CREATING AND ENTERING FILTERS
roly,
This is probably a good time to discuss using REW to create recommended filters and then entering them into the BFD and then re-measuring to check the results. I'm assuming that the filters will be entered into the BFD manually from its front panel. The BFD manual can be used to describe how to enter filters.
So at this point, you have taken a raw measurement with no BFD filters and saved it as a *.mdat file (using the pull down FILE / Save Measured Data Set). You are able to load that mdat file at any time in the future to work on your filters, but you may as well do some filters now when everything is hooked up and enter them into the BFD so you can test how they work.
Something you have to realize about REW is that it won't automatically create any suggestions for filters for dips that require gain. It simply suggests filters that will reduce the peaks down to the target line. This is smart since it has no idea if a filter with gain will work in a system, due to the fact that some dips in response are a result of a room resonance that won't budge no matter how much gain you throw at it.
In that case all you're accomplishing is to reduce your available headroom. So filters with gain are trial and error. If you have a dip, you can use REW manually to produce the best suggested filter with gain as if it would indeed respond. Then you enter that filter and do another response measurement and see the effect. Some dips simply respond exactly as REW predicts and some dips are very stubborn. Don't waste your time on them.
This part of the procedure requires some creativity on your part. REW will indeed automatically suggest some filters for you to enter (after pushing a few buttons) and shows on the graph the likely result after entering them. The predictions are remarkably accurate. If you just followed the advice of REW for filters you would remove all the worst peaks and your system would sound better. This just doesn't satisfy most people though. The fact is, after you have pushed the buttons to get the filters that REW recommends, you should use REW in the manual mode and play with the filters by adding your own and modifying the suggested ones until you get a smooth response predicted on the REW display. Then enter those filters into the BFD and see if the response is as REW predicts, and then tweak a bit to get it perfect.
Let me go through the procedure and show a few pictures.
Once you've taken the raw measurement and have saved that mdat file, it's time to let REW suggest filters.
I'll use my own system in the pictures. Here's my raw measurement and my target. My target is 80dB, but I have added a house curve to the target and have a crossover at 60Hz, so my display will look somewhat different than yours.
1. FIND PEAKS
Press this button and REW will find the peaks within the limits of the find peaks default entry of 20Hz to 200Hz. I usually set this first to 15Hz to 200Hz. Remember a BFD can only enter filters down to 20Hz, but I get better suggestions from REW with a lower limit of 15Hz.
REW will find the peaks and display them in a persistant chart in the left hand side of the graph. You can see by the little chart that the frequencies are at the peaks. I suggest
every time this chart pops up to remove it with a couple Cntrl-Shift-P's.
Here's my own systems raw measurement and the associated FIND PEAKS chart. It shows, for example that there is a 10.2dB peak at 34.5Hz.
2. ASSIGN FILTERS
Now press this button and REW will assign appropriate filters to attempt to remove the peaks it found. The suggested filters are displayed in the bottom right hand corner of the REW page. The green entries are the actual BFD front panel entry for the frequencies.
Below are the suggested filters:
3. OPTIMISE PK GAIN AND Q
Press this button to automatically adjust the gain and Q settings of the suggested filters to obtain the best match to the shape of the peaks and so create the flattest corrected response. The corrected predicted response is now displayed along with the graph of the filters. The display graph gets very busy now.
You can remove all the extra stuff on the graph now to get a better picture of what the predicted corrected response will be. Turn off Sub Measured, Filters and Filters+Target to reveal only the target and the SUB CORRECTED response. At this point, if you simply entered those suggested filters, your response would look fairly close to the predicted response that REW now shows.
Below is the predicted response if I used the filters suggested. See the filter numbers along the top of the graph - kinda handy.
GET CREATIVE
Here's where you play with the filters manually by adding your own and also modifying the suggested ones to arrive at the best and flatest response (without getting too crazy). I ended up with quite a few extra filters as you can see by the diagram below. When you modify and add filters be sure to press the
ADJUST PK GAINS after you do so......Play around with the parameters of each filter and see what it effect is on the response - REW dynamically shows you the results.
Below is the actual response (and very close to the predicted) that I achieved by the new and improved filters shown above. The predicted and actual were very close.
Once you like the look of the predicted graph it's time to enter the filters into your BFD and then do another REW response check to see if the actual response is as good as the predicted response. Where it will possibly differ is in the areas where a filter has gain.
Read the next line carefully because it's quite important!
When you do the new response sweep after the filters are entered into the BFD, you must temporarily clear the filters out of REW so they don't affect the new measurement. Just uncheck the boxes as shown below or the filters shown in REW will be applied to the new sweep measurement.
If you don't like the new response you obtained with the filters that you've entered into the BFD, you can reload your raw mdat file and then uncheck the filters and play some more. It can be several iterations (or not) before you're satisfied with the response. (You can see why the midi cable would be handy at this point). I don't have one myself, but I get why it's nice to have...
brucek