Mdacs and panning

[JAY X]

Hi Friends!

From a time to now i discovered that some famous mixing desks use mdacs for volume adjustement. I found some mdacs models on the net like Burr Brown pga 2310-11-20.

They are intended for volume control. But, ¿is it possible to use them for panning? that would be interesting.(Although expensive)

I found some THAT application notes that uses vca's for panning. In terms of audio performance  ¿which is the best solution? ¿vca's or mdacs?

thanks for your advise!

JAY X

 

[abbey road d enfer]

Hi Friends!

From a time to now i discovered that some famous mixing desks use mdacs for volume adjustement. I found some mdacs models on the net like Burr Brown pga 2310-11-20.

They are intended for volume control. But, ¿is it possible to use them for panning? that would be interesting.(Although expensive)

I found some THAT application notes that uses vca's for panning. In terms of audio performance  ¿which is the best solution? ¿vca's or mdacs?

thanks for your advise!

JAY X

I haven't tried the pga2310, but I have used the Dallas (now Maxim) 1882, which uses a very similar technology. In terms of performance, they can hardly be faulted, but in terms of complexity, it's another story; first, you need a microcontroller of sorts, and then you have to be very careful, as always in hybrid designs, of the interactions between digital and analog signals, grounding, decoupling, and so on. VCA's are marginally inferior in terms of S/N and distortion, and also require a not-so-trivial amount of glue to function.

 

[JAY X]

Thanks abbey road!

I think if i will take the mdac road...

Pga 2310 mdacs have 2 channels that can be controlled independently. But, what about panning? Do you think is possible to build a pan circuit around those ic's ?

Maybe with (among other components) a dual pot and reverse wiring? Increasing volume in one channel while decreasing on the other channel. ¡that's the basics of panning!

thanks again Abbey road!

JAY X

 

[abbey road d enfer]

Thanks abbey road!

I think if i will take the mdac road...

Pga 2310 mdacs have 2 channels that can be controlled independently. But, what about panning? Do you think is possible to build a pan circuit around those ic's ?

Maybe with (among other components) a dual pot and reverse wiring? Increasing volume in one channel while decreasing on the other channel. ¡that's the basics of panning!

thanks again Abbey road!

JAY X

The PGA2310 has independant controls for left and right channels, so clearly, using it with a common mono input signal makes it a pan-pot.

 

[TomWaterman]

They are only 0.5dB steps if I recall, so while will get close to some very useful pan curves, I think you'll struggle to hit true sin/cosine behaviour. However it'll probs work fine and you can program the micro to ignore  certain steps to create a 'taper' of sorts.

I looked into this pretty heavily with these parts last year.

SSL boast something like 0.03 or 0.003 dB accuracy on their mdac for faders in Duality.

 

[JohnRoberts]

It depends on why? In general MDACs or VCA will not beat a well engineered conventional PAN circuit for raw performance, especially if custom pot tapers are tooled up.

MDCs and VCAs do lend themselves to grouping, automation, and digital control.

I am currently experimenting with using a pair of VCAs to perform fader and pan functions so I can combine the raw output currents in a summing bus structure with unity noise gain.

I don't know that this is very practical but I'm mainly trying to amuse myself.

JR

 

[TomWaterman]

If you can live with 11 positions, you can make are really nice pan circuit with a switch controlling a pair of attenuators.

I just calc'd one up over xmas for a little project (or rather large project) I'm working on. I remember seeing an example here a while ago.

I got resistor values for -3dB centre tru sin/cosine match but then tweaked it to -4dB centre with a slight s-taper... trying to hit the middle of the -3, and -6 argument.

It sound be pretty good sounding when done. You could always switch the steps with MUXs... I wonder if the DG408 would sound good there?
-T

 

[TomWaterman]

BTW John, reading your stuff on the VCA is inspiring. Great work!!!

Excited to see what happens.
-Tom

 

[JohnRoberts]

If you can live with 11 positions, you can make are really nice pan circuit with a switch controlling a pair of attenuators.

I just calc'd one up over xmas for a little project (or rather large project) I'm working on. I remember seeing an example here a while ago.

I got resistor values for -3dB centre tru sin/cosine match but then tweaked it to -4dB centre with a slight s-taper... trying to hit the middle of the -3, and -6 argument.

It sound be pretty good sounding when done. You could always switch the steps with MUXs... I wonder if the DG408 would sound good there?
-T

If using a multi position switch to execute the pan, consider instead of using it potentiometrically (like a variable pad), use it to ratio up the resistor value feeding the mix bus. This way as you turn down the channel, you also reduce the noise gain the bus combining amp is operating at.

I figured out a long time ago, noise in the bus amp is not the only issue. Phase shift and distortion is also improved by reducing the noise gain.

The only downside to varying that resistance is it will complicate the math to properly differential that bus, but there are many consoles out there that don't bother, and minimizing the bus noise gain will minimize the impact of local ground potential differences.  Perhaps a forward differential stage to reference the channel output to the bus ground before the switch would be more practical than doubling the number of switch wafers and resistors to make it fully differential. This could be incorporated into a channel fader gain stage, or perhaps that fader too could be a switched resistance in series with the pan switches.

I bet that could make a very respectable performer, if you don't mind limited resolution.

JR 

 

[Rochey]

just clarification... the PGA2310/11/20 family are not mdacs.

They are digitally controlled amplifiers. Varying value resistors are changed in the path of an amplifier.

 

[TomWaterman]

Good point Rochey!

If using a multi position switch to execute the pan, consider instead of using it potentiometrically (like a variable pad), use it to ratio up the resistor value feeding the mix bus. This way as you turn down the channel, you also reduce the noise gain the bus combining amp is operating at.

True enough and good points John but I'm not planning on using a virtual earth summing stage...

My design is using a passive transformer balanced mix buss with buffering after the pan. 5 stereo mix busses are fed from impedance balanced channel outputs. Buffering the outs after the pan to ensure low source Z to buss to improve x-talk among other things.

-T

 

[JohnRoberts]

If you combine the signals with less insertion loss there is less make up gain needed after the summing bus to get back up to nominal levels.

Less gain is genally a good thing for circuit performance, but there's always multiple ways to remove the skin from the feline. 

JR

PS: It seems digital pots might be similarly applied. Need to think about that. +1

 

[juniorhifikit]

It depends on why? In general MDACs or VCA will not beat a well engineered conventional PAN circuit for raw performance, especially if custom pot tapers are tooled up.

MDCs and VCAs do lend themselves to grouping, automation, and digital control.

I am currently experimenting with using a pair of VCAs to perform fader and pan functions so I can combine the raw output currents in a summing bus structure with unity noise gain.

I don't know that this is very practical but I'm mainly trying to amuse myself.

JR

Did you ever come to any conclusions while fiddling with this concept?  I'm kind of playing with the same thing.  What are SSL using in the Duality I wonder?

 

[JohnRoberts]

It depends on why? In general MDACs or VCA will not beat a well engineered conventional PAN circuit for raw performance, especially if custom pot tapers are tooled up.

MDCs and VCAs do lend themselves to grouping, automation, and digital control.

I am currently experimenting with using a pair of VCAs to perform fader and pan functions so I can combine the raw output currents in a summing bus structure with unity noise gain.

I don't know that this is very practical but I'm mainly trying to amuse myself.

JR

Did you ever come to any conclusions while fiddling with this concept?  I'm kind of playing with the same thing.  What are SSL using in the Duality I wonder?

So far this is just a mental exercise. I have real work I must do first. I don't expect any VCA based solution to be audibly superior to say a 11 position pan switch (properly executed), or custom taper dual pots, that could reduce the sum bus noise gain to effect the pan law.  Where I believe my current summing approach has utility is if VCA fader automation is already in use.

From a practical cost consideration, real VCA consoles usually perform the fader moves with a single VCA per channel then apply a conventional pan pot and circuitry after. Back in the '80's I used per channel current sources to reduce sum bus noise gain, I still believe using VCA current outputs directly could deliver a similar benefit, very likely even more benefit, but with a cost complexity increase (while there are a few practical details to work out). In the real world of console design doubling the number of VCAs and increasing the surrounding complexity is not undertaken lightly. 

I haven't melted any solder over this (yet) but have spoken with a friendly console company, so who knows. I suspect the days of analog consoles and even digitally controlled analog consoles are numbered, but so are mine. Current source summing is one of my design preferences but everything has pros and cons.

JR

 

[juniorhifikit]

Does anyone know of a 10bit Mdac (like SSL uses these days)?  The PGA 2310 is only 8 bit.

 

[jdbakker]

Does anyone know of a 10bit Mdac (like SSL uses these days)?  The PGA 2310 is only 8 bit.

You're comparing apples and oranges. A 10-bit MDAC's control range is linear and spans ~60dB (plus 'off'), the PGA has a logarithmic control range spanning 120dB. See this for more.

JDB.

 

[juniorhifikit]

Yes, you're right - sorry.  SSL claims 10bit (1024 step) resolution from the fader to their DAW layer in HUI mode.

What I'm interested to find out is:  What is SSL using in their current consoles.  They claim they use "high resolution Mdac's" in their new consoles, as opposed to the VCA's which they once used (first dbx then That).  I can't find any component that has anywhere near the resolution that would be acceptable for console volume automation.

Anybody have service doc's for an AWS900, Matrix, or Duality?  ;D

 

[JohnRoberts]

I recall 8 bit(?) mdacs that specifically targeted volume controls with nonlinear laws. I haven't kept abreast of this niche so don't know what the current SOTA is. It seems there are lots of cheap linear 8-9 bit digital pots.

I kind of lean toward micro controlled VCAs (for additional reasons) but a good audio taper digital pot could be useful.

Maxim makes some 64 step http://www.maxim-ic.com/quick_view2.cfm/qv_pk/4661

I also found a 100 step LDPOT but no audio specs.


JR

 

[juniorhifikit]

I recall 8 bit(?) mdacs that specifically targeted volume controls with nonlinear laws. I haven't kept abreast of this niche so don't know what the current SOTA is. It seems there are lots of cheap linear 8-9 bit digital pots.

Yeah, I've browsed more than a few of them.  The best resolution I've found so far is 0.5dB steps

I kind of lean toward micro controlled VCAs (for additional reasons)

The "roll your own" approach with a VCA and a DAC of your choosing is the only way I'm familiar with.  If there is now a monolithic solution, as the SSL brochure leads me to believe, I'd be interested to see it.


**edit**

at a glance, the Intersil X911x line of "digital controlled pots" states 1024 steps...

 

[JohnRoberts]

[

The "roll your own" approach with a VCA and a DAC of your choosing is the only way I'm familiar with.  If there is now a monolithic solution, as the SSL brochure leads me to believe, I'd be interested to see it.


**edit**

at a glance, the Intersil X911x line of "digital controlled pots" states 1024 steps...

Linear...

JR