Virtual Earth Summing Questions

[JohnRoberts]

The usual 22pF compensation cap?

Yes I thought I was clear, OK I looked it up for you, from the data sheet 5534 with 22pF cap is 6 v/uSec slew rate, vs 13V/uSec w/o a compensation cap. Open loop gain gain at 10 kHz is  6V/mV (like 6000x)  decompensated vs. only 2.2V/mV with 22 pF. 

10 dB more loop gain margin reduces distortion and phase shift 10dB.. And the higher slew rate will impress your friends who don't know what it means in audio terms.

Most of the virtual earth summing schema I have seen, used the NE5532/LM833 op amp with a feedback cap of 100pF paired with a 10K or 22K feedback resistor.

Most people drink lite beer...

JR

 

[abbey road d enfer]

Note2: when operating decompensated, you can't use feedback cap across feedback resistor since that looks like unity gain or less (100% NF)  to the opamp at very high frequency so could become unstable without added compensation.

Most of the virtual earth summing schema I have seen, used the NE5532/LM833 op amp with a feedback cap of 100pF paired with a 10K or 22K feedback resistor.

5532/LM833 are unity-gain compensated. 5534, without the "usual" 22pF cap, is compensated for a minimum gain of 3. In a VE mixer, the HF gain above audio frequencies is determined ultimately by the ratio of the cap across the FB res divided by the parasitic impedance of the bus. If you want to take advantage of the higher BW and phase margin of a decompensated 5534, you have to make sure that the FB cap is smaller than about one half the bus capacitance. Since generally, you don't know the bus capacitance, you start with no FB cap at all and check for transient response and oscillations, then increase the FB cap until cooked to taste.

 

[electroslut]

Do you need pan? you need to look at the whole circuit to optimize your design, also the non-inverting input
of your summing op-amp (current summing) is connected to ground, earth, 0v
whatever you want to call it, the same place the rest of your studio's noise gets dumped.
If I was building a 16ch summing amp I would choose balanced (floating ground) voltage
summing. my 2 cents.

 

[metalb00b00]

5532/LM833 are unity-gain compensated. 5534, without the "usual" 22pF cap, is compensated for a minimum gain of 3. In a VE mixer, the HF gain above audio frequencies is determined ultimately by the ratio of the cap across the FB res divided by the parasitic impedance of the bus. If you want to take advantage of the higher BW and phase margin of a decompensated 5534, you have to make sure that the FB cap is smaller than about one half the bus capacitance. Since generally, you don't know the bus capacitance, you start with no FB cap at all and check for transient response and oscillations, then increase the FB cap until cooked to taste.

OK, thank you for explaining that

One last newbie question if you don't mind...

How do I connect the GSSL circuit to the output a summing amp shown in this schematic?
I can omit the GSSL input circuit right? But what about the selectable 27K (for unity)/15K (for unbalanced) resistor in front of the VCA's input? Can I omit that one too or do I need to change to another value?

Thanks again for the help guys

 

[abbey road d enfer]

You should connect the "Insert send" to the 27k and do without the 22k. In the GSSL, the 27k is there to provide 6dB attenuation in order to compensate for the 6dB gain provided by the balanced output stage. In fact, to be mathematically correct, it should be 30k, but I think the reason is that THAT (the mfgr of the VCA chips) says that higher than 27k can lead to oscillations. Another solution would be to use a 13.5k resistor in the I/V converter (the stage that follows) but it's not a standard value. You could also use 24k/12k, at the cost of (very) slightly increased distortion. Anyway, it's not a big deal, since there is a make-up gain control.

 

[metalb00b00]

Hi abbey, I need help again with this thread.

If I want to add a control gain to the summing amp circuit above (the one with an inverting "zero" impedance amp) using a 10K pot, I should:

1. Disconnect the 75R resistor from the op-amp output pin.
2. CCW pin to ground, CW pin to the op-amp output pin, and wiper pin to the 75R resistor.

Am I correct? If yes, do you know how much is the gain (roughly) in dB?

Thanks a lot for teaching me!

EDIT: Solved!

 

[abbey road d enfer]

Gain is P/R.
P depends on the rotation, it's a fraction of the nominal value.
If the pot is 10k and the input R is 10k too, max gain is 1.
If you want some gain in hand, you may either increase the value of the pot or decrease the value of the input res.

 

[metalb00b00]

I get a fader like operation with this set-up, however, I can't help but to feel that it is wrong, because the pot rotation range from a "somewhat" unity gain to +6dB is less than 10%. Very difficult to use

 

[abbey road d enfer]

That's because the 10k input resistor loads too much the 100k potentiometer. I guess the pot is a Log (or Audio) type. For 10k input impedance, the pot should be less than 5k. So you must either decrease the pot or increase the resistors, but increasing the resistors increase noise.
An alternative is to use a Lin pot without changing the values; loading a pot is called law-steering. When you load a lin pot with a significantly lower value, the law becomes hyperbolic, which is close enough to a fader type behaviour. The advantage in doing that is that Lin pots have a much tighter law than Log types, which can have as much as 6dB difference at mid-rotation. Typically, Lin types have less than 2dB tolerance.

 

[metalb00b00]

You're right. I forgot that the 10K resistor is used to set the input impedance. I got mixed up about using 100K for another reason. Tried with a 1K linear pot in one of the channel, and the unity gain spot is right in the middle of the pot rotation.

However, I think I shouldn't mess with the summing amp, and so I wanna do something like in this schematic. Is it a safer/better solution, or am I just adding another complexity when the previous attempt was already good enough?

The pot is 10K linear.

 

[joaquins]

As abbey road said, you should use 100K linear as in first schem, the second one is adding a cap and a op amp, so more THD there... if you don't have it you can have some DC offset, depending on the op amp could vary, I'd put something like 8k2 resistor from non inverting to ground bypassed with some film cap, value is not critical, just for decreasing noise, something between 1µ and .1µ should work ok. If you want 10dB boost you could decrease pot and inverting input resistor value as you want, also non inverting resistor to something between max Z and min Z seen by non inverting...

Max is when pot is at min or max, just feedback in parallel with non inverting input, min is with pot at mid point, so 1/4 pot value in series with non inverting input and all this parallel with feedback

So something between ((Ri//Rf)+((Rpot/4+Ri)//Rf))/2 would be a nice value, for min offset in all the gains, if you want min offset when muted or max just use Ri//Rf and if you want min offset with some gain you'll call 0dB calculate Z to this position.

In my mixer I use active gain in summing and a inverting amp to deal with inverting in summing amp, in the summing I use a servo DC feedback, the inverting amp is fixed gain, so I just put the non in input and feedback in parallel values and use as said before and have very little offset... Now I'm thinking in using transformer summing for my busses to create the master so the transformers will do the inverting task and this amps will be out.

JS

 

[abbey road d enfer]

Both are correct.
I would suggest you add lead compensation caps on the FB resistors on the summing and inverting stages, particularly on the summing stage, because the bus is capacitive due to the wiring.

 

[metalb00b00]

Both are correct.
I would suggest you add lead compensation caps on the FB resistors on the summing and inverting stages, particularly on the summing stage, because the bus is capacitive due to the wiring.

Sorry, that was me being lazy not to draw them in the schematic