Console VU meter causing THD problem

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

living sounds

Well-known member
Joined
Jul 26, 2006
Messages
3,931
Location
Cologne, Germany
Following suggestions from very helpfull and knowledgeable forum members I've lowered THD in my console signal path from channel line input to master output to essentially my soundcard's level (first harmonics visible for a 1k sinewave start ca. 110db below signal peak), via heavy local decoupling and building a "dirt ground".

Sounds great, but now I've discovered that switching the VU meter to display the output signal results is raising THD by a considerable amount. This does also happen when the meter gets fed the PFL signal.

I should also mention that all signal path caps have been bridged, excluding the inaudible ones like those going to the meter, of course. The 2uf caps before the meter have been replaced with 4,7uf.

All caps, including the TL074 driving the meter are bypassed locally via three 10uf electrolytics (+/- to ground and rail-to-rail) as well as a rail-to-rail 100nf ceramic. Decoupling the TL074 didn't make a difference though.

Any ideas how to fix this? Thanks!
 

Attachments

  • 200B_master.png
    200B_master.png
    436.7 KB · Views: 119
Regular (unbuffered) VU meters do this due to the non-linearities in their rectifiers. How much distortion is created depends principally on the output impedance of the driving device.

Cheers

Ian
 
living sounds said:
Thanks! So they skimped on an additional TL074 and a handfull of caps and resistors here?

Real VU meters were designed to hang across 600 ohm terminations so probably made even more distortion than you are seeing now.

yes, you should be able to clean it up with a simple buffer.

Another tidbit, when buffering a mechanical VU meter. make sure the buffer can pass adequate signal swing. I once tried to get away with buffering a real VU meter bridge using the somewhat reduced voltage from my LED meter bridge's power supply rails. Visible tracking errors were noticeable to experienced users (with handy VU meters nearby fed with the same audio signal to compare).

I was never a big fan of mechanical VU meters (but the customer is always right). If you are going to use them for more than ornamentation, you don't want to clip the buffer in front of them prematurely to realize proper ballistics. 

JR 
 
Thanks. So would a TL074 or LM348 after the dc blocking cap in the schematic with the signal going into the positive input and the negative input connected to output suffice?
 
living sounds said:
Thanks. So would a TL074 or LM348 after the dc blocking cap in the schematic with the signal going into the positive input and the negative input connected to output suffice?

That should do it so long as it is run from the same supply as the driving circuit.

Cheer

Ian
 
Upon reviewing that schematic I may have another thought about what may be going on.

? Is the distortion only at high level?

If I read the schematic correctly it looks like the meter feed is already opamp buffered.

Those opamps are configured in an inverting topology. That is mostly inconsequential to the meter that doesn't care about polarity, but the input impedance of an inverting gain stage changes dramatically if the output clips. As long as the output is unclipped, it will hold the - input at virtual earth, and the input impedance looks like a fixed resistance. The moment the output clips, the input impedance rises in a step function to the input resistance "plus" the feedback resistance. So this step change in load will talk back into the source impedance of the previous stage. Which appears to be some FET switches.

While it may take cutting some traces, changing those opamps to non-inverting topology may clear up that particular distortion.

You should be able to confirm this mechanism before cutting traces, if the distortion cleans up for levels that do not clip that opamp stage, that is the smoking gun.

or not....

JR
 
Even at low levels (as low as I can see above the noisefloor) the meter adds distortion, most prominently the 3rd harmonic, and the 2nd harmonic is as low as the 6th.
 
One possible mechanism is the meter driver opamp is dirtying up the rails around it. You could try to isolate it from the more sensitive parts of the circuit with separate damping resistors (at least 10ohm) and some 10-22uF caps. Try with one channel first.

In my mixer project for example I had to do this to overload/signal LED driver opamps. You could easily hear the LED's click in the audio band.
 
Kingston said:
One possible mechanism is the meter driver opamp is dirtying up the rails around it. You could try to isolate it from the more sensitive parts of the circuit with separate damping resistors (at least 10ohm) and some 10-22uF caps. Try with one channel first.

In my mixer project for example I had to do this to overload/signal LED driver opamps. You could easily hear the LED's click in the audio band.

Decoupling the meter op amp didn't make a difference at all. And all decoupling caps go to a dirt ground, too.
 
living sounds said:
Decoupling the meter op amp didn't make a difference at all. And all decoupling caps go to a dirt ground, too.

Exactly, increased decoupling alone shouldn't help. It doesn't isolate the opamp from the rest. Those damping resistors are the crucial bit. In fact, do you even have damping anywhere in the channel strip currently? Or have you simply added lots more capacitance directly on the rails? If you have, it's like you now have a telephone line between all the opamps, and trust me they will talk.
 
I see R103-R106 in the schematic now. These are damping resistors if the concept wasn't clear enough. Only two damping stages in a system of this complexity is a bit optimistic from the designer. Or more likely: he had a tight budget for electrolytics since each damping stage eats one.

That meter driver is a "high current" opamp stage compared to summing and fader buffers. My estimation is that it dips the rail voltages of these sections enough to create that THD you are seeing. The PSU regulator can't keep up since it cannot "see" through those damping stages.

The fix is completely separate RC's (damping stages) for the meter driver opamps. 20ohm and 22-47uF are probably fine. You can dump their dirt to the same garbage drain you are already using for the rest of the system.
 
JohnRoberts said:
Can you measure the distortion upstream or before the JFET switches.

Those are not zero distortion.

JR

Because of the topology it's not easy to test while on and I'd have to build a sort of "probe" for the sound card input, since the scope won't show this still low distortion.

But the distortion can be turned on and off via the switch circled at the bottom. And the whole circuit part encircled in green including the JFET switches is unaffected by this switch. So I would presume that indeed the problem should be caused by the meters.
Wrong?
 
Kingston said:
Exactly, increased decoupling alone shouldn't help. It doesn't isolate the opamp from the rest. Those damping resistors are the crucial bit. In fact, do you even have damping anywhere in the channel strip currently? Or have you simply added lots more capacitance directly on the rails? If you have, it's like you now have a telephone line between all the opamps, and trust me they will talk.

I've decoupled everything with electrolytic caps, which - according to the discussion in the other thread - provided sufficient inherent resistance to provide damping.

From my measurements and just listening to the console the whole exercise (with the dirt ground, of course) cleaned everything up nicely, and even TL072 sound great in there now.

I'm also building a better PSU.
 
Kingston said:
The fix is completely separate RC's (damping stages) for the meter driver opamps. 20ohm and 22-47uF are probably fine. You can dump their dirt to the same garbage drain you are already using for the rest of the system.

So a 20ohm resistor each in series with the +/- supply lines for the TL074?
 

Latest posts

Back
Top