Buffered mult. Best for sending pitch cv to multiple OSCs?

Anything modular synth related that is not format specific.

Moderators: Kent, luketeaford, Joe.

User avatar
ben_hex
Super Deluxe Wiggler
Posts: 6724
Joined: Sun May 12, 2013 12:58 pm
Location: West Yorkshire, England

Buffered mult. Best for sending pitch cv to multiple OSCs?

Post by ben_hex » Wed Aug 21, 2013 12:26 pm

So here's the deal ...

I'm sending midi out of my soundcard (from computer) to an external Kenton pro solo mk II midi-cv unit. It has pitch and gate cv out plus some more.

I want to take the pitch cv and send it to my 3 Oscillators ... An AFG, e350 and a braids.

Will using a buffered multiple be the best way to accurately distribute the same cv source to all 3 modules?

The other options would be stackables, not been home to try this so though I'd ask then just order a buffered multiple if needed.

Feel free to suggest buffered multiples too.

User avatar
z3r01
The Mighty Souls
Posts: 2028
Joined: Tue Aug 30, 2011 5:28 am
Location: Singapore

Post by z3r01 » Wed Aug 21, 2013 1:12 pm

With multiples that are not buffered, with every split, there would be a slight droop. So, if you require the tuning between the 3 oscs to be tight, a buffered multiple would be advisable, so there won't be any loss in voltage. :tu:

User avatar
Dave Peck
Super Deluxe Wiggler
Posts: 4779
Joined: Thu Mar 24, 2011 1:22 pm
Location: SF bay area

Post by Dave Peck » Wed Aug 21, 2013 2:53 pm

Keep in mind that this issue of passive versus buffered mults is mostly an issue if you tend to CHANGE the number of oscs that the CV signal is being sent to. For example, if you always send the CV signal to three oscs, and if you calibrate your three oscs while they are all getting the CV signal, they will all work fine. The problems can come up when you add oscs, or make a patch that requires temporarily disconnecting the CV from one or more oscs.

User avatar
burdij
Veteran Wiggler
Posts: 720
Joined: Mon Dec 28, 2009 4:38 pm
Location: Wisconsin US

Post by burdij » Wed Aug 21, 2013 3:21 pm

Knobcon 7 - 2018 Photo Album
While website and store updates are in progress, access the Grove Audio Synth Store
Tesla Science Center http://www.teslasciencecenter.org

User avatar
ben_hex
Super Deluxe Wiggler
Posts: 6724
Joined: Sun May 12, 2013 12:58 pm
Location: West Yorkshire, England

Post by ben_hex » Wed Aug 21, 2013 5:13 pm

Thanks everyone, tried it earlier and although got them in tune first then multiplied up a 1v/oct pattern there was the expected pitch drift. Was kinda cool as it was a detune lead style thing I had going. Definitely no good for accuracy though.
Dave Peck wrote:Keep in mind that this issue of passive versus buffered mults is mostly an issue if you tend to CHANGE the number of oscs that the CV signal is being sent to. For example, if you always send the CV signal to three oscs, and if you calibrate your three oscs while they are all getting the CV signal, they will all work fine. The problems can come up when you add oscs, or make a patch that requires temporarily disconnecting the CV from one or more oscs.
Thats a great idea! Didn't even cross my mind! Gonns try that tomorrow, multiple the 1v/oct pitch tune them all then lay and see if it tracks properly. Guess it should, as you said. I won't be adding more oscillators after initial patching.
All DivKid video series to be found here!
http://www.youtube.com/divkidvideo

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3318
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Post by Graham Hinton » Wed Aug 21, 2013 5:32 pm

The loss in voltage is caused by a series resistor in the output of the CV source(s). If there isn't one you don't need a buffered mult and if there is one a buffered mult doesn't compensate the loss. Effectively you end up with the VCOs tuned to 0.99V/octave.

Here's how to use our Trimmer module to correct and drive completely different V/oct VCOs. You only need one output (and a passive mult) if they are all the same standard.

Image

User avatar
Dave Peck
Super Deluxe Wiggler
Posts: 4779
Joined: Thu Mar 24, 2011 1:22 pm
Location: SF bay area

Post by Dave Peck » Wed Aug 21, 2013 7:58 pm

ben_hex wrote:Thanks everyone, tried it earlier and although got them in tune first then multiplied up a 1v/oct pattern there was the expected pitch drift. Was kinda cool as it was a detune lead style thing I had going. Definitely no good for accuracy though.
Dave Peck wrote:Keep in mind that this issue of passive versus buffered mults is mostly an issue if you tend to CHANGE the number of oscs that the CV signal is being sent to. For example, if you always send the CV signal to three oscs, and if you calibrate your three oscs while they are all getting the CV signal, they will all work fine. The problems can come up when you add oscs, or make a patch that requires temporarily disconnecting the CV from one or more oscs.
Thats a great idea! Didn't even cross my mind! Gonns try that tomorrow, multiple the 1v/oct pitch tune them all then lay and see if it tracks properly. Guess it should, as you said. I won't be adding more oscillators after initial patching.

...It may not work at first, depending on how the oscs were previously calibrated. You may need to recalibrate (not tune, but calibrate) each osc after connecting your 1V/OCT CV signal to all of them. Like this:

1. Connect the 1V/octave keyboard CV to all three oscs via a multiple.
2. Get a fourth osc source that you know is properly calibrated to produce exact octaves, like a digital osc (or get a fourth osc that just drones on one pitch and is not connected to the keyboard CV). You'll use this fourth osc as a reference.
3. Now one at a time, turn up just one of the three oscs in the modular along with this fourth reference osc and play up & down the keyboard, checking the tuning at each octave. Adjust the 1V/OCTAVE calibration trim pot (not the front panel tuning knob) on your analog osc and keep checking it until it tracks the fourth reference osc exactly and stays perfectly in tune over several octaves.
4. Repeat this for each of the analog oscs, one at a time.
5. Now they will all be calibrated to produce notes exactly one octave up, two octaves up, etc. when you play the keyboard, even if the actual 1V/OCTAVE CV signal is getting slightly reduced by sending it to them via a passive mult. You have compensated for this by doing the calibration to a reference osc while they are all receiving the CV signal the same way they will be getting it during normal use.

User avatar
ben_hex
Super Deluxe Wiggler
Posts: 6724
Joined: Sun May 12, 2013 12:58 pm
Location: West Yorkshire, England

Post by ben_hex » Thu Aug 22, 2013 5:48 am

Another good tip Dave Peck although if I calibrate that way then only want to use one OSC won't the calibration be out?
All DivKid video series to be found here!
http://www.youtube.com/divkidvideo

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1823
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Thu Aug 22, 2013 6:23 am

ben_hex wrote:Another good tip Dave Peck although if I calibrate that way then only want to use one OSC won't the calibration be out?
Yes.
What is needed is a buffered mult that can provide a little bit of gain to compensate for the loss due to output resistance. Graham's is one example.
Then you don't need to calibrate your VCOs and can leave them set to 1.00V/oct.

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
ben_hex
Super Deluxe Wiggler
Posts: 6724
Joined: Sun May 12, 2013 12:58 pm
Location: West Yorkshire, England

Post by ben_hex » Thu Aug 22, 2013 6:29 am

I'm not totally with that set up neil.johnson not sure I really get it. Plus Graham Hinton Trimmer module is big and lot of money.

Would a Trimmer module just take the original pitch CV and then put that into the Mutliple? So it's a bit hotter say 1.05Hz/oct then the multiple trims that back to 1Hz/oct?
All DivKid video series to be found here!
http://www.youtube.com/divkidvideo

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1823
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Thu Aug 22, 2013 6:47 am

ben_hex wrote:I'm not totally with that set up neil.johnson not sure I really get it. Plus Graham Hinton Trimmer module is big and lot of money.

Would a Trimmer module just take the original pitch CV and then put that into the Mutliple? So it's a bit hotter say 1.05Hz/oct then the multiple trims that back to 1Hz/oct?
All of this problem is due to the poor design decisions taken early on in Euromodules, and continued to this day. For a robust CV output you want it to be able to control a reasonable number of CV inputs without any droop whatsoever. Unfortunately that is not the case, more the rare exception.

The problem is due to the CV output having some series resistance. Lets say it is 1k. Now take your buffered mult, it will have an input resistance of, lets say, 100k. The result of these two resistors will be an CV output that is now 100/101 = 0.99V/oct. It may not sound like much, but you're now out of tune.

As a bodge you need something that can apply a little bit of gain to the CV as it passes by. In this example you need a gain of 1/0.99 = 1.01.

In an ideal world, you wouldn't need this - the CV output of your Kenton box would be able to drive several 100k CV inputs without droop. Speak to Kenton about this.

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3318
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Post by Graham Hinton » Thu Aug 22, 2013 7:17 am

ben_hex wrote: Would a Trimmer module just take the original pitch CV and then put that into the Mutliple? So it's a bit hotter say 1.05Hz/oct then the multiple trims that back to 1Hz/oct?
Yes for the first part, but the multiple is not doing the trimming, it is the interaction of output impedances with varying load impedances and that changes with every patch.

If the source of CV has feedback so that it corrects its output then there is no problem and patching can be done with passive mults and no errors. However the whole principle of voltage control is that any module may be a source so that means getting rid of all the series output resistors everywhere. If you had an LFO making a trill on a VCO and set exactly to a minor third, then if you patched another VCO in they would be out of tune in exactly the same way. If you had a portamento module and wanted to insert it between pitch source and VCO it would introduce an error too.

The problem with buffered mults is that they introduce the same error at their inputs and then repeat that. They only reduce the amount of error rather than eliminate it. There are some with floating inputs, but their outputs thrash between power rails when patched.

I only gave the Trimmer example because I had the diagram handy and it illustrates the problem. Trimmer was designed to front end our SwitchMix and PinMix modules so that pitch accurate True Summing can be done. Eg a keyboard, a sequencer, a pitchbend, LFOs and envelopes may be summed to a variety of VCOs in any combination on a matrix without going out of tune.

So you can choose between between modifying every module in your system or using something that compensates for their design limitations and that isn't a buffered mult. It doesn't have to be my modules either, but it does have to be something that follows that principle.

Don't accept tracking over three octaves as an improvement over two when you should expect at least six as standard.
Last edited by Graham Hinton on Thu Aug 22, 2013 7:23 am, edited 1 time in total.

User avatar
ben_hex
Super Deluxe Wiggler
Posts: 6724
Joined: Sun May 12, 2013 12:58 pm
Location: West Yorkshire, England

Post by ben_hex » Thu Aug 22, 2013 7:22 am

thanks neil.johnson starting to make sense now.


Graham Hinton cheers for the input. Again helping me understand it all. I've seen the SwitcMix and Pin Matrix, which makes sense with the Trimmer.

Got a bit of Wiggling in and not getting a massive amount of detune between 3 OSCs so I'm happy. In reality too if I'm using 3 OSCs I want a bit of detune anyway, wider thicker sounding. If I wanted straight and very solid I'd just use one or two if I wanted octave up effects. Ultimately I'll buy a Vermona qMI so I can run out 4 channels of midi anyway which would eliminate any multiple pitch CV problems.
All DivKid video series to be found here!
http://www.youtube.com/divkidvideo

User avatar
flo
Space Jack
Posts: 5005
Joined: Tue Jul 10, 2012 4:37 am
Location: Zurich

Post by flo » Thu Aug 22, 2013 8:55 am

ben_hex wrote:not getting a massive amount of detune between 3 OSCs so I'm happy. In reality too if I'm using 3 OSCs I want a bit of detune anyway, wider thicker sounding.
The problem is not so much the detuning, but rather that your tuning goes flatter (for all oscs) the higher you try to get them to track...

User avatar
soundwave106
Super Deluxe Wiggler
Posts: 1728
Joined: Thu Dec 16, 2010 10:16 pm

Post by soundwave106 » Thu Aug 22, 2013 9:32 am

neil.johnson wrote:The problem is due to the CV output having some series resistance.
As I understand it, that resistance on the outs is put in place for short circuit prevention, right? So it's more a "design decision" I guess.

Practically I have not noticed any detuning errors using buffered mults, I suppose errors may show up with larger octave ranges, but that is not a typical Eurorack application for myself.
Last edited by soundwave106 on Thu Aug 22, 2013 3:26 pm, edited 1 time in total.
BC | SC

User avatar
tojpeters
Super Deluxe Wiggler
Posts: 1266
Joined: Mon Feb 06, 2012 10:34 am
Location: Mt.Shasta, Ca

Post by tojpeters » Thu Aug 22, 2013 10:06 am

I use the MFOS CV/GATE distributer schematic. My analog keyboard is only 4 octaves but it tracks true no matter if i'm running 1 VCO or 4. Midi keyboard with CV converter does 6 octaves and seem true but i don't think i've ever put the tuner on it on the upper range. seems in tune to me. I use an old Peterson strobe tuner,very accurate. you can tune to exactly 3% flat. 100k input impedance, no resistor on the output. Easy build, i've done 4 of then in the last couple days as part of a project. I'm no E.E. and welcome the thoughts of those who know more that i.

User avatar
Dave Peck
Super Deluxe Wiggler
Posts: 4779
Joined: Thu Mar 24, 2011 1:22 pm
Location: SF bay area

Post by Dave Peck » Thu Aug 22, 2013 11:52 am

ben_hex wrote:Another good tip Dave Peck although if I calibrate that way then only want to use one OSC won't the calibration be out?
Depends on exactly what you mean by 'only want to use one osc'. If you make a patch that only needs one of your audio oscs, you can simply not patch the audio outputs of the second & third osc to anything and you'll only hear the first one, but leave the 1V/octave CV patched to ALL of them so the keyboard tracking will be unaffected. But if you want to use one of the oscs for standard keyboard tracking / audio range use, and also use the others in the patch, but WITHOUT keyboard tracking (you want them to drone, or you want to use both of them as LFOs with no keyboard tracking) THEN you need to disconnect the 1V/octave CV from those two oscs and leave it connected ONLY to the first osc, and THAT might affect the tracking of the first osc.

But all of this can vary from system to system, depending on things like the number of oscs and your specific source of the keyboard CV. For example, my synth has eight oscs all driven from an Encore Expressionist midi-to-CV converter, usually all from one CV channel, in unison. They are patched via a Dotcom passive Q146 "Normalization" buss distribution module, which sends the keyboard CV signal to all of them via a large fan-out harness behind the panel. I have several other dedicated LFOs so I usually just leave all eight audio oscs connected to the keyboard CV, but sometimes I'll use a couple of the eight oscs for LFOs or droning, and when I disconnect two of the eight from the keyboard CV, the keyboard tracking of the remaining six is unaffected.

User avatar
ben_hex
Super Deluxe Wiggler
Posts: 6724
Joined: Sun May 12, 2013 12:58 pm
Location: West Yorkshire, England

Post by ben_hex » Thu Aug 22, 2013 1:37 pm

Been experimenting today with only sample and hold, random LFO waves sent to the pitch and I can get them to play nicely.

Need to get the computer and midi to cv hooked up to properly test it out.

For more info ... I'm my running one voice (ie one part with 1-3 OSCs no droning etc) for use in production. Not using drones and multiple layers.
All DivKid video series to be found here!
http://www.youtube.com/divkidvideo

User avatar
ucacjbs
Wiggling with Experience
Posts: 349
Joined: Sat Feb 02, 2013 11:22 am

Post by ucacjbs » Sat Aug 24, 2013 12:30 pm

soundwave106 wrote:
As I understand it, that resistance on the outs is put in place for short circuit prevention, right? So it's more a "design decision" I guess.
As Graham alludes to,above, it is possible to design the output stage to include the short-circuit protection resistor without suffering the problems being discussed here. In short, you put the short-circuit protection resistor inside the feedback loop of the output op-amp. By doing that, you get the best of all worlds. As an aside, the popular TL07x op-amps used in a number of modular designs have in-built short circuit protection; a lot of modules that have the problematic (i.e. outside of the feedback loop) 1k output resistor don't even need them!

User avatar
ben_hex
Super Deluxe Wiggler
Posts: 6724
Joined: Sun May 12, 2013 12:58 pm
Location: West Yorkshire, England

Post by ben_hex » Sat Aug 24, 2013 12:44 pm

Well for now all I have is a Doepfer A-180-2 multiple to that just does whatever it does.

The results so far have been good going, pitch cv in - hold a note - tune each OSC - then play. Aiming for the middle note in the pattern to ensure there is minimal drift much higher or lower than needed. Tuning to the lowest or "root" note of the pattern hasn't been as useful if the pattern spans a couple of octaves. I like the slight detune and thickening sound though, by the time I'm stacking 3 OSCs that's what I want anyway.
All DivKid video series to be found here!
http://www.youtube.com/divkidvideo

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3318
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Post by Graham Hinton » Sun Aug 25, 2013 7:44 am

ben_hex wrote:Well for now all I have is a Doepfer A-180-2 multiple to that just does whatever it does.

The results so far have been good going, pitch cv in - hold a note - tune each OSC - then play. Aiming for the middle note in the pattern to ensure there is minimal drift much higher or lower than needed. Tuning to the lowest or "root" note of the pattern hasn't been as useful if the pattern spans a couple of octaves. I like the slight detune and thickening sound though, by the time I'm stacking 3 OSCs that's what I want anyway.
Reading between the lines of this statement, it seems that you are using "tuning" to mean getting a best fit with VCOs of unknown calibration rather than actually calibrating them which is what I and others replying to you mean. Am I right?

If you want to play in tune then VCO calibration is a prerequisite. All VCOs have an adjustment to set the "scale", but it may vary in resolution, i.e. some use a simple preset and others use multiturn ones. Some VCOs have an additional preset to adjust high frequency tracking too, but whatever you have the technique is the same. First you have to decide what your reference voltage is, if you are playing a keyboard then that is your reference. You can check whether you are getting 1V for every octave with a voltmeter, but it doesn't matter if this is not exact as long as it is equal steps. Then the VCOs have to be individually adjusted to this CV. Start by playing the bottom note and an octave above and adjust the VCO calibration preset (not the front panel knobs) until you hear an octave change. Ignore the absolute pitch for now and just concentrate on octave changes. If you are using a tuning meter and one note is, say, 30 cents out go for 30 cents out on the other note too. Once you have an octave interval play two octaves , then three and four making finer adjustments. Keep playing all the octave notes to get the best fit to the keyboard and some VCOs need an HF adjustment at the top. Once you have the scale set, retune the VCO to A=440 using the tuning preset if it has one or the front panel knob. Repeat for each VCO individually and then try them all together.

This is a bit of a laborious process, not helped by having to gain access to the module presets, but it only has to be done properly once. That is where Trimmer is an advantage because you only have to take that out of the rack and each channel has 25-turn presets where the VCOs may only have a single turn. It will even cope with non-1V/oct standards like Buchla and EMS and make those track other systems.

Some people seem to think that having VCOs in tune makes them phase locked and therefore exact tuning is undesirable. This is not true, you always get beating between good unsynchronised VCOs and if you don't they have a design problem. Locking on each other is often caused by poor power distribution. The beat rate doubles for every octave higher and you should be able to achieve a slow one at the high end and a very slow one at the bottom. Expect five octaves tracking on a five octave keyboard as normal and another couple with transposition. Then get your desired amount of detuning by using the front panel fine tune knob.

One problem of using several discrete VCOs of different makes or bought at different times is that you cannot rely on the factory settings or know what they were tuned to. The maker might be using a CV source with a 1k resistor and so were really setting it up to 0.99oct/volt. Also many use 100k 1% input resistors on the 1V/oct input when they should be better than 0.1% and you cannot tell by looking whether they were hand selected or not. This will limit the tracking range and is another area that Trimmer is designed to improve by correcting for the differences.

Just like a guitar, if you want to play a synthesizer as a musical instrument then learning to tune it first comes with the territory.
Last edited by Graham Hinton on Tue Oct 01, 2013 6:44 am, edited 1 time in total.

User avatar
ben_hex
Super Deluxe Wiggler
Posts: 6724
Joined: Sun May 12, 2013 12:58 pm
Location: West Yorkshire, England

Post by ben_hex » Sun Aug 25, 2013 1:01 pm

Thanks Graham Hinton that wasn't what I was thinking but insightful and definitely something I need to properly check out. I think seen as I've been getting great tracking from the AFG for a while and recent additions of the e350 and the braids have tracked nicely against it I haven't given it any thought.

On the note of calibrating oscillators ... will the e350 and braids need it since they are digital modules?
All DivKid video series to be found here!
http://www.youtube.com/divkidvideo

User avatar
Fleggy
Wiggling with Experience
Posts: 310
Joined: Thu Jun 13, 2013 12:24 pm
Location: Dublin, Ireland

Post by Fleggy » Sun Aug 25, 2013 1:11 pm

Found this thread to be a massive help cheers Graham. :tu:
LFO'S BEFORE HOES
_______________________________________
Current set up

Instagram

Twitter

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1823
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Sun Aug 25, 2013 4:25 pm

Most opamps these days have some form of internal short circuit protection. The output resistors are necessary to stop the opamp from oscillating due to load capacitance (cables, etc). In practice you need about 22-75 Ohms to keep a TL072 from oscillating. Where DC accuracy is required that resistor should be inside the feedback loop so that any voltage drop across the series resistor is compensated for by the action of the feedback. It certainly doesn't need to be anywhere near 1k0. I think that value is based on some historical convention or misunderstanding.

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3318
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Post by Graham Hinton » Sun Aug 25, 2013 4:42 pm

ben_hex wrote: On the note of calibrating oscillators ... will the e350 and braids need it since they are digital modules?
All oscillators need calibrating to work with whatever you are controlling them with.

Some digital oscillators have limitations because their designers don't understand how analogue tuning is done. I can't comment on the e350 because I can't find a manual online (Paul?), but a quick glance at the Braids manual reveals the same problem as the Cyclebox: it is sampling too small a voltage interval, 2V for the Braids and 1V for the Cyclebox. This means that any error is magnified by 5/2 or 5/1 respectively when you play it with a 5 octave range keyboard which is enough to make it out of tune.

The only tuning algorithm that would make any sense to me would be to give it the highest and lowest note voltages and let it achieve the best fit between them. You shouldn't have to tell it what a volt is, you want to tell it what your keyboard range is and it shouldn't expect a toy keyboard. Like all DSP devices, it is only as good as the software writer.

Post Reply

Return to “Modular Synth General Discussion”