Inconsistent Results

Hi Guys

REW is giving me inconsistent results. It is as if the filters become corrupt or my BFD intermittently refuses to filter. If I repeat the same frequency sweep without changing anything, my results are one or the other of these two attachments (seemingly random).

Anyone else ever experienced this? Not sure what to do next. Thought I had produced a pretty good curve with 6 filters and now this

Any advice gratefully received

Thanks

Al

[brucek]

The top graph appears as if the filters were turned off and the bottom looks filtered...

As long as the Filter IN/OUT light is on, then it should be working unless there is a fault in the BFD. I suspect you've turned the power on and off...

brucek

Then there could be a fault in the BFD. Nothing has been changed between the two readings. Just now I thought it was because I had inadvertently set a second filter at the same frequency as another but I am still having the problem so that can't be the reason. I wonder if the filters can become corrupt. Maybe I should reset the BFD to factory settings before putting the filters in again. I recall reading that there is a way to do this but don't know the procedure. I expect that it is in the manual and will take a look.

I do hope that the unit is not faulty because new 1124p's are virtually impossible to obtain in the uk now.

al

Edit> Did a reset but still have the same problem.

[brucek]

You haven't recently changed the cables in the rear have you?

No, I haven't recently changed the cables in the rear.

Some more information about the problems I've been having...

The top attachment here shows the unfiltered curve. As you can see I have a big dip between 30 and 40 hz. The sub is currently in the front right hand corner of the room. The only other location for it is a little further left near the right hand speaker, or somewhere in between. I have moved the sub around and still get that dip in my listening position, which cannot parctically be changed.

I have attempted to equalize according to Ayreonaut's house curve that he recently posted. I used 12 filters to produce the second curve attachment. That was my first attempt which began with transposing the curve upwards to bring the bottom of the trough to a decent level before making some pretty major cuts elsewhere. The filters are as follows:

Filter ISOFreq BW Gain
1 49 10 -35
2 31.95 3 -17
3 83 6 -15
4 68 5 -9
5 35.8 2 8
6 57 4 -10
7 97 10 -20
8 29.35 4 -4
9 33.65 3 -2
10 44.5 5 -5
11 78 2 -3
12 25 20 -9

As you can see, there are rather a lot of quite severe cuts so I thought I'd have a go at doing it again with fewer filters. The result is the two curves in my first post - the REW plot alternates between the two. The filters were as follows:

Filter Freq BW Gain
1 32 8 -12
2 50 40 -8
3 63 15 -9
4 40 5 +4

Could the inconsistency be due to filter 2, i.e. having a cut over a large bandwidth (40)? Filter 2 (sometimes) has a very positive effect on the curve as a whole with the large bandwidth. But it seems that the four filter curve is sonically unstable in my room :holycow:

The 12 filter curve seems to give consistent results. However it requires the sub gain to be turned up from about 10 o'clock (unfiltered) to 3 o'clock, which is nearly maximum! I am concerned the 12 filter curve has too many large cuts and will have a negative effect on the available headroom and sound of the sub.

I would appreciate any thoughts that you may have, apart from the rude ones. My wife is certainly having a few thoughts about the amount this equalisation lark I expect that I need to try again and aim for fewer filters with a stable curve. One other question... is sweep the most accurate? The results seem to vary quite a bit for the different tone types, e.g. 1/6 oct versus sweep.

Thanks again

Al

[brucek]

I would appreciate any thoughts that you may have

I'm kinda lost. Your initial post indicates that when you measure twice with no changes, you get two completely different results.
Your last post isn't consistent with the inital post at all?

I have moved the sub around and still get that dip in my listening position, which cannot parctically be changed.

Then you are stuck with it. Trying to make massive cuts isn't the answer. You end up with no signal. The answer is to filter the peaks within reason and leave the rest alone. Keep any gain to a minimum.

is sweep the most accurate? The results seem to vary quite a bit for the different tone types, e.g. 1/6 oct versus sweep

Use the sweep. It's the most accurate. The next version of REW won't even include the manual or auto octave tests.

Could the inconsistency be due to filter 2, i.e. having a cut over a large bandwidth (40)? Filter 2 (sometimes) has a very positive effect on the curve as a whole with the large bandwidth.

You can have any filter with any size bandwidth you want. You can have them all at the same frequency if you want. No problem. There is no such thing as inconsistency. It either works or it doesn't because there's a fault or a error in the inputting of the filter.

brucek

Yes my initial post indicates that when I measure twice with no changes, I get two completely different results. For example, if I load up the four filter curve and run a sweep 10 times, without making any changes between the sweeps or even entering the room (pc is next door), about 3 out of 10 curves will appear unfiltered. The rest follow the target almost perfectly. I have no idea why this is so.

Sorry to confuse by adding the 12 filter curve to the discussion.

[brucek]

about 3 out of 10 curves will appear unfiltered

That just shouldn't occur. I could press the measure button on my system 10 times and you wouldn't be able to tell how many times I did it. It would look like one single line....

You've got a fault I think. Could even be an intermittent cable in the REW setup.

brucek