Modular Digitally Controlled Analog Console Format

[Grummy77]

Hey all. I'm not sure if or how much of a discussion on this has happened at all here, but a quick search it didn't come up.

Basically the idea of starting a format (potentially based off of 500/51x) that incorporates a broad system for digital control of analog gear. There's a new thread on GS right now with some cool ideas being thrown around. Here'a a link to that thread: http://www.gearslutz.com/board/so-much-gear-so-little-time/867829-chat-about-future-digitally-controlled-analog-console.html#post9407029

Ideas (not mine) that are appealing are a 52L series. One that incorporates bacwards compatibility to 500 & 51x but adds another four pins 19,20,21,22 but utilizing both sides of said pins for a total of 8 more connections. Making possible a Cat5 or Cat6 connection. The next step would be implementing the software to go along. Sort of like the Bettermaker 502 EQ but on a larger more open format scale. HUI was mentioned as well.

I'd be curious what some of the smarties like Jeff Steiger would think of stuff like this. Since he's already well on his way to designing a 51x modular console already. I would imagine some good ideas could get brainstormed on here. Pointing out obstacles and then hopefully solutions to those obstacles. Think how sweet it would be to use all or most of the kind of analog hardware found in 500 & 51x but all controlled like plugins for speedy recall.

Seems like the technology is out there to make this happen at a reasonable cost.

Cheers.

 

[Echo North]

I'm growing more and more obsessed with digitally controlled analog.

Great discussion.  Wish I had something to add other than I think about it everyday

 

[Brian Roth]

I, too became quite interested in digitally controlled analog maybe 20 years ago (egads!) when I saw a demo of the then-new Euphonix console system at an AES show.

Bri

 

[Andy Peters]

I suppose there are several ways to approach this.  It depends on how much intelligence you want to embed in each module.

One way to approach it would be to have a master slot (along with the power supply) which has the interface to the DAW. The DAW interface can be whatever you like: Ethernet, USB, etc. The master slot will have a master processor which is the bridge between the DAW or other controller and the modules in the chassis.

There's a 3.4 Mbps variant of I2C which might be fine for connecting the modules in the chassis to the master processor. I2C makes sense for a few reasons. One is that it's only two wires (plus ground). Another is that the logic on the individual modules can be reduced to a PCF8574 or a PCA9555 I/O expander for the simple stuff. Also most of the small micros which support I2C can act as bus slaves, so your master can send messages to the modules and then each module's processor handles the heavy lifting. Maybe you need something simple to just set some relays or switches. Maybe you need to control several MDACs or a preamp using SPI. Choose the amount of module intelligence needed.

Also I2C allows for a 10-bit bus address, which should be enough to cover any reasonable amount of slaves. Or, one could choose a controller processor which has more than one I2C bus. The slaves don't care.

So that's the interface between modules and the controller. The hardware is easy. As usual, it's the software and firmware which will consume all of the design time. Ideally, the end user shouldn't have to do any microcontroller programming at all -- he just plugs the module in to the chassis, and the controller processor recognizes that it's there and enumerates it. (OK, that implies hot plug capability, which maybe we don't want to support yet.)

This is taking a page from USB. Perhaps one needs to define device "classes" such as "preamp" and "EQ" and "compressor," each of which implements particular controls, and the module designer can choose whether or not to implement any particular control. If a module is a "channel strip," which includes preamp and EQ, perhaps what's needed is a "composite" device (again, stolen from USB) which looks like both types to the controller.

The system is powered up, its processor scans the slots to see what's out there (each module responds to a "what are you?" query), and then it waits for host communications. The DAW (or other external control system) connects to the system using whatever interface (Ethernet, USB, etc) and queries it to find out what's in the box. Now the DAW "knows" what is supported.

Then it's a matter of the DAW sending standard messages to the controller, the controller interpreting them and sending messages to the correct module.

Simple, right?

-a

PS: I'd include a 3.3 V digital supply rail and return to the backplane connector. This will let you power the logic on each module without requiring a regulator on each. Yes, 3.3 V and not 5 V, because nobody uses 5 V any more.

 

[KDE]

Posted my ideas on gearslutz. Take a look at the Neve Genesis console as it has digitally controlled 1084's on each channel.

 

[Andy Peters]

Posted my ideas on gearslutz. Take a look at the Neve Genesis console as it has digitally controlled 1084's on each channel.

Who's got time for Gearslutz?

-a

 

[joaquins]

It would be great! I'm thinking about build a digital controled faders with 2 vactrol, one for the audio and one for feedback, it would allow me to manage the 12 I want with my arduino mega, I already make one channel proto and get nice results, I still have to change a couple of things (log converter at output and drive led with current and no with voltage) in order to improve usuable range.

For me the idea was around my head since I start building guitar pedals... MCP41XXX kind seems usefull but I don't know they U/I limitations and audio apps limitations... I guess at least should be usefull to set non signal chain parameters (ratio, treshold, VCA gains, etc) and this'd be nice to use soft-automation.

The other option is to have some automatic controls and others that are analog controled but digitaly sensed in order to have an internal sense to recall with precission. I think ssl used a system ly this somewhere but I can't remember...

JS

 

[JohnRoberts]

Digitally controlled analog has been around for decades. I recently worked on a project (for another company) that they did not finish, but i had a functional working prototype. This product was an automatic mixer, a type of mixer that automatically adjusts the gain of multiple inputs based on how many people are talking and how loud. This is a mature product category but the technology is usually executed fully in the analog domain or completely digital. I did a mixed technology design where digital was used to do what it does well (math) and the audio path remained fully analog with the gain controlled by DPOTs.  When I laid out my prototype I populated it with both DPOTs and  VCAs fed by DACs for gain control. The DPOTs worked so well I never fired up the VCAs (i was surprised since I figured the VCAs were lower risk, just more expensive. The DPOTs surprised me by working better than I expected for this application)..

I used SPI to talk to the DPOTs but that's 6 of one half dozen another compared to I2S many parts are available in either.

I have some limited experience in the analog side of this digital controlled analog. I do not have the skills required to come up with a computer interface. My thinking is something like hosting an IP stack so any web browser can interface with and control the analog hardware... I have seen micro's with the IP stack, but haven't looked into it much beyond that.

Of course it makes sense to interface with something like pro tools(?).

I guess one question is what is the desired result? Equivalent to analog path but with digital control?  Modern mic preamps with digital gain control make that part easy. The rest is not rocket science but not trivial, and potentially expensive.

My digitally controlled analog automatic mixer was cost effective because even an all analog design required VCAs and multiple log conversions.  The comparison between a  conventional analog console and digitally controlled analog console will not be as attractive, and a digital console minus the control surface is getting cheaper every day.

JR

 

[JPK]

I think the big problem with digital is the ultra fast obsolescence. If you build a digitally controlled analog device today, 10 years later, the analog part will still be working perfect but the digital part won't communicate anymore with the future computers: new connectors, new standards, new protocols, ...

In 2001, I bought a Sony DMX R100 digital mixer and a Manley Massive Passive EQ. Today, 12 years later, I could sell the Manley with very little loss. The Sony has lost 90% of its value (the automation was stored in floppy disks!!!). I think this is a good reason why many of us prefer 100% analog devices.

 

[Andy Peters]

I think the big problem with digital is the ultra fast obsolescence. If you build a digitally controlled analog device today, 10 years later, the analog part will still be working perfect but the digital part won't communicate anymore with the future computers: new connectors, new standards, new protocols, ...

In 2001, I bought a Sony DMX R100 digital mixer and a Manley Massive Passive EQ. Today, 12 years later, I could sell the Manley with very little loss. The Sony has lost 90% of its value (the automation was stored in floppy disks!!!). I think this is a good reason why many of us prefer 100% analog devices.

Yes, storage media obsolescence seems to be the biggest problem with anything digital. Hey, who's got a Neve with the Necam automation that ran on 8" floppies?

But other than storage, I think that a lot of digital standards have had remarkable longevity. MIDI is one excellent example. USB evolves but remains backwards compatible with the earliest versions. Ethernet is pervasive. Newer WiFi standards still support the older standards. Even RS232 still lives, as you can buy a USB-to-serial dongle.

Certainly there are a few standards that have gone away -- the old PC-AT slot format, IDE and SCSI for hard drives, Apple Desktop Bus, PS/2 mouse and keyboard.

OK, you can't win.

Now what about those old pieces of analog kit which require a 600-ohm termination in this day and age of bridging interfaces?

-a

 

[Talljoe]

There hasn't been much talk about data coming the other way. I would also like to see some amount of feedback in the protocol. For instance, meters on the 1176. It's unlikely that I'll place a rack of knobless gear within arms reach, so having meters come back would be useful.  And of course need the ability to read the current values and states.  The rack could do in/out metering via a/d automatically.

Going deeper to implementation details with a 51x, would we want to include, say, a +3.3v rail?  I think it's a good idea but it will also complicate the 51x power supply by adding a 6th voltage standard. Would we require the +/-24v rail since the pins would exist?

An interesting possibility would be a self-contained 51xD rack with a small screen that allows you to set the values of any module in the rack without needing a computer (inspiration from the RME UFX). Of course, for this I'm envisioning a bunch of racks without a tactile interface (which would make things much cheaper; I'm looking at you, Grayhill switches).

(edit) P.S. I'm not sure a discussion about whether or not this technology should be built is relevant, unless you're afraid it will supplant all existing analog gear and bring us close to SkyNet taking over the world.

 

[Andy Peters]

There hasn't been much talk about data coming the other way. I would also like to see some amount of feedback in the protocol. For instance, meters on the 1176. It's unlikely that I'll place a rack of knobless gear within arms reach, so having meters come back would be useful.  And of course need the ability to read the current values and states.  The rack could do in/out metering via a/d automatically.

I2C is a bidirectional bus, so implementing feedback is possible.

Going deeper to implementation details with a 51x, would we want to include, say, a +3.3v rail?

That's what I suggested.

An interesting possibility would be a self-contained 51xD rack with a small screen that allows you to set the values of any module in the rack without needing a computer (inspiration from the RME UFX). Of course, for this I'm envisioning a bunch of racks without a tactile interface (which would make things much cheaper; I'm looking at you, Grayhill switches).

The "master controller" could have a user interface, in addition to or instead of a host control interface. Obviously if the idea is to have the unit be a "hardware plug-in" for a DAW then it needs the host interface.

-a

 

[joaquins]

The compatibility thing doesn't seem a problem, if MIDI, enthernet and USB are aplied on the first model we should be nice for many years and if something change and all MIDI USB and Ethernet dissapear it's just adding another microcontroler that manage connection between old and new standard.

I did talk about feedback, not for meters that would be nice but if not everithing digital controled but digital sensed to total recall. If digital pots are good enoght doesn't seem needed to use this aproach.

JR, how good do they are (DPOTs) to use in audio path? or VCA is absolutley needed to a project like this?

My vactrol with feedback was smooth and transparent but needs a ADC and a DAC (PWM) and 2 vactrol cells for each attenuator (8 bits may be enoght but the converter is needed) The comparator could be analog so no ADC needed... But I didn't experiment with it because I has all ADC aviable then and not used for anything else...

I think a control surface could be future added there to stand alone or DAW control but could be totally independent and much smaller too to have in the desk and the mixer it self in the other room.

JS

 

[JohnRoberts]

JR, how good do they are (DPOTs) to use in audio path? or VCA is absolutley needed to a project like this?

I was pleasantly surprised by how good the modern DPOTs worked in my application but for the record, this was just a one off prototype that never made it to a serious test bench for specifications**.

In my judgement (not proved) DPOTs can deliver better specs than VCAs, while VCAs will have an edge for continuos adjustability, most noticeable at deep attenuation where DPOTs suffer from finite number of discrete steps.  For my application (an automatic mixer) the important gain range was 0dB to -15dB, so again DPOTs were just fine. Signals already attenuated 15dB below louder signals were not going to be bothered by step size of gain tweaks.

My vactrol with feedback was smooth and transparent but needs a ADC and a DAC (PWM) and 2 vactrol cells for each attenuator (8 bits may be enoght but the converter is needed) The comparator could be analog so no ADC needed... But I didn't experiment with it because I has all ADC aviable then and not used for anything else...

Micros with 8 or more PWM are relatively cheap. Good enough A/D  for a control path are also cheap up to 12 bit. While the complexity discourages me. Back in the 70s I did some bench work looking for alternatives to VCAs (that were not as good as today) And i was able to linearize the gain of dual JFETs or center-tapped LDRs but overall they did not work well enough, in my judgement to make them practical substitutes for the better VCAs back then.

I think a control surface could be future added there to stand alone or DAW control but could be totally independent and much smaller too to have in the desk and the mixer it self in the other room.

JS

I have ideas for several dig controlled analog products but lack a handy interface. One old pet project I'd like to make a digitally controlled analog compressor, with tens of controls to make it extremely flexible and capable of mimicking other popular comps. Only practical with a computer interface (that I do not have worked out).  One concern about my uber-compressor is that without a user friendly interface, all the adjustability is a liability instead of a benefit. 100 ways to make it sound bad. 

JR

** re DPOT performance  the first popular criticism is signal swing, but this is silly in the context of A/D convertor that can only handle single digit peak voltage.

The second concern is the internal switches but this is similar technology to the digitally controlled mic preamps from a couple vendors that spec very good distortion numbers. I also have ideas for alternate circuit topology to mitigate any distortion caused by these internal switches, but i do not have a good enough test bench to explore this. 

A third concern is step size.. You can pay up for more bits but the steps are typically linear so always jumpy at bottom end. IMO this is not important for many applications, for ones that are, two DPOTs can be cascaded to increase the apparent resolution.

PS: I also have ideas for how to squeak more performance out of VCAs but I hear more people chasing after old VCAs because the want the old dirty path, than people complaining that the new ones are not clean enough.

 

[joaquins]

...

PS: I also have ideas for how to squeak more performance out of VCAs but I hear more people chasing after old VCAs because the want the old dirty path, than people complaining that the new ones are not clean enough.

I've seen a couple of the lasts, in case it means something...

What you are saying is the VCAs still be limited step but it's not a problem because dB/V scale, right?
What about parallel resistor with the DPOTs like the old taper trick to make it log?

I've used 2 LDR and led when I tested my circuit because I didn't has real vactrols near, and the tracking was really good, I don't remember now but in the time I measure tracking between both cells and have nice results... I did have trouble with lower steps though I think that with some log conversion maybe one diode and one resistor is enoght and driving the led with a controled current source should work, I was driving the led from the PWM output directly and between off state and minimun was a huge difference, last 10/15dB was great.

JS

 

[Andy Peters]

What you are saying is the VCAs still be limited step but it's not a problem because dB/V scale, right?

No ... the VCA control port is an analog input, so there are no "steps." Now if you drive it from a DAC, then, yes, you add steps to the control voltage. But a 12-bit DAC gives you 4096 steps, much more than the usual 128 taps on a DPOT. (Yes, you can buy DPOTs with 256 or 512 or 1024 steps, for more $$$, of course.)

Whether you need the finer resolution is application dependent.

One argument against DPOTs is their tolerance is often 20%. When used potentiometrically, that's perhaps not an issue. But if you're trying to set a particular gain or you need to target a particular resistance (for a filter, perhaps) across multiple units, they can't really do it. A VCA driven by a DAC with a solid reference will do much better.

I suppose you could use dual DPOTs and use one for feedback and assume that the two in the package are well matched, but now you need an ADC and a mux (or multiple ADCs) and some scanning mechanism, and now your design has gotten really complex.

-a

 

[JohnRoberts]

...

PS: I also have ideas for how to squeak more performance out of VCAs but I hear more people chasing after old VCAs because the want the old dirty path, than people complaining that the new ones are not clean enough.

I've seen a couple of the lasts, in case it means something...

What you are saying is the VCAs still be limited step but it's not a problem because dB/V scale, right?
What about parallel resistor with the DPOTs like the old taper trick to make it log?

I haven't thought about this much, but from a quick pass it seem like the parallel shunt resistor would help reduce step size at the bottom, but make the step sizes at the top more pronounced.

I've used 2 LDR and led when I tested my circuit because I didn't has real vactrols near, and the tracking was really good, I don't remember now but in the time I measure tracking between both cells and have nice results... I did have trouble with lower steps though I think that with some log conversion maybe one diode and one resistor is enoght and driving the led with a controled current source should work, I was driving the led from the PWM output directly and between off state and minimun was a huge difference, last 10/15dB was great.

JS

Do what floats your boat...  I didn't like my bench results a few decades ago, but that was a few decades ago and the alternatives since then have gotten better.

JR

 

[JohnRoberts]

What you are saying is the VCAs still be limited step but it's not a problem because dB/V scale, right?

No ... the VCA control port is an analog input, so there are no "steps." Now if you drive it from a DAC, then, yes, you add steps to the control voltage. But a 12-bit DAC gives you 4096 steps, much more than the usual 128 taps on a DPOT. (Yes, you can buy DPOTs with 256 or 512 or 1024 steps, for more $$$, of course.)

and the DAC output can be smoothed by an RC so indeed gain changes are continuous. The DPOT will always be discrete steps. That said there are techniques to coordinate these step changes with audio zero crossings to reduce impact since any step times 0V makes a 0v perturbation.

Whether you need the finer resolution is application dependent.

One argument against DPOTs is their tolerance is often 20%. When used potentiometrically, that's perhaps not an issue. But if you're trying to set a particular gain or you need to target a particular resistance (for a filter, perhaps) across multiple units, they can't really do it. A VCA driven by a DAC with a solid reference will do much better.

I suppose you could use dual DPOTs and use one for feedback and assume that the two in the package are well matched, but now you need an ADC and a mux (or multiple ADCs) and some scanning mechanism, and now your design has gotten really complex.

-a

Yup, I scratched out some high performance designs and I ended up using way too many DPOTs and opamps per simple circuit block. The products would be cool, but VERY expensive. I've never been good at selling the premium fluff... I'm more of a bang for the buck guy. 

The elephant in the room is WHY?  It is pretty hard to make DPOT and/or VCA control as good as a simple $0.50 pot, let alone better... And digital audio paths are not as flawed as people think. Give me a 16 x 24b multiply and I'm happy.

YMMV

JR

 

[joaquins]

You are right abot VCAs...

The price side, that's why I said to use sensing with double pot, one wiper to use in the circuit and the other to digital sense and store the value, then you have to turn it till the light is green and you are in the same spot you left it the last time you mix was beeing made.

JS

 

[Rochey]

From a control interface perspective may i suggest midi? All our DAW's already support it. Recallwould bea snap.