where is the noise coming from

[pucho812]

Wasn't sure if this would go in the lab or drawing board.
What are potential sources for increased noise floor and THD in a large format desk?
Using our studio desk as an example, I have been working on the studio desk and taking measurements with the AP.
On any given channel going from the channel input(mic or line) to the direct out, I have a measured noise floor of around -80dBu and THD+N of around .007%. This is with a +4dBu output signal, line input of +4dBu, and or a mic level signal so that output is +4dBu. These are numbers are nice. I have seen better in other equipment such as the Avalon gear with a -91dBu noise floor, but these numbers are good considering the equipment was designed in 1976 and maintained well for a long time.
However if I take that same channel and assign it to the stereo bus, or to an aux or to a multitrack bus, using the same signals and level I measure an increase in noise floor to around -50dBu and THD+N to .5%. I am lucky in the fact the noise is just hiss. But I am wondering why there is such a drastic increase. I know from the channel to the stereo bus there are extra active electronics, buffers, gain stages, etc but I can't wrap my head around why there is such an increase in noise. From there wondering what could be addressed to improve things.

 

[JohnRoberts]

I have written one or more articles about this but none this century.

In a console, or any active gear, noise is generally associated with high voltage gain. So logically mic preamps are an obvious suspect. Next comes sum bus amps with multiple channels being summed together. The noise gain of a typical virtual earth sum amp is N+1, so summing 16 channels is like running a voltage gain of 17x.

Is the noise floor increase the same if you assign that input channel to the L/R bus, as to an aux send, or a sub master bus. I have seen measurable differences between those even in well designed consoles.

Different bus topologies can exhibit varying noise floors.

JR

PS: There is typically a +10dB fixed gain stage post channel fader so that will elevate noise floor some.

 

[pucho812]

I have written one or more articles about this but none this century.

In a console, or any active gear, noise is generally associated with high voltage gain. So logically mic preamps are an obvious suspect. Next comes sum bus amps with multiple channels being summed together. The noise gain of a typical virtual earth sum amp is N+1, so summing 16 channels is like running a voltage gain of 17x.

Is the noise floor increase the same if you assign that input channel to the L/R bus, as to an aux send, or a sub master bus. I have seen measurable differences between those even in well designed consoles.

Different bus topologies can exhibit varying noise floors.

JR

PS: There is typically a +10dB fixed gain stage post channel fader so that will elevate noise floor some.

What is interesting is the following input source of mic or line on any given channel yields around the same results at the direct outputs, bus outputs, and auxes provided that the level be that at output we have +4dBu. Between the stereo bus output vs the multitrack bus assignment noise level is negligible, within a dB or two and the THD+N is about the same. There is a difference between the stereo bus and multitrack bus outputs VS the aux main outputs. There is also a difference in auxes, enough to be noticeable when compared to each other but not enough to be an issue in the grand scheme of things.
Yes, I do know about the 10dB gain stage at the fader in most desks. Ours has that option on the channels but also like most desk the stereo bus fader tops at unity.

I'll have to trace out some things and see what is what to fill in the hand drawn schematics of those sections.

 

[Brian Roth]

How many channels are connected to the buses under investigation?

Bri

 

[pucho812]

How many channels are connected to the buses under investigation?

Bri

for testing purposes only 1 channel assigned, but the desk is 48 channels with 24 bus, stereo bus and 24 monitor split desk
. usually when mixing just

 

[Brian Roth]

Older designs often had every input always connected to every bus even if not used. Hence the noise level would be increased at all times.

Here is an exercise to think about:

Assume 10K resistors feeding into the bus and a unity gain bus summing amp (10K feedback resistor).

In effect, 48 channels feeding the bus via the 10K resistors results in a result of 208 Ohms in the equation. 10K feedback R and 208 Ohms calcs to a noise gain of 48X which is 33 dB of "noise" gain.

Bri

 

[JohnRoberts]

I won't speculate about the design but asked if the L/R bus had a higher noise floor than aux or subs that would not have all channels assigned.

JR

 

[RSRecords]

Wouldn’t this depend on the summing architecture? Or Is obviously a virtual earth configuration? I’m genuinely asking!

 

[JohnRoberts]

Wouldn’t this depend on the summing architecture? Or Is obviously a virtual earth configuration? I’m genuinely asking!

That's why I asked more than once what is the bus noise floor from the different summers. In a typical split console there is a main or L/R bus that everything can be assigned to, sub masters that generally only get one or two sources assigned to those buses, and aux /efx sends that get less than all feeds assigned.

In my console article I wrote back in 1980 I discussed different sum bus approaches and more. old console article caveat I spelled bus wrong back then.

JR

 

[Sorr]

John, I read your article with interest. What sort of current sources would you use today?

 

[JohnRoberts]

I have discussed this here before. With decades of hindsight I wouldn't change the general strategy (5 resistor synthesized op amp current sources). With cost no object there are better newer op amps but the TL07x series I used was around 3uV input noise so not bad. The last console I used this in Peavey/AMR production series has schematics published.

here is cut from the full schemo current source in fact here is a thread with way too much discussion. big mix thread while this thread veers off into sundry directions.

Back to the topic of my current source summing, I really wish there was a practical need for a modern version but guess what? Using digital technology we can combine unlimited numbers of inputs with no loss making a fancy sum box moot.

JR

 

[Newmarket]

Back to the topic of my current source summing, I really wish there was a practical need for a modern version but guess what? Using digital technology we can combine unlimited numbers of inputs with no loss making a fancy sum box moot.

I know this is your view on this ie Digital Summing makes it moot at least.
I'd say that there are still instances where "Top Notch" summing is a goal in analogue environment without ADC / DAC conversion.
Maybe not for a sizeable commercial concern but maybe for "niche" applications and esp DIY.

 

[JohnRoberts]

I know this is your view on this ie Digital Summing makes it moot at least.
I'd say that there are still instances where "Top Notch" summing is a goal in analogue environment without ADC / DAC conversion.
Maybe not for a sizeable commercial concern but maybe for "niche" applications and esp DIY.

For combining a modest number of channels the noise gain of a well executed virtual earth sum bus using modern very low noise technology op amps should be OK.

My last big un had over 100 feeds into the L/R bus so noise gain was a concern to me.

JR

 

[Newmarket]

Yes. I know it's not really a problem. But there is a certain logical elegance to the what I'll term "True Current Source" summing that appeals.
Maybe I'll DIY it someday. But running only a small desk for analogue summing there is a lack of incentive. That does sort of reinforce your comment

 

[JohnRoberts]

Yes. I know it's not really a problem. But there is a certain logical elegance to the what I'll term "True Current Source" summing that appeals.
Maybe I'll DIY it someday. But running only a small desk for analogue summing there is a lack of incentive. That does sort of reinforce your comment

We kicked this tire around the block for a while over on Wayne's forum years ago.

For a modest number of channels we could just about feed the current outputs of VCAs directly into a VE bus - input. Modern SOTA VCAs (like THAT) have respectable noise and distortion at unity gain, and when attenuating also reduce their noise output. This would also obviate the need for the +10dB post channel fader gain stage. The most obvious concern is the output capacitance of multiple VCAs in parallel hanging off a VE bus, but I speculated that using a 2 pole active LPF as the combining amp could lump that capacitance together into one of the two filter caps.

Note: I was never interested enough to melt solder, but I did kill a bunch of brain cells thinking about it. To repeat I really wish this was a viable stand alone product but in my judgement I would have to exaggerate the need for this to market it, and I don't roll that way.

JR