OTB Mixer Using Current Summation

[juniorhifikit]

I have given some thought to this and using a simple microprocessor interface, something like my VCA summer could receive control from protools or DAW? could be cheaper than knobs and controls. I wonder if this wouldn't offend the sensibility of OTB analog mixer proponents? A hybrid analog/digital box, to allow an analog sum under digital control.

My sense is we want to use our old toys, and digital plug-ins don't yet accurately mimic how they overload? If when the plug-ins get it right, OTB becomes moot. Note: this includes a plug-in for parallel processing.

JR

PS: Yes I know Paul Wolff from waaay back.. He needs to charge enough that people value his product, and being a small volume mfr he needs to charge more than a larger company for similar BOM.

Yep, you got it. This is what I'm fiddling with in Eagle. I've learned to mix without knobs and faders for the most part, so tactile control is not something I'll be including in my widget. It's also the expensive part. This is why this thread is so interesting to me, and why I've been asking so many SSL related questions (as you may recall from other forums). This started as a frankenstein job someone asked me to do for them, but it got me thinking. You guys are all in a whole other league from me, as I'm no designer. I really benefit from the ideas you toss around in these threads!

Thanks!

p.s. Paul's product is great and he's put a lot of time/energy into it. It's priced fairly, but is out of my reach; and many others at my "B" level as well I suspect.

[juniorhifikit]

Not to make things more complicated, but in regards to keeping the parts-count low, has anyone given any thought to doing the same thing with an MDAC? Something like a Maxim Max532, or perhaps an Analog Devices SSM2018 VCA which already has differential inputs?

I was looking a little more at the cost/performance differences between the THAT 2180 and it's surrounding components, and the Analog Devices SSM2018T, which is slightly more expensive but already has a differential input and it's own control voltage buffer amps built in. In the spirit of KISS, can anyone of larger brain comment on the advantages/disadvantages of the two?

http://www.scottgreiner.com/AD_SSM2018T_VCA.pdf
http://www.scottgreiner.com/2180data.pdf

Their spec's seem comparable at first glance (though again, negating the current summing idea...)

[ricardo]

John (JR), early on in this very old & long thread, you intimated that you had a cunning way to build in a 2nd order LP filter into the mixing amp with only minimal extra bits. Any chance of even a Bob Pease type sketch?

[JR.]

John (JR), early on in this very old & long thread, you intimated that you had a cunning way to build in a 2nd order LP filter into the mixing amp with only minimal extra bits. Any chance of even a Bob Pease type sketch?

I think I've published the topology before in one of my sundry discussions around the WWW. I used the very standard 2 pole MFB LPF topology. The C to ground is the stray bus capacitance, with perhaps some real capacitance in parallel for good measure.(In production you don't want any circuits that oscillate when not connected to the bus or whatever so it can be tested outside the console). The gain is the input current times the feedback R. The inner RC pole is set pretty HF so the first C doesn't develop much V in use. The input R is not used or connected to GND, with the VCA currents summed directly to the first C node.



I have used this bus amp filter approach with non-VCA current source summing so I know it works, but don't ask me to remember what values I settled on decades ago.

In hindsight when used with VCAs this need to be looked at specifically to keep the AC voltage at the first C to ground node << +/-.5V to not interfere with the VCA innards. that could saturate and distort with too much voltage there.

Note: you may want to keep the input R but tied to ground to A) make filter Q and frequency calculation easier, and B) to provide a DC path for VCA outputs if bus is cap coupled.

CAVEAT: I have never used this sum amp filter with VCAs so this is unproved in that combination and might not work... but I expect it can be made to work with thoughtful (very high frequency poles) tuning of filter. Note the filter doesn't need to filter anything audio, just keep the sum amp stable and happy despite bus capacitance. If the VCA gets a little dirty up above 50kHz, that's life.

JR

[ricardo]

I used the very standard 2 pole MFB LPF topology.

Excuse my tiny beach bum brain but can I confirm for this application, not all the Rs in the diagram are the same and neither are the Cs?

[JR.]

You should be able to find design equations for MFB in filter design texts, If you can't I can dig out mine from old Burr Brown Operational Amplifiers book.

Typically the input R and output R back to that input node are the same value for unity gain inverting response below tuning. The Cs are generally not the same value and that is how you tweak the Q, the inner R affects Q and pole frequency which is all caps and the Rs (may only be two of the Rs, I need to look it up, but not tonight.

JR
try here...

http://www.science.unitn.it/~bassi/Sign ... loa049.pdf

[ricardo]

I think I've published the topology before in one of my sundry discussions around the WWW. I used the very standard 2 pole MFB LPF topology. The C to ground is the stray bus capacitance, with perhaps some real capacitance in parallel for good measure.(In production you don't want any circuits that oscillate when not connected to the bus or whatever so it can be tested outside the console). The gain is the input current times the feedback R. The inner RC pole is set pretty HF so the first C doesn't develop much V in use. The input R is not used or connected to GND, with the VCA currents summed directly to the first C node.

All dis zillion kHz square waves make me want to look at serious band limiting. CD, being the most common source of high quality music these days, simplifies things with Nyquist 22.05kHz.

But the Golden Pinnae brigade are asking for fs = zillion kHz when they master.

Does this result in CDs which sound different? Of course is does! The reason is simple. Most Sample Rate Converters, especially those in the expensive DAWs, are terrible and alias loadsa HF stuff back into the audible band. And there are loadsa zillion kHz recorders with abysmal anti-aliasing filters too.
[/rant with my microphone hat]

I wanted to ask John if he had some rough ballpark figures for "..the stray bus capacitance..". I'd like to do some work on John's virtual earth summer with 12dB/8ve LP. Insufficient steam power in Cooktown for me to melt solder but my abacus dun need steam.

[JR.]

Simple answer I don't know. It seems it would depend on lots of stuff. A bus in a console that is 6' long should be expected to have more than one 6" long.

FWIW I never really intended for the MFB 2 pole LPF to be a practical band limit filter, but more a way to manage the stray capacitance benignly (i.e. relatively flat at 20kHz and/or low phase shift at 20kHz if possible) any actual roll-off would be up where it would mainly bother the bats.

The MFB is a very opamp friendly topology. Input is passively LPF and opamp operates as an integrator so very hard to overload.

I would probably empirically determine bus capacitance by starting with some arbitrary values and seeing what it takes to make it stable and go from there.

The overall feedback R is what it needs to be to line up with VCA output current, the inner R needs to be reasonably small so it doesn't add a lot of noise, the input R may not even need to be there. If it is stable with a small feedback cap and no extra shunt to ground that is a good start.

If you build one up, let the circuit tell you what to do by how it acts... it is so long ago since I did this I don't recall precisely where it ended up.

JR