Wet/dry Parallel Processor

[Fragletrollet]

Hello,

I have a tube opto compressor that I've built that I want to add a wet/dry function to. But I also am building an MTC where I would love to include a Wet/dry function on a couple of the inserts.

I am trying to conceptualize what would be needed. We're combining the pre-processor audio with the processed audio, presumably through a potmeter/switch. Do we add bus resistors to both channels, and a summing amplifier after the wiper of the pot? And either a electronic balanced receiver ala opamps to debalance and buffer the signal, and the same on the other end? Could be discrete opamps and output trafo, or something like that124x/164x chips, or if just buffering and no need for (de)balancing just a unity gain opamp chip on input (summing amp buffers output ?)

Or maybe use something like the JLM Dingo, balanced receiver but then use the THAT VCA to control the volume in opposite directions between the boards with dc from the dual pot, then opamp/trafo driver on the output?

I'm sure you wizards could come up with a simple pcb alà the M/S Mini board that would be very handing not just for your line of Mastering tools, but also for other DIY builds. There is quite alot of interest around the old Send'n'Blend project that unfortunately seems to be dead (although the schematic and self-etch files are available... I've just not come around to start etching on my own yet. Thought I'd skip a step and just learn PCB design... but what a rabbithole )

Thanks for any input, guidance or ideas :=)

[mediatechnology]

Thanks for asking.

You can do it with pots.
A single channel blender can be made with two pots wired contra-rotating with the wipers actively-summed.
As one signal increases the other decreases.

You can also make a poor man's version with a single pot fed at both ends and use the wiper as the output.
It's sort of like a pan control but backwards.

Stereo, done the proper way, requires a 4 gang pot or stepped switch.

I haven't built a board to do this because most MEs want to use a stepped switch (as opposed to VCAs) and the active circuitry to sum them is super-simple.
The heavy lifting, and cost, are the switches.

[mediatechnology]

I've decided to do a wet/dry parallel mixer board.

It's a pretty simple circuit that splits the feed to the device, drives a four section stepped switch "pot" and sums the wet/dry.
A relay will switch from a Mix mode to an Add mode where the dry is fixed in level.

The summing stage will have a +/- 5 dB makeup adjustment to trim the final output level.

I'll be posting a schematic in a few days.

[Fragletrollet]

Very cool! I think this utility board will go nicely with the other mastering processors. Looking forward to learning more about it!

[mediatechnology]

Thank you.

This would be super easy and cost effective if done with VCAs but I'm thinking most MEs would prefer using stepped switches.

Mix Mode requires 4 switch sections, Add only requires two.

There is a significant cost for a 4 section stepped switch or a 4 gang pot:
Goldpoint's pricing, without resistors installed, is $191 US for a 24 step 4 section and $221 for a 47 step. https://goldpt.com/prices.html

If gain make-up is added then another 2 section switch is required.

If done with VCAs it would only require a single-section pot and a second one for make-up gain.
VCAs in SIP have delivery delays which show no sign of letting up in the long term.

What do you think would be the most agreeable approach for the ME?
VCAs or switches?

A board using customer-supplied switches would be super-simple and low cost.
It could be jumperable for fixed or variable make-up gain.
The form-factor would be about the same size as the MS Mini: https://ka-electronics.com/shop/index.p ... uct_id=115

I could probably do a VCA-based one in the same form-factor and make it so that it could be built with VCAs or switches...

[Fragletrollet]

As I wrote in my first post, using VCA's is the way I was planning to use. Can't really defend my choice with other than that it's a way to have a solution with a cheap pot/single deck switch (just using say 0-5V and it's inverted copy to control two sets of vca's)

I generally find the 2180's to be transparent (especially when not modulated with varying CV), but one might argue for a solution without VCA's that might be even cleaner? But yes availability in these days is something to keep in mind. I guess you could design a discrete VCA? But I guess the performance of the THAT chips will be hard to beat without resorting to supertrimmed transistors (?)

Personally I won't be buying expensive 4 deck switches to be able to omit the VCA's...

[mediatechnology]

Thanks for the reaffirmation that VCAs would still be OK.
If it were VCA-based it would be relatively easy to provide an external send and return for someone using a stepped switch.

If VCAs are used the wet and dry VCA current outputs could share an op amp for the I-V conversion.
For Add mode the dry VCA could be bypassed altogether.

[mediatechnology]

I've gotten feedback offline that made me think I should make both a VCA and a stepped switch option available.

Some MEs would go the extra mile and used stepped switches to eliminate the VCAs.

What I'm grappling with is the control law and number of VCAs required.
The implementations I've seen have a linear control law with the center being a 50/50% mix.
That's easy to do with a stepped switch or multi-section linear pot.

With THAT VCAs the control law is exponential.
DN-120 shows us how to make a constant power pan or cross-fader (or one that's 6 dB down at center) using diode-shaping.
(See: http://www.thatcorp.com/datashts/dn120.pdf)

Linear control of THAT VCAs can require a second VCA to log the control voltage making the overall response linear.
See: viewtopic.php?f=6&t=892
and https://proaudiodesignforum.com/images/ ... nuator.pdf
and https://proaudiodesignforum.com/images/ ... ontrol.pdf

The signal path would require 4 VCAs to mix L and R from the direct and processed path.
An additional VCA, or maybe two, would be required to linearize the control.
A less accurate log converter using a BJT could be used as well.
That's 5 or 6 VCAs (or 4 VCA plus transistor array) plus supporting op amps for the control voltage path.

A constant power crossfader made using VCAs might actually have a better control "feel" than a linear one.
It would however not operate the same as one made with a pot or stepped switch.
Another disadvantage (seen in DN-120) is that the cutoff attenuation at end stop varies considerably with temperature.
It cannot go to full -100 dB cutoff.

Thinking out loud I'm wondering if a simpler solution - a 4 section linear pot - would be a more viable option than 6 VCAs for crossfading.
I think I'm going to have State Electronics quote a 4 section 10K linear pot. https://www.potentiometers.com/select_custom.cfm

If a 4 section pot were used it would still be less expensive that a 4 section switch and the circuit for both would be identical.

Thoughts?

[terkio]

The THAT VCA: I recall it has two control inputs.
They are both linear giving Db gains. One input goes increasing the gain, the other one goes decreasing the gain.
If this is right a single linear pot could feed the 4 VCAs control inputs to achieve the wanted result.

[mediatechnology]

Thanks Terkio.

Yes, the VCAs have differential control that easily allows one VCA to have inverted Ec and with respect to the other.
That part is easy.

What you want however is 6 dB attenuation in the middle of pot rotation.
Looking at one VCA you need -∞ to -6 dB on the first half of rotation and -6 dB to 0 on the second half.
That is not a linear dB/V taper.

Let's say 0 to +5V is 0 dB to -100 dB.
With linear dB/V control 2.5V is -50 dB.
We need -6 dB at 2.5V.

Look at DN-120.
It uses a diode shaper and steering diodes to control Ec.



To make the above a cross-fader the lower VCA's current output sums into U3A.

The required taper is not linear in dB/V.



In the case of a crossfader emulating counter-rotating pots the overall taper in the commercial units I've seen are linear.
The graduations range from 100%/0% at CCW "Dry" rotation to 50%/50% in the middle and 0%/100% at full CW "Wet."
The steps are 10% e.g. 90/10, 80/20 etc.

You can't get that with an exponentially-controlled VCA.
Normally expo (linear dB) works for you in this case it works against.

So the options appear to be either a diode breakpoint generator like the above or linearization of the control voltage with a log converter.

The complexity of a VCA-based option starts to make a 4 gang pot attractive.