OTB Mixer Using Current Summation

[mediatechnology]

I am speaking from the cuff and don't know if the PICO has a easy to implement feed forward mode. Since the VCA is feeding a summing bus, feedback topology is not available

The Pico is a feed-forward topology. I wouldn't put a compressor in the unit however.

I can see that with 100 dB cutoff how one could reach the diode drop potential at Ec.

[JR.]

Thanks.. Just to reiterate my thoughts, I am a huge fan of KISS, and when it comes to audio paths, less is more.

If someone "should" happen to desire applying any kind of dynamics processing to signals passing through our OTBc (I thought it needed a punchier name), the best of all worlds is perform some external side chain processing and simply add that to the internal control voltage. The nice thing about log domain gain control is you just add dB gain/atten and get your desired result. Plus you avoid the multiple audio I/O steps of running through insert interfaces and cumulative VCA passes, etc.

To that end... providing the hooks to extract a single ended line feed from the input buffer, that can be sent out, and a CV input voltage that can be returned and added to the overall fader/pan commands seems promising. This CV can be processed at some reasonably scaled up voltage, say 10dB/V or 6.6dB/V so interface ground errors will be scaled down with the CV.

This should not cost much overhead if any, and could be handled through a simple TRS jack. The external side chain could be a depopulated PICO comp, or variant on that... My "shiny hammer" could also interface if i ever were to make one...

The overhead cost could be as little as a build out resistor and one or two input resistors with some PCB pads. If the scope of this grows too onerous it should be nipped, but as long as the impact as minimal, it seems promising.

JR

PS: I designed a comp back in the early '80s and I performed a full range comp, a hard limiter, a downward expander (gate), and a de-esser, all with only one VCA by adding the sundry computed side chain CVs together.

[mediatechnology]

John -

I think anyone wanting/needing dynamics on a per-channel basis could just do it ahead of the line input. Having a CV insert seems feature-creepish and I doubt anyone would use it.

For the added interconnect overhead there would be a lot more mileage if it just had a TRS insert send/return in the audio path between the line input and the pan/mute/current-out stage. There one could patch a Pico Comp, LA-3 or EQ.

For combined master fader, bus gain trim, automation and dynamics the SSL bus compressor is a good case study in using one set of VCAs for all of the aforementioned. I don't think we should do that here either though just keep the analog path short and sweet. That's my two cents though...

[JR.]

Agreed... the beauty of a line input mixer, is that a per channel EFX could be patched in series with the input. Adding an insert point requires another line level receiver.

From a purist POV, one less VCA in any path is GOP (good on paper) but will be limited to a very small subset of possible external side chain solutions. Most serious users probably have their favorite collection of dynamics boxes.

I consider adding a resistor or two minimal feature creep. and other applications include interfacing digital automation. This wouldn't be a big deal to cut in after the fact if desired, so like Emily Litella used to say... never mind.

JR

[mediatechnology]

We put a bunch of simple build options in the original 10 LED GR meter: Diode rectifier, constant current, Vreg etc. That board generated more support questions, due to it's options, than all the Pico Compressors combined.

And you're right that a balanced insert adds another line receiver.

I think simple is best: Line Recever-VCAs-Mix Amp with the only op amps in the channel being (hopefully) a dual or quad for CV.

[emrr]

A Most serious users probably have their favorite collection of dynamics boxes.

Probably the single most important thing to keep in mind, IMO. there would be the rare case where all-inclusive would be desired, but honestly, people will want this sort of thing as outboard mixing from a DAW, and lots of processing will be done in software there before hitting this. Or DAW to hardware to mixer.

The more important place to consider inserts these days seems to be submix busses and master, and that's another giant feature creep can o' worms probably best avoided. Sorry I brought it up.

[JR.]

I dusted off my thinking cap, last night, and I can't figure out why not just add a simple unity inverter to split the CV for symmetrical drive?

While true, the noise of the first opamp will be fed to the second opamp and add coherently. It seems more appropriate to use relatively low noise opamps for these buffers instead of adding 2 more opamps per channel. Especially since we should use relatively low noise parts anyhow.

JR

[mediatechnology]

I dusted off my thinking cap, last night, and I can't figure out why not just add a simple unity inverter to split the CV for symmetrical drive?

That's what I was thinkin' too. Looks like an MC33179/MC33079 quad would seem ideal or two duals.

[JR.]

I found a (on) data sheet and they look OK I think... the data sheet had some two collector transistors in the schematic, and some, what appear to be, mislabeled data graphs but i ASSume the parts work...

If we lay out PCB for duals it will be easier to substitute other devices if we change our mind later...but quads could be a tighter layout.
JR
EDIT----- undo undo...
=========

Ok, JP who is lurking has prodded me to rethink this... He went through this exercise before the the THAT guys...

The CV is effectively in series with the base of the VCA transistors so noise there does matter...

I am now leaning toward using simple resistive pads dividing down a several volt CV to the mV level instead of low gain buffer opamps. The impedance of a simple resistive divider can be a few tens of ohms and we can parallel this with a bunch o'capacitance so HF Z will be adequately low.

Voltage Control
The 2180 Series VCAs are designed to be operated
with zero source impedance at pins 2 and 3, and an
infinite source impedance at pin 4. (Pin 4 should be left
open.) To realize all the performance designed into a
2180, keep the source impedance of the control voltage
driver well under 50

With a passive pad we should be able to easily get CV down under 50 ohms... and this will be much quieter than even the best opamp.



==================
Edit #2.. I Went to the source and have exchanged email with Gary Hebert at THAT.

It appears noise and impedance at CV ports are important...

I have some ideas,,, but need to think about it some more.

JR

[JR.]

pan.JPG (89.24 KiB) Viewed 14070 times

OK, this isn't finished but getting closer...

Using a low noise transistor as an emitter follower I can pad down the opamp noise, and deliver a pretty low output impedance.

The approach is to still keep the opamp in the DC loop for CV accuracy but the AC feedback around the opamp is tied to the opamp out. That node gets divided down another 20dB to the base of the low noise follower. At this point the source impedance is rbb of transistor and base impedance divided by HFE.

The cap around the opamp can be an electrolytic since there is a well defined DC voltage bias.

I changed the mute to just switch fader to full atten...

I am still unhappy that I don't have a clamp (yet) so full pan attenuation "and" full fader attenuation could result in too much atten. but I hope to come up with something easier than adding another opamp. I have some ideas for that I need to soak in beer...

This is still a work in process..

JR