AudioCircle
Industry Circles => GR Research => Topic started by: THROWBACK on 20 Apr 2019, 11:07 pm
-
I have a pair of OB LS-9s buttressed by a pair of Rhythmik OB subs (3 stack per side), driven by A370 PEQ servo amps. The problem is that the amps have 17 different switch positions. Moreover, the effects are interactive. The versatility of this arrangement is both a boon and a curse.
Can someone point me to a step-by-step procedure along with an indication of how one is to know when the step is optimized? I will be using Stereophile Test CD-2 with 1/3 octave warble tones and a hand-held PAA3, hand held SPL meter (I could use either flat or C weighting).
E.G. (and I'm making this up)
1. set the 17 controls in the following positions to start (list)
2. attack phase first.
a. with SW LOW PASS switch set to 80/24, send 80Hz signal to both main and sub.
b. starting at zero, rotate DELAY/PHASE control until minimum indication on meter
3. etc.
As I said, I was making this up. But I have been fussing with this combo for a long time by meter and ear and, although my system sounds very good, I am still not sure I have it as right as my room will allow.
Any ideas? (BTW, I sent essentially this same message to Rhythmik this afternoon)
-
I was never able to get things really dialed in with an SPL meter. I could get close but that was it.
To get things dialed in better I got a mimiDSP U-Mik1 (~$80 at Parts Express) and downloaded the free REW software. This made dialing things in a lot easier and was the first time I was able to get phase set correctly. The graphs make the process more intelligible and systematic. You can see the effect your changes have. Something you can’t do with an SPL meter.
If you want to go this route, I’d be happy to describe the steps I follow.
Mike
-
Good idea, Mike. In fact, I have the mic and have downloaded REW, but have never used it. I'd love to get your steps to proper phase. I have a hunch that that is the first thing you should set. I'm further guessing that you should measure the phase of your main speakers then match the subs to that.
Thanks,
Chuck
-
Every component change, cable swap, placement and toe-in experiment, tweak, or even seasonal variation often requires messing with those settings. The settings can change with each recording. And if you're really anal, the settings can be a bit different for each amp if you dial them in separately as you should. At some point, you gotta set it and forget it. Regardless, it will still be some of the best bass you've ever heard, so find some solace in that.
I've found that the most important setting is phase. I rarely fool with the others.
-
Good idea, Mike. In fact, I have the mic and have downloaded REW, but have never used it. I'd love to get your steps to proper phase. I have a hunch that that is the first thing you should set. I'm further guessing that you should measure the phase of your main speakers then match the subs to that.
Thanks,
Chuck
Chuck,
I don’t set phase first. As Early B says, each speaker is done separately.
The first thing I do after determining a starting placement and toe in (by physical measurements for placement and toe in for center image) is find the crossover point. This is done one channel at a time. Turn the sub off and run a full range sweep of the LS9. You can run three or more and average them if you like. If there are any major dips you will need to move the LS9s or possibly add some diffraction. Once there are no major dips, look at the graph and find out where the LS9s are 6dB down.
Turn the amp to the LS9s off and turn the sub amp on. For the initial settings I use:
PEQ - off
Phase - 0
Crossover - pick a setting close to what you think it should be based on the previous step
Volume -12o’clock
Line in/ low pass - AVR/12
Rumble filter - Off
Extension Frequency - 20
Damping -High. These are just the starting points I use
Now run a sweep from 15Hz - 250Hz. If the base is significantly elevated or depressed compared to the LS9s, turn the volume up or down as needed then run another sweep. Once the average volume of the sub is close to the LS9s, determine where the sub is -6dB on the high end. You want the sub to be -6 at the same point the LS9s are -6. Adjust the crossover control in the direction you need to go and run another sweep. Continue making adjustments and running sweeps until the sub is -6 at the same frequency the LS9s are -6.
Now compare the average output of the sub with the average of the LS9s again. If the averages are not the same, adjust the sub’s volume control again until the averages are the same. Now repeat the crossover adjustment until the sub’s -6 point is the same as the LS9. Then Check the average volumes again. Keep going back and forth until both the sub and LS9 have the same average volume and -6 point. These adjustments are all with the sub. The LS9 has not been turned back on.
Next look at the sub’s graph and see if there are any peaks or valleys. If there are any valleys deal with these first as the only way to deal with them is to physically move the sub. If you have to move the sub to fix a valley, go back and recheck the average level and crossover point as necessary. If there is a peak, turn on the PEQ and start adjusting the frequency, width, and gain until you have tamed the peak.
Now turn the LS9 back on and run a full range sweep with both it and the sub playing. If the crossover point is smooth, you’re done with this speaker. If it is not smooth, start adjusting the phase on the sub until it is.
Now do all that again with the other channel. The settings on the two sub amps will likely be close but not the same.
This will get them dialed in relatively flat.
Then I turn both channels on and run a full range sweep to see what they look like together. The shape of the two channel graph is usually very close to the individual channel graphs, just 4-6dB higher.
Now listen to music for awhile then experiment with the 50/24 and 80/24 slope positions and the damping positions to determine which you like best. You can also experiment with the extension settings.
I hope all this makes sense.
Mike
-
Wow, Mike! Did you get any sleep last night or did you stay up all night writing your wonderfully detailed response? Thank you.
It's a lot to digest, but over the next week or so, I will follow your procedure exactly. I noticed something, though. EARLY B (thanks!) says that phase is the most important parameter; you, however, never even mention it except to say that you don't look at phase first.
Now it may be speaker dependent. The LS-9s consist of twelve 6-inch mids/woofers in a stack 80” tall. Thus, each element is at a different distance from the observer’s position. I haven’t done the math, but it seems to me, therefore, that phase between the tall array and the short, floor-standing sub can be set only approximately.
Still, my intuition tells me that should be done first.
Now if I can only find a local, REW-savvy, bribeable (lunch and/or a glass or two of good wine), patient, audio nut to help me . . .
-
Chuck,
REW is a very powerful and free tool so it in and of itself requires some study. There is a learning curve with it. And it does a whole lot more than room based measurements. Having an REW savvy individual will be a good thing, along with 1 or 2 others to help with moving speakers and subs. None of yours are light. Hopefully your room allows for different sub locations, that’s one of the biggest impediments I have seen with 2ch enthusiasts.
Best,
Anand.
-
Chuck,
I do mention phase. It comes into play when you turn the LS9s back on. For example, in my room after I determined the -6 point for the NX-Otica and finished adjusting the sub to match there was an issue. You would expect the two speakers to reinforce each other at the -6 crossover point and boost that point up to flat. However, when I played the Otica and sub together, there was a dip at the crossover point. They were no longer crossing at the frequency where the Oticas were -6 but where they were -10. They were partially cancelling each other out at the crossover frequency.
This meant the sub and Ottica were out of phase with each other. That’s when I adjusted the phase control until the dip at the crossover point was gone. Now the sub and Otica were in phase with each other.
As Anand says, REW is powerful. I’m the first to admit I’m no expert with REW. What I’ve described is about all I know how to do with it at this point.
Mike
-
Chuck,
Here are 2 threads, one from Mike (mlundy57) and the other from Ed (ebag4), where they go through what they did for sub integration:
https://www.audiocircle.com/index.php?topic=159295.msg1701189#msg1701189
https://www.audiocircle.com/index.php?topic=159409.msg1702157#msg1702157
Best,
Anand.
-
Anand (PM) asked me to post my results so far. I related to him an old Chinese proverb: "Man with one watch knows what time it is; man with two watches never sure."
I have a PDF of my results so far using "two watches." They are certainly different, but roughly comparable. Neither is terrible, but I can't figure out a way to post it here. One was taken using warble tones from Stereophile Test CD 2 as measured by my hand-held SPL meter (PAA3 Personal Audio Analyzer). The other made a couple of weeks ago by an audio buddy using pink noise and his HP Audio Analyzer.
But - - being a hobbyist and all - - I would like to see what could be done using REW, with which I have no experience. I'm looking for a local REW-savvy, bribeable (lunch and/or a glass or two of good wine, or . . . ), patient, audio nut to help me.
-
Chuck,
You can take a screen capture of the PDF and save it as a JPG in Paint or other picture editor. It copies it to the clipboard on a Windows machine.
Here is an example with Paint.
(https://www.audiocircle.com/image.php?id=193695)
-
Previous "two watches" results of integration effort.
(https://www.audiocircle.com/image.php?id=193697)
If this works, thank (blame?) Mike Lundy.
Chuck
-
It DID work! But blame HAL (sorry, Rich), not Mike.
-
Chuck,
Glad it worked.
Here is what REW will do for measurements. This can be done repetitively so you can see changes in response as you dial in the system.
This is just a measurement I did while learning the adjustments on the HX800-12OB servo amps at about 1 meter. Similar to the A370PEQ. This is just seeing what happens in the room with just the sub first.
There is a preferences tab where you setup your sound card and calibration data file for the mic system. It also lets you check the system with a level test to start out.
(https://www.audiocircle.com/image.php?id=193703)
-
Previous "two watches" results of integration effort.
(https://www.audiocircle.com/image.php?id=193697)
If this works, thank (blame?) Mike Lundy.
Chuck
This graph has me confused. It seems to be saying an ideal room response is 8dB less at 12kHz than at 25Hz. Extending the graph out it would be even farther down at 20k.
Am I misinterpreting the graph or am I missing something else?
Mike
-
Mike,
Look online and read about “Target Curves” from Toole et al. Several other acousticians, Geddes, etc..have commented on the same.
Best,
Anand.
-
Some info on House Curves, Target Curves, etc...
https://www.hometheatershack.com/forums/rew-forum/96-house-curve-what-why-you-need-how-do.html
Best,
Anand.
-
Good job, Anand. Great discussion of "house curve."
FY (and others')I, I had a golden-eared buddy over a couple of days ago to help fine tune my system. We listened to a bunch of music - - vinyl & SACD; classical and jazz. We found that the sound was so source-dependent, we ended up not changing a darned thing.
I'd still like to do an REW analysis though . . .
-
I haven’t digested the info on house curves yet, I’t’s all I can do to get ready for LSAF and my day job right now. I’ll spend more time reading up on this once things calm back down.
What I’ve read so far looks interesting but I’m not sure how to accomplish any of them. I can boost the bass easy enough but since top end is passive it seems like it is what it is.
Mike
-
Here is more:
http://www.aes.org/e-lib/browse.cfm?elib=17839
Of course if you already own Toole’s “Sound Reproduction...” book, the 3rd edition, you know all this stuff.
Strongly recommended.
Best,
Anand.
-
I haven’t digested the info on house curves yet, I’t’s all I can do to get ready for LSAF and my day job right now. I’ll spend more time reading up on this once things calm back down.
What I’ve read so far looks interesting but I’m not sure how to accomplish any of them. I can boost the bass easy enough but since top end is passive it seems like it is what it is.
Mike
Top end is intrinsic to the design of the speaker. It’s the reason constant and controlled directivity are important points of discussion as the designer of said speaker takes that into consideration.
Best,
Anand.
-
I believe most speakers roll off at the top at least a little bit just due to being in a room, unless the room is totally undamped. The problem is for digital systems run through SW programs like JRiver that allow equalization. The temptation is to make the system measure flat. If you do, in most cases it will sound way too bright.
-
FWIW, Here is a recent REW plot (1/24 octave smoothing, Cross Spectrum Calibrated USB Mic) of some OB subs mated to Magnepan 20.1 bass panels measured at 5 feet.
LP/HP XO point is at 40Hz 48db (black vertical line). Note the flat line to the left of the 40Hz line.
2 - 12" subs are in stacked cabinet configuration similar to the Orion 4's.
Both drivers are facing up, stacked on top of each other, but firing out of phase to average driver travel non-linearity, center driver weight and cancel cabinet resonances. There is one larger "port" in the front and 2 smaller "ports" in the back.
On other tests in this cabinet configuration with a higher XO point, it remains significantly flat to @ 55Hz which is good for my application, but not if you are wanting to drive them higher into the woofer range. Adding a notch filter past 55Hz extended its range a bit.
The top plot is done with a PEQ370 amp and the bottom plot is done with an ICEPower ASP1000 amp using a software Low Shelf Filter.
(https://i.imgur.com/oHnzEXI.jpg)
Settings on the PEQ370 are:
- Dampening: High
- Rumble Filter: ON
- Phase: 0
- AVR-12 - Using external XO
- Volume: minus 5 clicks from 12 O'Clock
- Bass Extension: 14Hz
- No PEQ
REW and multiple overlayed measurement plots was very helpful in determining the optimal PEQ370 settings for bass extension, phase matching, level matching and minimized distortion. Using an SPL meter and my ears would have resulted in less than optimal results.
Update: Image of cabinet configuration:
(https://www.hifizine.com/files/2012/03/Linkwitz-W-frame.png)
-
emailtim,
Thanks for contributing with your personal measurements. Nice extension to 15hz.
What would be your solution(s) for the area to the right of the 40Hz line, ie 40Hz to 250Hz or so?
Best,
Anand.
-
emailtim
(https://www.audiocircle.com/image.php?id=194028)
-
emailtim,
Thanks for contributing with your personal measurements. Nice extension to 15hz.
...
Best,
Anand.
Poseidonsvoice,
I was surprised to see how "flat" the sub response was in that cabinet but higher frequency range extension seems to be limited. I could get a few more Hz bass extension if I disabled the Rumble Filter, but REW indicated the distortion measurements went up accordingly so I left it off. Someone more knowledgeable of cabinet and driver characteristics could probably come up with more optimized cabinet dimensions than what I did. I measured it both in the garage when making the cabinet and then in the house. I have some 1.5" MDF laminated up and am ready to make some H-Frames out of them to test next. Have to make a router dado jig first. Wish I had access to a nice CNC.
... What would be your solution(s) for the area to the right of the 40Hz line, ie 40Hz to 250Hz or so? ...
I don't exactly understand your question. The area to the right is the actual response of both planar bass panels including the room response measured at 5 feet. This will change as you move the mic around the room. Smoothing is set at 1/24 and the plot is zoomed in so it is showing a lot of the warts. If I set the smoothing to 1/12, 1/6th, 1/3rd, ... the plots flatten out more. I have not tried to create any PEQs for the rest of the plot yet.
As for strategy, I would try to understand what are causing the specific anomalies (furniture, location, reflection, room mode, etc.). Fixing the cause is usually better than resorting to a PEQ because it usually fixes the anomaly in all locations of the room and not just the mic location. I have noticed you have to examine multiple smoothing settings in REW analysis because a given "smoothing plot setting" may give you a false impression that is an artifact of the smoothing algorithm else you will start chasing ghosts under the microscope.
I measured the drivers by themselves and together to see where the rolloffs, summing and cancellations of the XO were. This is also how I determined the best phase settings.
The XO is at 40 Hz and the slopes are symmetric at 48dB/octave. That means drivers would be down 48dB at 20Hz and 80Hz (off the bottom of this plot at 20Hz and 80Hz). It also means the phase of both drivers should match after going through both XO with the 48dB slopes (providing their starting phase is the same).
-
emailtim,
Thanks for contributing with your personal measurements. Nice extension to 15hz.
What would be your solution(s) for the area to the right of the 40Hz line, ie 40Hz to 250Hz or so?
Best,
Anand.
PoseidonsVoice,
FWIW, here is the same plot with 1/3rd octave smoothing versus 1/24th octave smoothing. I don't know if there is a "standard/customary" smoothing value to be used when posting plots here.
(https://i.imgur.com/3cMRNAV.jpg)
-
Hey Chuck,
I got your message on this, but also noticed the post you made here so I figured I'd let some others give their feedback first, and then I have been so busy getting ready for the show that I haven't gotten back to it yet.
Ideally you need to take some measurements to dial it in perfectly.
The number one error in measuring them is everyone tries to measure left and right speakers and subs at the same time. This just causes cancellation effects from different distances and time delays created by multiple sources. Just measure left speaker and left sub individually and then together to get the phase, crossover, and amplitude correct.
The damping settings will have an effect on imaging to some extent. Don't assume that high just means tighter or better control and thus better sound. You may get better sound on low damping if the extension filter is set low, and better imaging.
And don't get caught up in trying to measure distortion figures. Measuring or trying to take any distortion measurements outside of an anechoic chamber is futile. The actual distortion numbers will be at or below the room noise level. So true data will be covered over by the room noise levels.
I'd also like to comment on implementing the servo woofers. I tend to prefer the H or W frame design. I like the H frame best as it divides the side panels into smaller areas and maintains a larger opening. The W frame leaves a much larger panel area on the front side between the drivers and the side panels have to be increased to thicker material to compensate for the pressure created in that area.
I'm using 1.25" think material on my H frames and lined with No Rez to eliminate and control any panel flexing and resonances, and that is about what it takes.
Going to a Ripole design like the one pictured unnecessarily loads the drivers and causes the amp to use more power to maintain the linear output levels. So it eats into amp headroom. It also will limit the upper range quite a bit. But more importantly is that it allows for very large panel areas in front of each driver. To do away with panel flexing and resonances that this alignment will cause will require panel thickness in the 2.5" to 3" thick range. Most of the designs I have heard using this alignment and 3/4" plywood is what I'd call a buzz box. So I don't recommend it.
(https://www.hifizine.com/files/2012/03/Linkwitz-W-frame.png)
-
...
Going to a Ripole design like the one pictured unnecessarily loads the drivers and causes the amp to use more power to maintain the linear output levels. So it eats into amp headroom. It also will limit the upper range quite a bit. But more importantly is that it allows for very large panel areas in front of each driver. To do away with panel flexing and resonances that this alignment will cause will require panel thickness in the 2.5" to 3" thick range. Most of the designs I have heard using this alignment and 3/4" plywood is what I'd call a buzz box. So I don't recommend it.
(https://www.hifizine.com/files/2012/03/Linkwitz-W-frame.png)
Hi Danny,
FWIW, that is not a Ripole cabinet. I believe it is called a "Linkwitz W-frame" according to the Subwoofer Origami page with drivers firing out of phase to null cabinet resonances/jumping (force cancellation) which an H-Frame cabinet does not do while averaging out the forward/backward driver movement characteristics and centers the driver weight to the middle of the cabinet. https://www.hifizine.com/2012/12/subwoofer-origami/ (https://www.hifizine.com/2012/12/subwoofer-origami/)
(https://www.hifizine.com/files/2012/03/Linkwitz-W-frame.png)
This is supposed to be a RIPOLE (matches patent) with facing drivers firing in phase much closer together and much smaller cabinet volumes (loading the drivers as you mentioned) (motors/magnets protruding out the sides).
(https://www.hifizine.com/files/2012/03/Ripole.png)
The Subwoofer Origami pages calls this an M-Frame (push-pull) which you refer to as a W-frame.
(https://www.hifizine.com/files/2012/12/M-frame-pp.png)
Is there a definitive reference on the cabinet naming conversions to avoid confusion ? We know what a Ripole is because it was named after the inventor (Axel Ridtahler's DiPole) in his patent but there seems to be some ambiguity with the W and M frame naming conventions.
-
Ah, thanks for the correction of terminology.
The issues that I pointed out though still remain. There are still significant pressures in each area that flex and excite each panel.
-
Ah, thanks for the correction of terminology.
The issues that I pointed out though still remain. There are still significant pressures in each area that flex and excite each panel.
I would think the internal pressures of the Linkwitz W-Frame are less than a RiPole because the cavities and ports are much larger.
What are your impression of my first 1/24th octave plot with respect to the subs flat frequency response. Is that typical for a H-Frame cabinet, or is it more of a hump. Note, it has no EQ other than the Hardware/Software Low Shelf filters and the XO point.
(https://i.imgur.com/oHnzEXI.jpg)
P.S.
Also, why would you ignore distortion measurements ? Isn't distortion the reason for using a Servo (to eliminate it) ?
-
I would think the internal pressures of the Linkwitz W-Frame are less than a RiPole because the cavities and ports are much larger.
It is less than the pressures of the Ripole, but still very extreme compared to an H frame.
What are your impression of my first 1/24th octave plot with respect to the subs flat frequency response. Is that typical for a H-Frame cabinet, or is it more of a hump. Note, it has no EQ other than the Hardware/Software Low Shelf filters and the XO point.
It looks good. I am surprised it isn't humped up more from the loading.
Also, why would you ignore distortion measurements ? Isn't distortion the reason for using a Servo (to eliminate it) ?
Distortion measurements outside of an anechoic chamber are impossible to get any meaningful data from. Even the manual for my Clio measuring system states such. Since it is impossible to separate from the noise floor it has little value. And to be honest even taken correctly in a chamber the data is just a drop in the bucket compared to settling time and panel resonances. Those are 100 times greater and more significant. When you are laying power to these things and really moving the drivers pretty hard then those issues get real. And the servo system really speeds up settling times. Cabinet resonances have to be addressed in other ways.
-
It is less than the pressures of the Ripole, but still very extreme compared to an H frame.
It looks good. I am surprised it isn't humped up more from the loading.
Distortion measurements outside of an anechoic chamber are impossible to get any meaningful data from. Even the manual for my Clio measuring system states such. Since it is impossible to separate from the noise floor it has little value. And to be honest even taken correctly in a chamber the data is just a drop in the bucket compared to settling time and panel resonances. Those are 100 times greater and more significant. When you are laying power to these things and really moving the drivers pretty hard then those issues get real. And the servo system really speeds up settling times. Cabinet resonances have to be addressed in other ways.
Speaking of settling time - is that why the Neo10 panels sound so good? Because they are driven by magnets from the front and rear simultaneously and thus have a much faster return (settle time) vs. a cone/spider setup?
-
Danny -- for those of us who already have our servo subs built with only 3/4" MDF and would like to address the cabinet resonance issue, would it be helpful to add "decorative" wood or stone panels to the sides of our cabinets?
-
Danny -- for those of us who already have our servo subs built with only 3/4" MDF and would like to address the cabinet resonance issue, would it be helpful to add "decorative" wood or stone panels to the sides of our cabinets?
It sure would.
-
Thanks, Danny
Ideally you need to take some measurements to dial it in perfectly.
I'll be taking REW measurements shortly
The number one error in measuring them is everyone tries to measure left and right speakers and subs at the same time. This just causes cancellation effects from different distances and time delays created by multiple sources. Just measure left speaker and left sub individually and then together to get the phase, crossover, and amplitude correct.
That's the sequence I was looking for: 1. phase; 2. crossover; 3. amplitude. Otherwise it's easy to chase your tail.
The damping settings will have an effect on imaging to some extent. Don't assume that high just means tighter or better control and thus better sound. You may get better sound on low damping if the extension filter is set low, and better imaging.
You're right: I had assumed that. I'll try messing around with damping.
And don't get caught up in trying to measure distortion figures. Measuring or trying to take any distortion measurements outside of an anechoic chamber is futile. The actual distortion numbers will be at or below the room noise level. So true data will be covered over by the room noise levels.
OK.
Regards,
Chuck