OTB Mixer Using Current Summation

[mediatechnology]

John - I found the article I was thinking about: "Digital vs. Analog Summing" ProAudio Review July, 2007.

I can't find it on their website, the search function there seems very, very slow. Google can't find it either. I can send you a scan of it. I agree that the kit business isn't going to fund retirement. But, it would make a nice open source project. Don't know how that would help you though.

[JR.]

I am willing to lend my engineering support to a proper current source summer. My last 24 bus efforts at Peavey demonstrated merit with some 120 feeds to L/R bus.

I think the use of 2x VCA for channel L and R feeds to master offers a ton of promise in absolute performance. I would preface this all with the well worn caveat that one mic channel will dominate the noise, but in sum buses phase shift and distortion also matter.

I am only half joking when I see this as a THAT showcase project.. that balanced line receiver, a pair of VCA's, VCA bus amp, and THAT balanced output.

Maybe throw in a really hip PK/VU meter for good measure.

JR

[emrr]

It's a very interesting idea. Not one I may be able to contribute anything to, but will watch with interest.

[mediatechnology]

John - Thanks John. If you want to draw something up I can build and test it on a small scale proof-of-concept basis. Just the basic summing core is all we'd probably need.

Doug - You could help greatly by providing a feature set. Be thinking about that. Once we prove the summing concept we'd need that.

Guys pardon me if I'm incorrect, but the three topologies of OTB summation we currently see in the market are, active virtual ground (either SE or Balanced), resistive/passive and transformer. Did I forget any?

BTW I may get brave and move the last few posts into the all THAT all the time thread or rename this one. For now keep responding here.

[JR.]

Cool... now I guess I need to deliver...

My general concept is KISS... no EQ no auxes, no inserts, just input XLRs, input level fader and pan, a bus master fader, then output XLRs.

Per input EQs or dynamics, can be patched in series with line inputs, overall EQ or dynamics in series with output.

The current source bus, makes expanding by adding on more modular sections trivial since currents don't care about ground voltage potentials. So perhaps a block of x8 input channels is a good manageable size, with the x8 input one PCB and the bus master + output a second PCB. Perhaps put PS and other glue on the master PCB.

I would like to use inexpensive pots to generate DC control voltages which are crunched to generate the pan laws for two VCAs per input. Probably need to anticipate audio taper faders, since linear faders are less common. The master level fader could interact with a master bus VCA to reduce bus gain along with master level, providing a perhaps useful dynamic scaling of gain structure between summing a bunch of hot inputs, or more moderate input levels.

Meter if used might be physically in separate metal sub chassis, as I see this bus system possibly pushing SOTA noise floor, especially for larger numerical bus structures. Since the output of this will presumably be fed to digital for mastering, the PK/VU meter seems appropriate (IMO).

I would be interested in some user input on features like pan law? I lean toward -3dB law, but it seems more or less could be accommodated, perhaps with resistor selection during build.

It probably makes sense to develop this in stages... first combine VCA outputs to fixed virtual earth summer, then experiment with VCA to VCA combining. I need to think about how wrapping a VCA around the sum bus interacts with my 2 pole LPF to mitigate bus C. I have used a VCA in a master bus before but that was a couple decades ago, so I don't recall all the details.

JR

PS: perhaps a future enhancement might be a digital interface so something like pro tools could send level and pan commands to this analog sum box, where the commands are converted to DC voltages, but i don't need to get ahead of myself.

[mediatechnology]

JR - Here's something you might look at WRT pan law: http://www.thatcorp.com/datashts/dn120.pdf

I do think one might need an insert pre master fader dunno. Would anyone really need linear-action faders? Many of the OTB I see are rotary which would allow linear pots and linear dB control law. It would seem like the whole audio section could be on one board as well with the VCAs sitting right there with the summing amp and DC from each "channel" brought up to the master board. It might be big though once one considers the output pan law tapering circuitry having a bigger footprint than the two VCAs.

I like Roger's idea of being able to automate the CV with midi. Wouldn't have to be part of this project - just having access to read the pots and output the CV would be enough.

[JR.]

I'm still thinking about the problem this is supposed to solve, namely ITB summing. So letting this interface with and be controlled by some recording program software almost let's it drop in place of the flawed digital path, and that interface could also provide some cheap fader and pan automation.

Does anybody know how they trypically code pan controls in digital land. Probably 0-127 with 64 panned center?

Thanks, I was going to look up that pan app note... My initial response, which I recall from the first time I looked at that is "too many parts" for that reasonably simple function and it probably requires precision resistors, precision PS rails, or both.

I want the pans to be dead nuts on, if this is a quality effort. I'm inclined to at least explore coming up with a different pan control circuit. My new shiny hammer would make it trivial using a micro, but I don't think I want to throw a micro in the middle of uber low noise circuitry and a bunch of D/As are not cheap.

I may look at something based around a LTP. That starts out with a natural +6dB/0dB/-infinity law, so I just need to doodle on that +6 dB end some to get different pan laws (ever optimistic).

======

While this design is still very soft, the trick I am pondering of putting a VCA at the master summer makes the concept of "pre-master fader" insert moot. There is no nominal 0VU master bus output in existence that doesn't follow master fader commands. I guess if there's a demand for this, a switch could re-arrange the master circuitry so an additional master bus amp runs at fixed gain, and a separate VCA stage supports the master fader.

Since I'm posting this on the "picocompressorforum" it's probably a fair assumption that "everybody" wants and needs a linked stereo comp on the master.

This is all academic if the benefits I am speculating about for using a VCA master don't pan out. if someone is using a comp on the final mix, they probably aren't going to lose too much sleep over forfeiting the lowest possible bus noise floor anyhow. I do see this lowest possible bus noise as a worthwhile paper target, as an engineering exercise more than having practical value.

If one wanted to be cute, a comp/limiter/whatever could be built into that master VCA stage. To be even cuter, apply compressor attenuation to the master VCA and compressor gain, globally to the input VCAs. But I'm getting way beyond my KISS prerequisite. I like the idea of using separate boxes for all the different studio tricks people will likely want to use, and keeping the mixer spartan.

JR

[JR.]

OK, here's the part where I admit I was wrong.. The LTP gives me close to the actual transfer function I want for gain in current output, but the VCAs accept a linear control voltage for log gain change so the real control voltage needed is one linear slope above 0dB (pan center) and a non-linear slope below 0dB with increasing attenuation at pot extremes. The LTP gives me the second half of this but not the first... The app note uses one pair of diodes to steer currents for a relatively fast but linear initial slope (for attenuation, then a second pair of diodes shunting a resistor to speed up that slope for end point attenuation.

Vbe and temp co errors are only seen at tens of dB attenuation so not a big deal. I'm still not ready to just cut and paste this circuit, but I appreciate it a little more.

JR

[JR.]

I like it...KISS

How about borrowing another of my old concepts from when I was mixer product manager at Peavey... I would put extra features on just a couple "super" channels. This keeps cost-complexity down but provides the capability for one or two important inputs.

Actually 12-13 position rotary switches aren't that expensive and I like the repeatability and precision of switches over pan pots (but I'm not the customer here). Maybe a mix of real pan on two inputs, switches on rest? Big switches on some, small switches on other???

Good idea, is a continuous pan that important??

JR

[JR.]

Keep the ideas coming...

I would be careful about feature creep, I see this as mainly an exercise in how good can a summing bus be made. If there is demand for yet another full featured console, extra functional blocks could be added on.

Just like sub mixing stems in a crappy digital environment improves the overall mix, I could see an 8x2 high quality mixer, be used to build several sub mixes, then later combine them.

That suggests another idea, perhaps an input mode where the mono inputs with pan, could alternately accept stereo inputs for level only. This would allow pre building up to 8 stereo sub mixes, then finally mixing them down. 8x8x2, or 64x2

With minor tweaks, a builder could configure two sets of boards as an 8x2 with 2 aux sends. The extra balanced line input on the second input board set could be repurposed as the second input for a stereo line input for the later summation of stereo stems. PCB jumpers could provide a node for adding EQ or insert modules for the more adventurous builders. Adding a single stereo line input to bus, on the master section PCB could provide two EFX returns if board set is doubled up, or an extra stereo input for basic mixer.

We may need to come up with a matrix of features after we explore all the ways this would likely be used.

I still caution KISS since this will not be very cost competitive against full featured mixers, and few will expect a kit to be this good. The more that is added to it, will distract from the primary mission,

JR