Active Summing mixer using THAT Ic's and some...

[radiance]

Sorry for these vague comments, but I haven't followed this project closely.

Hearing hum when output is turned down, but not while turned up, if not just other noise masking the hum, may be a ground referencing issue between master local ground and output ground. If the output feed is properly differential from the master it should be silent when turned down. I'm not sure which schematic is in use, but if it's the one with simple follower buffers, perhaps they could be used to provide differential between master and sundry output grounds.

This is not the case. So far the output has always been full on. It's hard wired cause I don't have a master fader/pot yet.

Also, if the gauge of the ground wire is making an audible difference that suggests to me there is noise current flowing in that wire that is being reduced. So that ground wire is not just a reference but perhaps part of the ripple filtering path, so voltage will not be the same along that path.

Not sure if I understand what you mean here but let me explain how I did the grounding.
In this project , every channel, master module, monitor etc. etc. has it's own ground wire to a star ground point (bus bar).
Take for example the channel input module. It has a ground plane for all power decoupling caps and on the other side a star ground for all audio ground. Ground plane and star ground meet at one point on the pcb from which the thick ground wire originates which goes to the bus bar. The bus bar goes to the PSU (which is in a separate case btw) 0V and ties to the chassis in the PSU case. The chassis ground from the mixer goes to the chassis ground in the PSU. All XLR pin 1 go to chassis ground from the mixer.
Relay, lamp and LED power has it's own ground wire all the way to the PSU where it connects to chassis ground AND to 0V from the PSU.

[JR.]

Sorry for these vague comments, but I haven't followed this project closely.

Hearing hum when output is turned down, but not while turned up, if not just other noise masking the hum, may be a ground referencing issue between master local ground and output ground. If the output feed is properly differential from the master it should be silent when turned down. I'm not sure which schematic is in use, but if it's the one with simple follower buffers, perhaps they could be used to provide differential between master and sundry output grounds.

This is not the case. So far the output has always been full on. It's hard wired cause I don't have a master fader/pot yet.

Also, if the gauge of the ground wire is making an audible difference that suggests to me there is noise current flowing in that wire that is being reduced. So that ground wire is not just a reference but perhaps part of the ripple filtering path, so voltage will not be the same along that path.

Not sure if I understand what you mean here but let me explain how I did the grounding.
In this project , every channel, master module, monitor etc. etc. has it's own ground wire to a star ground point (bus bar).
Take for example the channel input module. It has a ground plane for all power decoupling caps and on the other side a star ground for all audio ground. Ground plane and star ground meet at one point on the pcb from which the thick ground wire originates which goes to the bus bar. The bus bar goes to the PSU (which is in a separate case btw) 0V and ties to the chassis in the PSU case. The chassis ground from the mixer goes to the chassis ground in the PSU. All XLR pin 1 go to chassis ground from the mixer.
Relay, lamp and LED power has it's own ground wire all the way to the PSU where it connects to chassis ground AND to 0V from the PSU.

I guess I'm confused because in an earlier post you say

Also, it's not noise but a slight psu hum.
I don't hear it with master & headphone volume cranked up full but still

If you don't hear it with volume up, and you can't turn the volume down, when do you hear it?
----------

If I understand your ground approach you are connecting each channel's local audio ground to the local PS ground at one common point, and route that though a single wire to another single point (bus bar ?) and from that through a wire to external PS. I presume PS ground currents must flow through this ground wire to get back to PS so there will be a voltage potential between channels and central ground point (this is typical).

Assuming every channel is clean wrt to it's local ground, and the signals from there are sent differentially for combining to the master you are fine. Your active summer part one shows a differential bus structure, part two is not differential. So using #1 the sundry different local input grounds will get ignore by differential bus, so the signal at part #1 should be clean.. The signal needs to differentialled between this ground to the outputs. This is done inside the THAt output chips as long as their ground is coming from master local ground, not output grounds. The unbalanced outputs, could benefit from differential (resistor divider in + inputs) referencing.

All grounds are relative. You have local grounds at every input, the master, and unbalanced outputs.. that need to be referenced to each other.

Your noise plot looks very quiet,,, I may just be confused by your comments,,, disregard if your have a hum free noise floor. It looks like you have a lot of time and effort invested into this. Good job.

JR

[mediatechnology]

Uh, is that monitor a CRT? It looks like the image is wall-to-wall so it's probably an LCD. If it's a CRT it could create significant radiated and cinducted interference. I often pickup 15.575 kHz and ~59.94 Hz measurement spurs at very low level that I think are conducted through the power line from a neighbor's CRT TV in the next house. I don't see those kind of emissions as much from LCD monitors but the scan rates are similar at the 60 Hz refresh rates for VGA monitors. I can't remember: Are you in a 50 Hz country? It looks like there may be small (tiny!) spurs at 50 Hz 6th and 7th harmonic.

It could also be conducted through the sound card ground (tied to the mixer most likely) from the computer's ground which is why increasing the conductor size from the mixer to the PSU improved it. The AC leakage current through the computer power supply's "Y" line filter capacitors is usually quite high. I measured a small ~100W switcher the other day that produced 300 uA leakage into the safety ground.

Regardless the levels you are seeing are very low. So low that I wouldn't worry. A very impressive project and nice clean build.

[radiance]

If you don't hear it with volume up, and you can't turn the volume down, when do you hear it?

Sorry, was not very clear at this point. What I did was first listen to the output on the headphone output with master & headphone volume maxed. I did not hear any noise or hum. Than, I connected the master to my DAW and boosted the signal till I stasted to hear noise and some PSU hum. I know, I should not do this and just don't worry, but I was curious what was to be heard there...
----------

The unbalanced outputs, could benefit from differential (resistor divider in + inputs) referencing.

Do you mean the unbalanced internal connections?

It looks like you have a lot of time and effort invested into this. Good job.

JR

Thanks! Took a lot of time indeed. Lot's of reading....

Uh, is that monitor a CRT? It looks like the image is wall-to-wall so it's probably an LCD. If it's a CRT it could create significant radiated and cinducted interference. I often pickup 15.575 kHz and ~59.94 Hz measurement spurs at very low level that I think are conducted through the power line from a neighbor's CRT TV in the next house. I don't see those kind of emissions as much from LCD monitors but the scan rates are similar at the 60 Hz refresh rates for VGA monitors. I can't remember: Are you in a 50 Hz country? It looks like there may be small (tiny!) spurs at 50 Hz 6th and 7th harmonic.

That's an LCD alright. My wife used to have a CRT. It was behind a 1 foot concrete wall but my Fender Rhodes piano could not be recorded when her CRT was on. Man, these things radiate....

[JR.]

The unbalanced outputs, could benefit from differential (resistor divider in + inputs) referencing.

Do you mean the unbalanced internal connections?

All of the internal connections are unbalanced.

On you schematic you show some balanced outputs, and unbalanced (like head phone, meter, etc).

I meant the "unbalanced outputs".

=======

To explain further.. all audio signals are relative. If balanced (or differential) they are the difference between the signal on the + lead and the signal on the - lead. If unbalanced (or single ended) it is the difference between the signal on the + lead and local audio ground. This nomenclature (audio ground) is a source of confusion as grounds handle much more than audio currents so are easily corrupted.

One powerful capability of operational amplifiers is the ability to differentially process signals, so you can take an audio signal relative to it's local ground where it's coming from, and then translate that to be relative to the local ground where it's going. Calling these all grounds implies they are all at the same potential,, they are not. Ground is a broader concept not a singular voltage node.

In your mixer the balanced outputs take the signal in the master relative to the ground in that master section and convert that to a positive and negative polarity output signal, this differential output doesn't care what the ground is at that output.

The unbalanced outputs, are relative to their output grounds so it does matter that the signal in the master section relative to the master section ground gets translated to be relative to the output grounds at those outputs. This is often as simple as adding a few resistors to operational amplifiers already in the circuit.

==========

Quick monitor noise story...

Back when I was working at Peavey I was helping some junior engineer troubleshoot an unusual 1 Hz component in the noise floor of a large powered mixer. This was in a work area with a bunch of small engineering offices side by side. Maybe 10' square rooms. I determined that the noise was actually the beat frequency between the CRT monitor on his desk, and the same model CRT monitor in the next office several feed away. Turning off one monitor or the other stopped the noise. Apparently the raster frequencies were very close but 1 Hz apart.

JR

[radiance]

Thanks for the comments JR! (is that John??) That was very good reading.
This project has been on hold due to my move from the Netherlands to France but I've finally found some time to finish this.
Earlier on in this thread you mentioned:

The unbalanced outputs, are relative to their output grounds so it does matter that the signal in the master section relative to the master section ground gets translated to be relative to the output grounds at those outputs. This is often as simple as adding a few resistors to operational amplifiers already in the circuit.

I do understand this but I don't understand the part of "adding a few resistors"...
Here a simplified picture of the connection between the master pcb and the output/buffer pcb..


Is it as simple as adding a resistor between fader and + input, a resistor between ground and - input and a resistor in the feedback path of the receiving opamp?

[JR.]

I do understand this but I don't understand the part of "adding a few resistors"...
Here a simplified picture of the connection between the master pcb and the output/buffer pcb..


Is it as simple as adding a resistor between fader and + input, a resistor between ground and - input and a resistor in the feedback path of the receiving opamp?

Yes, pretty much. The potentiometer complicates it a little, but a pair of equal value resistors, one from the wiper of the pot to the + opamp input, and the second from there to the output jack ground, and a second pair of equal value resistors, one from the sum bus local ground to - opamp input, with equal value feedback resistor. This will differential the signal and subtract out any ground errors, forward referencing the signal from the master to the output ground.

The one catch in this for your circuit is that the potentiometer presents a varying source impedance with pot rotation that will imbalance and degrade the differential. If the resistors used in the differential are large wrt the value of the pot this error will be minimal, and always improved by use of differential vs, not. (In console design you generally try to execute differentials between low impedance signal nodes).

Note: for the sake of discussion the level pot is grounded at the "master" ground. This way forward referencing will be most accurate at full cut or full up, where people notice such things. If the pot was grounded to the output it would kill completely but have zero CM rejection at full up.

Keep in mind this is a general discussion, if your chassis layout is very small there may not be large voltage differences between nodes, but in my experience there are always differences between local grounds.

JR (John Roberts)

[radiance]

I'm not sure what you meant by "sum bus local ground"...Something like this? Resistors of say...100K??

[JR.]

I'm not sure what you meant by "sum bus local ground"...Something like this? Resistors of say...100K??

Yes sort of...

A good practice to transfer layout grouping information between circuit designer and packaging engineer, in the schematic label the Master ground as say G1 and output ground as G2, connect all of these sundry local grounds to PS ground or chassis ground with 0 ohm resistors. In reality they are all connected together but in practice this helps you keep track.

Back to your schematic you have one resistor too many. Delete the resistor going straight down from the - input of your output opamp. Then connect the resistor coming straight down from the + input of your opamp to the output ground (say at output jack). The node now labelled as ground become G1 or master ground, and resistor from + input going to ouput jack is now G2. Both of these also connect back to your PS or chassis ground.

What is important is that ground current flow within a given node is just the local current and doesn't cause errors. Global current flow and voltage potentials between these sundry ground nodes doesn't matter because signal being passed from local area to area is relative, so signal integrity is maintained. Even the dreaded ground loop, doesn't matter as long as local nodes are relatively tight.

Sorry if this sounds a little obscure.. in many cases we could be talking about microVolts of noise but I've seen higher across large consoles not that there's an easy way to measure it. I prefer to just design it away.

The final reminder that this isn't perfect, using 100k resistors will reduce the loading effect on pot impedance. Since worst case source impedance from 10k fader is 2.5k (@ -6dB), 100k resistors will allow 1.2% of ground errors through (-38dB) at that worst case setting. Unless you have truly horrendous ground problems this will be very good, but now we are introducing noise from the Johnson (thermal) noise of the 100k resistors. Not to scare you, and without doing the math I predict this noise will be insignificant in use, I only mention it to be thorough.

JR

[radiance]

Thanks John, I'll try this tomorrow..Hope there's some room left on the pcb...otherwise I'll be making some kind of socket that plugs into the existing IC socket and that caries those extra resistors + the original opamp