"Potentiometers" on new DSI synths - *Rant Warning!*

Discussions about anything analog, digital, MIDI, synth technology, techniques, theories and more.
Shanesaw
Active Member
Active Member
Posts: 324
Joined: Wed Jun 08, 2005 9:44 pm
Gear: Ultranova, Minibrute, Microbrute, Spark, microKORG, microX,
Location: Long Beach, California

"Potentiometers" on new DSI synths - *Rant Warning!*

Post by Shanesaw » Sat May 22, 2010 6:18 pm

Are these potentiometers or are they ultra-hi resolution encoders? They certainly feel a lot better than the old encoders but there's still stair-stepping happening... at least on the tetra there is. If you crank the resonance up so it accentuates the overtone series then slooooooly sweep the filter, you'll hear it. Is it because these pots are integrated into a DCO type synth that this happens? If these were truly potentiometers (cutoff and resonance), we wouldn't see the values change in the display (0-127). The values are infinite in an analog system no?
I know, I know... you can't please everyone and it may seem that I'm nit-picking but every time I try any synth, I try to find it's short-comings and although DSI claims that these are "real potentiometers", they sure act like encoders.
Do other DCO synths with potentiometers act this way? I don't recall if the 106 did this.
I don't know, maybe I just want a DCO synth to behave like a VCO synth but don't you think that this takes a lot away from a synths sound sculpting capabilities because you just can't get between a digital 1 and 0? Perhaps this is the reason why some people prefer VCO over DCO.
I should clarify that I have no issue with the overall sound of the tetra.
Also, I performed the same test using the FutureRetro XS (VCO) and there's absolutely none of this going on. The transition from one partial to the next is absolutely smooth.
"There was never a notion that a synthesizer would be used by itself for anything" - Robert Moog (1934-2005)

User avatar
pricklyrobot
Synth Explorer
Synth Explorer
Posts: 1742
Joined: Sun Feb 19, 2006 6:37 am
Location: Austin, TX

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by pricklyrobot » Sat May 22, 2010 7:16 pm

It's not a VCO vs. DCO thing. Any synth with patch memory will have some stepping on the controllers, because they're all getting processed through a CPU. The degree to which the stepping is audible depends on the processor speed/quality, and probably several other things depending on the exact architecture of any given synth.

Your XS has no patch memory; the pots are directly varying the voltage going to various sound-generating components of the synth, hence no stepping.

Someone with more tech knowledge can probably give you a more thorough explanation, but I think that's it in nutshell.
Arturia MiniBrute -- bits o' Euro -- Sammich SID -- E-mu MP-7 -- Korg ER-1 -- Thingamagoop

User avatar
calyx93
Active Member
Active Member
Posts: 450
Joined: Tue Aug 16, 2005 5:48 am
Gear: PPG Wave 2.2, μWave (x2), Roland JP6+V-Synth+S-220, SCI P5+600, Ensoniq SQ-80, Kawai K-3+K-3m, ARP Odyssey+Omni2, Emulator II+HD, Ensoniq Mirage.
Location: Winston-Salem, NC

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by calyx93 » Sat May 22, 2010 7:27 pm

Most every analog polysynth (new and old) with memories does this - the pots aren't directly connected to the analog circuitry. Their positions are scanned by the CPU and via a DAC they're given a voltage value for the analog circuit to use. Depending on the resolution of the particular parameter, you'll hear some stepping when physically turning the knobs - resonance on filters tends to be the most noticeable parameter.

Totally normal. But hey - it beats having to turn an encoder several times 'round to close a filter :)

(added this comment during the second poster's comment - hence the duplicate info)
“I have stretched ropes from steeple to steeple; garlands from window to window; golden chains from star to star, and I dance.” - Rimbaud

User avatar
b3groover
Senior Member
Senior Member
Posts: 965
Joined: Thu Apr 10, 2008 7:41 am
Real name: Jim
Band: THEO
Location: Lansing, MI
Contact:

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by b3groover » Sat May 22, 2010 8:10 pm

It's due to the limitation of MIDI, a protocol which was developed in the late 70s / early 80s and is hampered by its resolution. There is only a maximum of 128 values for any controller within the MIDI protocol. Any device using MIDI has to adhere to the protocol, so that's why you're hearing stepping.

We really need a new MIDI standard that replaces the now 30 year old MIDI 1.0 spec that is backwards compatible with the spec but features exponentially more resolution. That would solve the stepping problem.

User avatar
Jinsai
Junior Member
Junior Member
Posts: 131
Joined: Sat Apr 07, 2007 12:07 am
Gear: Alesis Ion
JX-3P w/Kiwitechnics
Nord Lead 2
Poly Evolver Keyboard
Prophet 12
Tempest
Assorted basses, guitars, software
Band: Sid Luscious and The Pants
Location: San Francisco, CA
Contact:

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by Jinsai » Sat May 22, 2010 8:54 pm

OP - do not confuse the oscillators with the filters. They're completely independent.

B3groover is not entirely correct. MIDI (developed by Dave Smith of DSI!) allows for more than 128 steps of resolution for controls and parameters via NRPNs and other means. The Alesis Ion, for example, uses MIDI controls but has no audible stepping.

Many of DSI's front panel controls are not "smoothed", because the intention is to provide fast response. The Evolver manual states that if you turn the filter knob, you may hear stair-stepping. But if you assign filter to the modulation wheel and adjust the mod wheel, you won't hear stair-stepping...however, there's a tiny bit of lag in the response of the filter because the value is being smoothed.
--
"A musician is someone who's heard a lot of music" - Pierre Boulez

User avatar
Yoozer
Expert Member
Expert Member
Posts: 1390
Joined: Sun Oct 09, 2005 1:31 pm

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by Yoozer » Sat May 22, 2010 8:57 pm

I'd be surprised if the encoders would go through the hassle of translating everything via MIDI. It'd be much faster to address the registers directly.
"Part of an instrument is what it can do, and part of it is what you do to it" - Suzanne Ciani, 197x.

User avatar
b3groover
Senior Member
Senior Member
Posts: 965
Joined: Thu Apr 10, 2008 7:41 am
Real name: Jim
Band: THEO
Location: Lansing, MI
Contact:

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by b3groover » Sat May 22, 2010 11:21 pm

Jinsai wrote:OP - do not confuse the oscillators with the filters. They're completely independent.

B3groover is not entirely correct. MIDI (developed by Dave Smith of DSI!) allows for more than 128 steps of resolution for controls and parameters via NRPNs and other means. The Alesis Ion, for example, uses MIDI controls but has no audible stepping.
I admit that my post was based on my own understanding of the issue and I'm no synth designer. So why do certain synths have stepping? Why not increase the resolution of the data encoders to eliminate it? My assumption is because those values have to be compatible with MIDI, unless the designers purposely do not allow the user to access those parameters via MIDI. But I guess I'm wrong.

Wasn't this the biggest gripe with the Voyager and the inspiration behind the Voyager Old School?

I don't see how a fast response from the control would relate to stepping or no stepping.

User avatar
StepLogik
Expert Member
Expert Member
Posts: 1493
Joined: Fri Dec 29, 2006 4:16 am
Gear: Gauntlets of Ogre Power, Vorpal Sword, Chain Mail +3, Ring of Invisibility, Staff of the Magi, Boots of Speed
Location: Boston, MA

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by StepLogik » Sun May 23, 2010 12:06 am

Does anyone know if the DSI "pots" are actually pots or are they just single-turn absolute-value encoders? Does the pot conversion require a firmware upgrade?

If they are just single-turn encoders generating 12- or 16-bit values, then the inherit quantization of the values would definitely be a source of stepping. It would be great to see some schematics to see Dave's implementation of the "pot edition" versus the encoders. Is each encoder sequentially scanned or is it interrupt driven? That is, whenever an encoder is turned then the CPU knows to read a new value off the encoder data bus. If it is sequentially scanned, then that can be done VERY quickly for encoders. However, if you are using actual pots, then you have to "slow down" the scanning in order to let the ADC settle as you sample the value off of each pot. Such design compromises could be the source of stepping in the DSI instruments. Considering the amount of "work" that Dave is doing in software, I wouldn't be surprised if he is using the interrupt-driven approach which saves CPU cycles but creates a headache when sampling analog pots.

Also a quick note on MIDI - when using 7-bit values in MIDI (such as CC's) then it is necessary to scale those values to whatever internal bit depth is used. This is very simple to do in the CPU's firmware, but again results in stepping. By using sysex or other MIDI standards you can achieve higher resolution but you can still get some stepping due to the slower data rates of the MIDI protocol but its quite subtle.

Smoothing algorithms can help with these problems. They create a slight lag in the audible effect, but reduce the stepping effect by not allowing "jumps" in values that are placed on the DAC to control the analog circuitry. I think this is typically done in software but someone more knowledgeable might know of instruments that use hardware smoothing.

User avatar
RD9
Active Member
Active Member
Posts: 536
Joined: Tue Mar 02, 2010 4:24 am
Location: NY/NJ

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by RD9 » Sun May 23, 2010 12:17 am

Shanesaw wrote:Are these potentiometers or are they ultra-hi resolution encoders? They certainly feel a lot better than the old encoders but there's still stair-stepping happening... at least on the tetra there is. If you crank the resonance up so it accentuates the overtone series then slooooooly sweep the filter, you'll hear it. Is it because these pots are integrated into a DCO type synth that this happens? If these were truly potentiometers (cutoff and resonance), we wouldn't see the values change in the display (0-127). The values are infinite in an analog system no?
I know, I know... you can't please everyone and it may seem that I'm nit-picking but every time I try any synth, I try to find it's short-comings and although DSI claims that these are "real potentiometers", they sure act like encoders.
Do other DCO synths with potentiometers act this way? I don't recall if the 106 did this.
I don't know, maybe I just want a DCO synth to behave like a VCO synth but don't you think that this takes a lot away from a synths sound sculpting capabilities because you just can't get between a digital 1 and 0? Perhaps this is the reason why some people prefer VCO over DCO.
I should clarify that I have no issue with the overall sound of the tetra.
Also, I performed the same test using the FutureRetro XS (VCO) and there's absolutely none of this going on. The transition from one partial to the next is absolutely smooth.
Would it be possible for you to post an audio clip somewhere?

Shanesaw
Active Member
Active Member
Posts: 324
Joined: Wed Jun 08, 2005 9:44 pm
Gear: Ultranova, Minibrute, Microbrute, Spark, microKORG, microX,
Location: Long Beach, California

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by Shanesaw » Sun May 23, 2010 1:29 am

Here's the Tetra first, then the XS. I swept the filters a smoothly as possible. The image also reveals the stair steps quite nicely.
http://soundclick.com/share?songid=9178332
Image
"There was never a notion that a synthesizer would be used by itself for anything" - Robert Moog (1934-2005)

User avatar
RD9
Active Member
Active Member
Posts: 536
Joined: Tue Mar 02, 2010 4:24 am
Location: NY/NJ

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by RD9 » Sun May 23, 2010 1:52 am

Thanks for posting that.

Interesting... In my Evolver manual there's a note about the use of smoothing in the filter's frequency cutoff control, specifically to address "stepping" during manual sweeps. I wonder if this is the same issue. Is it possible that smoothing is also required in the resonance control, but that it wasn't applied? I wonder if there's an OS update.

User avatar
guillermotin
Newbie
Newbie
Posts: 86
Joined: Wed Jan 21, 2009 4:37 am
Location: Somewhere in the outer ring...

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by guillermotin » Sun May 23, 2010 2:14 am

Shanesaw wrote:Here's the Tetra first, then the XS. I swept the filters a smoothly as possible. The image also reveals the stair steps quite nicely.
http://soundclick.com/share?songid=9178332
Image

Nice examples, thank you!

I'm now considering a XS...

:)

User avatar
Stab Frenzy
Moderator
Moderator
Posts: 9723
Joined: Tue Jun 06, 2006 5:41 pm
Gear: Eurorack, RYTM, Ultranova, many FX
Location: monster island*
Contact:

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by Stab Frenzy » Sun May 23, 2010 3:05 am

Shanesaw wrote:Are these potentiometers or are they ultra-hi resolution encoders?
StepLogik wrote:Does anyone know if the DSI "pots" are actually pots or are they just single-turn absolute-value encoders? Does the pot conversion require a firmware upgrade?

If they are just single-turn encoders generating 12- or 16-bit values, then the inherit quantization of the values would definitely be a source of stepping.
It wouldn't matter if they're pots feeding voltage to an AD converter or digital encoders, the end result is a numerical value which controls the filter value. In fact if they were 12 or 16 bit values you wouldn't hear the stepping at all, the Ion and Micron have 11 bit encoders and there's no noticeable stepping from them.

I don't have any issues with stepping on my MEK cause I don't expect the cutoff knob to behave like a direct analogue control. If I want to directly tweak the cutoff without stepping I assign the mod wheel to it, but generally I tweak the filter env attack and decay times to adjust the filter. That might sound weird to you but it's actually more natural sounding than a direct filter sweep, filter sweeps sound pretty cheesy to me.

User avatar
StepLogik
Expert Member
Expert Member
Posts: 1493
Joined: Fri Dec 29, 2006 4:16 am
Gear: Gauntlets of Ogre Power, Vorpal Sword, Chain Mail +3, Ring of Invisibility, Staff of the Magi, Boots of Speed
Location: Boston, MA

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by StepLogik » Sun May 23, 2010 4:28 am

Stab Frenzy wrote:It wouldn't matter if they're pots feeding voltage to an AD converter or digital encoders, the end result is a numerical value which controls the filter value. In fact if they were 12 or 16 bit values you wouldn't hear the stepping at all, the Ion and Micron have 11 bit encoders and there's no noticeable stepping from them.
Its not just about the bit depth. The rate at which you acquire that numeric value can also affect the perception of stepping. If the rate is low, then the knob can travel a considerable distance between samples resulting in stepping if there is no smoothing function.

Encoders basically come in two varieties - absolute and relative. The absolute value gives you an actual number - higher as the knob is turned to the right. The relative encoders put out a pulse when the knob is turned in one direction and a different pulse when the knob is turned the other way.

The relative encoders require that the CPU "count" the pulses and add/subtract from the initial value. Whenever the value changes (knob is turned) the CPU can update the DAC and change the sound. If the CPU updates on every pulse (unlikely) then you would have an extremely smooth response since the values being shifted onto the DAC don't have significant gaps in value.

Furthermore, with this scheme, you can control the "sensitivity" or the velocity of the knob by dividing or multiplying the pulses from the encoder. If you want knob movement to be very slow (multiple turns to sweep the full filter range) then you can divide the pulses - say, for every 10 pulses add/subtract 1 from the current filter cutoff value in memory. The reverse is true as well, you can say for every 1 pulse, add/subtract 10 from the current filter cutoff value in memory. This would give you less "range" on the encoder.

The relative encoders are nice because they require less hardware and don't have the hassles of sampling from the ADC (settle times, etc). Furthermore, they can be interrupt driven so that the CPU can concern itself with other tasks instead of scanning controls that may or may not have been changed.

If this is the technique being used, then shifting paradigms to a traditional pot-scanning methodology could result in some pretty serious stepping issues for all kinds of reasons. I'm thinking he didn't use relative encoders because if that were the case it would be fairly easy to change the response velocity with a simple firmware update. The fact that the PE versions require new hardware leads me to believe something else is going on (besides the fact that the original encoders were flimsy and felt cheap warranting their upgrade).

I do wish I knew how Dave implemented it in his instruments. If I knew the technique then it would be fairly easy to figure out where the stepping is coming from.

User avatar
EmptySet
VSE Review Contributor
VSE Review Contributor
Posts: 474
Joined: Sun Jul 02, 2006 10:33 am
Gear: MS20, x0xb0x, JV-1080, JX-3P, JX-8P(PG800), Juno Stage, VFX-SD, VL-1, ARP Omni I, D-50, DW6000
Location: Illinois, USA

Re: "Potentiometers" on new DSI synths - *Rant Warning!*

Post by EmptySet » Sun May 23, 2010 5:31 am

b3groover wrote: I admit that my post was based on my own understanding of the issue and I'm no synth designer. So why do certain synths have stepping? Why not increase the resolution of the data encoders to eliminate it? My assumption is because those values have to be compatible with MIDI, unless the designers purposely do not allow the user to access those parameters via MIDI. But I guess I'm wrong.

Wasn't this the biggest gripe with the Voyager and the inspiration behind the Voyager Old School?

I don't see how a fast response from the control would relate to stepping or no stepping.
Jinsai is right, MIDI supports values outside the 128 (8 bit) range using NPRNs. The problem is that patch storage is digital.

Post Reply