Simple active summing mixer.

[radiance]

[quote author="Harpo"][quote author="radiance"]Why is that?[/quote]
There is no level adjust within your insert loop.
As drawn, a That1243 drops level by -3dB, but a 1246 by -6dB.
As the 1646 has +6dB gain, your insert loop will be at unity using the 1246 instead of the 1243.[/quote]

Ah, I see. Actually, I was thinking of using the That 1600 in place of the 1234.
I'll have a look at the datasheets one more time...

 

[radiance]

Updated my schametic taking your suggestions into account.

R1 is bothering me though...First I thought I understand it's purpose when John Roberts explained it like this:

[quote author="JohnRoberts"]

The 50k to ground is only needed to provide DC discharge path when cap coupling is used.

JR[/quote]

But that Samuel Groner said it is not needed...

[quote author="Samuel Groner"]RThere is a small current flowing from/into the opamp inputs, which you need to provide a path for. I think that the 49k9 and R6 are meant to provide this, but note that in both cases there is another DC path (the 10k feedback resistor for the first case and the OPA604 output for the second) which allready provides bias.
[/quote]

I guess I'm entering territory where there is no absolute right, just different ways of doing things. I think I leave a space on the pcb for R1 and see what works best...

Thanks all! ok..opamp bias reading...

 

[JohnRoberts]

R1 is not needed in that circuit as the 10k input resistor provides a DC path to charge/discharge caps. The 49.9 ohm R doesn't appear to be doing anything useful either.

When passive components are in series you can change the order of them without changing circuit behavior. In that circuit if you change the order of the 49.9 and 100uf cap you now have, 100uf cap>49.9 ohm>10k. Without the 49.9k inserted, this looks exactly like a 100uf cap>10.0499k R. The only thing the 49.9 ohm resistor is doing is creating a small gain error (-10/10.0499).

If the 49.9k is left in, it creates another small loss as it forms a pad/divider with the 49.9. I'd lose both.

JR

 

[burdij]

I agree with Samuel. The DC path to ground for the right side of the capacitor is provided through the two 10K resistors back to the output of the OPA604. Keep in mind that the output of an operational amplifier is usually a fairly low impedance with respect to ground. In addition, for AC, the resistor is shunted by the low output impedance of the DOA990 in series with the 50 Ohm resistor.

Offset problems can occur more frequently when the positive input lacks a DC path to ground. The negative input is actively being forced to ground by current supplied through the feedback resistor from the output, the effect of the so-called "virtual ground". In this case, the positive input is grounded.

Another thing you might want to consider is eliminating A10. Since you are feeding the positive inputs on all following devices, you really don't need the buffer before them. One less amp, one less crossover distortion source.

 

[radiance]

Thanks ...All makes sense now :grin:


[quote author="burdij"]

Another thing you might want to consider is eliminating A10. Since you are feeding the positive inputs on all following devices, you really don't need the buffer before them. One less amp, one less crossover distortion source.[/quote]

I thought of that as well but it seems that the That 1646 likes to be drivin by an opamp so...well see.

 

[burdij]

I guess you will need that amp. I just checked the datasheet for a THAT1646. It has a 5K input impedance. Shunting the 10K log control with that low an impedance will change it's law.

 

[jeth]

[quote author="Svart"]
3. You should really think about adding some kind of high value load resistor before your mute switch. I don't think it's good design to leave an output from an active device unterminated, especially when switching in/out another load.[/quote]

In a recent project I experienced the pops and clicks on mute switches as mentioned by svart in a later post on this thread... It was suggested to me that i try switching the source to ground via a smallish resistor (I used 2K) .. The idea suggested was that the resistor would allow the coupling cap pre mix resisitor to discharge.

The results in my circuit were an improvement, though not perfect and there are still audible pops on the switch if muted during lf signal peaks. Is this likely to improve by switching source to ground via a higher value resistor?
Or would anyone suggest a different switching method?
The circuit has level pots between the mute switch and mix R, could it make sense to mute by shunting both pot wiper and mix resistor to ground?

 

[JohnRoberts]

Ah clicks... Lots of engineering and caps to get completely clickless.

First the easy one. Whenever you hard switch an audio signal, with signal playing there will be an audible HF click when signal is interrupted at anything other than a zero crossing. Some sophisticated processing devices make a point of only stopping or starting at zero crossings.

A blocking capacitor, without a discharge path when switch is open, will rise up (or down) to its input side voltage due to internal leakage with no place to go.

Connecting a blocking cap output that is discharged to 0 Vdc to an opamp -input through input resistance will cause a small click equal the opamp input offset voltage, x the noise gain, this is small but if objectionable both sides of the selector switch need to be cap coupled. A more obscure fix, is to discharge the blocking cap to the opamp's - input voltage. The best way to do that depends on specifics of actual circuit.

Many capacitors, perhaps ultimately degrading audio paths, have been added to eliminate these small mostly cosmetic clicks. But paying customers perceive clicks as worse than blocking capacitors. :roll:

JR

 

[jeth]

Thanks John for a good overview.. unfortunately in my application the switches are going to be used creatively to bring signals in and out of the mix during performance, so minimal clicks are a must. Perhaps I can split the blocking cap using two double value caps, one on the source output pre switch and pot and another post pot, would have the advantage of giving protection against any pot noise and means there is dc blocking both sides of the switch.
My main doubt is actually this.. it seems the norm is to disconnect the mix R from thew source and connect it to ground.. when and why is grounding the source or both source and mix R a better option?

 

[JohnRoberts]

There's a technique for that too...

Imagine if you had a switch that clicked with music, but you ran that signal after the switch to a tone control circuit with the treble turned down... since the click is mostly HF, the treble control would attenuate it. Unfortunately along with the signal too. But since we know how much we are cutting the highs after the switch, we just add some symmetrical HF boost in front of the switch. The music comes out flat, but the switch click gets rolled off... :grin:

You give up some HF headroom in the channel equal to the HF pre boost, but with modern circuitry you can realize significant declicking with good s/n. I've used this before several times and it works. just take care to use identical boost/cut EQ.


JR

 

[jeth]

Thanks again John.. very interesting that, a bit more complex than I'd hoped for, though there are eq circuits in that area of the signal chain so maybe I can work that in somehow.
I remain confused regarding simple switching arrangements and the whys and wherefores of switching either source or mix R to ground..if there is a problem with unterminated sources then why is it almost standard practise to just lift or ground the mix resistor (at source end)?

 

[JohnRoberts]

[quote author="jeth"]
I remain confused regarding simple switching arrangements and the whys and wherefores of switching either source or mix R to ground..if there is a problem with unterminated sources then why is it almost standard practise to just lift or ground the mix resistor (at source end)?[/quote]

In general you don't want to hang long PCB traces (antennas) off opamp - inputs. The same antenna connected through a resistor will do less damage and stray capacitance will also have less impact on opamp stability.

JR

 

[jeth]

Ok.. thanks again, that makes sense, hadn't looked at it like that in the light of my mixer being small and the master mix bus being a couple of inches long, looks like I can stick with the source shunted to ground. Any advantage in a larger value resistor than the 2K I'm using, or perhaps omitting the resistor and simply switching the wiper of the pot that follows the switch to ground?

 

[radiance]

Ok, made some progression on the monitor section.

Any thoughts/opinions on the mono summing thing?

....and part 2...

Is C1 needed? I added C1 to prevent both volume pots from getting scratchy since there are not much DC blocking caps in the mixer so far.....

Here a link to the Cmoy headphone amp website...

http://tangentsoft.net/audio/cmoy-tutorial/

 

[Samuel Groner]

Is C1 needed?

Yes. The 0.1 uF and the according 100k resistor not though.

Not sure if a OPA604 is a clever solution for a headphone amp... Very weak output stage.

Samuel

 

[radiance]

Thanks Samuel :thumb:

[quote author="Samuel Groner"]

Is C1 needed?

Yes. The 0.1 uF and the according 100k resistor not though.[/quote]

Hm...did not give it much thought ....I just copied that from the Cmoy schematic


[quote author="Samuel Groner"]Not sure if a OPA604 is a clever solution for a headphone amp... Very weak output stage.

Samuel[/quote]


I will experiment with the opamp in the headphone section....dunno why I used an opa604 there....I think it was a leftover from when I only wanted to use a +/- 24V psu. Might opt for an opa2134 there....

 

[Samuel Groner]

Might opt for an opa2134 there...

Depends on the headphone impedance you'd expect to drive. Sennheiser HD600 et al is no problem with a standard opamp; for 32 ohm cans you want a buffer for low distortion though.

Samuel

 

[McCroskey42]

Since you're already using THAT 1646's for the main outs, why not try mediatechnology's 1646 headphone driver? The posts here have been nuked, but it's pretty well documented over in the Pico Compressor forum. Sounds awesome, too! (listening to Dark Side of the Moon through it right now, actually...)

 

[radiance]

[quote author="McCroskey42"]Since you're already using THAT 1646's for the main outs, why not try mediatechnology's 1646 headphone driver? The posts here have been nuked, but it's pretty well documented over in the Pico Compressor forum. Sounds awesome, too! (listening to Dark Side of the Moon through it right now, actually...)[/quote]

Thanks, I'll definitely check that. Need to buy a lot of THAT chips soon so I might reach a nice pricebreak.

 

[wolfgang]

Hello radiance!

Any news on this project?

Very interesting thread!

Regards,
Wolfgang