There has been some discussion in the past about the importance of a capacitor to be a specific value and how much of an audible effect the value has on how it sounds.
I will use measured responses to verify the variances in value changes.
The test speaker in this case will be an A/V-1.
This first measurement is shown just as most companies show responses (at least those that do publish response curves) on a 10db scale.
This is the most common and makes the speakers response look really nice and smooth.
We seldom ever even look at our response curves on this 10db scale. It does not allow a close enough look at the response.
We typically use a 5db scale for better perspective.
Shown below is the response curve on a 5db scale.
These made it into the +/1db level from 500Hz and up, and nearly make it into that range from 200Hz and up.
Now let's see how much difference a cap value change will make.
Below is the response curve just as above, which used a 6.8uF cap in the tweeter circuit and the difference in response when a 6.0uF cap is used in the circuit instead.
Note the difference in response. The greatest difference was made at 2,273Hz, and the difference was .54db. That is about 1/2 of a db difference in going from one standard cap value to the next.
This was a cap value difference of .8uF (a whole value change).
Differences in the drivers themselves can typically make as much or more difference than this.
How hard is it to hear 1/2 of a db difference centered at one point in the response? Pretty hard. In fact it is really hard.
If listening to a pair of speakers that has one cap value in one and one cap value in the other then the musical information in that area would be 1/6th as loud in that area. This would simply shift the image slightly toward one speaker by only a few degrees.
If listening to a mono pair of speakers this way and trying to compare them, then most people would never hear the difference in these two speakers.
For the average person to hear a difference there needs to be about a 3db difference in output level in an area for a difference to be heard. At leasts that is what has been said to be a known standard by members of the industry.
I feel that a trained ear could detect as little as maybe 1.5db difference in output in a given area.
I am talking about a change in a given area here and not a tipping up of one end or the other.
Yet some people have publicly stated that differences in cap values down to the third decimal place over can be heard. Third decimal place over? .001?
Let's look at this more closely.
First lets zoom in on that lasts response curve.
Yep, this is the same impressively smooth response just zoomed in to a 1db scale. Looks rough now huh?
Now I have shot it again with the 6.8uF cap in place and then I have bypassed it with a .1uF cap. Not a .01 or a .001uF cap, just a .1uF cap.
Yep there are two separate lines there.
Look close and you might see that at 2,273Hz the difference in the response is .04db.
That is 4% of 1db.
If you look real close you can see another variation around the 1,780Hz range that is equally as big.
This could simply be just differences unrelated to the cap change.
I found that there can be as much as .02 to .03db variance from one response to the other when using the same caps and just making back to back measurements. Most of the variances are typically in the lowest frequency ranges where slight changes in air pressure can make slight differences.
Hmmm, now I wonder who out there can hear a .04db change?
De-winding a cap down to a specific value that is accurate down to .001uF is pretty tough too.
For one there is only one piece of equipment made that is accurate to that low and it cost $10,000.
Furthermore, variances that low are not stable. Changes in temperature, barometric pressure, even holding it in your hand will cause greater fluctuations than .001uF.
Our Clio will read to the third decimal place but is only accurate to the second. Readings in the third decimal place over are never stable and move constantly as the signal is passed through them.
Now on a different note, de-winding a pair of caps down to 6.8uF will make them sound the same, but allowing one cap to be a 6.8uF cap and matching it with a pair of caps with one being a 6.7uF bypassed with a .1uF will not sound the same at all even though both values are 6.8uF.
Now many properties that effect the sound are changed by the by-pass cap, but the sound will not be changed because of the change in value.
Difference are also very apparent between different types of caps too, but that's a different story.
So whether the caps used in a design are 6.20's, 6.15's, or 6.29's it will have little effect on the response.
What is important is that the caps used in both speakers are close to one another in value for the left and the right speakers.
If you have ever ordered a kit from us and ordered the Sonicap upgrade then you can see that each cap has been measured and the exact value to the second decimal is written by hand on each cap.
Caps are then selected to be matched in pairs that are often vary near to exact values.
If you have ever ordered our one of our A/V series kits then you will have noticed that the tweeters have been measured, tested, and each tweeter comes with its own response curve.
They are then matched into pairs with near identical responses.
If you have ever ordered fully assembled speakers from us then I am sure that you have noticed that each completed speaker has been measured and tested and comes with its own response measurements. Overlapped measurements of the pair are also provided so one can verify the closely matched pair of speakers.
Matched left and right speakers are very important for good imaging.
If your brand X speakers didn't come with it's own set of measurements and you would like to see just how accurate they are or how close they match one another, then you can send them here to GR Research for measuring.
It's a free service we have provided for years.
Just cover the shipping both ways, and we'll provide you the results.
Happy listening.
