MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

can proper "unity gain" be achieved when mixing CV
MUFF WIGGLER Forum Index -> Modular Synth General Discussion Goto page 1, 2  Next [all]
Author can proper "unity gain" be achieved when mixing CV
MarkH
whats up guys
I have a question about mixers.
For mixing (audio or CV) I use my dotcom q112 and q113 4-channel and 8-channel mixers. But I find that if I send a pitch CV (1V/Oct) into one of these mixers, with volume all the way clockwise ("unity gain"), and then into a 1V/Oct VCO, the resulting note is flattened as I go up octaves. So "unity gain" on these mixers isnt quite unity gain. Im wondering, can proper "unity gain" be achieved in a mixer, if the mixer is dedicated to mixing CV's, such as the MOTM 830?
Thanks!
Mark
CLee
I try to send my 1v/oct sources straight to a 1v/oct input, then send anything modulating or offsetting it to another VCO in. Just to avoid this problem.

To get accurate unity gain from a mixer the module needs to be designed to do just that. If it's built with "unsorted" 1% resistors it can be up to 2% off (if the feedback resistor is 1% too high and the input resistor is 1% too low for example). That's worst case but it shows that if you want to build an accurate mixer you're going to need to hand select your resistors and probably have some trimmers.

Also, if your modules have a 1K output protection resistor that's not part of the feedback loop that gets added to the gain equation in the mixer and introduces a gain error.
PrimateSynthesis
MarkH wrote:
whats up guys
I have a question about mixers.
For mixing (audio or CV) I use my dotcom q112 and q113 4-channel and 8-channel mixers. But I find that if I send a pitch CV (1V/Oct) into one of these mixers, with volume all the way clockwise ("unity gain"), and then into a 1V/Oct VCO, the resulting note is flattened as I go up octaves. So "unity gain" on these mixers isnt quite unity gain. Im wondering, can proper "unity gain" be achieved in a mixer, if the mixer is dedicated to mixing CV's, such as the MOTM 830?


Yes, such proper unity gain can be achieved in a mixer if it is designed for control voltages. The problem is that the design and parts selection that would make it voltage accurate, would make it too slow or too noisy for audio. So most mixer modules are a compromise between the two.

It's also why I think all of these "powered multiples" aren't such a good idea. If they were for precise for DC they wouldn't be have the best sound for audio, and vice versa. Like most mixers they are OK for both, but not as clean a passive multiple.

The MOTM 830 is rare in that it uses an OP-285, a special op-amp specifically designed for summing DC-coupled audio. So not only is it very fast and low noise, it also has very little offset.
Graham Hinton
MarkH wrote:
whats up guys
I have a question about mixers.
For mixing (audio or CV) I use my dotcom q112 and q113 4-channel and 8-channel mixers. But I find that if I send a pitch CV (1V/Oct) into one of these mixers, with volume all the way clockwise ("unity gain"), and then into a 1V/Oct VCO, the resulting note is flattened as I go up octaves. So "unity gain" on these mixers isnt quite unity gain.


That is because you lose 1% of the voltage going into the mixer and another 1% coming out due to the 1k outputs/100k inputs. Pitch CV sources should not have 1k series resistors in their outputs, but some do creating the dilemma of whether VCO modules should have their inputs tuned to 1V/oct or 0.99V/oct. Often it is really the latter.

Quote:

Im wondering, can proper "unity gain" be achieved in a mixer, if the mixer is dedicated to mixing CV's, such as the MOTM 830?


Any mixer can be unity gain, it either requires precise resistor matching or a trimmer in the design irrespective of whether it is intended for CV or audio.

PrimateSynthesis wrote:

The problem is that the design and parts selection that would make it voltage accurate, would make it too slow or too noisy for audio. So most mixer modules are a compromise between the two.


Where are you getting that idea from? Precision resistor values do not effect the speed or noise.

Quote:

The MOTM 830 is rare in that it uses an OP-285, a special op-amp specifically designed for summing DC-coupled audio.


All operational amplifiers work down to DC. Some have higher offsets, but these may be easily trimmed out. The Minimoog used crappy 741s for its pitch CV summing. You only need closely specified parts to eliminate adjustments (and therefore skilled labour) in production.
PrimateSynthesis
Graham Hinton wrote:

Where are you getting that idea from? Precision resistor values do not effect the speed or noise.


I never said it did. But the type of op-amp, as well as the circuitry around it, does effect speed and noise. The compensation, whether internal or external, limits slew rate.

Graham Hinton wrote:

All operational amplifiers work down to DC. Some have higher offsets, but these may be easily trimmed out. The Minimoog used crappy 741s for its pitch CV summing. You only need closely specified parts to eliminate adjustments (and therefore skilled labour) in production.


Yet, the 741 is still crappy. You seem to be ignoring the difference between an ideal op-amp (infinite input impedance, etc.) and op-amps in the real world. Input current and input offset voltage also vary with temperature. So it is not simply a matter of trimming it out with closely specified parts or skilled labour. Generally speaking, op-amps designed for high speed have higher bias currents, BJT inputs can have an Ib 1,000 times higher that FET inputs. Which is just one reason among many why there are so many different op-amps.
CZ Rider
MarkH wrote:
Im wondering, can proper "unity gain" be achieved in a mixer

You can trim a CV mixer to any gain using a combination of inputs and attenuators. If a little more gain is needed to get exactly 1v/oct, using a second input of the same CV attenuated down should get you there. Or any scale can be dialed in like 2V/oct or .5V/oct.
I tune Moog 901's that was all the time by adjusting the scale of the incomming CV so I can get a 6 octave span. The 901's have 7 CV inputs, so using that doubling method with an attenuator is the easy way to do it.
lvoemachine
For unity gain for pitch purposes, STG's active multiples has a unity gain mixer. There's no knobs but you can use 2 sources. I use it a lot to transpose sequences and to do chord changes (depends if you quantize before or afterwards).
Graham Hinton
PrimateSynthesis wrote:

I never said it did.


Not in so many words, but you implied it in your statement that accuracy would compromise audio performance. Please explain how, with figures.

Quote:


But the type of op-amp, as well as the circuitry around it, does effect speed and noise. The compensation, whether internal or external, limits slew rate.


Again your statements need qualifying with real figures showing how that effects tuning accuracy.


Quote:

You seem to be ignoring the difference between an ideal op-amp (infinite input impedance, etc.) and op-amps in the real world.


No, I'm not ignoring anything including the OP's question. A unity gain mixer would not solve his problem, it's a system problem not a mixer design or op amp problem.
CLee
I don't see any reason that a mixer using a TL072 can't with proper matching of resistors and a trimmer pot be trimmed for a gain of 1.000.

Yes, it may not have the best output offset specs, but we're not building life-or-death medical equipment here...
negativspace
I've had perfectly fine results with 0.1% input resistors and a dual inverting op-amp design with a trimmer for gain on the second stage. TL072s or not.
PrimateSynthesis
Graham Hinton wrote:

Not in so many words, but you implied it in your statement that accuracy would compromise audio performance.


No, I didn't. Unless you think there is only one way to design a circuit, and resistors are the only parts. seriously, i just don't get it

Graham Hinton wrote:

Again your statements need qualifying with real figures showing how that effects tuning accuracy.


That would be quantifying, not qualifying. Regardless, if you refuse to understand the variables, I don't see how that would help.
PrimateSynthesis
CLee wrote:
I don't see any reason that a mixer using a TL072 can't with proper matching of resistors and a trimmer pot be trimmed for a gain of 1.000.

Yes, it may not have the best output offset specs, but we're not building life-or-death medical equipment here...


I put TL072's in my Oakley MultiMix. Since it is a general purpose design (and if I recall, it doesn't have a constant input impedance). The TL072 is a good choice of general purpose op-amp, as long as you keep it off the negative rails. It works well for both voltages and audio, but isn't the best for either.

My point was regarding the supposed advantage of powered multiples. The outputs of most modules can already drive multiple inputs. In order to improve on that, it would need to be optimized for voltage accuracy or audio quality. If it were optimized for voltage accuracy, it wouldn't be good for audio. So in order to do both reasonably well, it's a compromise that is possibly worse than using a passive multiple.

Although you're right that it might not make a difference. So why bother with a powered multiple in the first place?
PrimateSynthesis
503 double post confused
neil.johnson
Well, CV is not exactly DC, is it? A pitch CV will have fast edges at note change events, whether from a keyboard or step sequencer for example. Add in an envelope perhaps. You probably need a step response of sub-ms without overshoot. Noise is less of an issue as a pitch CV typically feeds some sort of integrator so low-level wideband noise on a pitch CV will tend to be averaged out (within reason, obviously).

For DC you're more concerned with offset and gain drift with time/temperature/etc. Absolute gain and offset can be trimmed, or minimised with precision components (a precision low-offset op-amp and resistors can sometimes be cheaper than a cheap op-amp plus trimmers when you consider parts cost, PCB area, labour time to set the trimmers and so on).

For AC you're concerned with distortion and noise; we don't hear slow-changing offsets, and rarely does gain need to be precisely controlled (within 0.5dB is usually good enough).

As Graham alluded to, many modules have quite a high output resistance due to series decoupling resistors isolating/decouple op-amp outputs from cable capacitance. These tend to be around 1k, I think mostly for historical reasons (22R would be sufficient in most cases). Fine for modulating signals (LFOs, etc) and audio where the absolute amplitude is not too critical, but for pitch CVs coming from devices which supposedly generate 1V/oct outputs they really should put the decoupling resistor inside the feedback loop, a technique that is common sense for both DC and AC applications.

DC-critical inputs should ideally be buffered first, to present a constant input resistance. The usual 100k-pot-with-100k-wiper input circuit varies from 100k to 50k input resistance as the pot is swept from 0 to 10 ("min" to "max", or whatever you want to call it). You can do tricks to minimise that without resorting to input buffers, but I guess not many designers see that as important hmmm.....

Substantial output resistors will produce CV scaling as inputs are connected to outputs. Passive mults make this worse with every load plugged in. For example, as has been pointed out in other threads a 1k output feeding a 50k input (input attenuator set to 10, typical input circuit) will scale the CV by 0.98. With a passive mult feeding two CV inputs both set to 10 the situation is a rather dire scaling by 0.96. And so on. And this scaling will change as input attenuators are adjusted.

Neil
MarkH
Thank you very much guys,
I appreciate all your replies.
I certainly have a better understanding now
of why the Pitch is getting flatter when I put it through a mixer.
Mark
PrimateSynthesis
neil.johnson wrote:

Substantial output resistors will produce CV scaling as inputs are connected to outputs. Passive mults make this worse with every load plugged in. For example, as has been pointed out in other threads a 1k output feeding a 50k input (input attenuator set to 10, typical input circuit) will scale the CV by 0.98. With a passive mult feeding two CV inputs both set to 10 the situation is a rather dire scaling by 0.96. And so on. And this scaling will change as input attenuators are adjusted.


Right. While I've seen plenty of designs without series resistors outside the feedback loop. You could avoid that situation with a powered multiple designed specifically for voltages controlling VCO's. Otherwise, if the offsets are going both ways, wouldn't it be better to use a passive multiple, and only have to adjust one output?

Anyway, while I think we agree here, and I appreciate your knowledge and insight. Let's say there is a VCA output using a nice audio op-amp, low resistor values to reduce noise, etc. for the best sound. All that goes out the window going into an all-purpose powered multiple...

Or you have a quantizer output using a slow precision op-amp, integrator in the feedback path to compensate for cable capacitance, etc. for the most stable VCO accuracy. Then stick it in a powered multiple using TL072's...

Or you have a clock output using a high-current 555 or discrete transistor built to smack up a whole studio full of gear. Only to connect it to a powered multiple that can drive less of a fan....

It's fine if some people want to think powered multiples are the greatest thing since jesus sliced bread. But any design that could handle both CV and audio would be a compromise in performance.
PrimateSynthesis
MarkH wrote:
Thank you very much guys,
I appreciate all your replies.
I certainly have a better understanding now
of why the Pitch is getting flatter when I put it through a mixer.


No problem. Remember, we're always here, if you ever need to start a esoteric discussion that has fuck all to do with your original question thumbs up
neil.johnson
PrimateSynthesis wrote:
MarkH wrote:
Thank you very much guys,
I appreciate all your replies.
I certainly have a better understanding now
of why the Pitch is getting flatter when I put it through a mixer.


No problem. Remember, we're always here, if you ever need to start an esoteric discussion that has fuck all to do with your original question thumbs up

Oooohhhh. Little bit tetchy are we? Fixed your grammar for you as well, just for the lulz!

Neil
neil.johnson
PrimateSynthesis wrote:
neil.johnson wrote:

Substantial output resistors will produce CV scaling as inputs are connected to outputs. Passive mults make this worse with every load plugged in. For example, as has been pointed out in other threads a 1k output feeding a 50k input (input attenuator set to 10, typical input circuit) will scale the CV by 0.98. With a passive mult feeding two CV inputs both set to 10 the situation is a rather dire scaling by 0.96. And so on. And this scaling will change as input attenuators are adjusted.


Right. While I've seen plenty of designs without series resistors outside the feedback loop. You could avoid that situation with a powered multiple designed specifically for voltages controlling VCO's. Otherwise, if the offsets are going both ways, wouldn't it be better to use a passive multiple, and only have to adjust one output?

Modules tend to have adjustable inputs rather than outputs. But otherwise yes if you have a module that can scale its output by something like 1.02 then you could compensate for loading (for 1k source, 50k load, adjust for other combinations).

Quote:
Anyway, while I think we agree here, and I appreciate your knowledge and insight. Let's say there is a VCA output using a nice audio op-amp, low resistor values to reduce noise, etc. for the best sound. All that goes out the window going into an all-purpose powered multiple...

You seem to be assuming that the all-purpose powered multiple is lower quality than the VCA. Maybe they all are. Maybe only the ones you have looked at are. The MOTM one might be good enough....?

Quote:
Or you have a quantizer output using a slow precision op-amp, integrator in the feedback path to compensate for cable capacitance, etc. for the most stable VCO accuracy. Then stick it in a powered multiple using TL072's...

Then don't use a low-cost powered multiple.

Quote:
Or you have a clock output using a high-current 555 or discrete transistor built to smack up a whole studio full of gear. Only to connect it to a powered multiple that can drive less of a fan....

Ummm...yeah...whatever rocks your boat I guess.

Quote:
It's fine if some people want to think powered multiples are the greatest thing since jesus sliced bread. But any design that could handle both CV and audio would be a compromise in performance.

Well first you need to define what that performance is before you can say if it is being compromised. But I guess Jesus would be rather good at slicing loaves of bread.

Neil
daverj
MarkH wrote:
whats up guys
I have a question about mixers.
For mixing (audio or CV) I use my dotcom q112 and q113 4-channel and 8-channel mixers. But I find that if I send a pitch CV (1V/Oct) into one of these mixers, with volume all the way clockwise ("unity gain"), and then into a 1V/Oct VCO, the resulting note is flattened as I go up octaves. So "unity gain" on these mixers isnt quite unity gain. Im wondering, can proper "unity gain" be achieved in a mixer, if the mixer is dedicated to mixing CV's, such as the MOTM 830?
Thanks!
Mark


Many of the answers here have bounced all over the place, but have hit on the probable reason you are seeing slightly less than "unity" gain.

It's actually fairly easy for an adjustable CV mixer to achieve unity gain by simply setting it's gain slightly higher than unity and then using the level controls to attenuate the signal down until it hits unity gain. You might be able to modify your existing mixers fairly easily by simply replacing the feedback resistor on the final op amp in the circuit with a slightly larger value. That would bump up the gain above unity. Then you could dial in the actual gain you wanted.

That might not be the ideal situation for you if you were looking for some automatic way to always achieve unity with your mixes. But doing that requires taking into account many of the things mentioned so far in this thread, like the output resistance of the modules being used, and without modifying everything in your system that might not be practical.
Graham Hinton
PrimateSynthesis wrote:

Let's say there is a VCA output using a nice audio op-amp, low resistor values to reduce noise, etc. for the best sound. All that goes out the window going into an all-purpose powered multiple...


Or any other module full of TL07xs and 100k resistors. I'm not sure what VCAs you mean, but to appreciate them the rest of the system quality has to match.

Quote:

It's fine if some people want to think powered multiples are the greatest thing since jesus sliced bread. But any design that could handle both CV and audio would be a compromise in performance.


No-one was talking about buffered mults except you. They have been discussed before and my assessment is that they are not good enough for pitch CVs and not necessary for anything else, just something amateurs like to build.

As Neil correctly points out, pitch CVs comprise transient steps and modulations making their bandwidth, noise and distortion requirements very similar to audio. The real problem is that in a modular system it is reasonable to expect to be able to insert modules in a CV path for mixing and processing and still expect it to be in tune over at least 5 octaves afterwards. The only way that is going to happen is if every module in the chain can deliver the correct output voltage into any load. That means no 1k output resistors. The OP's problem was caused by his CV source so however accurate or high quality a mixer is it won't compensate for a corrupted input voltage.
PrimateSynthesis
neil.johnson wrote:
Little bit tetchy are we? Fixed your grammar for you as well, just for the lulz!


I have no idea what tetchy means, I was just making a joke for jacking the thread.

neil.johnson wrote:

You seem to be assuming that the all-purpose powered multiple is lower quality than the VCA. Maybe they all are. Maybe only the ones you have looked at are. The MOTM one might be good enough....?


I don't think MOTM made one. But yes 8_)
PrimateSynthesis
Graham Hinton wrote:
They have been discussed before and my assessment is that they are not good enough for pitch CVs and not necessary for anything else


Then I have no idea what you were arguing about seriously, i just don't get it
neil.johnson
PrimateSynthesis wrote:
neil.johnson wrote:

You seem to be assuming that the all-purpose powered multiple is lower quality than the VCA. Maybe they all are. Maybe only the ones you have looked at are. The MOTM one might be good enough....?


I don't think MOTM made one. But yes 8_)

Sorry, yes, slightly crossed paths, there was earlier reference to the MOTM-830 dual mode mixer. That said, looking at the PCB for it (see here http://www.wiseguysynth.com/larry/motm/830/830.htm ) I can make some reasonable guesses about the circuit, and it looks like it has 1k output resistors outside the feedback loop and the usual 100k-pot-plus-100k-wiper-resistor circuit that gives an input resistance that varies between 50k and 100k depending on where the wiper is, which will interact with any source impedances, and both together will result not only in less-than-unity gain, but also interactions with other modules if a passive mult were used to share a CV from a similarly-impeded source with this mixer and other modules. Which brings the discussion back on-topic It's peanut butter jelly time!

Neil
PrimateSynthesis
neil.johnson wrote:

Sorry, yes, slightly crossed paths, there was earlier reference to the MOTM-830 dual mode mixer.


I mentioned it regarding parts selection because it uses an op-amp that is designed for summing DC-coupled audio. So it's an excellent choice for an audio mixer, and still better than most audio op-amps for summing control voltages. While the type of op-amps best for a mixer solely designed for voltage accuracy, are no good for audio.
MUFF WIGGLER Forum Index -> Modular Synth General Discussion Goto page 1, 2  Next [all]
Page 1 of 2
Powered by phpBB © phpBB Group