ETF and BFD graphs.. (1 Viewer)

[Ranga]

Finally managed to get a reading from the ETF 5.0 software - forgot to turn ON the RS Meter - :b
Link Removed
Here's a graph of the Manual measurements that I took -
Link Removed
Looks pretty similar. I am worried about the nulls at 32Hz and 63Hz. For this measurement, the crossover was set at 80Hz in the processor (fixed).
As a preliminary exercise, I set 3 filters on the BFD using Preset 4 -
1 - Freq – 50, Fine – 8 , bandwidth – 30, gain - -12
2 - Freq – 80, Fine – 0 , bandwidth – 10, gain - -7
3 - Freq – 100, Fine – 0 , bandwidth – 20, gain - -5
Here is a graph with these filters on -
Link Removed
It looks like the filters 1 and 3 did not really make a difference. Is it because my BANDWIDTH is too wide?
Secondly, after I set the filters, I needed to increase the volume on my sub's (25-31PC) amp to balance the left and right channel readings in ETF. Is this correct?
Would like some suggestions on filter values and number of filters.
Ranga

 

[Ranga]

:frowning: :frowning: :frowning:

 

[Mike Matheson]

Ranga,

My experience playing around (still limited so far) has been that the bandwidth numbers I compute using the standard formula stuff ("Q" and all that) tend to be too wide.

I wind up setting ETF to its "Sequential Data Acquisition" mode so that I can play with my BFD settings and then run a sweep and see how the graph was affected. This method has greatly helped me understand what the heck is going on with the BFD. You can make small changes and see what (if anything) changes.

Small note--when you jump around on your BFD between filter sets, note the little green LED in the bottom right corner of the display. It indicates the filter set you've jumped to is loading. Wait until that light is out before running a sweep.

Mike

 

[Ranga]

Mike

Thanks. Keep them coming.. I am going to try that this evening..

Ranga

 

[brucek]

Ranga,

Yep, I'd say the graphs look very similar, but you have to remember to compare apples to apples. You've wisely altered and matched the vertical scaling, but you have to remember that the horizontal scaling is quite different on the two readouts. This will change the appearance of how the data is presented.

A graph of course can have any scale you like to present data in many different ways. It doesn't change the data, it changes the appearance. You can have linear divisions with non-linear or logarithmic values, or you can have logarithmic values with linear divisions, or linear values with linear division etc, etc.

I believe when a graph is made of the results of octave-band sound pressure level measurements rather than individual linearly increasing frequencies, the frequency scale is commonly divided into equal intervals. This would be the case of your Excel graph were each successive measurement is a sixth octave higher. This compresses the graph properly as the frequency increases along the lower axis. Alternately if you had measured instead at 10, 20, 30, 40, 50, 60, 70, 80 etc, then you would have needed to use a logarithmic X scale.

The scale that ETF has chosen for its graph is a linear value with linear divisions. Personally I like to see more expansion on the lower frequencies, so the Excel graph is a little easier to work with, but since you have ETF running, you should likely use it...

Anyway, I would say your bandwidths are too wide. Your data says your first filter is at 50 with fine 8. That's 55Hz. I suspect you meant 50Hz with fine - 8, thats 46Hz.

If so, you have a 46 Hz filter that is 30/60th or 1/2 octave wide. That's very wide. It's full effect or "absolute bandwidth" will still be felt 1/2 octave above and 1/2 octave below. A 1/2 octave below 46 is around 32Hz. Yikes.

The majority of the energy will be in its "relative bandwidth" though, which is 1/2 octave centered about the center frequency of 46Hz at the 3dB down point. But its effect is still felt right down to 32Hz - way too wide.

I think your best bet is to start with the 46Hz peak and try to eliminate it first while experimenting with one filters parameters to get the feel of their effect. Since you have ETF working, it should be easy to observe.

brucek

 

[Ranga]

brucek

It looks like my best option is to take a new measurement, and then use the "Sequential Data Acquisition" mode to experiment with one filter setting at a time.

How does the overall unfiltered response look?

Ranga

 

[brucek]

Doesn't look like you should have any trouble, but you never know until you play with it.

The SDA mode is very helpful. The immediate response you get back from altering a filter helps a lot.

brucek

 

[Rick Radford]

brucek,
Ranga has 2 valleys at 35 and 63Hz. What do you think about a mild boost for them... say, 3 to 4 dB or so, and cut the rest of the freqs down to that level (of the slightly boosted freqs).
Perhaps:
1) a 10dB cut at 21Hz (experiment with the BW.. but probably something like 6 to 10/60ths octave, I'd guess.
2) Then another 3dB cut at 28 Hz (with another narrow filter, ~6/60).
3) Then a 4dB boost at 35Hz (play with the BW in SDA mode).
4) And a 13dB cut at 46Hz, maybe 15-20/60ths wide,
5) and finally a 3dB boost at 63Hz, fairly narrow BW.
Ranga, start at the lowest freq and work up. After you tweak a filter to your satisfaction, take a new reading before adding another filter.
Since the application of a filter can affect the response curve, calculating all the filters from just a base measurement can be problematic, I'd think. Thus, take a base reading and save it. Run SDA and apply beginning filter parameters. Tweak 'til you're happy and save that parameter. Take a new reading with the filter(s) applied and determine your next filter's parameters.
I saved the graphs as each new filter was applied so I could see how the response curve was affected with the application of each filter (overlay mode).
Just as an example, here's an ETF graph of my sub (no RS cal file applied) with the non-EQ response in blue (baseline) and my final EQ (6 filters) in red (EQ 02.07).

Here's the baseline and EQ result using the Excel graph (with RS cal applied):

Filter parameters are:
FilterFrequencyBWGain
F1:25-8/606/60-3
F2:32+8/606/60-6
F3:50-8/606/60-10
F4:50+8/603/60-7
F5:80+1/6010/60+3
F6:40+0/604/60-1
** Sonnie.. if you see this, the ETF graph is a bit large. Can you fix it? (Sonnie put some of my info on his site since I have problems with GeoCities).

 

[Dennis B]

Ranga,

Try increasing that gate time to, say, 200 ms or 300 ms to get a more detailed plot. Your peaks are wide and you can't really tell what's the center frequancy of them. Such wide peaks usually mask the "true" excited frequencies. Another very useful thing to do is apllying multiple time slices to the LFR to check out the resonant modes of your room. Check out ETF's tutorial in the help section or at their website to see how to do this. Mess with them first and as Rick said, start from the lowest frequency.

Rgds.

 

[Wayne A. Pflughaupt]

Mike,

If so, you have a 46 Hz filter that is 30/60th or 1/2 octave wide. That's very wide. It's full effect or "absolute bandwidth" will still be felt 1/2 octave above and 1/2 octave below.

In other words, a so-called 1/2-octave filter affects that much above and below the filter center.

Regards,

Wayne A. Pflughaupt

 

[brucek]

Rick says:

Since the application of a filter can affect the response curve, calculating all the filters from just a base measurement can be problematic

Absolutely true. Right on. Even if you forget about the fact that you cannot tell if a null or peak will respond to equalization from simply looking at an unequalized graph, the fact is as Rick indicates, a filter at one frequency has the ability to modify the response at another frequency.

The SDA mode in ETF gives you this insight. From my experience, if you remove a peak with one filter, this area can then change its response somewhat when you add another filter that is outside that area. I suspect this effect comes from the varying phase effect that each additional filter adds to the mix.

As others have said, it seems to be the best method to simply start at the lower freqencies and add a filter and observe its effect and then add a second filter and so on...

brucek

 

[Ranga]

I missed Rick's and Dennis's comments when I was applying the filters.
Here is how my frequency response looks now -
Link Removed
Filters applied -
Link Removed
Questions -
1. As you can see, I had to boost 2 frequencies - 34Hz and 66Hz. When I applied all the filters, I noticed another valley at 92Hz and applied a small boost to it although I may get away without it. I understand that it is better to cut the peaks and minimize the boosts. Considering that I have used just 5 filters, do you think I may get better results by using more filters to cut the peaks which may help removing the nulls?
2. My filters are set for both L+R channels (couple mode). Is this fine?
3. I used the RS calibration file provided by brucek in ETF. So I did not add the correction values in Sonnie's XLS spreadsheet to plot the graph. Is this correct?
Ranga

 

[brucek]

Ranga,

3. I used the RS calibration file provided by brucek in ETF. So I did not add the correction values in Sonnie's XLS spreadsheet to plot the graph. Is this correct?

Correct, if you got the values to enter into the Sonnie graph from your ETF graph which was corrected by the rs.cal file. You don't want to correct the readings twice..

brucek

 

[Ranga]

How does it sound ??? -

The sub localization issue is gone and I am playing it louder than before although the 25-31PC's volume is still around the 10-11 'o clock position.
I haven't played around with the sub's phase and ETF. I had manually set the phase before the ETF came in. Do you think I should leave it alone?
Also, I tested the new settings with some music and Floyd's The Wall DVD. I am going to test some more.. Before that I think I need to re-calibrate my system as the peaks have been eliminated.
Thank u all..
Ranga

 

[Rick Radford]

>I need to re-calibrate my system as the peaks have been eliminated<
Absolutely! You may find the bass to be significantly enhanced at lower volumes. I know I have.
Now I gotta sort through Vince M's lengthy thread about DTS/DD pad here. I'm still scratching my head on this.