Building the Forssell ACN schematic.

[babyhead]

Hello-

I have sold my mixing console and want to build this:

http://www.forsselltech.com/downloads/schematics/Summing%20Buss2.pdf

with reference to this:

http://www.forsselltech.com/downloads/design_discussions/summing_buss.pdf

24 channels, 10 with faders.

As I understand it, each channels ground would pass though a 1k resistor. To unbalance the signal I was planning on a 10k resistor between + & - on the switched channels and a 10ka pot with the wiper going to a l/r-c switching network on the fader channels. For the layout, all mute grounds would collect at the - input and pass though the 1k to a single ground bus. I think I am going to build all of this on the front panel.

Should I lower the value of Rsum for the channels with faders to compensate for the level difference? Or maybe raise Rsum on the straight in channels to make up for the lack of a fader?

As for the output, I'd like to substitute 49.9k with a 10ka pot with the wiper tied to a Jensen 10k:10k to balance the signal and correct the phase.  Will this mess with the servo?

What kind of problems can I be running into?

The most it will ever have to drive is my Rosetta.

Thanks all. This should be fun. His opamps sound wicked.

 

[babyhead]

Shameless self serving bump.

My main query involves the layout. It seems like the pan/mute grounds would node to the - input and then pass through Rgrnd (1k) to the summing ground bus. Can anyone tell me if I understand this correctly? Otherwise Rgrnd would be a single resistor 40ohms.

Also, Hardy reduces Rsum to 3.16k on channels with what I assume is a 10ka pot. With JH, Rsum is 10k, with FF it's 15k, so adjusted Rsum for a channel on the FF circuit with a 10ka pot is..... 4.74k? Shot in the dark.

One thing at a time, but this is getting cued up.

 

[Svart]

I've read through your posts about 3 times and I *think* I understand what you are asking..   :-\

As I understand it, each channels ground would pass though a 1k resistor.

I think you might be taking the schematic a little too literally.  FF is only showing you the portions of the schematic that he is concerned with.  The Rgnd resistors are not the only ground to the channels.  you are referencing the whole Non Inverting input node to the characteristic impedance of the bus in a very similar way to the signal bus.  You want the opamp to see all of the traces and impedances as equal therefor canceling out the effects of the signal traces themselves.  Kinda like using balanced cable instead of unbalanced cables. 

He is pseudo-balancing the bus.

You need to be aware of the values of Rsum and why they are different.  This has nothing to do with the pot value at all.  Fred's opamps are FET input opamps.  They don't need a lot of current on their inputs like BJT opamps need.  This allows you to increase the value of Rsum and lighten the load on the opamps that are driving Rsum on each channel.  Just because an opamp isn't outputting a signal doesn't mean that the output section isn't active.  An opamp's output section is using power to keep the output signal at zero or whatever it's supposed to be.  When you feed a signal out of an opamp on say channel 1(of a 24 channel mixer), then that opamp is attempting to drive the output sections of the other 23 opamps.  This would be impossible if you didn't have current limiting resistors like Rsum on each channel.  Going higher in value reduces the amount of power each opamp outputs in order to drive the bus with 23 other loads driving it as well.

This can do a lot for your signal but you have to be very careful to match the loads carefully.  This is why FF mentions the equation for figuring out your summing resistor for a given load and number of sources.  Going too high or too low for a load will cause general wonkyness.

I would use a dual opamp, one side for gain before the fader and the other side for a unity gain buffer into Rsum.  For noise structuring you want your gain before your attenuation.

I urge you to read Fred's discussions on summing amp design, they explain what I've just touched on with greater detail.

 

[babyhead]

Svart-

Thank you for your reply. I was wondering why FF would draw the schematic as such when he could have written it with a single resistor. As I was rereading the discussion I came across this:

The ability of the opamp to reject these common mode signals is called the common mode rejection performance of the opamp and is often expressed as a ratio (CMRR). (When expressed as a ratio, it is compared to the voltage gain of the amplifier.) If we can get the ground noise to be the same kind of noise present at each channel, we can use this CMRR to help reject the overall noise.
This is done by running a ground trace along with the summing buss signal trace and tying to each channel’s ground via a resistor

It seems like he was referencing the schematic to this layout trick.

I am not looking to make anything too elaborate. My main mission is to be able to parallel my outboard gear to the main mix in PT. I have seen simple mixers unbalance the input with a 10k pot with the wiper switched to rsum. No buffers. If I can get away with only using the 992s with servos, that would be ideal.

I understand the output fader question I had as explained in another thread. Unity will do. Impedance balance will do. This makes the schematic stock...

Gracias Amigos...