Linearize 'log' VCA's response

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
Igor Dockx
Posts: 11
Joined: Fri Dec 26, 2008 5:10 pm

Linearize 'log' VCA's response

Post by Igor Dockx » Wed Jun 14, 2017 1:31 pm

Hi All,

Some time ago I built a linear VCA using two sections of the SSM2164 quad VCA, following an old schematic, see attach. It works great.

I am wondering if the same trick can be done using two THAT 2181 VCA's.

Ultimately my goal is to build a linear VCA to allow me to add VCA automation to some channels of my mixing console. I do want to keep the 'audio log' P&G faders, but have them control the VCA, which then controls the audio.

There have been a number of examples on this forum of squeezing max performance out of That VCA's by using differential control voltages. So ideally I'd like to come up with a circuit that benefits from differential drive and has a linear gain constant.

Thanks for your thoughts!

Best Igor
Attachments
Linear_VCA_edn-magazine.pdf
(73.79 KiB) Downloaded 91 times

User avatar
mediatechnology
Posts: 2797
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Linearize 'log' VCA's response

Post by mediatechnology » Wed Jun 14, 2017 3:05 pm

Hi Igor. Its been awhile since we've heard from you. Welcome back.

I've seen that circuit before and its very clever.

I haven't done exactly the same thing with 2180/2181s but I have done log converters using them in the feedback loop which is the IC1 portion of the linearizer. I've also made them expo converters by putting a DC reference current into the VCAs audio input and using Ec as the expo input.

So I think what you want to do, with minor changes ought to work to get linear VCA control.

The 2180/2181 invert current polarity so you may need to change the polarity of the reference current. The compensation of the 2164 is also different.

The key to using any of the THAT VCAs as log or expo converters is to make sure that the current output sums into a virtual ground and not elevated to a DC potential. The VCA output current mirrors have limited range into anything other than ground. IC1 and IC2's inverting input is at virtual ground in the circuit so in this case its not an issue.
Read Damore v. Google and ask yourself the following questions:
Are Googlers really the people you want in charge of your search results, browser, data or operating system?
Would you be comfortable riding in a Google self-driving car knowing that Googlers might want you dead just because they don't like you?
Would you trust a Googler with your children?

User avatar
JR.
Posts: 2140
Joined: Sat May 24, 2008 7:21 pm

Re: Linearize 'log' VCA's response

Post by JR. » Wed Jun 14, 2017 3:18 pm

I am tempted to ask why, but the customer is always right. :lol:

One obvious downside is using two VCAs to perform one gain move.

You mention that you want to add VCA automation, how are you converting and storing the fader position info? Many real console automation system use a fader with a linear track.

Audio taper faders are not very good down in the hop off region but perhaps good enough..

I have used a similar trick to linearize gain elements on my bench back in the 70s. One time I used a dual N-ch JFET... and tracking was fair, another version used a center-tapped CDS cell that was grounded in the middle. I do not know how precise a THAT vca will be for small DC voltages. The op amp DC offsets could be a source of error.

I'd be tempted to throw a microprocessor at it, but that's my answer for almost everything. :roll:

JR

Igor Dockx
Posts: 11
Joined: Fri Dec 26, 2008 5:10 pm

Re: Linearize 'log' VCA's response

Post by Igor Dockx » Thu Jun 15, 2017 3:12 pm

Hi Wayne and John,

I spent a couple of hours behind my computer today and decided to give LT Spice a try. I found a 2181A model and did some simulations.

I can proudly report that Wayne's assumptions were correct; I managed to obtain linear VCA control with two 2181's using the schematics in my first post and inverting the reference current.

I also simulated a variation with 'differential drive' and it works as well; I buffered the output of IC1 using G=1 and G=-1 opamp sections; the non-inverting one feeding the Ec- ports of both VCA's and the inverting one feeding the Ec+ ports. Looks good on the sims, will breadboard as soon as time allows.

John,

I am looking for a way to add VCA automation to my EMT broadcast mixing console. I can buy an Automan system for a reasonable price. Automan is just a computer connected to A/D and D/A converters, developed to be connected to SSL 4000 / 6000 desks equipped with VCA's (google 'Automan Efforts' for more details).

I just don't want to throw out all my existing P&G audio log faders, so I'd rather recycle them and have them control the 0>+5V DC to feed the VCA's, which is read / stored / reproduced by the automation.

I thought this could be an interesting learning experience.

Thanks for your help, I'm sure more questions will pop up as I move along...

PS I can post the LT Spice project if that's of interest to anyone...

Best Igor

User avatar
mediatechnology
Posts: 2797
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Linearize 'log' VCA's response

Post by mediatechnology » Thu Jun 15, 2017 3:45 pm

Glad it worked!
I also simulated a variation with 'differential drive' and it works as well; I buffered the output of IC1 using G=1 and G=-1 opamp sections; the non-inverting one feeding the Ec- ports of both VCA's and the inverting one feeding the Ec+ ports.
Glad you tried that.
I've done that before to reduce control feed through. It's worth the extra op amp.
I was using the VCA as a modulator and the reduction in AC feed-through was significant.

In automation requiring high attenuation, differential drive permits higher "off-ness" since single-ended drive hits a Vbe threshold on the Ec port around -98 dB. Differential drive for the same attenuation only requires 3 mV/dB/port. IIRC JR has done this as well.
Read Damore v. Google and ask yourself the following questions:
Are Googlers really the people you want in charge of your search results, browser, data or operating system?
Would you be comfortable riding in a Google self-driving car knowing that Googlers might want you dead just because they don't like you?
Would you trust a Googler with your children?

User avatar
JR.
Posts: 2140
Joined: Sat May 24, 2008 7:21 pm

Re: Linearize 'log' VCA's response

Post by JR. » Fri Jun 16, 2017 9:47 am

Igor Dockx wrote:
Thu Jun 15, 2017 3:12 pm


John,

I am looking for a way to add VCA automation to my EMT broadcast mixing console. I can buy an Automan system for a reasonable price. Automan is just a computer connected to A/D and D/A converters, developed to be connected to SSL 4000 / 6000 desks equipped with VCA's (google 'Automan Efforts' for more details).

I just don't want to throw out all my existing P&G audio log faders, so I'd rather recycle them and have them control the 0>+5V DC to feed the VCA's, which is read / stored / reproduced by the automation.

I thought this could be an interesting learning experience.

Thanks for your help, I'm sure more questions will pop up as I move along...

PS I can post the LT Spice project if that's of interest to anyone...

Best Igor
OK, I'm too busy (lazy) to research Automan but will ASSume its a 1:1 V-in to V-out.

I am still uncomfortable as a design engineer with decades of experience counting parts cost (I worked at Peavey, enough said).

Committing a VCA to convert an audio taper pot voltage to a log gain control, seems like overkill (read expensive). You kind of want an anti-log conversion, scaled for the VCA control law. Back in the day I would do this with a cheap transistor array, for your budget you could throw a premium dual transistor at it.

Let me chew on this a while... If using two VCAs works and this is one-off why worry about money. :lol:

JR

User avatar
mediatechnology
Posts: 2797
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Linearize 'log' VCA's response

Post by mediatechnology » Fri Jun 16, 2017 10:06 am

I agree that you could do the same thing with a transistor.
On a one-off project or test fixture the VCA linearizer makes it easier.

In this case it is required over a large number of channels so a transistor logger seems almost as simple and definitely lower cost.
Read Damore v. Google and ask yourself the following questions:
Are Googlers really the people you want in charge of your search results, browser, data or operating system?
Would you be comfortable riding in a Google self-driving car knowing that Googlers might want you dead just because they don't like you?
Would you trust a Googler with your children?

User avatar
JR.
Posts: 2140
Joined: Sat May 24, 2008 7:21 pm

Re: Linearize 'log' VCA's response

Post by JR. » Fri Jun 16, 2017 1:26 pm

Probably two transistors, one to log the rectified voltage from the pot, and a second transistor biased up to 0dB to subtract from the first and generate a + and - dB control voltage, for gain at the top of the fader, and attenuation at the bottom... (not unlike the old TS-1 dB meter guts, but without the extra complexity to generate a dB count). Since this is operating on DC there is no concern about op amp loop gain which makes that simpler, OTOH DC offset errors could get significant wrt deep attenuation.

A modern dual transistor should work, and perhaps a precision op amp to manage errors near the bottom of the pot travel.

If commanding lots of attenuation, symmetrical differential control voltage will keep the VCA happier (JP used that in their console fader VCAs).

JR

User avatar
JR.
Posts: 2140
Joined: Sat May 24, 2008 7:21 pm

Re: Linearize 'log' VCA's response

Post by JR. » Sat Jun 17, 2017 5:01 pm

OK here is crude conversion of 0-5V from audio taper fader (0V = full attenuation, 5V = +10dB gain).

91k pull up resistor should deliver 0V (unity gain) control voltage when fader V is roughly 1.6V (-10dB from full scale) , but can be scaled up/down that's what prototypes are for.

Output will be invalid if 0-5V fader voltage is ever negative, so never do that.
fader.jpg
JR

PS: I'm not sure about scaling of log generated by transistor junctions vs, VCA control V... IIRC it's around 6mV/dB but it has been decades since I actually designed something like this (while I have doodled around).... something for you to figure out the details...you could use a 4th op amp section to buffer a pad or scale up.

User avatar
mediatechnology
Posts: 2797
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Linearize 'log' VCA's response

Post by mediatechnology » Sat Jun 17, 2017 7:47 pm

Thanks for posting the schematic John.

The VCA scale is 6.1 mV/dB typically vs. the 3 mV/dB of the junction so the output buffer would need a gain of about 2.
Read Damore v. Google and ask yourself the following questions:
Are Googlers really the people you want in charge of your search results, browser, data or operating system?
Would you be comfortable riding in a Google self-driving car knowing that Googlers might want you dead just because they don't like you?
Would you trust a Googler with your children?

Post Reply

Who is online

Users browsing this forum: Bing [Bot], Google [Bot] and 6 guests