Sum bus theory and practice

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Sum bus theory and practice

Post by JR. »

As promised here is a new thread about summing. While I already have one or more threads kicking around about my current source summing, we can use this to discuss sundry practical approaches (like new THAT chip etc.).

For now I will provide a link to my old console article which IMO still holds water almost 30 years later, and basic opamp theory is always useful.

http://www.johnhroberts.com/des_art_1.pdf

This article will give you something to chew on, please try to digest this first.

Happy Thanksgiving,.

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: Sum bus theory and practice

Post by mediatechnology »

John - Thanks for posting this.

To our readers:

The summing bus discussion in John's article begins about here: http://www.circularscience.com/des_art.pdf#page=4

I did figure 13 with a 1570/1246 and it worked pretty good. I used a servo to bias up the bases to remove the -Vbe offset at the emitters of the 1570. I've also done it with a 1512 and it worked equally well but with lower CM rejection. (The 1512's internal diff to SE amp is untrimmed.)

FIgure 14 http://www.circularscience.com/des_art.pdf#page=6 ties into our thread here: viewtopic.php?f=6&t=183
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Sum bus theory and practice

Post by JR. »

Yes, that is a very respectable sum bus amp approach. The transamp topology (some call this Cohen.. I don't) buys you an open loop gain characteristic that increases as you throw more inputs onto the bus increasing the noise gain, so we get a constant high, loop gain margin for very low phase shift and low distortion (more important IMO than noise which is dominated earlier in the chain).

But please read the first several pages of the article.. that is basic opamp circuit design 101...

JR

PS: I have some new ideas about summing even since this (it has been almost 30 years) but the info shown here is certainly adequate to make a summing bus, better than the rest of the path.

PPS: Wayne,, I have made the full article one link, so the page calls just redirect to the full article.
Cancel the "cancel culture", do not support mob hatred.
juniorhifikit
Posts: 139
Joined: Sun Oct 04, 2009 8:37 am
Location: Oakland/Paris

Re: Sum bus theory and practice

Post by juniorhifikit »

mediatechnology wrote:
I did figure 13 with a 1570/1246 and it worked pretty good. I used a servo to bias up the bases to remove the -Vbe offset at the emitters of the 1570. I've also done it with a 1512 and it worked equally well but with lower CM rejection. (The 1512's internal diff to SE amp is untrimmed.)
Any chance you could post the schematic of your build with the THAT components for us less experienced folks? :D
juniorhifikit
Posts: 139
Joined: Sun Oct 04, 2009 8:37 am
Location: Oakland/Paris

Re: Sum bus theory and practice

Post by juniorhifikit »

JR had said in another thread:
I am only half joking when I see this as a THAT showcase project.. that balanced line receiver, a pair of VCA's, VCA bus amp, and THAT balanced output.
...and I found this while looking at the THAT 1600 data sheet:
Image

...but in regards to some ideas posted in other threads about current summing using a vca in the NFB loop of the summing amp; and JR's paper about current summing and trans-amp design; It's all a little unclear to me. Could the THAT 1606 in the above drawing act on it's own as the summing bus in this current summing concept?
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Sum bus theory and practice

Post by JR. »

I am not familiar with what's inside the 1606 but I am guessing some precision resistors, so not appropriate for a bus amp as it is expecting voltage inputs.

My current summing bus concept would involve combining multiple 2180 outputs into a single virtual earth opamp (the 4570 in that schematic).

One aspect I need to prove, is that i can manage the accumulated output capacitance of tens (hundreds) of VCAs. Note: it would also involve two VCAs per channel, one per bus for L and R pan functionality.

I have hypothesized using a 2 pole active LPF topology for the virtual earth sum amp, so this bus capacitance becomes one of the two filter poles and is unconditionally stable.

I think I described the filter in more detail in one of the other threads, while final values are still subject to bench testing and adjustment.

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: Sum bus theory and practice

Post by mediatechnology »

Here's one version of the summing amp I did where the servo samples the emitters:

Image

This one samples the output to pin it to DC baseline:

Image

The bases, pins 7 and 10, are biased up one Vbe in both examples.

The 120 R represents approximately 16 assigned inputs using conventional resistive build-out.
My current summing bus concept would involve combining multiple 2180 outputs into a single virtual earth opamp (the 4570 in that schematic).

One aspect I need to prove, is that i can manage the accumulated output capacitance of tens (hundreds) of VCAs. Note: it would also involve two VCAs per channel, one per bus for L and R pan functionality.
The current outputs of the 2180 are ideal for driving the 1570's emitters directly. IIRC the output capacitance of the 2180 is about 15 pF. The 2162 is lower maybe 3-5 pF? This may require some clever tricks to avoid particularly with a CFB amp when a large number of channels are required.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Sum bus theory and practice

Post by JR. »

I am apprehensive about putting any capacitance on the emitters of CFB amps.. I have been thinking of a more conventional opamp for the current sum amp, since it is operating at much reduced noise gain, and must be unity gain stable to support the capacitive feedback to mitigate input capacitance, or MFB filter topology.

The 1570 looks fine for conventional resistive sum bus.

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: Sum bus theory and practice

Post by mediatechnology »

I am apprehensive about putting any capacitance on the emitters of CFB amps.. I have been thinking of a more conventional opamp for the current sum amp, since it is operating at much reduced noise gain, and must be unity gain stable to support the capacitive feedback to mitigate input capacitance, or MFB filter topology.
True. Hadn't really thought about that angle for the current-based approach. The emitter is certainly not the place for stray capacitance and the noise gain issues of conventional resistive summing help the 1570 to shine in this app. When I did the circuits above (I should probably show the 1512 version) I had in mind a conventional resistive bus structure. Those circuits were quick proof-of-concept. I'd like to see someone tune them.

Just off-the-cuff but what about added series inductance from the bus(es) to the emitter(s) in the current sum approach? I had pretty good results with inductors in series with both Rgain legs in preamps. bruno2k also used them in his preamps. Wouldn't that help to isolate the VCA output C from the CFB summing node at the emitter?
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Sum bus theory and practice

Post by JR. »

I have used a few ferrite beads at mic inputs but never tried inductors to decouple bus capacitance. In principle it should work, not unlike the inductors inside the classic Deane Jensen opamp (i.e. tuned well above audio range). My sense is that there is not enough natural inductance in a physical bus to do any good, due to distributed capacitance between adjacent buses or interleaved grounds (while I concede I haven't looked at the characteristic impedance of ribbon cables and common physical bus approaches).

I have actually bread boarded the MFB LPF into a console bus years ago, and it worked, while I don't recall the cut off frequency I ended up with.

JR

PS: While an obscure factoid, I did once encounter a small 4 bus mixer where the physical bus just happened to make a remarkably effective RF antenna, and we were getting rectification in just that bus amp, nowhere else inside the mixer. Since it was a modest number of channels and in full production at the time, I made a running fix by swapping out the bus opamp with a BiFET, that had a large enough input Vth (input linear range) to not rectify the RF.

PPS: I have actively bootstrapped the shield when using gain pots at some distance from a CFB mic pre to reduce shield capacitance effects. I have never been a big fan of inductors or transformers in high performance audio paths, while I have used more than I can count in constant voltage install products.
Cancel the "cancel culture", do not support mob hatred.
Post Reply