L/R level accuracy tolelerance

how much is to much? 

the L/R output level differences of an analog gear,  how much can be tolerate?
see the attached pic please,  what can cause  0.2  difference between L and R RMS level? (loopback test)
thank you

not enough info but that doesn’t look bad to me

Small level difference may shift the image off center to L or R. Our brain uses loudness to localize sound sources.

A small fraction of dB should not be very audible.

For mastering you want to be dead on, for general playback how does it sound to you?

JR

PS: One of the few things that mechanical VU meters do better than LED displays is provide fine resolution around 0VU for zeroing a desk. 

Depends on the nature of the test but 0.2dB difference in rms with no difference in peak value could simply be slightly more noise in one channel.

Cheers

Ian

simonsez said:

how much is to much? 

the L/R output level differences of an analog gear,  how much can be tolerate?
see the attached pic please,  what can cause  0.2  difference between L and R RMS level? (loopback test)
thank you

Click to expand...

I see that much difference comparing AD/DA converter channels in loopback tests.  You see much more looking at stereo analog faders or pots, and it changes with position. 

Very small differences will not audible though, just trying to get better matching.
Sometimes small  differences number on the DAW meter cannot just ignore.
and I thought mechanical VU meters is less accurate than digital meters,  especially when try to get accurate L/R matching/adjustment.

ok, so same dbfs, 0.2 rms differences is really matter? or it still  tolerable in the analog world?

0.2dB is 1%.

Analog devices are rarely matched to 1%.

any number of things can cause a level difference. Any difference that small is nothing to worry about.    About the only way to get better then that is to test and measure every component to be 100% identical. When I did my manufacturing bench stint we made a mastering eq that was street price of 15,000.00 dollars. But every position on our step switches were exactly identical.  so left and right were .1dB or better in difference. To achieve that took a lot of time and effort.  What fun we had matching capacitors to be the same capacitance and so on. Is it worth it? I would say so,  but it was very time consuming.

The  Igor Sontec I built is very tight tolerance like described above. That one took 100 hours to build. The second unit looks like it will come in at about half that. Still very time consuming to build.

That $15,000 comes up fast with that much labor, expensive switches and having to buy 3x-4x the parts to get matches.

Gold said:

The  Igor Sontec I built is very tight tolerance like described above. That one took 100 hours to build. The second unit looks like it will come in at about half that. Still very time consuming to build.

That $15,000 comes up fast with that much labor, expensive switches and having to buy 3x-4x the parts to get matches.

Click to expand...

tell me about it. we would solder table the main board and daughter boards.  The stepped attenuators and switches, were all hand built.  Then you can factor in which version as we did 2 types(Japan version and regular version).  Take for example the LF boost or cut, we had 1dB steps up to 10dB for boost or cut.  it worked well but the folks in Japan wanted half dB steps for the first 4 positions and then dB for the remaining 6 positions on boost or cut.  we did other things that were Japan only versions that you would not find in the  regular versions. All of it related to what each step was doing on boost and cut at each band of which there were 4.
Then after all the switches  and attenuators were wired in,  we had hours of testing and calibration of trim pots to match as closely as possible.  It takes it's time, but if you want it just right.. Factor in labor times to do all the wiring and switches along side calibration and what looks like a lot of money for a unit is easily explained.

I even came up with some time saving measures to tweak the switches.  Normally I’d figure out resistor values by using a pot and measuring to come up with a value.

This time I twisted together resistors to form the strings. That way it was easy to substitute different values . Just untwist  and put in a new value. 

The  resistors for frequency were from the Harpo spreadsheet and matched what I measured. The gain switches had to be figured out. The twisting and untwisting saved a ton of time over soldering and desoldering a switch. It also didn’t trash a switch.