new project: small home recording console

[Dimitree]

thanks for the precious advices!

First, from what I see your Aux sends are going to short the Aux busses when turned fully down affecting the bus by the position of the wiper relative to ground

this part is confusing me.. All the schematics I've seen when multiple Aux were used, were almost identical to what I did, as far as I can understand. There was a pair of tracks (one before and one after the fader) going to a switch, then the potentiometer wired as voltage divider and last the mixing resistor. What am I missing?

Second: Your send and return for the Insert leaves the send/return open

that was intended, but as you mentioned sidechain processing, I reckon it was a mistake, so I will use half-normalling send-returns.

you seem to have opted for balanced return which makes it tricky if not impossible normalling unbal out to bal return - do you really need balanced return?

I thought I had drawn both the send and the return as balanced. I thought the "Zero impedance output" that I used on the Send (and for Direct Out too) was impedance balanced. But probably I misinterpreted that configuration.
I don't think I need balanced send and returns. It's just that since I'm trying to use quad OPA1679 everywhere I can (the more ICs I get, the less I pay for them) I had two spare opamps so I thought about making the insert balanced.

 

[RoadrunnerOZ]

this part is confusing me.. All the schematics I've seen when multiple Aux were used, were almost identical to what I did, as far as I can understand. There was a pair of tracks (one before and one after the fader) going to a switch, then the potentiometer wired as voltage divider and last the mixing resistor. What am I missing?

What value is the final resistor? What value is the Aux pot?
The problem you have is that if the send resistor value from pot wiper to the Aux bus is low then it acts as a shunt to ground - with all Auxes turned down for say Aux 1 on all but one channel, then you have a lot of resistors in parallel on Aux 1 bus to ground.
So for even a 10KΩ feed resistor with 15 out of 16 going to ground you end up with 666.66Ω to ground. With say the pot top going to the bus using a 50K pot then you’d have a constant 3,125Ω with 16 in parallel and the master Aux bus level doesn’t change with wiper positions on channel Auxes. You’d need to have a reasonably high feed resistor to the wiper if reversing so you don’t drag down the channel level as the pots are turned down so it would be advisable to have a pre and post op-amp for Aux send before the send pots. See example of Mackie mixer:

 

[Dimitree]

thanks for this schematic, I didn't know this approach, I always saw the aux pots connected to a bus via 20K log pots and 10K bus resistors, and their signal being taken directly from the fader and/or after the postfade amp, just like I drawn.
Now that you let me notice that, I re-read the articles by Steve Dove - Console Design, and I noticed that indeed he used a pair of opamps before the aux sends.
This looks different from the Mackie approach that you linked. Mackies seems to invert the signal (probably they invert it again after the summing?) and drive the bus with the pot after the summing resistors. Steve Dove shows the signal attenuated before the opamp, then boosted again before hitting the sends.
Which one is better, and why, in your point of view?

Also, do you see any issue with the PFL switch I drawn? does it load the signal too?

 

[RoadrunnerOZ]

The Mackie method doesn’t have varying input loading on the master Aux bus, just on the output of the Aux drive op-amp on each channel. Maybe a slightly better way to go.
Some consoles use a different approach again by having each Aux on each channel driven by an op-amp with the input to the op-amp fed by wiper of Aux level pot. Neve 5106 Aux module (note odd numbered pot shown on one board - the other even numbered one is on another board, op-amps are all on the one board - 2 board sandwich)

 

[RoadrunnerOZ]

Also, do you see any issue with the PFL switch I drawn? does it load the signal too?

If I’m seeing correctly, yes I think so.

 

[ruffrecords]

I think there is some confusion about the way buses work. First it is not clear to me whether you plan to use a virtual earth bus or a passive bus.

If it is a virtual earth bus then the bus impedance is already extremely low ( maybe a few ohms or tens of ohms) so adding at most 600 ohms in parallel with it make no difference.

If passive mixing is being used then you need to ensure that the AUX sends for instance are driven from a low impedance source. I that case the worst case source impedance at the pot wiper is 25% of the pot value. So the bus load varies between the bus resistor (AUX at zero) and bus resistor +25% of the AUX pot value. In passive mixing the bus impedance is simply the parallel combination of this impedance from all channels. The problem is the variation of source impedance as the control is operated will vary the bus level of other channels i.e. there is interaction between controls. This is not difficult to mitigate. Attached is a document I wrote about this back in 2015 for the EZTubeMixer project.

Edit: A potentially more serious issue is the worst case loading of the driving op amps.

Cheers

Ian

 

[RoadrunnerOZ]

It would be good to see what your Aux mix bus schematic entails.

 

[Dimitree]

I think there is some confusion about the way buses work. First it is not clear to me whether you plan to use a virtual earth bus or a passive bus.

If it is a virtual earth bus then the bus impedance is already extremely low ( maybe a few ohms or tens of ohms) so adding at most 600 ohms in parallel with it make no difference.

If passive mixing is being used then you need to ensure that the AUX sends for instance are driven from a low impedance source. I that case the worst case source impedance at the pot wiper is 25% of the pot value. So the bus load varies between the bus resistor (AUX at zero) and bus resistor +25% of the AUX pot value. In passive mixing the bus impedance is simply the parallel combination of this impedance from all channels. The problem is the variation of source impedance as the control is operated will vary the bus level of other channels i.e. there is interaction between controls. This is not difficult to mitigate. Attached is a document I wrote about this back in 2015 for the EZTubeMixer project.

Edit: A potentially more serious issue is the worst case loading of the driving op amps.

Cheers

Ian

Thanks Ian, I’m starting to read the document you attached in the hope I will understand the matter,
Sorry for the confusion/ignorance, how can you determine that the bus impedance is already too low? I still haven’t picked any bus resistor values so this sentence

(If it is a virtual earth bus then the bus impedance is already extremely low)

puzzles me. Obviously I’m still missing a lot of information to understand.

Also, about your question regarding the mixing topology, I’d like to have a Neve-like summing mix on the main L-R mix (so trasformers and make up gain amps), and a simple virtual-earth on the groups and aux summing mixers.

 

[ruffrecords]

Thanks Ian, I’m starting to read the document you attached in the hope I will understand the matter,
Sorry for the confusion/ignorance, how can you determine that the bus impedance is already too low? I still haven’t picked any bus resistor values so this sentence



puzzles me. Obviously I’m still missing a lot of information to understand.

There are two basic types of summing: passive (Neve) and virtual earth (VE). In passive summing we assume each bus resistor is driven from a low source impedance (compared to the bus resistor). So, for ac, the sending end of each bus resistor is effectively connected to ground. This means that the bus impedance is therefore simply the parallel combination of all the bus feed resistors. So if your bus feed resistor is R and there are N of them on the bus, the bus impedance is R/N. The bus loss (level on the bus compared to the level feeding the bus resistor) is equal to N. So if you have 10 feeds the loss is 10 times which is 20dB. This is the amount of gain make up you need to restore the level. As an example, 10 feeds of 10K to a bus gives a bus impedance of 10K/10 = 1K

In a VE system an op amp creates the bus impedance. Basic op amp theory tells us that in an inverting configuration the effective impedance at the -ve input of an op amp is the feedback resistance divided by the op amp open loop gain. Op amps usually have enormous open loop gains; more than 80dB is common. This means that a 10K feedback resistor looks like 10K/10,000 or one ohm so the bus impedance is indeed very low compared to the Neve style passive summing

Also, about your question regarding the mixing topology, I’d like to have a Neve-like summing mix on the main L-R mix (so transformers and make up gain amps), and a simple virtual-earth on the groups and aux summing mixers.

Not a problem. You just need to be aware of the low source impedance requirement of the Neve style.

Cheers

Ian

 

[abbey road d enfer]

why they usually put rotary potentiometers on the AUX return levels? to spare money on the faders? or is it because AUX return is more comfortably controlled with a rotary control?

Mainly because it saves space. AUX returns generally do not require the kind of control and fiinesse associated with faders.

 

[abbey road d enfer]

2: (MUTE) should be shutting off everything

Debatable. A compressor would not react predictably it suddenly hit by a blast.
I've always had AUX sends Inserts wired Pre-Mute.

 

[abbey road d enfer]

Ok.
First, from what I see your Aux sends are going to short the Aux busses when turned fully down affecting the bus by the position of the wiper relative to ground

I don't see that. In both extreme positions, each send point is seen as a resistance eaqual to the injection resistor, since the pot sees zero-ohm at each end.
In the middle position, the bus sees Rinj + 1/4th the pot value.
Typically, with a 10k pot and 10k injection resistors, it varies from 10k to 12.5k. I wouldn't worry one second about it.

 

[abbey road d enfer]

The problem you have is that if the send resistor value from pot wiper to the Aux bus is low then it acts as a shunt to ground - with all Auxes turned down for say Aux 1 on all but one channel, then you have a lot of resistors in parallel on Aux 1 bus to ground.

This is a well-known fact, and it's been dealt with for decades now.
The problem with the Mackie approach is that it requires a specific taper for smooth operation, and the noise gain of the summing amp varies significantly, which results in possible instability.
The benefit is that when many pots are down, the summing amp noise is reduced. It allows using sub par opamps in this position.

 

[abbey road d enfer]

Also, do you see any issue with the PFL switch I drawn? does it load the signal too?

PFL always loads signal, except if you add a buffer. Nobody does.

 

[RoadrunnerOZ]

Debatable. A compressor would not react predictably it suddenly hit by a blast.
I've always had AUX sends wired Pre-Mute.

Why? You’ll still be hearing the effect coming through the mix or if using Aux send for headphone mix the headphones still get signal from the muted channel. (A snare hit to a compressor is a sudden blast) Neve 5106 Aux send for example is post channel cut/A/B switches and selectable pre or post channel fader.

 

[RoadrunnerOZ]

First, from what I see your Aux sends are going to short the Aux busses when turned fully down affecting the bus by the position of the wiper relative to ground

I don't see that. In both extreme positions, each send point is seen as a resistance eaqual to the injection resistor, since the pot sees zero-ohm at each end.
In the middle position, the bus sees Rinj + 1/4th the pot value.
Typically, with a 10k pot and 10k injection resistors, it varies from 10k to 12.5k. I wouldn't worry one second about it.

Sorry if that’s a bit misleading - the short to ground I was referring to is via the injection resistor, not a dead short. The point I was making is the positions of Aux sends on other channels varies the overall Aux level in a passive system

 

[abbey road d enfer]

Sorry, I meant Inserts, not AUX sends.
I've corrected post #31

 

[abbey road d enfer]

Sorry if that’s a bit misleading - the short to ground I was referring to is via the injection resistor, not a dead short. The point I was making is the positions of Aux sends on other channels varies the overall Aux level in a passive system

So-called "passive" mixing indeed suffers gain variations when the bus load varies, but it's also the case with the "Mackie" arrangement, even worse actually.

 

[abbey road d enfer]

I thought I had drawn both the send and the return as balanced. I thought the "Zero impedance output" that I used on the Send (and for Direct Out too) was impedance balanced. But probably I misinterpreted that configuration.
I don't think I need balanced send and returns. It's just that since I'm trying to use quad OPA1679 everywhere I can (the more ICs I get, the less I pay for them) I had two spare opamps so I thought about making the insert balanced.

Many mixers have unbalanced sends AND returns, capitalizing on the fact that cable length are relatively short and that many outboards are balanced. Connecting a balanced output to an unbalanced input, or conversely an unbalanced output to a balanced input, although it doesn't make the connection balanced, transforms it into a "ground-sensing" connection, provided the wiring is done with balanced cable, with unbalance done at the unbalanced side.
I've commissioned many studios with that arrangement and it has seldom been a problem.
The main reason for unbalanced inserts is that it saves one connector and a few parts, mainly to save real estate on the connector panel and on the PCB.
For balanced sends, you have the choice of using a cross-coupled circuit (2 or 3 opamps), a dedicated line driver ($), a ground-sensing output, or an impedance-balanced output.
I've found a ground-sensing output to work quite satisfactorily, if you accept the fact that the max send level is about +20/21 dBu, vs. +24/25 for the CCC or dedicated line driver. Impedance balanced is fine too, and costs nearly nothing.

 

[ruffrecords]

Sorry if that’s a bit misleading - the short to ground I was referring to is via the injection resistor, not a dead short. The point I was making is the positions of Aux sends on other channels varies the overall Aux level in a passive system

But in a well designed passive system the variation in minimal and inaudible.

Cheers

Ian