Fully balanced mixer design??

[Pat Redmiles]

Hi all. I'm messing around with mixer design, and I'm wondering if any of you have evaluated fully balanced designs vs. unbalancing the signal in the mixer and working with that signal, rebalancing it prior to the outputs. Keeping the signal fully balanced is, of course, a PITA, doubling component counts and requiring exotic things like four-gang pan pots. But, is there an inherent increase in sound quality due to keeping it fully balanced? Certainly, noise is less, but I'm concerned about transparency and transient response and that sort of sound quality considerations. Some of the best-sounding gear I've heard uses fully balanced signal processing, so I'm thinking how that might apply to a mixer. Any big-name consoles use this topology? Thanks!!

[TRW]

It has merits but usually gets outweighed by cost. Doubles all component counts.

I believe Mike @ Wunder is sporting a fully balanced routing module with balanced pan pots, claiming he had an advantage of high kill levels on hard pan due to the differential design, I imagine some form of polarity cancelling panner?

Personally, its worth trying maybe but in my mixer design I couldn't afford it!

Good luck.
-Tom

[mixedupsteve]

Originally Posted by Pat Redmiles

Keeping the signal fully balanced is, of course, a PITA, doubling component counts and requiring exotic things like four-gang pan pots.

Plus all the components need to be perfectly matched with each other and remain so through the life of the equipment.
I've never heard of any improvement in sound quality in such a design.

[JohnRoberts]

Originally Posted by Pat Redmiles

Hi all. I'm messing around with mixer design, and I'm wondering if any of you have evaluated fully balanced designs vs. unbalancing the signal in the mixer and working with that signal, rebalancing it prior to the outputs. Keeping the signal fully balanced is, of course, a PITA, doubling component counts and requiring exotic things like four-gang pan pots. But, is there an inherent increase in sound quality due to keeping it fully balanced?

Not that I know of or could imagine.

Certainly, noise is less,

How? (OK, in theory you now have two parallel audio paths so the max voltage is now twice the nominal IC process limit of 38v or so. Since the noise floor of the two parallel paths is similar and incoherent, you pickup +6dB - 3dB= 3dB better S/N. Nice for the marketing wonks to write up in the literature, but the S/N of the simple line level audio paths are not the limiting noise constraint in console design. )

but I'm concerned about transparency and transient response and that sort of sound quality considerations.

Twice as many opportunities to do something wrong.

Now the entire internal path needs to be precisely matched to each other and wrt stray reactance to ground. You can't just trim it out at the final output since there are so many possible permutations for signal flow internally. The practical expense is not just two times the internal components but the need for high precision components throughout, or internal trims to rebalance at the end of each unique internal path.

Some of the best-sounding gear I've heard uses fully balanced signal processing, so I'm thinking how that might apply to a mixer. Any big-name consoles use this topology? Thanks!!

Not that I know of.

There may be consoles using a balanced bus topology, specifically for the bus summing section, due to the number of different local audio grounds involved and physical distance spanned. There is merit electronically in doing this section differentially, but calling it "balanced" will impress customers far more in the marketing effort than just calling it differential.

OTOH making the ground side of a differential summing bus with lower impedance resistors will reduce their Johnson (thermal) noise contribution, [EDIT: it also reduced the noise contribution of the active summing amp's input noise current times the termination impedance. /EDIT ] so this "unbalanced" differential bus topology is lower noise and IMO superior to a balanced bus (I have used this approach in the past).

I'm sure opinions vary but electrons don't have opinions.

JR

[TRW]

Great post John!

I just don't think its worth the cost or design time.

-Tom

[Matt Syson]

Hi
There are many desks that use 'balanced' or at least symmetrically driven mix busses, the AMEK BC3 and 9098 desks among them. Since it is advantageous to at least have 'partial' balanced bussing, to reduce the amount of rubbish that the physical layout that a large console 'attracts' it is a relatively simple step to actively drive signals to the bus from the channel strips. As stated earlier this benefits noise and headroom. To have the EQ 'balanced' would be a nightmare, even the old Pultec designs use transformers to unbalance and then rebalance the actual EQ network.
Activities like aux level pots would have to be very expensive pots or proper attenuators to maintain 'balance' as their operation must not 'unbalance' the sources.
Anyone can make a single channel strip work. Stereo is also feasable but when you get into more than a couple of channels all the problems suddenly multiply if you want decent performance.
Matt S

[JohnRoberts]

Just to continue this dialog, the bus summing amp is one of the critical throttle points for console performance (mic preamps another). Noise is the obvious manifestation of the difficulty, but you can also get phase shift and distortion associated with this heavy lifting of combining several tens of signals.

It is perhaps worthwhile understanding what balanced is and isn't. Balanced terminations or signal paths are basically the same impedance to ground in both + and - legs, then differentially combined before using. The result is always single ended at the final output. You could carry a differential signal all the way to your loudspeaker, because the speaker itself is a differential receiver and will ignore any common mode signal present on both leads. The primary reason for "balancing" external interfaces is so noise interference will be picked up equally in both signal legs, for perfect cancellation in the following differential receiver. Inside a console (at least the ones I've seen), we are fully inside a metal chassis, so the only interference we should need to worry about is what we create. Crosstalk is always a concern inside high performance consoles, but in my experience this is almost always layout related. A fully balanced internal path in theory should also cancel crosstalk, but I will probably have bad dreams just thinking about this.

There is nothing wrong with trying to pick up 3dB of S/N in a bus (from the coherent signal vs incoherent noise), but I am suspicious that in practice the real improvement is less, and this is mostly a cosmetic concern as one mic channel at nominal gain will swamp this noise floor in a well executed design. (I would not ignore phase shift and linearity of bus response which does not get swamped by input noise.)

This is probably TMI for the OP, but a subject that console designers have pondered for decades. Design philosophies differ and a well executed "balanced" or symmetrical bus is fine if budget is not a concern.

JR

[Matt Syson]

Hi
AHAH the old 'metal chassis trick', where you use steel panels to 'attract' stray magnetic field and 'concentrate' them onto imperfect mix buses. Any form of hum / interference rejection is handy even a meagre 20dB at power line frequencies. Of course trying to get rid of 16KHz CRT monitor noise is fun too, as steel doesn't work so well at this frequency.
Signal following ground (or vice versa) reduces interferance pickup (and radiation) but gets a bit mind boggling when you have more than a decent handful of channels with signals being required in several places. Some desks use the 'brute force' plan of 3/4 inch diameter copper bars to keep things 'sane' but then with attention to detail a 'differential' mix bus can reduce the need for the third mortgage on the copper.
All amplifiers are 'differential' inputs, although not necessarily impedance balanced. An amplifier magnifies it's input signal with reference to the ground it is presented with (the difference). If the ground is 'jumping' but the input is tied to something else, it will simply magnify the ground jumps.
Matt S

[JohnRoberts]

I don't know about metal chassis being a "trick" or completely adequate especially for that one studio in VA located in the main beam of an AM antenna. While in that case the problem was coming in the inputs, and outputs.

The last console I did (RIP) was some 20 years ago and that used (IIRC) steel panels up top and a couple layers of aluminum below (bummer).

I actually used a big old hunk of copper bar stock for a ground bus in my very first console way back when, but I'll freely admit now that was out of ignorance. Grounds need to be understood and finessed, not brute forced (differentials are your friend).

----

My best computer monitor interference story was in a more mundane product. I was helping a junior engineer chase a 1 Hz noise out of a powered console (MI product). It was difficult to imagine where 1 Hz was coming from but it finally made sense when we could make it go away by turning off the engineer's computer monitor. "or" the monitor in the office next door. The 1 Hz was a beat tone between the two monitor's raster, so instead of looking for 1 hz we needed to clean up the HF rejection.

JR