Scaling VCA Cross-Fader Circuit

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

thermionic

Well-known member
Joined
Jun 3, 2004
Messages
1,671
Hi,

A client has asked me for a particular option that will either involve a) having open wallet surgery from State Electronics and buying a (subject to MOQ + set up cost) pot that I can't find elsewhere; or b) using a pot I can get at Mouser etc, but having to implement a DJ-mixer style cross-fader. At this point I'm opting for option b, as the additional noise and THD from the VCA is not an issue as it's for a cue-ing circuit, not main signal path.

However... The Mouser part only comes in a single-gang 10K Lin variant. I notice most VCA cross-fader schematics seem to use a dual gang 100K for the CV input...

The circuit I'm using is like this one, but I've buffered each CV input with a Voltage Follower: http://www.djzone.net/pg/0101/graphics/vca.gif

I'm using the SSM2164 (I need to cross-fader stereo 'a' signal with stereo 'b'): http://www.analog.com/static/imported-files/data_sheets/ssm2164.pdf

As the VCA stalwarts will have realised at this point, the issue I'm having is 'scaling' the pot. If I want complete cut-off, then each signal fades in and out at the extreme end of each CCW / CW pot position, with 'no man's land' in the middle. If I want a 3dB point at pot half-travel, then I sacrifice complete cut-off.

I've scoured the Internet, but no schematics avail (before you suggest trying the usual sites, please take a look for yourself; it appears that scaling a VCA cross-fader is regarded as a trade secret).

I'm now thinking that I might need to put a logarithmic amp in place of the Voltage Follower that feeds the SSM2164 CV inputs. There will also likely be a need for 'law bending' resistors on either side of the pot wiper.

I have tried all manner of resistor values in the current circuit.

Before I spend even more time with a log amp, can anyone here tell me if I'm likely to succeed down this route? I realise that the single-gang 10K pot requirement makes things a lot more difficult. There are various VCA cross-fader circuits out there, but they typically use 100K-ish track values, in a dual-track fader.

By the time I've figured this out I'll probably have been better off poneying up to State Electronics for a custom pot (like I did when I last needed one of these), but I hate being beaten. If the Brain's Trust here can tell me I'm on the right track with a log amp, just that'll be good to know - better than beating a dead horse!

Thanks in advance.
 
A Linear (not dB) VCA seems more what you want.

Take '3080. +/-15V, run 33K into Iabc pins and feed them complementary +/-15V control voltages. You get 0dB at +15V, -6dB at zero, and -infinity below -14V. Some small tinker (+7V clamps) should give -3dB at equality.

Of course '3080 has not-good THD/hiss ratio. There may be other linear VCAs?

There's also digital potentiometers and a Basic Stamp or PIC to do-up the bits.
 
Thanks, much appreciated. 

Although I've been generously given 4 potential start-from-scratch solutions, having already made an SSM2164 strip-board prototype with buffered outs and buffered CV pins, is there any point in pursuing the CV buffer amp into Log amp conversion idea? Isn't there a good chance that this will work?
 
Making a custom pot for this seems like a significant step. It has been years since I looked at this (DJ cross-fade laws) but IIRC there are different philosophies about what is optimal.

Using a linear cross-fade pot and then defining the cross fade law with extra circuitry is how I would do it.  Perhaps budget a 3 dB drop or less for both at mid position, with steeper drop-off past that.  Doing this external to the physical pot gives you the flexibility to vary or offer different cross-fade options.

Tooling up a special pot means you only have one solution and it better be right since it will be hard to change after you have a pile of them in stock.

JR

PS: Look at diode break circuits for ways to dial in different slopes for different regions of pot travel.
 
You really have to define your curve. Yes, you probably won't know until you build it and sit at the desk a day. But right now we are shooting blind.

What do you want at 50%? 0dB? -3dB? -13dB?

What do you want at 25%/75%? -6dB? -20dB?

What do you really need at 0%/100%? The linear-dB VCA never gives literally ZERO output, - infinity dB. It may be that -60dB is good-enuff, or that you need -100dB, or that you can't be happy without a dead-short.

The SSM spec says 100dB attenuation. With simple linear CV this is likely to be -50dB at 25%. While -50dB is not silent, it is not much either, giving the audible impression that the last quarter turn doesn't do much.

You say that State Electronics fixed you up last time, but give no details about that solution.

In short-- I have two feet of snow coming Sunday. Come over Monday and plow my driveway. You are sure to plow a wrong curve and end up in my hole or missing the house altogether. I can do it because I have *some* idea what the driveway curve should be.

BTW, "The circuit I'm using is like this one" is an illegible thumb-nail (on my browser). It does seem to have a cleverness, but maybe I'm squinting too much into it.
 
Thanks.

>80dB would be a decent amount of 'offness'.

-3dB at 50% travel would be ideal

A conventional DJ crossfader curve would suit perfectly

The thumbnail doesn't give values. It's very basic. There is a resistor being fed +ve which is either side of the fader track. The end of each track feeds CV. The wiper is grounded.

The solution from State Electronics is a passive crossfader. I could do it with a bazillion dual pots. The problem here is that State (and Mouser, but only in a single-gang 10K) are the only sources I can find that sell a pot that fits the mechanical spec for the project (trust me, I've tried!).  Hence the 2 options of either paying insane $ to State, or working out a VCA option.
 
thermionic said:
Thanks.

>80dB would be a decent amount of 'offness'.

-3dB at 50% travel would be ideal

A conventional DJ crossfader curve would suit perfectly
It has been decades sine the last time I looked at this but IIRC -3dB at 50% was not optimal.  Dropping two incoherent sources to half power does not give a full power blend, and DJs don't want to lose loudness during the crossfade... If anything getting a little louder at 50% when both are playing near full up, is probably more acceptable.  My recollection is that the cross fade was steeper and closer to the middle...

You probably want to start gradually turning up (mixing in) the opposite side right away, but not turn down the side that is playing until closer to the midpoint.   

I agree 
The thumbnail doesn't give values. It's very basic. There is a resistor being fed +ve which is either side of the fader track. The end of each track feeds CV. The wiper is grounded.

The solution from State Electronics is a passive crossfader. I could do it with a bazillion dual pots. The problem here is that State (and Mouser, but only in a single-gang 10K) are the only sources I can find that sell a pot that fits the mechanical spec for the project (trust me, I've tried!).  Hence the 2 options of either paying insane $ to State, or working out a VCA option.

Custom passive cross fade will limit the control laws you can generate far more than VCA  voltage, and you are stuck in case your customers change their mind.

JR
 
Thanks, John.

This project is for a cue path, not the main output. Of course, I'd prefer to keep the signal as pristine as possible, but it's not as crucial as what will be going to the main outs.

The '50%' crossover point in a DJ crossfader is contentious from what I've seen, with various companies' options differing in ideology. I'm sure that, in my design, it'll be easy to tweak the exact x-over levels finely. The problem at the moment is getting the right CV law so they work crudely!

 
thermionic said:
Thanks, John.

This project is for a cue path, not the main output. Of course, I'd prefer to keep the signal as pristine as possible, but it's not as crucial as what will be going to the main outs.

The '50%' crossover point in a DJ crossfader is contentious from what I've seen, with various companies' options differing in ideology. I'm sure that, in my design, it'll be easy to tweak the exact x-over levels finely. The problem at the moment is getting the right CV law so they work crudely!

OK the cue is less critical... FWIW back in 1978 I designed a DJ mixer kit...  I used a circuit trick to generate an "automatic" cue mix that allowed the DJ to line up the next track. I just subtracted the output that was panned to and playing, from both inputs, so the one that was playing dropped out, and the opposite channel that wasn't playing was always up in the cue output.

I didn't have to deal with the cross fade law because my mixer used two rotaries requiring two hands to execute a cross fade but you had full control of how they mixed.  (FWIW back in those days DJ mixers used rotaries not slide pots. Like the old Bozak ).

JR
 
thermionic said:
Hi,

A client has asked me for a particular option that will either involve a) having open wallet surgery from State Electronics and buying a (subject to MOQ + set up cost) pot that I can't find elsewhere; or b) using a pot I can get at Mouser etc, but having to implement a DJ-mixer style cross-fader. At this point I'm opting for option b, as the additional noise and THD from the VCA is not an issue as it's for a cue-ing circuit, not main signal path.
Most disco mixers use a passive system that is not too dissimilar from the so-called passive pan-pot. See attachment.
By varying the ratio between the pot value and the RS value, you can achieve almost anything you want.
 

Attachments

  • X-fade.JPG
    X-fade.JPG
    14.6 KB · Views: 14
abbey road d enfer said:
thermionic said:
Hi,

A client has asked me for a particular option that will either involve a) having open wallet surgery from State Electronics and buying a (subject to MOQ + set up cost) pot that I can't find elsewhere; or b) using a pot I can get at Mouser etc, but having to implement a DJ-mixer style cross-fader. At this point I'm opting for option b, as the additional noise and THD from the VCA is not an issue as it's for a cue-ing circuit, not main signal path.
Most disco mixers use a passive system that is not too dissimilar from the so-called passive pan-pot. See attachment.
By varying the ratio between the pot value and the RS value, you can achieve almost anything you want.

Thanks. I know the circuit very well. I used it in the last version of the project in question. Alas, this time, because of the mechanical constraints, I have a choice between sourcing another unusual dual-gang pot from State and opting for the same circuit I used before; or implementing a VCA, whereupon I can use a single-gang pot off the shelf (preferable, for the reasons John asserts - not least being for the client, who won't have to pony up to State should he accidentally yank a shaft off!). I have stacks of dual pots on the shelf - but I can't reach for them!

The current stage is at sticking with the 2164 and putting a log amp in the SC. However, it will need scaling - so I'm not on the home straight yet. 
 
The cross-fade 'law' is an interesting factor....I learnt some interesting leasons whilst designing a couple of radio stations in the UK that included live dj sets as a major part of their night time schedules.  Various "STAR" DJ's requested/demanded specific DJ mixers..stating that they could not/would not work with other units.  Following various comments about prima donna's we investigated the issue and found the favoured units had similar crossfade laws....I guess it come down to 'feel' just like my choice of keyboards.

The original  units have now been replaced 4 times and the later units have switchable laws.. We stuck with the Pioneer DJM series btw.

Rane have a useful note @  http://www.rane.com/pdf/ranenotes/Evolution_of_the_DJ_Mixer_Crossfader.pdf  Just a little partisan but why not?

tc
 
Back
Top