Mungo Enterprises d0, dual channel Delay

[jnlkrt]

here's the d0-related part of a résumé-ish post about my Mungo and d0 experience from another thread. i know it's kind of a cross-post, but i feel like it might be a worthy contribution to this topic

here's some thoughts, findings and things that John pointed me at in regards to the d0:

the control ranges are HUGE, and allow for all kinds of weird settings which no other echo/delay device would even allow. it's a very raw and open design. what one needs to understand is that even though it has 'hifi' specs, the control ranges allow for a lot of error. common 'hifi' digital delays just block out these errors by restricting the ranges and settings. this way one can easily come to the conclusion that the d0 is indeed not 'hifi', since it allows you to do all kinds of settings which essentially are just plain wrong (in a traditional echo and mod fx context).

the slew rate is super essential to get a clean sound. if i understand correctly, putting voltage control over the delay time without all kinds of glitches is a huge challenge in general. slewing the control signal is pretty much mandatory, so most manufacturers put it in the background at fixed or automated settings. the d0 gives you complete control over that, and thus allows for lots of errors.

most of the time when the d0 was behaving erratic, i could solve it by finding the slew rate sweet spot for the current application. even with audiorate modulation it often helped to introduce a bit of slew. there is also definitely a relation between the correct slew rate and the delay time range.

on a related note, the attenuator settings for the delay time CV inputs are crucial even when there is nothing connected. if you have them turned up a lot (which can easily happen accidentally due to the zoom settings) and the slew very low/off, then you essentially have a random high freq modulation of the delay time, caused by the amplified noise of the input. this causes all kinds of glitches and a higher noise floor.

in the end, i found some of the d0 patches i did were really rewarding, but the playability and a direct interface of my instrument have top priority. i want to see where my knobs are at right away, ESPECIALLY with a digital module and huge control ranges. i guess a coarse/fine arrangement would serve me better. if there a lot of fiddling around and finding sweet spots and double checking generic knobs, than i'd rather do something in Max and create a dedicated interface with Lemur.

there definitely is no alternative out there in regards to audiorate modulation and audiorate delay times, and i was looking for something like the d0 ever since i got the modular. however, as mentioned above, the interface of my instrument is very important to me, so i guess my search continues. i sold the d0 to Navs, and am really looking forward to what he will do with it and what his findings are.

[m-m]

Thanks very much for posting this jnlkrt, especially since you've now sold the module.

on a related note, the attenuator settings for the delay time CV inputs are crucial even when there is nothing connected. if you have them turned up a lot (which can easily happen accidentally due to the zoom settings) and the slew very low/off, then you essentially have a random high freq modulation of the delay time, caused by the amplified noise of the input. this causes all kinds of glitches and a higher noise floor.

This is something I've been wondering about. I always thought the modulation attenuators were normalled to something like 1v so you could use them for fine tuning the delay time. But having them outside 12 o'clock with no slew does, as you say, distort the audio, although the delay time isn't modulated. Where does this high freq modulation come from

[jnlkrt]

Thanks very much for posting this jnlkrt, especially since you've now sold the module.

on a related note, the attenuator settings for the delay time CV inputs are crucial even when there is nothing connected. if you have them turned up a lot (which can easily happen accidentally due to the zoom settings) and the slew very low/off, then you essentially have a random high freq modulation of the delay time, caused by the amplified noise of the input. this causes all kinds of glitches and a higher noise floor.

This is something I've been wondering about. I always thought the modulation attenuators were normalled to something like 1v so you could use them for fine tuning the delay time. But having them outside 12 o'clock with no slew does, as you say, distort the audio, although the delay time isn't modulated. Where does this high freq modulation come from

interesting point, now i remember i read about the offset normalization for the attenuator and i think i even used it, but then john made some remark about amplifying input noise when nothing is connected.. maybe i got it wrong. mhhhh.

[gddfp]

This is something I've been wondering about. I always thought the modulation attenuators were normalled to something like 1v so you could use them for fine tuning the delay time. But having them outside 12 o'clock with no slew does, as you say, distort the audio, although the delay time isn't modulated. Where does this high freq modulation come from

Why would the CV attenuator be normalled to -/+V when there's that nifty Zoom function ?
Also, I don't understand how a non-connected CV input could in any way influence the audio signal, let alone add distortion or noise to the input ? Control voltage & audio circuitry should be completely separated, unless this is not the case with a digital delay...

How about a nicely detailed flowchart of the d0, John ? What goes where, what does what, bipolarity of controls, AC/DC-coupling, clipping levels, min/max values, etc ?
A bit less obfuscation ain't gonna hurt anyone, isn't it ?

And also, I'd like to know if there's a good & quick way of setting the d0 to a kind of default state; a blank slate, as it where ? The best position [of knobs] to start from, including the appropriate fiddling of the Zoom...

cheers,
_g

[jnlkrt]

Also, I don't understand how a non-connected CV input could in any way influence the audio signal, let alone add distortion or noise to the input ?

as i tried to explain, if you amplify the noisefloor of the CV input (i think the CV input "attenuators" of the d0 provide quite some gain actually) then you get a random modulation of the delay time, which is at audiorate if the slew is turned down. this random modulation of the delay time is creating noise and glitches in the audio path. however, i might have gotten this wrong since there is a DC offset normalled to the inputs. dunno. what's important is that if you don't shut those attenuators off completely (hard to do since they are polarizing and obviously no center detent due to zoom) you get all kinds of erratic results.

[gddfp]

Also, I don't understand how a non-connected CV input could in any way influence the audio signal, let alone add distortion or noise to the input ?

as i tried to explain, if you amplify the noisefloor of the CV input (i think the CV input "attenuators" of the d0 provide quite some gain actually) then you get a random modulation of the delay time, which is at audiorate if the slew is turned down. this random modulation of the delay time is creating noise and glitches in the audio path.

I'm sorry, but this goes against everything I know about modular synthesis; unless there's something [odd] with the CV inputs/attenuators I have not been told (by ehm, a users manual, or John).

A CV attenuator, by principle, does not "provide gain", unless it's configured such that the input voltage would allow for "more voltage" than the ideal CV range (which is kinda pointless). In general a CV can extend the range as set by a manual control, but is always limited to the min/max range of the functionality of the device itself.
Unless the CV input is a signal gain control, I fail to see how it can amplify something which isn't there, or at least shouldn't be there. In this case, the CV inputs control delay time (in addition/subtraction to the manual controls, I assume), so when there's nothing connected, it can't/shouldn't amplify a noise floor at all. Unless you're dealing with exceptionally noisy power rails, there really shouldn't be any noise at all at a CV input. If there is, then it's badly designed.

Furthermore, I also fail to see how the Slew rate control has anything to do with the "audio rate" of a CV input. It just slews the raw digital delay signal, and has no direct connection to the delay rate. Correlation yes, but no control.

Contradicting my own [analog] logic: IF there's such a thing as the [non-connected] CV attenuators influencing the delay time -- and I'm not saying there isn't -- I really would like to know where this is coming from. Perhaps this is a side-effect or even a feature of a digital delay unknown to me, or perhaps it isn't. In the former case I'd love to get a detailed explanation, in the latter I'm inclined to think it could be a design fault.

_g

[Yohda]

This is common on digital modules : noise on the input. Even very low amplitude noise can alter the parameters.

[jnlkrt]

1. i pointed out that i am not sure about this..

2. i don't know if the input stage of the CV inputs has just an attenuator or an actual gainstage which provides some gain. i wrote "attenuator" since i don't know if it's just an attenuator

3. the latter might also have been implemented digitally after the conversion, we don't know this.

4. if there is digital gain, you could amplify the converter noise which definitely is always there..

5. even if it's an analog gainstage (given there is one) there might definitely some noise in the input stage which could be amplified.

4. i assumed if you slew a noise signal at a low slew rate then you get a random modulation at a much lower rate.. is this wrong?

5. the slew is applied to the delay rate CV input and offset, not the audio signal, so there is indeed a direct relation.

[gddfp]

This is common on digital modules : noise on the input. Even very low amplitude noise can alter the parameters.

On the Signal Input, I guess so. Maybe. Dunno
But on a CV input too ?

[Yohda]

But on a CV input too ?

Yes. CV, audio that's the same thing (electricity)

[gddfp]

1. i pointed out that i am not sure about this..

I didn't assume you are. That's why I want to know where this is coming from.

2. i don't know if the input stage of the CV inputs has just an attenuator or an actual gainstage which provides some gain. i wrote "attenuator" since i don't know if it's just an attenuator

Well, it's definitely a bipolar attenuator, so it's not just a passive jack-to-pot thing.
How exactly it is implemented, we don't know... And I'd like to know, since I noticed something very odd at those CV inputs (but first need to investigate that further).

3. the latter might also have been implemented digitally after the conversion, we don't know this.

Indeed, we don't. Why ?

4. if there is digital gain, you could amplify the converter noise which definitely is always there..

That would mean the CV input is converted to the digital domain before it's attenuated, isn't it ?

5. even if it's an analog gainstage (given there is one) there might definitely some noise in the input stage which could be amplified.

Are we talking about the same thing here ? I was under the impression that you're talking about the CV inputs, not the signal inputs.

4. i assumed if you slew a noise signal at a low slew rate then you get a random modulation at a much lower rate.. is this wrong?

More or less not. ;-) It depends on the spectrum of the noise. If it's pure white noise, slewing that is not going to have much effect, except perhaps attenuating it completely down. Most [analog] slew processors aren't fast enough to "follow" an infinite frequency signal.

5. the slew is applied to the delay rate CV input and offset, not the audio signal, so there is indeed a direct relation.

\
I hope it is not ! If you'd slew the CV input, one couldn't get "immediate voltage changes", and the CV signal would be completely altered. That surely can't be meant that way.
As far as I understand, the slew rate slews the delayed signal.

Then again, I could be completely wrong. The d0 is still a bit of mystery to me, and I need more quality time with it.
Until very recently, anything digital within my analog systems was a bit of a nono for me. Call me old-fashioned.

[jnlkrt]

4. if there is digital gain, you could amplify the converter noise which definitely is always there..

That would mean the CV input is converted to the digital domain before it's attenuated, isn't it ?

yeah, this is very possible i think, considering the input "attenuators" are controlled by the zoom. would be pretty inefficient and expensive to convert the digital zoom control values to control a real analogue attenuator. the more i think about all the issues, it makes more and more sense to me that it's all digital.

5. even if it's an analog gainstage (given there is one) there might definitely some noise in the input stage which could be amplified.

Are we talking about the same thing here ? I was under the impression that you're talking about the CV inputs, not the signal inputs.

yes, the CV inputs. there could very well some gain on the CV inputs, why not? very possible since John likes giant control ranges.

4. i assumed if you slew a noise signal at a low slew rate then you get a random modulation at a much lower rate.. is this wrong?

More or less not. ;-) It depends on the spectrum of the noise. If it's pure white noise, slewing that is not going to have much effect, except perhaps attenuating it completely down. Most [analog] slew processors aren't fast enough to "follow" an infinite frequency signal.

yeah, that makes sense.

5. the slew is applied to the delay rate CV input and offset, not the audio signal, so there is indeed a direct relation.

I hope it is not ! If you'd slew the CV input, one couldn't get "immediate voltage changes", and the CV signal would be completely altered. That surely can't be meant that way.
As far as I understand, the slew rate slews the delayed signal.

no the slew is definitely on the CV inputs and also the delay time offset controls. basically on everything which controls the delay time, both on the panel as well as external CV inputs. that's for sure.

you can obviously not slew at all by fully turning it down.

however, as mentioned earlier, you need to apply some slew to the control signals in almost every application since if you allow full too quick and abrupt changes of the delay buffer you'll just get glitches etc..

[jjclark]

I haven't played with the Mungo D0, but I can maybe help you understand glitching caused by noise on delay time control signals.

Suppose you had a 1-second delay, and feed a 1KHz 10Vp-p triangle wave through it. The slope of the triangle waveform is +/- 20V/msec.
Now suppose you modulate the delay time with a signal that gives a full-scale variation (so the delay time can be modulated from 0 to 1000 msec). If there is a noise step in the modulation signal of 1/2000th (1/20 %) of the whole range there will be a 0.5msec change in the delay time. Doesn't sound like much, but on a 1KHz triangle it can result in an output change (glitch) of 10V!
Even if the noise step is 1/2,000,000 of the whole range there will be a glitch of 1/10V on the output, which would be clearly audible. In the digital world 1/2,000,000 is 3 bits of noise in a 24 bit signal. A Digital-to-analog converter can easily produce this sort of noise level.

The way to avoid this is to slew the delay time control signal so that glitches turn into frequency shifts instead of step amplitude changes. Some may find the frequency shifting caused by the slew objectionable, but the alternative is a noisy output.

[jnlkrt]

[...] The way to avoid this is to slew the delay time control signal so that glitches turn into frequency shifts instead of step amplitude changes. Some may find the frequency shifting caused by the slew objectionable, but the alternative is a noisy output.

thanks for the background details! this backs my initial statement that all digital delays with CV control have some slew applied to the control signals. is the slew adaptive or fixed in common designs?

[gddfp]

I haven't played with the Mungo D0, but I can maybe help you understand glitching caused by noise on delay time control signals.

Suppose you had a 1-second delay, and feed a 1KHz 10Vp-p triangle wave through it. The slope of the triangle waveform is +/- 20V/msec.
Now suppose you modulate the delay time with a signal that gives a full-scale variation (so the delay time can be modulated from 0 to 1000 msec). If there is a noise step in the modulation signal of 1/2000th (1/20 %) of the whole range there will be a 0.5msec change in the delay time. Doesn't sound like much, but on a 1KHz triangle it can result in an output change (glitch) of 10V!
Even if the noise step is 1/2,000,000 of the whole range there will be a glitch of 1/10V on the output, which would be clearly audible. In the digital world 1/2,000,000 is 3 bits of noise in a 24 bit signal. A Digital-to-analog converter can easily produce this sort of noise level.

The way to avoid this is to slew the delay time control signal so that glitches turn into frequency shifts instead of step amplitude changes. Some may find the frequency shifting caused by the slew objectionable, but the alternative is a noisy output.

All clear now. Thanks for the explanation !

_g

[Navs]

As promised, here are some sounds and thoughts on the d0:

http://navsmodularlab.blogspot.de/2013/ ... -demo.html

When unpatched, both the CV inputs and Z outs carry ca. 3V & +/-500mV respectively. No idea why they are there (although I guess it's to do with the Zoom and the way control is extrapolated) but if you don't eliminate these by nulling the pot or plugging in a dummy jack you will get distortion. Stands to reason, especially if you have feedback patched for multi-echos or Karplus.

The d0 omits the filter often found in the audio path of delays, allowing you to patch one in or not. The Cwejman MMF-1 is good as a transparent correctional tool, the Plan B M12 is instant vactrol dub.

The slew affects the control input, not the audio output. You can verify this by nulling the slew (fully CW) and running your CV through an external slew. This is a cool patch anyway.

[boramx]

^those sounds are mighty sweet.

[jjclark]

[...] The way to avoid this is to slew the delay time control signal so that glitches turn into frequency shifts instead of step amplitude changes. Some may find the frequency shifting caused by the slew objectionable, but the alternative is a noisy output.

thanks for the background details! this backs my initial statement that all digital delays with CV control have some slew applied to the control signals. is the slew adaptive or fixed in common designs?

I should point out that this analysis is for the case where you implement delay time changes by changing the position of the delay line read pointer relative to the write pointer. Another way to change the delay is to change the rate at which the reads and writes are being done (as is done in the Modcan dual delay module in resampling mode). This doesn't create any glitches since all of the delay line samples are still being read out in order, with no jumps. The drawbacks of this approach are that the pitch of the material still in the delay line gets shifted, and the sampling rate can get lower, and hence lower quality. The resampling causes a filtering of the signal resulting in a muffled sound. Also, the range of delay modulation is limited since you really don't want to drop the sampling rate too low, and will have a limit on the high end of the sampling rate.

The Modcan delay allows you to select between the two approaches to obtaining delay modulation, with the problems and advantages of each.

[digidandy]

As promised, here are some sounds and thoughts on the d0:

http://navsmodularlab.blogspot.de/2013/ ... -demo.html

Whatever's been said about the peculiarities of the mungo series (I own the g0 myself): That is a mighty impressive demo!

[m-m]

as i tried to explain, if you amplify the noisefloor of the CV input (i think the CV input "attenuators" of the d0 provide quite some gain actually) then you get a random modulation of the delay time, which is at audiorate if the slew is turned down. this random modulation of the delay time is creating noise and glitches in the audio path. however, i might have gotten this wrong since there is a DC offset normalled to the inputs. dunno. what's important is that if you don't shut those attenuators off completely (hard to do since they are polarizing and obviously no center detent due to zoom) you get all kinds of erratic results.

Yes of course, there are A/D converters on the CV inputs as well. So the digital noise floor could get amplified and then start to modulate the delay time. It could well explain what I have experienced.

Very cool of you to keep posting in this thread

And what an excellent demo Navs

[Navs]

Cheers, guys!

[digidandy]

Has anyone heard anything more about the preset expander mentioned earlier in this thread? Is it getting nearer release?

[Boddhisattva007]

Cheers, guys!

Hi Navs, do you plan on possibly doing a g0 demo as well?

[Navs]

Hi Navs, do you plan on possibly doing a g0 demo as well?

Don't have it, but I'll do the entire range of Mungo modules if John wants to send me them

[bendedavis]

Has anyone heard anything more about the preset expander mentioned earlier in this thread? Is it getting nearer release?

For anyone who hasn't seen my other post here, this module is very close to being finished. I'll have the PCBs next week.

viewtopic.php?t=100402&highlight=