Proper way to match the gain of 2 VCA in a stereo volume control?

elskardio · Aug 20, 2014

Hi guys,

I've been experimenting with VCAs for a stereo volume control. I have pretty good results so far but... the gain of both VCAs is always off by 1 or 2 dB. To correct this I simply replaced one of the input resistor (R2 or R5) for a trim pot and adjusted it until both channels give the same readings. Is this the proper way?

Thanks for your input

JohnRoberts · Aug 20, 2014

It works but is not ideal IMO.

2dB seems like a pretty large error. First you might want to try using 1% resistors.

If you still need to trim the gain, you can add a gain offset into the other EC port of either VCA. Drive the other control port with a similar 51 ohm R to ground and probably tens of K from trim pot to tweak gain.

I would trim for gain match at 0dB (unity) which is 0V at EC port.

JR

elskardio · Aug 20, 2014

Hi John, thanks for the quick response.

I have 20k trimpots on hand. I'll try the following for the gain offset. Is this what you had in mind?

Thanks

JohnRoberts · Aug 20, 2014

Close but more like a 20-30k resistor in series between the pot wiper and the EC port.

JR

elskardio · Aug 20, 2014

Like this?

JohnRoberts · Aug 20, 2014

Yes, but now add back the 51 ohm resistor to ground at the EC port. You need to divide the control voltage down to mV level.

JR

elskardio · Oct 9, 2014

Hi John,
The gain offset at the EC+ port works but the VCA's harmonic distortion increase a lot!

Even if I tweak the Symmetry, nothing change...
Any idea?
Thanks

abbey road d enfer · Oct 10, 2014

elskardio said:
Hi John,
The gain offset at the EC+ port works but the VCA's harmonic distortion increase a lot!
Even if I tweak the Symmetry, nothing change...
Any idea?
Thanks

Is your schemo up-to-date? You need to add a resistor between the Gain offset trimmer and the EC+ port. And I believe you should add a 51r at the Ec+ port of the other channel.
BTW, I've always thought that the value of 51r should be reconsidered in view of the performance. I've never used more than 22r in this position.
There is a significant advantage (and some drawbacks) in driving the Ec ports with a near-zero impedance such as provided by an opamp. Then the offset control can be applied to the opamp, not directly to the Ec port.
Finally, I note that the gain control range is +/-40 something, which is not really convenient for a volume control, particularly on the + side. I believe a range of -50/+10 would be more suitable. That is obtained by increasing the value of the 3.01k from the wiper to the Ec- port to about 4k and offsetting the Ec voltage by adding a resistor (about 6k) between V+ and Ec-.

Beware that with the resistor values indicated here, you are deemed to have problems with the potentiometer. Assuming +/-15V rails, the dissipation in the 5k pot is about 0.2W, which is pretty close to the max for many commonly available pots (1/4W). In addition the current through the wiper will certainly quickly destroy the fragile conductive surface.
One option would be using a wire-wound or cermet pot, but that would still leave a serious problem: the scratching noise of the wiper. It is customary to put capacitors across potentiometer-derived control voltages. You would need to install an electrolytic or large film cap (about 10uF) from wiper to ground to tame the noise. Beware that the control voltage is changing from positive to negative so you need to a bipolar or non-polarized type.

In view of all these considerations, I would advocate the use of much larger value pot and res, delivering the control voltage to the Ec ports via a voltage-follower. It needs to be a low noise type, such as a 5532.

JohnRoberts · Oct 10, 2014

As Abbey mentioned too much impedance at the EC port can increase distortion. The EC ports are connected to the bases of the VCA transistors so a small fraction of AC audio signal current will flow there, so the voltage from that current flowing into the termination impedance will modulate the gain causing distortion. .

The THAT corp spec sheet advises driving the EC port with less than 50 ohms. I am a little apprehensive about driving directly with an op amp since noise injected into the control port will modulate the gain, so indeed a quiet op amp should be used. FWIW in Gary Hebert's (THAT design engineer) AES paper they showed a AD797 driving the control port for absolute lowest noise measurements, but that is probably overkill for most applications.

While this may be too much information here http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=211&start=120 is a link to a 17 page long thread where I investigated some consequences of control port termination. IIRC I probably drove the control ports with 5534 and there was an added RC snubber (91 ohm x 1 nF) to account for op amp rising output impedance at HF.

FWIW this was mostly a mental exercise as I never made that product,,, did I mention that analog audio is obsolete? 8)

JR

joaquins · Oct 10, 2014

JohnRoberts said:
...
FWIW this was mostly a mental exercise as I never made that product,,, did I mention that analog audio is obsolete? 8)

JR

How do you take your music from your hard drive to the air?

What about adding a cap in the EC ports? that would reduce the AC in there decreasing the noise, but I don't know if it would need to be too slow to change the level, I think impractical cap sizes are needed to be fair and that would be a problem but if using an opamp, driving a divider with the pot so it's fast enough and low noise enough as a compromise the cap would help, and maybe you could live with a bit more THD at lower frequencies.

JS

ricardo · Oct 10, 2014

It's 30+ yrs since I did serious work on this but from the descriptions I suspect the THAT VCAs are similar to the Allison 'blue' VCAs I worked with.

In dem days, we took scrupulous care to match the resistance present to the Ec ports to minimize distortion. My memory is vague but we might have driven them differentially too.

But you should refer to the THAT datasheets & articles for chapter & verse. What I do remember clearly is that you definitely want to do your matching at '0dB' gain.

JohnRoberts · Oct 10, 2014

joaquins said:
JohnRoberts said:

...
FWIW this was mostly a mental exercise as I never made that product,,, did I mention that analog audio is obsolete? 8)

JR

Click to expand...

How do you take your music from your hard drive to the air?

I don't. After having live bands in my living room for parties, the old hifi just doesn't quite cut it, so I rarely even fire it up. I have some small computer speaker that are good enough for most casual listening

What about adding a cap in the EC ports? that would reduce the AC in there decreasing the noise, but I don't know if it would need to be too slow to change the level, I think impractical cap sizes are needed to be fair and that would be a problem but if using an opamp, driving a divider with the pot so it's fast enough and low noise enough as a compromise the cap would help, and maybe you could live with a bit more THD at lower frequencies.

JS

I linked to about 17 pages of discussion. I'm pretty sure I discussed using caps in that old thread.

To provide a cap that exhibits low impedance < 50 ohms at 20 Hz requires 150 uF or so. Considering modern manufacturing and PCB constraints op amp buffers are not out of the question. People I know doing premium VCA paths buffer the EC. As I mentioned THAT in their AES paper used a low noise op amp buffer to drive that node for low distortion plots..

JR

PS: I just looked back at another project i worked on in 2012 where I did a parallel prototype development using VCAs or DPOTs. At prototype stage the DPOTs worked so well I abandoned the VCAs, but in that proforma design I was using a resistor divider to drive EC port with 47.5 ohm in ground leg of the divider, with an electrolytic cap in parallel. But since I never fired it up I have no specific bench observations.

JohnRoberts · Oct 10, 2014

ricardo said:
It's 30+ yrs since I did serious work on this but from the descriptions I suspect the THAT VCAs are similar to the Allison 'blue' VCAs I worked with.

I used Buff VCAs in one compressor/limiter/gate back a hundred years ago, The modern THAT VCA is significantly better than even last generation THAT VCAs thanks to process improvements that improve the symmetry of NPN and PNP parts performance.

If you think about the distortion mechanism related to EC base current it might cancel out at unity gain but at different gain settings I don't expect the base current to be identical at EC+ and EC- ports. While I have not conformed this on the bench, the way a current ratioing VCA works the further away from unity gain the more imbalanced the base current in the EC ports. So even if you could get a first order cancel at unity gain, that would degrade, as you increase or decrease gain.

The general design advice seems to be keep the total impedance at those EC ports low. Preferably grounding one, and terminating the other with <50 ohm.

Note for elskardio, Since you have two VCA control ports connected in parallel, the impedance advice should be <25 ohms to be consistent with THAT <50 ohm advice for one VCA.

In dem days, we took scrupulous care to match the resistance present to the Ec ports to minimize distortion. My memory is vague but we might have driven them differentially too.

But you should refer to the THAT datasheets & articles for chapter & verse. What I do remember clearly is that you definitely want to do your matching at '0dB' gain.

Again I would expect the standard circuit with 1% resistors to track a lot better than 2 dB at unity gain (0V) control V.

JR

abbey road d enfer · Oct 11, 2014

elskardio said:
Hi John,
The gain offset at the EC+ port works but the VCA's harmonic distortion increase a lot!
Even if I tweak the Symmetry, nothing change...
Any idea?
Thanks

Can you be a tad more specific?
Do you mean you can actually hear distortion, or is it a measurement?
At what gain settings?
At what input and output levels?

dfiction · Feb 8, 2016

Hello, wow it's been a long time since anyone posted here. But I'm going to give this a try anyway!

Do you think you could provide a schematic on how to implement the voltage follower? I'm a little confused—where does the 51R go in concert with the follower?

best,
daniel

abbey road d enfer said:
Is your schemo up-to-date? You need to add a resistor between the Gain offset trimmer and the EC+ port. And I believe you should add a 51r at the Ec+ port of the other channel.
BTW, I've always thought that the value of 51r should be reconsidered in view of the performance. I've never used more than 22r in this position.
There is a significant advantage (and some drawbacks) in driving the Ec ports with a near-zero impedance such as provided by an opamp. Then the offset control can be applied to the opamp, not directly to the Ec port.
Finally, I note that the gain control range is +/-40 something, which is not really convenient for a volume control, particularly on the + side. I believe a range of -50/+10 would be more suitable. That is obtained by increasing the value of the 3.01k from the wiper to the Ec- port to about 4k and offsetting the Ec voltage by adding a resistor (about 6k) between V+ and Ec-.

Beware that with the resistor values indicated here, you are deemed to have problems with the potentiometer. Assuming +/-15V rails, the dissipation in the 5k pot is about 0.2W, which is pretty close to the max for many commonly available pots (1/4W). In addition the current through the wiper will certainly quickly destroy the fragile conductive surface.
One option would be using a wire-wound or cermet pot, but that would still leave a serious problem: the scratching noise of the wiper. It is customary to put capacitors across potentiometer-derived control voltages. You would need to install an electrolytic or large film cap (about 10uF) from wiper to ground to tame the noise. Beware that the control voltage is changing from positive to negative so you need to a bipolar or non-polarized type.

In view of all these considerations, I would advocate the use of much larger value pot and res, delivering the control voltage to the Ec ports via a voltage-follower. It needs to be a low noise type, such as a 5532.

abbey road d enfer · Feb 8, 2016

dfiction said:
Do you think you could provide a schematic on how to implement the voltage follower? I'm a little confused—where does the 51R go in concert with the follower?

Thevoltage-follower would drive both Ec- ports directly. The control voltage is 6mV/dB, so you need to scale down the voltage from the pot's wiper with a resistive divider. As I mentioned earlier, in order to have a usable control range, the voltage should be offset. the easiest way is adding a resistor in the -V connection of the pot.

dfiction · Apr 4, 2016

Hello,

Thank you so much for your helpful response! I'm tweaking a circuit and I thought I'd post it here, as I've created a circuit with +/-50 gain, and really I'd like to switch it to -100db to 0 db. So, now I'm trying to create an offset and I'm not quite sure how to apply that to the circuit I already have. Right now I'm dealing with significant gain and thus, i'm getting lots of distortion. I only need to attentuate gain in this circuit, not add any.

some aspects of my circuit:
1) my application calls for a single supply, so i'm biasing a virtual ground node at 6v, and naming that "VCC/2" in the circuit. I grounded the ec+ channels at this node.
2) I calculated my gain from this equation: http://www.wolframalpha.com/input/?i=12(56/(56+x))=.61 —I just wanted to show my work!
3) I'm using a buffer on the vca gain control (see) U$3B. Should I add my gain offset before or after the buffer?
4) how do I "match for 0db", according to John Robert's earlier advice?

best,
daniel

JohnRoberts · Apr 4, 2016

I am not enthusiastic about using a V/2 supply for a VCA that likes low impedances.

Since you are buffering the VCA EC signal with op amp why make the R23 only 51 ohms? maybe scale that higher.

The EC drive from the op amp may benefit from a RC on the output like U$3A...

THAT published an AES paper about that VCA and it likes low source impedances for control ports. (But I read it years ago so my memory may not be very precise).

JR

abbey road d enfer · Apr 4, 2016

dfiction said:
Hello,

Thank you so much for your helpful response! I'm tweaking a circuit and I thought I'd post it here, as I've created a circuit with +/-50 gain, and really I'd like to switch it to -100db to 0 db.

With this new schemo, that's what it should get, since you're driving the Ec- port with positive only control voltage.

So, now I'm trying to create an offset and I'm not quite sure how to apply that to the circuit I already have. Right now I'm dealing with significant gain and thus, i'm getting lots of distortion. I only need to attentuate gain in this circuit, not add any.

This is not as expected; max gain should be 0dB, min gain about -90. If you don't get it, check the control voltage; I suspect the VCC/2 reference is not stiff enough. With the gain pot at max (min gain), there is 10mA flowing in R23, then U3B is asked to drive the two EC- ports. The resulting current is again dumped, via the EC+ ports, into the VCC/2 reference, which may not be actually very stable. First you should scale up R22 & 23 by a factor 10. Note the gain pot's wiper may not like the idea of carrying 10mA!

3) I'm using a buffer on the vca gain control (see) U$3B. Should I add my gain offset before or after the buffer?

First you should find why the actual circuit needs offset. Then, if you still need to add some offset, it depends if it's a common offset (which you should apply before the buffer) or separate (L-R); in the latter case, you could offset via the EC+ ports. If not, you would need two CV buffers.

4) how do I "match for 0db", according to John Robert's earlier advice?

You should not really need to match anything, but if you're concerned about a fraction of dB, you should trim one of the FB resistors, on U2A or U2B.

Proper way to match the gain of 2 VCA in a stereo volume control?

Help Support GroupDIY Audio Forum:

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Member

Well-known member

Member

Attachments

Well-known member

Well-known member

Similar threads

Join the conversation!

Join today and get all the highlights of this community direct to your inbox. It's FREE!