KEF KC62 — An applied review and guide for setting up an audiophile desktop setup (Part XXV)

Friday is fun time. So let us dig more into the world of the audiophile. Last time, there was already a bit of reasoning on group delay and sums of frequency bands.

Today, after some weeks of vacation, also from my audio system, i now can come back with some fresh pair of ears.

This means the following:

  • I now know for sure that it is possible to identify 6.x as the perfect ballpark for delaying the 8030c’s. I know this from stable measurements around the listening place, remembering the calculated theoretical delay, and also by being able to make use of crossovers from 60 to 85hz without much difference in amplitude summation (and to some extent also phase and grouddelay).
    I know there had been some question marks around that, and for sure listening to some progressive tracks suggested ultra-high delays below 10ms. The calculated delay was around 6ms (SB 3000 dsp processing plus listening distance difference between sub and monitors); i somehow ended up with ±6.5ms now.
    This exercise tells us that physics is useful anyhow, and that deviations are not feasible just because GPT tells you so.
  • So there are 2 things absolutely paramount here: 1. get the delay right and 2. get the filtering right. The issue here is that those are entangled: when i got the delay a bit too high, changing a monitor peq @90Hz was addressing 83Hz etc. So both have to be tuned in a ping pong way.
  • Measuring at different points around the listening position and bein able to minimize the amplitude deviations in these is also a good sign of a good delay and filtering setup; this will increase your degrees of freedom in moving around in front of your audio system at listening position.
  • Another important point is some refined understanding of fletcher-munson and how to use them in practice. Here is is again:
  • so as one can see, for 80 phon (sine wave 1kHz), around +28db SPL are being needed to display the same volume at 20Hz. This can lead to a false conclusion such that your sub would have to be able to produce 108db to match the loudness perception (know that 80 phon equals 80db SPL, and it also equals roughly dbc, but ONLY @80Hz!!!).
    Now the issue with this assumption is that one would need to recall what you actually try to do: let’s say you are listening with 90dbc at max, 95% of the time. Now how does that translate to the necessity for sub performance? The point here is that the summed amplitude which is measured by a dbC meter eg. will translate 90dbC to around 80–85 phon (instead of 80dBC!!!). Now this means in layman terms that one need a lower subwoofer performance than suggested by fletcher munson (see table below):
dbC to Fletcher-Munson (phon avg.) table

As you can see, if you have a measured sum level of 80dBC, your system needs to deliver 80db SPL @80hz, 90 dbC demands 90dB SPL @80hz, and 100 dbC, which is really loud already, will demand 100dB SPL @80hz.
So @80dbC, only below 80Hz the frequency bins need to be louder than the summed level; this will also be the case above 1kHz, but only for louder levels above 100dB. Have a look at the Fletcher-Munson graphics again.

Now the header is the punctum here: to you need to look at the right point in the fletcher munson table to understand what you really would need (room compensation not included!).
Or in layman terms: if you want to listen to 90dbC max, it means you need to look up 82 (or, roughly 80 phon) values as also shown in the table. It means that for non-cinematic content, which is mostly musically content, you need about 100 dB SPL @30hz usw.

But what about room compensation? Let’s have a look:

Exemplified room compensation gains included for 30sqm room (estimation)

As one can see, the demand for 90dBC @30hz is reduced by about 6db, such that you now need 93dB SPL only, instead of 99dB. This is worth considering: 90dBC is already really loud, and above 100 dBC one reaches the danger zone for hearing damages.

Now one can connect this to CEA-2010 measurements, like provided by some very nice people online; those are the numbers as per 2 different measurements:

90.3@15Hz
96.6@20Hz
101.7@25Hz
[email protected]
114.6@40Hz
118.7@50Hz
120.7@63Hz
120.9@80Hz
120.8@100Hz
120.7@125Hz

source:https://docs.google.com/spreadsheets/d/1dU5OOnf3nVgctJszmfyBjaxK69dkXte6ZL6anVTW2_M/edit?pli=1&gid=834598950#gid=834598950

and then

89.9@16Hz
94.5@20Hz
102.1@25Hz
[email protected]
115.2@40Hz
119.0@50Hz
121.4@63Hz
120.1@80Hz
120.2@100Hz
122.0@125Hz

as measured by Erin:
https://docs.google.com/spreadsheets/d/18bz7z-xIlRJsC-bw6k4mHkuwv_uiGAMyEhgrTkjwdXc/edit?gid=407670234#gid=407670234

Note that the lowest value somehow indicate a free field level of 90Hz @±16Hz. This will become relevant in a minute; note that the implications are that a. the CEA-2010 measurements are freefield 2m distance, and b. my listening distance is about 1.2m and room compensation is adding to that. This will result in a precisely estimated output level, including room gain of about +10db, and a distance gain of 1.2m vs 2m, of about 104dB SPL peak. For permanent levels, one can subtract about -3dB, which results in about 103dB SPL. This results in a demand supply result as shown in the following graphs.

Now let us have a look at the lower boundary, remember that we need 16hz, since this marks the lowest tone of a church organ:

As one can observe, (second last points from left side == 16hz), 80dBC seems to be the last reasonable loudness where the system would render 16Hz equiloud. Not that one needs this every day, but this was the goal (see headline): 16hz — <what you still can hear, 16kHz in my case>. For ±23Hz, it is in exactly working out @90dBC (intersection between green and black curve).

So what if i add a 2nd SVS SB-3000? Let’s have a look again at the violet and gray curves:

The room gain is not so easy to assume as for one, but it would somehow give me around 10 more dB if i would like to, meaning, i can listen to an avg volume of about 90dbC without loosing 16Hz, assuming a gain of +6dB, which is somewhat optimistic. But even with +4dB gains, it will be a similar picture.

The main reason for considering a 2nd subwoofer is the negative peak @±35Hz (which is also shifting towards the listening place, where it is not measurable then); this is not an immediate flaw since it have a nice flat spectrum at my listening position, but one of the main reasons for a 2nd sub would be this dip, and i also expect some even flattened out group delay between 25 and 100Hz.

One last detail making the graph a bit more realistic: we need to subtract about -3dB for permanent listening levels, since CEA is peaking the output. What does that give then? Let’s have a look again:

As one can see now, for 90dBC, one SB 3000 is only sufficient until down to about ±23Hz. We leave the group delay topics out for now here, which are a different story at those depths. 2x SB 3000 would solve the issue here really until 16Hz .. let’s see.

Learn more about KEF KC62 — An applied review and guide for setting up an audiophile desktop setup (Part XXV)

Leave a Reply