  ---  CC: Brightness        
 001F390B   7  --     B0    4A    7C    1  ---  CC: Brightness        
 001F390F   7  --     B0    4A    7D    1  ---  CC: Brightness        
 001F3912   7  --     B0    4A    7F    1  ---  CC: Brightness        
   
 

Quite a difference already (though still surprising how many values the external controller is skipping, let alone the XW). But hang on. The XW is showing the results for three channels.. so the real comparison is the above external controller results, and the following XW results edited for direct comparison :-

 

  TIMESTAMP IN PORT STATUS DATA1 DATA2 CHAN NOTE EVENT               
 001B6A28   6  --     B0    4A    03    1  ---  CC: Brightness        
 001B6A47   6  --     B0    4A    31    1  ---  CC: Brightness        
 001B6A67   6  --     B0    4A    56    1  ---  CC: Brightness        
 001B6A87   6  --     B0    4A    7F    1  ---  CC: Brightness

 

It "MAY" be possible to turn the external controller knob a little faster, but it has more resistance than the XW, so in practical terms it would be easier to reproduce the XW results with little effort, whereas I took the 'fastest' of multiple attempts on the controller. I even moved the knob I was using to an endmost right hand side knob so I could have the easiest access to it. But even with a massive allowance for variables it is clear that the onboard controller is issuing vastly fewer parameters (In actual fact, there are 'timestamps' included, if anyone were so inclined to process these). This, then could !....... should ? explain the problem.

 

But wait. I then 'listened' to the results. The test was pretty pointless sonically, as the end result in both cases was an instant open  / shut filter. Okay, there is some usefulness in seeing the 'potential' for skipped values, but finally finally I thought I should turn the knobs at the fastest speed that still demonstrates stepping, then try to approximate this speed with the external controller.

 

To some extent you are lumbered with 'taking my word for it here' though I realise that this isnt really good enough. The problem is reproducing a similar enough speed, covering a similar enough range. At this point I do not feel I have produced results similar enough (in terms of speed and range covered) to provide any documented support of my findings. However, using my ears, and knowing that in practical terms I was covering speeds and ranges on both that would have certainly overlapped on a number of occasions, it is may view that there is STILL more going on, than merely the midi values supplied. I mean it still "could" be that. Probably is the more I think about it, as at these speeds the external controller isnt skipping many values at all. But I do find it odd that at speeds upwards from where I am starting to skip values an on the external controller there remains no stepping, but speeds low enough to skip very few values on the onboard controller, obvious stepping remains.

 

IMPORTANT NOTE

One last point that perhaps I should probably go back and add to the prior posts for completeness is this. During these final tests I used some extreme resonance settings (112 upwards, where 127 is max), which did actually clearly step with the external controller when changing the values very very slowly. With faster movements the stepping was reduced or eradicated, whilst it remained and was more intense with the onboard controller. This does rather temper all of my existing findings, though at this stage my original views essentially remain intact because the actual sounds being produced were not, in my opinion, musical to start off with. In other words I believe that the vast majority of people would be using the filter within the ranges that would reflect my initial findings. 

Link to comment
Share on other sites

CairnsFella,

 

Make sure you're not confusing filter stepping with normal filter behaviour. At high resonance, all filters, including analog filters, will exhibit an audible spike at each harmonic as the cutoff passes over it. That's what resonance is supposed to do. Stepping, on the other hand is, is a subtle “zipper noise” in the background that will be be there irrespective of resonance.

 

This is a common mistake: https://www.gearslutz.com/board/electronic-music-instruments-electronic-music-production/982585-filter-stepping.html

 

When I assign knob 1 to filter cutoff on my G1, I can hear stepping from about 11 o'clock to 1 o'clock, even at zero resonance, when I sweep the cutoff at a moderate speed. It's not so evident with very slow or fast sweeps. Amplifying the midrange frequencies around 2.5 kHz with EQ helps highlight this. I don't hear zipper noise when sweeping the filter with the mod wheel, or an LFO, or a filter envelope.

 

The XW filter does exhibit stepping, but it's not as bad as many think and there are ways around it.

Link to comment
Share on other sites

Based on your most recent observations I rescind my theory about smoothing. I now believe that what you are calling "completely smooth" is the "full" 7-bit resolution that is supported by most CC and NRPN commands (and many SYSEX - certainly for filter cutoff). I can indeed still hear _some_ stepping when using the modulation wheel to affect cutoff even when moving it very slowly to avoid missing steps (PS. even with moderate resonance, not high resonance), which is why I called it "smoother" not "smooth." I now suspect that the issue with SYSEX being not as smooth as CC or NRPN is that SYSEX transmits MANY more bytes in each message than a CC message or even an NRPN message. The XW likely can't quite keep up at whatever rate you are sending from the BCR2000. (It is also possible to max out the embarrassingly low bandwidth of the MIDI connection when transmitting SYSEX but I doubt you are doing that when moving only a single encoder knob on the BCR2000.)  The XW synth's processor is no speed demon, as evidenced by how easy it is to unintentionally slow down the sequencer. 

Link to comment
Share on other sites

5 minutes ago, BrettM said:

Make sure you're not confusing filter stepping with normal filter behaviour. At high resonance, all filters, including analog filters, will exhibit an audible spike at each harmonic as the cutoff passes over it. That's what resonance is supposed to do. Stepping, on the other hand is, is a subtle “zipper noise” in the background that will be be there irrespective of resonance.

 

This is a common mistake: https://www.gearslutz.com/board/electronic-music-instruments-electronic-music-production/982585-filter-stepping.html

 

Excellent observation. My last reply originally included a note about this but the site ate it unceremoniously and I had to start my reply over. It got a lot shorter! :)

Link to comment
Share on other sites

5 minutes ago, BrettM said:

When I assign knob 1 to filter cutoff on my G1, I can hear stepping from about 11 o'clock to 1 o'clock, even at zero resonance, when I sweep the cutoff at a moderate speed. It's not so evident with very slow or fast sweeps. Amplifying the midrange frequencies around 2.5 kHz with EQ helps highlight this. I don't hear zipper noise when sweeping the filter with the mod wheel, or an LFO, or a filter envelope.

 

The XW filter does exhibit stepping, but it's not as bad as many think and there are ways around it.

 

Using a knob or a wheel to sweep the cutoff introduces issues about the sampling rate of the control. But even if you take these out of the equation there is still something odd about the filter. In other words, I hear an obvious artifact even when the cutoff is swept using an LFO or an envelope. I mentioned it and described my workaround in this thread.

Link to comment
Share on other sites

Not entirely sure where to start.

 

The beginning I guess.

 

9 hours ago, BrettM said:

Make sure you're not confusing filter stepping with normal filter behaviour. At high resonance, all filters, including analog filters, will exhibit an audible spike at each harmonic as the cutoff passes over it. That's what resonance is supposed to do. Stepping, on the other hand is, is a subtle “zipper noise” in the background that will be be there irrespective of resonance.

 

How to keep this short (edit.. I didnt), yet still cover the ground. Hmm.

 

Okay. Yes I know what stepping is BUT your point raises a valid consideration, and to that end I perhaps should have stated "artifacts including stepping" throughout. Whilst the focus has been stepping, the "important note" probably did encompass a combination of these issues. That is say that the harmonic spikes are without doubt the noticeable element in play here, however the nature of the filter sweep means that these harmonics - or more specifically, the frequencies in which these harmonics reside - may well be coming in and out more harshly than they would with a finer resolution sweep. I am not about to get an oscilloscope on to this so I guess we will have to maintain our own opinions as to the reality of the situation here. But if we were to focus on stepping such that I exclude anomalies that may be mostly / entirely harmonic spikes, this would merely solidify my findings that external controls do not exhibit stepping. Whilst pertinent to the testing though, my primary point is that, regardless of the use of the term stepping or not, filter sweeps are exponentially better using an external controller via CC's or NRPN's.

 

9 hours ago, BrettM said:

When I assign knob 1 to filter cutoff on my G1, I can hear stepping from about 11 o'clock to 1 o'clock, even at zero resonance, when I sweep the cutoff at a moderate speed. It's not so evident with very slow or fast sweeps. Amplifying the midrange frequencies around 2.5 kHz with EQ helps highlight this. I don't hear zipper noise when sweeping the filter with the mod wheel, or an LFO, or a filter envelope.

 

I hope I have not appeared to suggest otherwise, but I have read much of what has gone before on this subject, including the use of modulation sources. You are obviously not alone though, re stepping via 'knob 1' and indeed it is this widespread acknowledgement of that issue that prompted my investigation with the use of my external controller. The point being that, should you have a controller available, it works much much (dare I add another much) better than using knob 1. The modulation sources were not in the scope of my post as they cannot truly replace this control, though they are useful to demonstrate what the filter (cutoff) may be capable of.

 

10 hours ago, BrettM said:

The XW filter does exhibit stepping, but it's not as bad as many think and there are ways around it.

 

This really had me scratching my head. I am going to have to disagree with the middle part of your statement, though it is sufficiently subjective it is perhaps not something that someone could legitimately disagree with. It really depends upon what YOU think that the MANY think. Certainly in my own view it is definitely as bad as "I" think.

 

As for ways around it.?? My initial thought here was to suggest... "hey, maybe I'll do some testing to find out if there is".... then to re-post my first three posts. Clearly I am not going to do that, but it doesn't sound as though you are convinced (and I'm not going to try to change your mind) of the significant difference when external CC or NRPN's are used. If one has access to a suitable controller, I cant think of a better 'way around it' than this.

 

10 hours ago, AlenK said:

(It is also possible to max out the embarrassingly low bandwidth of the MIDI connection when transmitting SYSEX but I doubt you are doing that when moving only a single encoder knob on the BCR2000

10 hours ago, AlenK said:

there is still something odd about the filter.

 

I am unsure you need to backtrack on any of your theories from my findings as I am equally unsure they 'prove' what is going on inside the XW (Edit. Perhaps I have misinterpreted this. I am now thinking that you initially believed that as a result of my testing the XW may be providing additional smoothing when being addressed by the BCR to - in effect - give a result emulating a finer resolution than 0-127. Whereas you now feel I am experiencing a conventional 0-127, in which case I understand you decision to rescind) though I do think the tests demonstrate quite clearly what you can expect out of it when using the different approaches I used.

 

As for still hearing some stepping (and I am going to exclude the extreme resonance examples in light of Brett's comments), it would be difficult for me to suggest the sweep becomes as smooth as butter / silk / (insert your preferred smooth thing) though I remain of the opinion that it becomes smooth 'enough' (subjectivity again). By this I mean if the filter performed this way by default, and I did not read any comments to the contrary, I may not notice any particular deficiency. None of this, though, contests your assertion that the filter has peculiarities. I actually did a blind test with my wife (who is not in any way musical or technical) yet she could identify which of the three sweeps I was using with ease. As she should really, as the differences are indeed obvious.

 

You may well be right about the sysex. If I think about it, when I am turning the knob (slowly or rapidly) I am likely decelerating ever so slightly at the extremities, hence the slight artifact at those extremes. Just a thought, not a theory. Given I am able to monitor what is being sent by the BCR though, I would be less convinced that the bandwidth would be an issue.
 

10 hours ago, AlenK said:

My last reply originally included a note about this but the site ate it unceremoniously and I had to start my reply over. It got a lot shorter!

 

Sorry for you. But also sorry we lost some of your excellent insight. I am often paranoid about this and often I will copy the contents of a post as I go along (and especially before submission) in case it g

 

10 hours ago, AlenK said:

I mentioned it and described my workaround in this thread.

 

I had seen this in the past, but I had not worked through it. Whilst the primary issues I raise in this thread concern the improvement found when using an external controller, and not what additional things can be done to further the improvement, I will still try this out at a later date. Edit (Sorry, that may sound like I did not think it relevant. Not at all. And of course it was a direct response to Brett's comments.)

Link to comment
Share on other sites

I guess this is about all that is left for me to add here (unless anyone suggests anything else).

 

In this example, I specifically chose a tone that 1. wasnt too fussy, and 2. was one of the ones that worked reasonably well (exhibited less problems) by default even with the onboard knob. Whilst choosing this does not emphasise the problem with the onboard knob, I believe it does show that even where the G1's own controller works reasonably well**, there remains clear improvements - other than the exception I highlight - with the external controllers.

 

In this .wav file I use knob 1, External CC's, External Sysex, and External NRPN's in turn. For each I performed a couple of slow sweeps, followed by faster sweeps.

 

Knob 1.

As I stated above, this is certainly not the worst case of stepping that I could establish, but steps remain evident in both the fast and slow sweeps. (Note:- the first BIG glitch is merely the filter jumping to the knob's position which was not in sync with the parameter setting having previously used an external control).

 

External CC's

In my opinion there are no 'obviously evident' steps in the fast or slow sweeps. Perhaps a detailed analysis of the waveform 'may' reveal something, however to all intents and purposes I feel this is artefact free. The consideration here (slightly irrelevant in this context) is that it is a relative control.

 

External Sysex

Similar (if not identical) to the external CC's in the slow sweep, but the speed of the fast sweeps here really show up that 'catch up' artefact (for want of a better term). What I didnt show it here, as obvious as the problem is, it disappears 'fairly' quickly i.e. at roughly half the speed of those fast sweeps. so not unusable, but then again why use this if there are better alternatives.

 

External NRPN's 

I cant tell any difference between this and the External CC's example (in terms of artefact's). Again, not pertinent but this is an absolute setting, which differentiates it from the CC's.

 

So there you have it.

 

IF, at the end of the day, one feels that the issues on tones such as these - where it is less prominent - are too small to be of concern, then that's great. Especially if all the tones you use are impacted as little. I'm certainly not trying to convince anyone to worry about something that they otherwise would not have been bothered by.

 

At the other end of the spectrum. If one feels that there are still artefacts within NRPN and CC externally controlled sweeps, once again, I am not trying to convince anyone that they shouldn't feel this way. I am merely reporting as I find.

 

Happily for me, I sit in the middle. In other words, I DO have a problem with the artefacts, even in less apparent tones such as these, but I am personally convinced that the external CC and NRPN controls all but eliminate the problem for the purposes of practical use.

 

** Note: - Whilst I intimate that the XW Controller (knob 1) performs differently on different tones, this was merely language of convenience. I do not believe that the knob performs any differently on any of the tones, nor do I believe that any tones are effected any less or more by the problem. What I am suggesting is that the issue is more audibly evident on some tones than others.. 

 

Filter example.wav

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.