Feedback Destroyer
- Thread starteravkid
- Start date
jkowtko
Well-Known Member
This is an old thread, but I'll continue the discussion as I need some quick feedback (no pun intended) on a couple of units --
My problem is that the center cluster is several feet behind the front lip of the stage (quarter-round flex theater with stage on floor) and we have not been able to move the speakers forward due to lack of gridwork above where the speakers really need to go (and lack of time to deal with these obstacles)
So, for the time being, I just pull down the volume of the center cluster on the wireless lav input channels (I'm using one of the aux outs for this) by about 10db or so. This allows the actors to go out to the front edge of the stage without feedback, and I can still keep the front wing speakers at full volume.
However I'd like to get that center cluster back up to full volume. I can throw a 1/3 octave EQ on the center cluster, but I'd rather use a 1/60 or 1/80 octave notch filtering feedback suppressor if it will work properly.
Does anyone have experience/comments with the Behringer DSP1100P or DSP1124P, or the DBX AFS224 or Driverack PA? Specific questions:
1) When running live filtering, do they apply the right amount of filtering, or too much?
2) Aside from the filtering, do the units introduce any noticable audio quality degradation or noise?
3) Does the balanced IO work well (since I will be plugging this in-line between the board and the self-powered speaker, I do not have an insert available on the board)?
Thanks. John
My problem is that the center cluster is several feet behind the front lip of the stage (quarter-round flex theater with stage on floor) and we have not been able to move the speakers forward due to lack of gridwork above where the speakers really need to go (and lack of time to deal with these obstacles)
So, for the time being, I just pull down the volume of the center cluster on the wireless lav input channels (I'm using one of the aux outs for this) by about 10db or so. This allows the actors to go out to the front edge of the stage without feedback, and I can still keep the front wing speakers at full volume.
However I'd like to get that center cluster back up to full volume. I can throw a 1/3 octave EQ on the center cluster, but I'd rather use a 1/60 or 1/80 octave notch filtering feedback suppressor if it will work properly.
Does anyone have experience/comments with the Behringer DSP1100P or DSP1124P, or the DBX AFS224 or Driverack PA? Specific questions:
1) When running live filtering, do they apply the right amount of filtering, or too much?
2) Aside from the filtering, do the units introduce any noticable audio quality degradation or noise?
3) Does the balanced IO work well (since I will be plugging this in-line between the board and the self-powered speaker, I do not have an insert available on the board)?
Thanks. John
If the problem is that the mic is in the speaker coverage then a feedback suppressor or notch filter probably isn't going to help that much. You are not dealing with narrow band room modes or resonances, you are likely dealing with a broader frequency range from the speaker getting directly into the mic(s). If so, you will likely end up needing to address a broader frequency range rather than narrow notches in order to really get much gain.
If you run the feedback suppressors with all filters 'live' or dynamic, then you will likely end up with them constantly cycling through, which can sound bad. If you do use a feedback suppressor you might want to try to 'fix' several of the filters and not go any narrower than 1/20 or 1/40 octave. However, a smaller, broader cut may actually work better in this situation.
If it is just one or two mics involved then you may want to think about addressing only those channels rather than the entire mix. You could simply use the relevant channel EQ on the mixing console to apply EQ for the mics when needed, just work out ahead of time what EQ needs to be applied when they walk downstage and apply it live. Or you could insert a feedback suppressor or EQ on just those inputs, Sabine makes some units specifically for this application.
The best bet if at all practical is closer micing. Halving the distance to the mic gives you 6dB more gain, which is why headset mics that are very close to the mouth can allow for so much more system gain.
If you run the feedback suppressors with all filters 'live' or dynamic, then you will likely end up with them constantly cycling through, which can sound bad. If you do use a feedback suppressor you might want to try to 'fix' several of the filters and not go any narrower than 1/20 or 1/40 octave. However, a smaller, broader cut may actually work better in this situation.
If it is just one or two mics involved then you may want to think about addressing only those channels rather than the entire mix. You could simply use the relevant channel EQ on the mixing console to apply EQ for the mics when needed, just work out ahead of time what EQ needs to be applied when they walk downstage and apply it live. Or you could insert a feedback suppressor or EQ on just those inputs, Sabine makes some units specifically for this application.
The best bet if at all practical is closer micing. Halving the distance to the mic gives you 6dB more gain, which is why headset mics that are very close to the mouth can allow for so much more system gain.
jkowtko
Well-Known Member
Brad,
The actors go to just behind the edge of the "published" sound field of these speakers. So they're not picking up full volume sound, maybe -10db or so coming from the side.
Right now I can more-or-less hear the feedback frequency ... it's been the same for the run of this show. So I figure it I can notch it out, and maybe a few others, the remaining frequency spectrum will still be below the feedback threshold and I'll be able to bring the volume back up a bit on the center speakers -- if I'm lucky, all the way to where I have the wings.
All of my lavs are Countryman B3s, very good smooth response and pretty feedback resistant, and I haven't detected any one unit is worse than the others. So for now I would just plan to filter them as a group (in fact I can't separate the lavs from the other input channels, board's too small, so unfortunately for now I'll have to filter the speaker itself.
Yes, I would plan to ring the house a bit to set a few fixed filters, then monitor during the first two shows to see what else pops up. If the suppressor applies more than a few filters during a show I would be inclined to shut it off for the remainder of the show and just lower the speaker volume back down.
If you've used feedback suppressors, other than the filtering, were you able to notice any degradation in sound quality?
Thanks. John
The actors go to just behind the edge of the "published" sound field of these speakers. So they're not picking up full volume sound, maybe -10db or so coming from the side.
Right now I can more-or-less hear the feedback frequency ... it's been the same for the run of this show. So I figure it I can notch it out, and maybe a few others, the remaining frequency spectrum will still be below the feedback threshold and I'll be able to bring the volume back up a bit on the center speakers -- if I'm lucky, all the way to where I have the wings.
All of my lavs are Countryman B3s, very good smooth response and pretty feedback resistant, and I haven't detected any one unit is worse than the others. So for now I would just plan to filter them as a group (in fact I can't separate the lavs from the other input channels, board's too small, so unfortunately for now I'll have to filter the speaker itself.
Yes, I would plan to ring the house a bit to set a few fixed filters, then monitor during the first two shows to see what else pops up. If the suppressor applies more than a few filters during a show I would be inclined to shut it off for the remainder of the show and just lower the speaker volume back down.
If you've used feedback suppressors, other than the filtering, were you able to notice any degradation in sound quality?
Thanks. John
jkowtko
Well-Known Member
The published nominal "coverage" simply defines the -6dB point for the overall response, which while it has some use for assessing whether a speaker can cover an area, does not provide a lot of information. The sound from a speaker does not stop at the defined nominal coverage angle, in some cases it may actually increase further off axis for some frequencies. The coverage of a speaker also typically varies dramatically over frequency, for example below crossover you may have very limited pattern control so that while the overall level off axis is down considerably, some frequencies may be down little if any in level. You really need to look at the actual polars or response balloons for a speaker to get a better idea of what really happens, especially at different frequencies.
Notching offending frequencies will help with GBF, but it is not uncommon is to notch out the worst offending frequency and then find others just a few dB below that, much like the old peeling an onion analogy. 10dB is a significant amount of additional gain, especially when the problem may caused by the speaker levels on the stage and there are multiple mics involved, which is why I am concerned about using a feedback suppressor. In order to get 10dB of additional gain you will likely find yourself needing to address multiple frequencies and/or a wider frequency range, while at the same time some frequencies may require much less than 20 to 40 dB in reduction a feedback suppressor would typically apply. For example, you may be able to put a small dip in the response over frequency range rather than having a bunch of very deep notches all assigned in that same range and also end up with a more stable system.
Especially as you feel the problem is the same every time, a multi-band parametic EQ may be a better choice than a feedback suppressor as it allows you a lot more options in the settings possible. However, it does require a little more work or experience since you are making manual settings and have several parameters to adjust. Some of the devices you mentioned include multiple bands of parametic equalization and I would probably try to get the gain desired using this fixed parametric equalization capability and then use active feedback suppression only to address minor changes that may occur as the actors move around, if at all.
The sound quality, as always, depends on the device. It is another box in the signal chain and thus enters into the resulting gain structure. Very steep, very deep notches may cause some ringing that is audible in some applications.
Here's a follow up.
We got two Sabine FBX 901's, and I really don't like them.
They are very loud when inserted into the signal chain, and because of that I hardly ever use them.
Learning how to better utilize the parametric equalizers on your channel strips is a much better way to curb feedback than using an outboard device.
We got two Sabine FBX 901's, and I really don't like them.
They are very loud when inserted into the signal chain, and because of that I hardly ever use them.
Learning how to better utilize the parametric equalizers on your channel strips is a much better way to curb feedback than using an outboard device.
This is like the joke of the boater who complains this his bilge pump will not keep up with the leak in his hull. The solution is not to get a bigger bilge pump it is to fix the leak. You need to either move the cluster, or make that section of the stage off limits. You are going to find that you keep reducing frequencies, that by the time you have finished you have effectively reduced the volume of the center cluster.
Sharyn
Sharyn
jkowtko
Well-Known Member
Okay, I got a cheap Behringer DSP1124P and tried it out on three shows this weekend. And ... it worked ... reasonably well, actually 
I connected the unit in-line from the board to the (self-powered) center cluster. Phil, I also noticed the noise when I first connected the unit, but then I changed the "nominal operating level" from +4 to -10db and the noise dropped out with no change in audio level. The Behringer units are probably nowhere near as quiet as DBX, but this adjustment worked for me. So if you haven't tried this already, it's possible this could solve your noise problem.
I was able to ring out the feedback on individual actors pretty easily, to the point where they could stand at the front lip of the stage and I had their mikes up +20db louder than I ever had before, with their voices just screaming through the house.
Dealing with lots of mics on stage and a band blaring in the background was much more difficult for this unit, so I found myself having to pull down the center output gain -3db or so just as a precaution. Since I was also getting a slight amount of reverb during normal dialogue, I left the center cluster gain down for all except for the songs.
So, for this unit I found myself having to babysit it a lot. It cut a lot of the feedback, but without adjustment could not deal with it all.
Now the one thing I do like about the Behringer is that it has great adjustment capability. Using the numeric display I can review and adjust the frequency, Q and gain values for any of the filters it sets, and I can selective set and freeze any of the filters of my choosing. So the unit also effectively gives you 12 fully parameter EQs per channel. (what I don't know is how good the EQ is, though). So on my second pass at using it I did a ring-out, checked the values of the 4-5 filters it set, and then widened them up a bit. Fyi the filters it did set generally had a 1/60 bandwidth, but a couple of them had 2/60, so there is some extra logic in there to widen the notch if needed. The gain it chose was anywhere from -6 to -12db. Also with the Behringer you can specify the sensitivity of the feedback detector, so it's a fairly tunable unit.
In comparison, the DBX does not have a numeric readout and therefore no manual filter adjust capability, so although the white paper describes what appears to be a very good algorithm for detecting feedback and applying filters intelligently, using the DBX would have to be a "set and forget", or "leap of faith" operation. My next goal is to try out the DBX and see if it does a better job.
I agree with everyone about the 1/3 EQ being a reasonable alternative here -- 1/3 octave is obviously better than dropping the speaker volume (what's that, an 8 octave EQ?) I do have a DBX 31-band EQ and can run it in-line with the feedback suppressor, so over time I should be able to identify the problem frequency areas and make adjustments to allow me to run full volume without issues.
Sharyne, yes the ultimate fix here is to remove the problem (i..e move the speakers downstage), but what I'm also realizing is that with a flex theater (which ours is) the stage ends where the audience begins so I also need to make sure the PA covers the front rows of seats. So unfortunately it's a balance between sound coverage and feedback avoidance, and I think there will always be a grey area in there where I have to walk on eggshells. As an engineer I always welcome a good challenge, and this one is no exception. So between standard EQ and some notch filtering I'm hoping to have this problem licked by the time our next show starts in September.
I connected the unit in-line from the board to the (self-powered) center cluster. Phil, I also noticed the noise when I first connected the unit, but then I changed the "nominal operating level" from +4 to -10db and the noise dropped out with no change in audio level
. The Behringer units are probably nowhere near as quiet as DBX, but this adjustment worked for me. So if you haven't tried this already, it's possible this could solve your noise problem.I was able to ring out the feedback on individual actors pretty easily, to the point where they could stand at the front lip of the stage and I had their mikes up +20db louder than I ever had before, with their voices just screaming through the house.
Dealing with lots of mics on stage and a band blaring in the background was much more difficult for this unit, so I found myself having to pull down the center output gain -3db or so just as a precaution. Since I was also getting a slight amount of reverb during normal dialogue, I left the center cluster gain down for all except for the songs.
So, for this unit I found myself having to babysit it a lot. It cut a lot of the feedback, but without adjustment could not deal with it all.
Now the one thing I do like about the Behringer is that it has great adjustment capability. Using the numeric display I can review and adjust the frequency, Q and gain values for any of the filters it sets, and I can selective set and freeze any of the filters of my choosing. So the unit also effectively gives you 12 fully parameter EQs per channel. (what I don't know is how good the EQ is, though). So on my second pass at using it I did a ring-out, checked the values of the 4-5 filters it set, and then widened them up a bit. Fyi the filters it did set generally had a 1/60 bandwidth, but a couple of them had 2/60, so there is some extra logic in there to widen the notch if needed. The gain it chose was anywhere from -6 to -12db. Also with the Behringer you can specify the sensitivity of the feedback detector, so it's a fairly tunable unit.
In comparison, the DBX does not have a numeric readout and therefore no manual filter adjust capability, so although the white paper describes what appears to be a very good algorithm for detecting feedback and applying filters intelligently, using the DBX would have to be a "set and forget", or "leap of faith" operation. My next goal is to try out the DBX and see if it does a better job.
I agree with everyone about the 1/3 EQ being a reasonable alternative here -- 1/3 octave is obviously better than dropping the speaker volume (what's that, an 8 octave EQ?) I do have a DBX 31-band EQ and can run it in-line with the feedback suppressor, so over time I should be able to identify the problem frequency areas and make adjustments to allow me to run full volume without issues.
Sharyne, yes the ultimate fix here is to remove the problem (i..e move the speakers downstage), but what I'm also realizing is that with a flex theater (which ours is) the stage ends where the audience begins so I also need to make sure the PA covers the front rows of seats. So unfortunately it's a balance between sound coverage and feedback avoidance, and I think there will always be a grey area in there where I have to walk on eggshells. As an engineer I always welcome a good challenge, and this one is no exception. So between standard EQ and some notch filtering I'm hoping to have this problem licked by the time our next show starts in September.
The problem also seem to be if I understand you correctly the center cluster is not lined up with the side speakers. to properly set this up you also need to have time aligned the cluster and sides so that the signal all appears to have come from the same plane.
When you use a eq to reduce feedback in a monitor situation it is one thing, and in most cases the monitors are positioned behind the mics. In front of house feedback reduction should be first resolved with placement, and then the eq is being used only to reduce frequencies that are being emphasized by the speaker system. In your case I would be concerned that you now have higher GBF but with a simple sound check it is difficult to determine the quality of the audio that this cluster now provides.
Using the center cluster to cover the front rows in what I think is your setup also is not a good idea, because you are likely to get complaints about higher sound levels than the audience wants, the sound from the distance cluster and the actors on stage will cause sound problems . Typically in your sort of situation small speakers are arranged across the front of the stage, and the level controlled such that the audience in the front rows gets the correct level but is not blasted out
I guess it is like the idea of pointing the speakers away from the audience to reflect on lanai panels, it "works" but you quickly learn that the quality of the experience is not what you are looking for.
I've used Sabine 901's they are a quick fix, are effective to a certain extent in a situation where you have little time or control, but I would ONLY use them on monitors.
Again everyone has their own theory and practice on designing systems , so...
Sharyn
When you use a eq to reduce feedback in a monitor situation it is one thing, and in most cases the monitors are positioned behind the mics. In front of house feedback reduction should be first resolved with placement, and then the eq is being used only to reduce frequencies that are being emphasized by the speaker system. In your case I would be concerned that you now have higher GBF but with a simple sound check it is difficult to determine the quality of the audio that this cluster now provides.
Using the center cluster to cover the front rows in what I think is your setup also is not a good idea, because you are likely to get complaints about higher sound levels than the audience wants, the sound from the distance cluster and the actors on stage will cause sound problems . Typically in your sort of situation small speakers are arranged across the front of the stage, and the level controlled such that the audience in the front rows gets the correct level but is not blasted out
I guess it is like the idea of pointing the speakers away from the audience to reflect on lanai panels, it "works" but you quickly learn that the quality of the experience is not what you are looking for.
I've used Sabine 901's they are a quick fix, are effective to a certain extent in a situation where you have little time or control, but I would ONLY use them on monitors.
Again everyone has their own theory and practice on designing systems , so...
Sharyn
Again, though, the problem has a more "correct" solution, namely properly selected and implemented frontfills or downfills. It's a matter of tight pattern control in the right spot.
I'm not quite sure what you mean by flex theatre, but if it's what I'm thinking, at least in part (the first row of the audience is seated on the same level as the stage?) you want to try towards downfills, which are harder to do tightly, but can be done. Remember, in that case, that you don't need to hit the audience's lap on-axis, just their ears!
Pattern control is very often overlooked in sound design, and it really can't be if your design is to be effective
jkowtko
Well-Known Member
As I'm getting my head above water on the million things that need to be done at this theater in sound, I'm starting to get a handle on the sound dispersion requirements.
The Mackie SR450s that I have hanging have only a 45 degree vertical dispersion. From their position about 12 feet up, the LF/RF speakers are just in front of the front row, which means a single 45 degree sweep won't cover the first and the last row. They are already tilted down a bit, but I'd have to tilt them a lot more and I still don't think the 45 degrees will cover front to back audience. Or, I can turn the speakers sideways to get the 90 degree side dispersion of these speakers.
Or ...
I spoke to one of the design guys at Meyer Sound (our lighting guys' son is the software engineer on Galileo product there) and they said they will come out to do a design review on our site Their UPM-1P loudspeaker has a 100x100 degree dispersion range, is much smaller than the Mackie and can easily be hung horizontally under the lighting grid, making it excellent for our theater application. One of these speakers should be able to reach front to back including downfill needs, and three of them could handle the entire quarter-round sweep of the audience seating area. Unfortunately Meyer products are extremely pricey compared to Mackie, but if we can get the money somewhere to fund this it would likely be worth it. Anyway, Meyer is about a half hour down the road so it's not a huge deal for them to come pay us a visit.
In the meantime, yes, I'll have to struggle with the dividing line between sound dispersion limits and rolloff, to try to cover that first row of seats but not the actors standing downstage ...
Back on the feedback suppressor front ....
I found a cheap used Behringer DSP1124p for $40, and I have it running passively to see what sort of feedback it picks up. So far a mix ... with single mic ringing exercises it came up with a few 1/60octave notch filters in the 600Hz-1kHz range. However when running it with 10 mics turned up and music blaring, I think it loses it's touch a bit (too deep, too wide, but still only 3/60 or 4/60). Still, this unit is great in that it will adjust width and depth of the notch filter as it applies them, you can set the sensitivity of the feedback control, and you can view the specs on every filter it applies -- frequency, width and depth -- and freeze them or set them manually and store in a separate program. So for me this is a great little tool to understand where the problems are occurring. I can then either set up some of these filters in parametric mode on the Behringer and leave them on, or I can use the DBX 231 to pull down those frequency areas a bit.
The Mackie SR450s that I have hanging have only a 45 degree vertical dispersion. From their position about 12 feet up, the LF/RF speakers are just in front of the front row, which means a single 45 degree sweep won't cover the first and the last row. They are already tilted down a bit, but I'd have to tilt them a lot more and I still don't think the 45 degrees will cover front to back audience. Or, I can turn the speakers sideways to get the 90 degree side dispersion of these speakers.
Or ...
I spoke to one of the design guys at Meyer Sound (our lighting guys' son is the software engineer on Galileo product there) and they said they will come out to do a design review on our site
Their UPM-1P loudspeaker has a 100x100 degree dispersion range, is much smaller than the Mackie and can easily be hung horizontally under the lighting grid, making it excellent for our theater application. One of these speakers should be able to reach front to back including downfill needs, and three of them could handle the entire quarter-round sweep of the audience seating area. Unfortunately Meyer products are extremely pricey compared to Mackie, but if we can get the money somewhere to fund this it would likely be worth it. Anyway, Meyer is about a half hour down the road so it's not a huge deal for them to come pay us a visit.In the meantime, yes, I'll have to struggle with the dividing line between sound dispersion limits and rolloff, to try to cover that first row of seats but not the actors standing downstage ...
Back on the feedback suppressor front ....
I found a cheap used Behringer DSP1124p for $40, and I have it running passively to see what sort of feedback it picks up. So far a mix ... with single mic ringing exercises it came up with a few 1/60octave notch filters in the 600Hz-1kHz range. However when running it with 10 mics turned up and music blaring, I think it loses it's touch a bit (too deep, too wide, but still only 3/60 or 4/60). Still, this unit is great in that it will adjust width and depth of the notch filter as it applies them, you can set the sensitivity of the feedback control, and you can view the specs on every filter it applies -- frequency, width and depth -- and freeze them or set them manually and store in a separate program. So for me this is a great little tool to understand where the problems are occurring. I can then either set up some of these filters in parametric mode on the Behringer and leave them on, or I can use the DBX 231 to pull down those frequency areas a bit.
