OTB Mixer Using Current Summation

[JR.]

I have a couple ideas for simpler approaches if we forgo a nominal voltage range input. In the case of a pan, I should be able to generate a -3dB (or whatever at mid point, with a bunch of cut at extremes. The fader cut can be added on top of that, but I wonder if there is some value to limiting the amount of cut. The VCA spec says -100dB total so maybe limit pan to -30 or -40dB?

I need to load some ideas into a excel spread sheet (while not drinking) to see what kind of pan law I get,...

I have been rethinking the MUTE, and removing the constraint of voltage control for everything, we can probably do a much better quality with a simple switch.

I don't see any parts saving advantage to using both ports of each VCA, not sure if its even practical... It appears using both ports is always more parts.

JR

[mediatechnology]

John - I think we would need full attenuation on pan. People are so used to that.

Maybe we should consider a "dual slope" pan law with -20 dB/V below center (-5V FS for -100 dB) and +1 dB/V above center (+5V FS for +5dB). The opposing channel could follow the inverse of this. The breakpoint, at 0V, would thus be ~5 dB down in the center WRT the ends. The above numbers are just examples, the scale above 0V/center could be 3 dB, 4 dB whatever set with resistors for 3, 4.5, 6 dB "constant power," "BBC" and "constant voltage" laws.

This could be done with diode steering with different gains in each quadrant. When panned center, the VCA is at it's optimum operating point (unity gain) and it's only being put 3-6 dB into gain when panned hard left/right. To the left or right of center it follows a smooth linear dB law just at different slopes.

I'd be willing to build one and see how smooth it sounds. My constant-voltage matrix panner (-6 dB center) was pretty darn smooth and it wasn't linear dB. Linear dB should be even better.

[JR.]

I'm busy today trying to ship some product before end of the year... but let me do some spread sheet work first. I can ballpark compare approaches and dial some values in on paper a lot easier than on the bench..

We can add a clamp at -100dB or whatever arbitrary max attenuation we decide upon. So we can command a lot of attenuation in the pan and the fader and let them simply add up until the max limit is reached.

It might be possible to just scale values so we let an opamp saturate to the rail to limit max attenuation, but that could cause some interaction between controls, and PS noise into control path, so not my first preference.

JR

[mediatechnology]

I'm busy today trying to ship some product before end of the year

You mean like "work?" I got a few orders this week too with end-of-year shoppers. Thanks guys!

[emrr]

beers for you both!

[JR.]

Ok.. I gave up on getting everything shipped before doing my end of year inventory since more orders came in today, but I got the old stuff shipped... Johnny Beer is in the refer getting ready for another wednesday night.

I am coming back to the design I posted a couple weeks ago with a few minor tweaks.

pan.JPG (50.6 KiB) Viewed 13107 times

I reduced the amount of fader kill since it was in excess of 100 dB

added clamp diodes to buffer to clamp VCA control port at approx 100 dB, so attenuation will add until it clamps.

added RxC on output if needed to deliver low Z to VCA

Added C across opamp to reduce noise and lower output Z, and de-click mute, pot noise, etc..

I would still consider using a real switch to disconnect VCA from bus to mute for even more kill, but this is more a numbers thing than really needed IMO.

Happy New Year all...

JR

[mediatechnology]

Thanks John for posting the drawing. Happy New Year to you and all!

[JR.]

The design friend of mine (JP) at the console company (APB) pointed out that VCAs can thump if hit with more than a diode drop at CV ports.

I was only thinking in terms of voltage delta between bases of the LTP, but there is a potential to get conduction in the base-collector diodes if those devices get into significant saturation. If those reverse collector diodes conduct, that saturation base current is corrupting audio input/output nodes.

If we want to command serious attenuation, we need to consider splitting up the control voltage into + and - control ports.

I vote for differential CV drive. (two more opamps). I would still clamp before we get to differential buffers so I need to revisit CV circuits.

better now than later..

JR

[Igor Dockx]

Hi John and all,

Happy new year!

Yes, let's get the best out of your design so one more vote for differential drive. Nothing comes for free; I think the use of two more opamps is justified here.

It might always be considered to make use of the That Analog Engines, which offer VCA, RMS-detector and three opamps in one package, in a later spin-off of the basic design you're doing now. That way a really compact PCB should be possible... Almost everything is already present for a per-channel compressor and bus compressor too.

But for now let's stay focused on the basic idea!

Thanks for allowing us all to follow your thinking in the desig process; I already learned a lot from reading all this.

Best regards,

Igor

[JR.]

Hi Igor, your vote is registered.. 2 : 0 for differential CV,

Regarding using the THAT audio engine, I'm not sure if that makes sense for every channel.

As you point out, there is merit to incorporating dynamics into these input channels, since we already have the VCAs in place. All that is needed is to add some side chain processing and controls.

In my "shiny hammer" thread you can see some of my thoughts about using an inexpensive microprocessor to make a very flexible digitally controlled analog comp. At this point it's putting the cart ahead of the horse, but I would leave hooks for add on PCBs that could grab the buffered input signal and then return some control voltage to the VCA CV buffers.

Rather than re-invent the wheel, expand the scope of this project (feature creep), and add more unknowns for what kind of sonic performance to expect. I would consider a variant on the PICO compressor that already has a good reputation, that could be added in, perhaps in groups of two.

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. I'll defer to the PICO experts here if it can be easily incorporated. If not I can help roll our own. I have designed a few comps over the years, but if we can use a comp with known behavior that seems the better choice (especially in light of the name of this forum).

JR