VCA theory question

[JR.]

Not sure I understand the question... Think of 0V (unity gain) as the starting point or home base for current ratioing VCAs. OV=0dB.. +X mV= +Y dB and -X mV = -Y dB (reversed direction from - EC port).

From looking at that chart you posted the VCA gain law is +6 mV/dB, so if you want +10dB of gain you could apply +60mV to EC+ and 0V to EC-, or 0V to EC+ and -60mV to EC-, or +30mV to EC+ and -30mV to EC-. In all three of those cases the control voltage relative to each other is the same 60mV so gain is same +10dB.

For modest amounts of gain/attenuation these control voltages are small, so it's fine to ground one EC pin and drive just the other, but as you see at -90dB of attenuation we are now talking over half a volt of control voltage. Rather than driving one EC port over half a volt above or below 0V, it is preferable to drive both, half that voltage away. Driving EC+ to -270 mV and EC- to +270 mV you get the same -90dB command but the VCA is much happier.

JR

[juniorhifikit]

I guess I was referring to this:

if the control voltage gets more than a diode drop away from 0V in either direction it not very good for the internal audio path symmetry

I'm not sure what you meant by "a diode drop" away from 0V.
As a first quick attempt at a bipolar buffer, I came up with this, though I haven't worked out the resistors for the gain of the buffer:

http://www.scottgreiner.com/buffer.png

[JR.]

That looks a little familiar,, good concept (I think).

By a diode drop I am talking in the ballpark of 500-700 mV... IMO far better to hit the control ports with + and - 300 mV than 600 mV on just one port.

Re your schematic, the RC to ground at the opamp outs may or may not be necessary depending on the opamps HF performance.

C4 filtering out digital noise (?) is in effect, dumping the full digital ground noise into the control voltage (at least before the pad dividing it down). Perhaps consider a two stage filter, with the first pole clamping digital noise to digital ground, then the second pole filtering out HF noise wrt analog ground.

The inverting opamp has an extra pole of LPF that the non-inverting doesn't... adding a cap to AGND from junction of r4 and r5 will filter some digital poop from both.

Adding a similar 2k and 100nF to the non inverting buffer will make it's output impedance more like the inverting (This last bit may be over design with no real utility.. In principle it would raise the source impedance which isn't desirable, I'm just thinking out loud.).

JR

[juniorhifikit]

That looks a little familiar,, good concept (I think).

yes, I...er...well...um... I learned from the best

By a diode drop I am talking in the ballpark of 500-700 mV... IMO far better to hit the control ports with + and - 300 mV than 600 mV on just one port.

Ah - thanks. That makes perfect sense now!

Re your schematic, the RC to ground at the opamp outs may or may not be necessary depending on the opamps HF performance.

Yes, I saw that in some of the THAT literature. They said it depends on the opamp. Do you think I would need it with a 5532?

C4 filtering out digital noise (?) is in effect, dumping the full digital ground noise into the control voltage (at least before the pad dividing it down). Perhaps consider a two stage filter, with the first pole clamping digital noise to digital ground, then the second pole filtering out HF noise wrt analog ground.

Hmmm... hadn't thought about the other side of that filter. I was only thinking about the speed of the CV change.

The inverting opamp has an extra pole of LPF that the non-inverting doesn't... adding a cap to AGND from junction of r4 and r5 will filter some digital poop from both.

By adding this extra filtering, could that negate the need for the above extra stage of filtering of the DGnd?

Adding a similar 2k and 100nF to the non inverting buffer will make it's output impedance more like the inverting (This last bit may be over design with no real utility.. In principle it would raise the source impedance which isn't desirable, I'm just thinking out loud.).

would the CV port see the 2K FB resistance in parallel with the 0.3 Ohm output resistance of the buffer amp? My calculation showed that would only change the impedance down from .3 Ohms to .2999 Ohms - did I miss a turn somewhere?

How do I post an image directly into my reply?



Oh cool! Like that!

[JR.]

The output impedance of an opamp is not a simple resistance, but a real resistance that gets decreased by a feedback factor or loop gain margin, In effect this output impedance is inductive as it rises with falling open loop gain..

I don't know if 5532 is OK without.. it is decent low impedance.

I think the HF impedance issue with control voltage is a spurious HF oscillation inside the VCA that would need to be checked for with a scope.

In any case the schematic showed c4 tied to digital ground... we want to keep any noise out of the control path and digital ground will not be the same as analog ground by definition.

JR

[juniorhifikit]

In any case the schematic showed c4 tied to digital ground... we want to keep any noise out of the control path and digital ground will not be the same as analog ground by definition.

JR

I'm a little unclear what they mean in the THAT application note:

R2 and C5, the filter
capacitor, should be physically near to the DAC,
and the grounded leg of C5 should return to the
Dgnd and Agnd pins of the DAC prior to connect-
ing to the remainder of Sgnd, as should the DAC's
decoupling capacitor. This will keep the transient
currents resulting from Vn1 in a small loop local to
the DAC.

Their R2 and C5 are labeled R6 and C4 in my drawing.

In a slightly related topic, does it make sense to have the analog ground isolated say 100 Ohms higher than the digital ground before tying them together at the chassis?

[juniorhifikit]

here's how M@ckie did it in their not-so-popular OTTO automation for the 1604 mixer:

[JR.]

Yup..interesting

The description sounds like they are bonding the digital grounds and analog (signal) grounds through the ground lead of C5. Note: analog ground on the DAC is only analog relative to the digital ground.

We definitely want the digital noise current to flow back to the digital ground.

I also prefer the two filter approach... with the first referenced to Digital ground, and the second to the analog ground.

It would be easier to follow if C5 was drawn connected to the DAC ground.. then we could see a one pole filter clamping digital noise to the digital ground, then we have a second 1 pole filter formed by the feedback cap on the opamp that is referenced to signal ground.

Regarding the crapie schematic, I will ASSume they have already filtered digital noise out of the control voltage feed before routing it around in an audio environment.

No, the grounds don't really need a resistor between them but they do need to only be connected once and with some forethought. It is a good discipline to think of every ground trace as a resistor, to better visualize how the currents are flowing in and around the circuit.

JR

[juniorhifikit]

Yup..interesting

The description sounds like they are bonding the digital grounds and analog (signal) grounds through the ground lead of C5. Note: analog ground on the DAC is only analog relative to the digital ground.

We definitely want the digital noise current to flow back to the digital ground.

I also prefer the two filter approach... with the first referenced to Digital ground, and the second to the analog ground.

It would be easier to follow if C5 was drawn connected to the DAC ground.. then we could see a one pole filter clamping digital noise to the digital ground, then we have a second 1 pole filter formed by the feedback cap on the opamp that is referenced to signal ground.

So ASSuming that C5 and R2 were located close to the DAC and the Agnd and Dgnd were tied together at that point; then the long trace resistance to Agnd at the buffer opamp would act to isolate the two? I'm not sure I understand how the feedback cap is referenced to signal ground.

Regarding the crapie schematic, I will ASSume they have already filtered digital noise out of the control voltage feed before routing it around in an audio environment.

Nope . Not that I see looking at the circuit board. There's no more filtering than what you see on the schematic.

[JR.]

So ASSuming that C5 and R2 were located close to the DAC and the Agnd and Dgnd were tied together at that point; then the long trace resistance to Agnd at the buffer opamp would act to isolate the two? I'm not sure I understand how the feedback cap is referenced to signal ground.

Just like I said to think of all traces as resistors, think of capacitors as dead shorts to AC (at HF). This means at HF, the top of the cap will be exactly the same AC signal as the bottom.

This characteristic of caps at HF means two things to us... first we can use this to eliminate HF noise relative to whatever ground we connect to. The second characteristic is that this HF noise will get dumped into the ground by that cap, and whatever current that is, will create finite voltages as it flows to it's final equilibrium.

If the output of the DA is relatively clean of digital noise, the first filter isn't doing that much work, so a final filter just to the signal ground is probably adequate, if the DA output is noisy, I don't want to dump that noise current into my clean signal ground.

A certain amount of it depends on what the DA looks like, and how it gets stepped between levels.

JR

This matters if we look at ground currents flowing into ground traces with real resistance. For a second cap filter tied to signal ground... If that secomg R