Old Motifator threads are available in the Archive.
langbehn
Total Posts: 11
Joined 11-17-2013 status: Regular |
Yep, It’s definitely mostly if not all the stepped physical indents on the filter knob that causes the stepping sound. (And I think that’s actually kind of funny, given the circles we’ve all run around in looking for other explanations). Assigning filter sweep to the mod wheel, I hear VERY little if any discrete stepping. As I wrote in an early post in this thread, I personally kind of like the stepped filter sweep for a lot of things (in some ways, it gives a sort of sample and hold or wavetable sweep-like effect) , but it ain’t like a classic analog synth. It’s great to realize that I can have it both ways, simply by assigning filter sweep to the mod wheel of continuous foot controller if that’s what I want. |
Bad_Mister
Total Posts: 36620
Joined 07-30-2002 status: Moderator |
Let’s throw a few facts in the mix here (just for giggles): forgive the length, didn’t realize how short on facts the thread was getting :-) hope it helps… Two posts… The CUTOFF KNOB and RESONANCE KNOB apply offsets to the actual filter settings, they themselves are not a filter. Repeat the knob marked “Cutoff” only applies positive or negative offset values to the (potential) eight filters in the selected Voice. If the Elements of the Voice have no filter assigned (THRU) the CUTOFF knob will do nothing at all. It is not, like some of you are obviously thinking, a filter itself. It only offsets values of the actual filters (plural) yes there are eight of them!!! 8 filters… (Bet you never owned an analog synth with eight filters!) Now a little filter application history so you can better understand what a filter is used for and what it should not be used for, just FYI. The role of the filter is to control the timbre, or tone of the sound as it proceeds forward in time. It is to mimic the harmonic content and the change of harmonic content over time. If you are a fan of analog synthesizers, first lets say this, mostly you are remembering very simple architectures with the filters. Take a classic analog synth, mostly it had an oscillator block that sent signals to the filter block which sent signal to the amplifier block which sent signal to the output. Whether you had a MiniMoog, Arp Odyssey, Prophet V, or Oberheim OB8xa, basically all your oscillators went through the filter block before going to the amplifier block to the outputs. You did not have eight complete synthesizer pathways… Each Element in your current Motif/MOX engine has an oscillator, has its own filter complete with its own filter envelope generator, its own Cutoff and Resonance, its own LFO, and to boot it is velocity sensitive. Its own amplitude generator… Each Element is a complete synthesizer pathway. That’s right ... Each is a complete synth engine. I never mind discussing synthesizers but I do have a prerequisite that you know what your talking about prior to engaging in the conversation.. Velocity Sensitive, let’s hover here for a brief moment. I tried to sit and think of all the velocity sensitive analog synthesizers that I owned and I quickly realized, yikes, I didn’t own any (I wasn’t a rich young musician) and let’s face it, there were simply not that many.
My memory is not perfect but I could think of three before my old-timers kicked in and I had to start scratching my head… Please help if your old enough to remember real old analog synthesizers that were velocity sensitive…
Then I start to scratch my head… They simply were not that plentiful and each was very expen$ive at the time. All the popular analog synths had basically one, two, four, five, six or eight oscillators… The double engines had ten and twelve oscillators… But all had just the one filter through which you routed your signal. On a few you could actually change it from the LPF to something other than than, but it was still just 1 filter per synthesizer. What I’m getting at is mostly you remember the role of the analog synth’s filter as a singular device that shaped the contents of the entire sound.and mostly did not concern itself with velocity (sadly). Fast forward to today’s modern synth engines… In particular the sample playback synth engine. Here, rather than geometric shapes as waveform sources (sawtooth, sine, pulse, square, etc.) the waveform sources are recording of actual instruments. The basic harmonic content is mimicked in the source wave, the filter can concentrate on controlling what happens to the sound as it proceeds through time, how the timbre changes as the note sustains. The detail now and the amount of control is extensive, on a per oscillator basis. If one particular filter is set to control a specific component within a Voice, and how it changes harmonic content over time, and you decide you want to sweep the filter like some analog synth jockey, then it is you who are misusing the filter… That is not how the filter was necessarily setup in this type of Voice. On just “any Voice” you can’t play the knob twisting synth jockey… Not on just “any voice”, that was my point!!! Say you have a Voice that is an acoustic piano. Well let’s start by saying analog synths doing acoustic piano was virtually an oxymoron ( they didn’t do acoustic pianos, really), on the Motif/MOX it is likely to have multiple filters each designed to control the harmonic content of a particular layer or component of the piano sound. When you turn the crank up the resonance you are cranking it differently on eight different filters, which might be set to different filter types entirely, and then you call yourself sweeping the frequencies with the Cutoff knob, you are actually sweeping the filters of eight different filters. Something you didn’t do, something you could not do on most of the ancient analog synthesizers. You hear zipper noise… Clear sign YOU don’t know what you are controlling, period. (Yeah it’s like its “jumping values”, ‘No sugar Sherlock’ is what we’d say in the Bronx.) |
Bad_Mister
Total Posts: 36620
Joined 07-30-2002 status: Moderator |
If you think you could control 8 filters on an analog, it was in your dreams. Want to set up a voice that plays like an old analog synth use less oscillators, and if you do use multiple Elements at least set the FILTER TYPES identical in each, this way when you move the CUTOFF and RESONANCE knobs they behave in a similar fashion to what you are attempting to emulate. I tired to think of an analogy to fit this, but none is ideal… If you think about the piano lid as a type of acoustic world FILTER, attempting to sweep a filter of a Voice where there are multiple different filters types ganged together would be as silly playing a Rachmaninoff piano concerto with one hand while wildly lifting and lowering the lid like some crazed wah-wah pedal… Just silliness. I asked the question and clearly the poster missed entirely my point… You do realize that the CUTOFF and RESONANCE knobs are offsets, they add and/or subtract vales to each of the eight filters banks. They are not themselves the filter. With this knowledge look at the particular Voice (your “pick any voice” statement shows that you don’t really understand what I’m trying to tell you, which is okay, but doesn’t change anything but clearly point out what you are not seeing) Compare the Motif/MOX engine with any other sample playback engine with eight oscillators per Voice… I’ll wait here… “Any voice” fallacy: what if you select a Voice where one filter already has significant resonance and the one next to it does not. What if the next Element has a HPF not a LPF moving the cutoff knob has the opposite effect on the HPF and LPF… This is why I asked if you know what you are changing when you twist the knob… If you do so and it makes a horrendous noise, so would wildly opening and slamming the piano lid during the Rachmaninoff piece… Clearly it is a case of DONT BE SILLY, DONT DO THAT! Move the filter wildly on a voice that is specifically programmed for it. Clearly if you see the Motif/MOX as an analog synth then make sure you understand the role of the filter first. Set it up to succeed. If all you remember is analog synthesizers then limit yourself to programming the filters in a similar fashion to how they actually work (you had only one filter, remember? Do not take a gang of different filters programmed to respond and accurately mimic the filter response of velocity and use it sweep like an analog… It may not work, surprise, surprise, surprise. This is like taking something designed to do one thing and doing something completely unintended… Like taking the boat’s anchor and deciding its a surf board, ain’t gonna happen. The development of the filter as a viable entity came to fruition on the sample playback engine, where its sensitivity to velocity is used primarily to mimic the acoustic reality where musical instruments get richer in harmonic content the harder or more forceful the energy used to cause the vibration. Your analog synth not so much unless you had CS80, Chroma, or T8… And even there you did not have multiple filters… Ganged together… The signal all went through a single filter. [Final note: somebody remind me this would make a good topic for a User Group discussion: programming like ancient synths] |
pkadidlo
Total Posts: 16
Joined 02-02-2006 status: Regular |
Phil, I enjoy you and your writing style! I’m 50 years young so I guess that makes me old enough to remember my first analog synth that had both a LPF and HPF (MS20). Probably why I’m a manual filter knob tweaked in the first place! I will defend my earlier comments by pointing out (again!) that the Motifs seem to have no issues adjusting “the offsets” of as many as 8 “filters” from the single knob assigned to “filter” on their front panel. So while the MOXF and Motif XF have the “same engine” they do not operate equally. But that’s okay. I’ll stick with the wonderful Motif and move on. |
pkadidlo
Total Posts: 16
Joined 02-02-2006 status: Regular |
5pinDIN, yes, I am sure they use rotary encoders on the MOX/F. And it shouldn’t matter whether or not the encoder has detents or not. A little history on rotary encoders: with the advent of programmable synths that still had pots/knobs there was often the issue of the position of the knob not reflecting the “stored” or current value of the parameter assigned to that knob (e.g. the filter cutoff). Many early programmable synth would “jump” to the internal value/position as soon as you turned the knob. Others operated so that the value of the parameter wouldn’t change until you had moved the pot knob to match the position that would correspond to that internal value, thus avoiding a sudden jump to a completely different value. This was a good initial workaround other than having to guess what that current internal value was so you knew which way to turn the knob to match it up. Enter the endless rotary encoder. Instead of using a pot with a finite range of min to max, the encoder simply sends out positive or negative values when you turn it counter-/or clockwise. The detented rotary encoder is meant to make incrementing up or down by a single digit or value easier than if the encoder has no physical feedback to help you turn it just the right amount. But the biggest advantage is that typically you are simply telling the software “I’m sending you a +/- value to add/subtract from the current internal value” (offsets, if you will). The new challenge became how to move through, say, 128 steps of range with out having to turn an encoder with maybe 16 detents 8 complete revolutions to go from say 0-127. With the old pot you’d get the entire sweep of the range between 7 o’clock and 5 o’clock. Much easier to make smooth broad changes. So some manufacturers programmed the rotary encoders to respond to the speed with which you turned the knob. If you turned it a click at a time it went by single digits. If you turned it fast it might jump more than the equivalent of 1 digit per detent, which is helpful when you have 128 or more values to cover. This could apply to some of the MOX’s parameters that have the detents. The only problem is that if the algorithm simply decides to jump to inc/decrementing higher values when turning the knob faster, it’s skips all those baby steps in between and makes huge changes in the parameter setting. You can also watch the displayed values and understand how hard the CPU is working to track all these sudden changes you throwing at it by twisting the knob. Perhaps some digital synths do a better job of interpreting the intent of the user making quick adjustments and interpolating the values that would be present if you turned the encoder slowly. It would be nice if they would allow you to experience the same smoothness of change that you get from the LFO or ENV when making offset adjustments from the front panel as well. Peace. |
anotherscott
Total Posts: 653
Joined 06-30-2010 status: Guru |
Bad_Mister, here’s a popular exception to your “all the old synths put everything through a single filter” scenario… The original Oberheims, i.e. 4-Voice, 8-Voice, where each voice consisted of its own pair of oscillators (in its own SEM module) and its own filter! But yes, your described scenario was the more common, probably outside the more esoteric modular stuff which you could configure six ways from Sunday. But back to the topic at hand… You make a very valuable point about how the knobs are offsets (perhaps affecting multiple filters simultaneously in different ways), rather than actual filter controls. But it does not address the phenomenon described that, whatever changes they are making appear to be done more smoothly on a Motif XF than on a MOX/F. And from the sound of it, it has nothing to do with underlying differences in electronics, but to the extent this phenomenon exists, it is simply a by-product of the fact that the Motif knobs are smooth-turning, whereas the MOX/F knobs have detents or notches which perhaps physically interfere with your ability to operate them as smoothly. Maybe the engineers felt there was some other advantage to this kind of control, or maybe those controls are simply cheaper and that was part of engineering the XF technology down to a lower price point. This actually reminds me of my first modern Yamaha board, the S30. I hated the notched feeling to the knobs because it seemed completely arbitrary in operation. That is, if you were altering a parameter, there was no one-to-one correspondence between the notch of the next detent and a single value increment in the desired parameter. It was very possible to turn the knob enough to feel a click and have it not be enough to change the value you were editing. To me, that completely defeated the theoretical advantage of the detents in the first place. If there’s not going to be a direct correspondence of a single notch click increasing/decreasing the associated value by one increment, I would much rather have a smooth operation knob.I don’t see the value of the notches otherwise, I actually find it an obstacle, creating a cognitive dissonance between the physical sensation and the actual operation being performed. It makes it feel like you’re operating a mechanism that is out of adjustment (and reveals the implied “precision” of the detent to be an illusion). The detents just felt arbitrary and disconcerting. |
anotherscott
Total Posts: 653
Joined 06-30-2010 status: Guru |
Good point, pkadidlo, about endless-rotary knobs functioning differently depending on the speed of operation. Tying that in with my previous post, I would suggest that endless rotaries should ideally not be notched. Though maybe it’s acceptable if the notches do at least correspond to an exact unit increment when they are operated slowly. |
MeMyselfAndI64
Total Posts: 201
Joined 11-12-2013 status: Enthusiast |
Funny thing!
And I saw this....on Page 143 in the Reference manual.
/ Knob Curve
The default is set to table 3 that means we can get smoother response on the knobs if we need it:) |
5pinDIN
Total Posts: 11891
Joined 09-16-2010 status: Legend |
Thanks for confirming that the MOX and MOXF use rotary encoders - that explains a lot. Since there are detents, the encoders apparently are incremental type having two outputs that are out of phase with each other. With that type of encoder, the pulses from the two outputs lead or lag each other (depending on direction of rotation), which is then interpreted by the software. While the detents do provide tactile feedback, they also help ensure that the encoder will be left in a rotational position that provides a stable output. Reliably reading the encoders can be tricky. Besides having to deal with variable speed of rotation, mechanical switch type encoders tend to have contact bounce, especially as they age. Yamaha has used that type of rotary encoder before - the A-series samplers that I own each have five of them. The algorithm used on those samplers didn’t deal well with bounce, often causing misreading of direction and value. For my own samplers, I added a series RC across each encoder output to suppress the unwanted pulses. If the bounce issue is dealt with exclusively in software, rapidly turning the encoder can lead to missed values (depending on the algorithm used), since the rapid pulses can be misinterpreted as contact bounce. |
anotherscott
Total Posts: 653
Joined 06-30-2010 status: Guru |
Thanks for that interesting tidbit, providing perhaps a technical rationale for using them even in situations where they might seem to be operationally undesirable from a user interface perspective. (Which also gets back to there perhaps being a cost advantage, as it could be more expensive to get comparable stability out of a smooth controller as in the XF.) |
5pinDIN
Total Posts: 11891
Joined 09-16-2010 status: Legend |
You may indeed have found something that could help to resolve the issue. It would seem that anyone who is unhappy with the result of doing a manual sweep should try one of the lower-valued settings (table1 or table2), which (if I’m correctly understanding what the manual says) should result in less skipping of values if the knob is rapidly rotated. Of course, previous experience with Yamaha manuals makes me a little leery of jumping to any conclusions… Â :-) |
5pinDIN
Total Posts: 11891
Joined 09-16-2010 status: Legend |
By the way, I just checked and page 144 of the MOX Reference Manual also explains this Knob Curve setting, so it’s apparently not something new…
...and apparently neither is the discussion (or some of the detractors):
|
anotherscott
Total Posts: 653
Joined 06-30-2010 status: Guru |
or try assigning the function to the mod wheel (or a pedal), as also mentioned earlier |
MeMyselfAndI64
Total Posts: 201
Joined 11-12-2013 status: Enthusiast |
That,s how I read it to....anyway I did one last test with MIDI-OX. First test! I had it on Table 3 (default) and turned knob “quite fast” and it skips. Second test! on Table 1 and it,s huge difference!! I would say that it don`t skip at all. But I leave it to the people that have big issues with this! |
Bad_Mister
Total Posts: 36620
Joined 07-30-2002 status: Moderator |
Not quite the way they work.. I think your understanding of the knobs is a bit off, as is your understanding of the filters. The “old knobs” (as on the Motif series) travel from 7 o’clock to 5 o’clock but that is not 0-127 when offsetting a filter. Is that what you think? If you have a complaint, at least lets define what it is exactly so you can move on. Say AS1 is assigned to Filter Cutoff.
12 o’clock = the current Cutoff value of the filter - depending on what the program has the filter doing ...could be set anywhere… And could be set significantly different for each Element.
Turning the knob clockwise will ADD values until the Cutoff frequency reaches maximum
If the filter on an Element which has its actual cutoff frequency parameter set to 0 on its range of 0-255.
All settings counterclockwise from 12 o’clock do nothing at all. You cannot close the filter beyond 0 (minimum) all values counterclockwise of zero are meaningless. When choosing a controller to perform a parameter you should choose one that works for your goal. The knob with a center detente only send 0-127 in the distance from 12-5 o’clock. If you want to have the most amount of movement to control the filter 0-127 you should choose a physical controller that does send those values 0-127… So the the controller’s range of movement matches your desired range of motion.
Those who find the wheel smoother are right because it does send 0-127 and the range of movement is full
So yes, if you use an offset knob you have selected a controller with half the resolution.
The knobs, when recalled default to their 0 Center position.
Both send 128 values, one just sends -64~0~+63, while the other sends 0~127.
Controllers that send minimum to maximum, 0-127 are as follows:
Controllers that send minimum to maximum with a center zero value are
Controllers that send Either/Or minimum/maximum include switches
I think if you analyze what is on order here versus your knowledge and memory of analog synths and things like knobs, there is a difference that should influence your decision about controlling these filters.
Designed primarily as musical filters (versus the old analog use case for its one filter) the filters here are optimized to control the timbre of the resulting sound. If you wish to get the most response to your controlling gesture we recommend you choose a controller that
In the in-depth article on physical controllers it is mentioned to always choose your assignment wisely, based on the knowledge of how it works now. If you are looking to control a filter continuously from 7 o’clock minimum to 5 o’clock maximum like on the analog synths of yesteryear, you will need to learn how to work it so that what you want as minimum is the filter setting that happens when your selected controller is at the minimum position, and you also probably want to set a distance you have apply the controller to get maximum useable result. (The whole feature of being able to program a minimum setting and a maximum setting is a function of more modern tech) this must be respected
But without getting into all of that… You cannot just pick “any Voice”, sorry about this but in today’s technology every Voice is not going to be served the same way, not every parameter is going to behave well in real time. Using a filter that was originally set to do something specific with velocity and now suddenly you want it to be a wah-wah, well its a little more complicated then that now. Not all oscillators go through that filter together.
|