Balanced Inputs

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
Post Reply
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by mediatechnology »

Thanks ricardo.

I'm going through my old "Ideas for Design" clippings and have found numerous gems.

"Composite Instrumentation Amp Extends CMRR Frequency Range 10X"

Image
"Composite Instrumentation Amp Extends CMRR Frequency Range 10X," Electronic Design, February 4, 2002.

See, nothing really is new. We keep re-inventing the same stuff. I've got at least three more examples of cross-coupled differential amps. I had considered the third stage but the CMR of just two (or a single 124X) is good enough. U3 doesn't have to have that good of a CMRR - I think I measure about 6-10 dB improvement when I originally tried it. If you really wanted to go over the top, the input legs of U3 could be made dual pots with an op amp for "balanced" gain trim. Though you normally couldn't get away with that from a CMR perspective with a single stage, with CMR already realized by U1 and U2, the poor CMR of U3 matters little.

The authors don't point out the impedance balance. For general-purpose INAs I'm not sure that matters as much as we think it might - I repeat might - matter in audio since the sources are often floating transducers. The balanced loading requirement may be a broadcast BBC thing. Jung and Birt seem to think it matters, Self doesn't buy it.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by JR. »

lets see a 3 opamp inst amp inside each package... 9 opamps total... pass....

Paralleling two actual "instrumentation" amps seems horribly redundant.

I started out reading ideas for design to absorb input, but after a while I became more critical.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by mediatechnology »

lets see a 3 opamp inst amp inside each package... 9 opamps total... pass....

Paralleling two actual "instrumentation" amps seems horribly redundant.
Lets clarify for our readers that John is talking about the AD623 shown in the example. I'd pass on 9 op amps too. The "point" is another cross-coupling example, not using the AD623 literally as a line input.

Image
"Composite Instrumentation Amp Extends CMRR Frequency Range 10X," Electronic Design, February 4, 2002.

Image
AD623 Simple Block Diagram

The first two stages, if they were THAT1240's as in the Super-Duper-Bal are only two op amps.
I started out reading ideas for design to absorb input, but after a while I became more critical.
Yep, some are not that great. Some are good ideas but poorly implemented. Some are truly great and have niche applications the author never intended. Some require looking at a bigger picture and not the part number used. Some are silly. Some cause you to ask: "What were they thinking?"

I admire the work of Williams and Pease. There's another guy by the name of Steve Woodward whose work I'll be posting elsewhere. Steve does some really clever things. In two of his articles Steve discovers the pan pot and shared gain topology. I think he may have been late to that party.

My point in posting the AD623 circuit is that a lot of examples exist for the cross-coupled differential input. There probably are as many reasons to do it as there are examples. Cohen was the first example of double-balanced that I've seen in audio. Cohen was an RF guy in his day gig. He may have "borrowed" the double-balanced diff amp from the double-balanced mixer. Who invented that?
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by JR. »

I guess I'm a little cranky today (every day?).

Yes an instrumentation amp has a differential input, so cross connected instrumentation amps are literally cross connected differentials, but there is no practical benefit to this topology other than selling more ICs for the app writer's employer. Nobody in their right mind would (should) do this, while I guess it serves as a smoking gun example for prior art (sort of).

Cross connected simple differential amps do provide real benefit in averaging out + and - input impedance differences. For instrumentation amps there are no + and - input differences to average out (or not supposed to be any).

JR

PS: yes I was a fan of the good regular submissions. I voraciously devoured this stuff religiously in my early days, but after a while I started seeing the same ideas getting recycled, and stuff like this (IMO a naked merchandising attempt that no real engineer should fall for).
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by mediatechnology »

I've got six Banker's Boxes, each 15" deep of IFD tear-outs. I've been through two boxes already culling out stuff to scan.

So far I've found at least 6-10 examples of cross-coupled diff amps that I wasn't even looking for.

As you might expect most of this stuff goes back into the box. I'll be posting a lot of it in "Document." Please don't think I'm obligated to defend them because I posted them. I particulalrly don't want to defend the pink noise generator made by scuffing a microphone across turntable felt. Yes, it was actually published... ;)
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by JR. »

maybe you can find somebody you can send that stuff to... :D

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by mediatechnology »

Thanks for what you John and crazynightowl (also a member) sent me.

Some stuff that I thought would be relevent still is. There are a few gold nuggets that make going through this worthwhile. I think I have four Audio Handbooks: Yours, crazy nightowl's, my NSC and a copy from Radio Shack. Having all of this scanned, indexed by Google, sharing them with others and then getting rid of the massive amount of bulk will be very rewarding. Finally, I get what the "portable" in Portable Document Format "pdf" really means. :D

The articles on how to design with 64K dynamic RAM that I saved don't seem as useful as they might have been. I'm thinking they might be "re-purposed" as recycled "post-consumer" content. Those will become the little cardboard thingies that cushion your cable modem when you open the box.

I've got one balanced input I found in an IFD that's so far out there I have to share. It's so "out there" that it might actually work. One driving leg is connected to a virtual ground node, the other driven leg "sees" a mirrored current source. It balances not input impedance, but input current. I culled it, and it's near the bottom. Clearly out-of-the-box thinking. But, in a number of ways not too different than the current loop interface a mic's emitter follower output stage is when driving a remote phantom pull-up resistor as the load.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by JR. »

It's not a crazy design target to desire zero net current between chassis (like transformers), so it's of interest....... but as you know there are so many flavors of quasi pro interfaces that you always have to be careful of what you might encounter that could trip up the best laid plans. On thing that my experience at Peavey developed in me was a sensitivity for what customer may try to connect to your interface (namely anything).

I've come to the conclusion that grounds are best treated like like sewer pipes and audio the clean hot and cold water pipes we want to keep pure.

=======

I wasted some time and brain cells in console design looking for ways to keep signal current deltas small to help reduce crosstalk. I always ended up with sewers and differentials.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by mediatechnology »

Me:
One driving leg is connected to a virtual ground node, the other driven leg "sees" a mirrored current source.
John:
It's not a crazy design target to desire zero net current between chassis (like transformers), so it's of interest.......
Your comment made me think that it might have merit as a balanced summing amp approach. And to be honest with you have no clue as to why. :? We've discussed current summation before...feeding an opposing current into the other leg may make some sense if it impedance balances the bus. One leg in an unbalanced bus is a virtual ground node. It would seem intuitive that the other leg should be forced to mirror it. Hmm...
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Evolution and Origin of the SuperBal Variable Gain Input

Post by mediatechnology »

"Amplifier and Current Source Emulate Instrumentation Amplifier," EDN, November 13, 2003.

Remember that Fletcher used the Super Bal in a mix amplifier. I suppose that the original driven reference leg summing amp, the "1972" whiteboard version, was fixed gain. The purpose must have been to equalize the input currents.

Birt (and Self) say the same thing: The driven reference leg is to equalize the input currents.

This equalizes the input currents:

Image

http://www.edn.com/contents/images/111303di.pdf#page=6

This time the INA134 is driving the reference leg and its reference leg is also being driven.

In figure 2, the output can be taken from IC2A. (Ignore the biomedical filter to the right in figure 2.)

Would this basic topology make a good current-domain mix amp?
Post Reply