Love Is Always Better the Second Time Around: The Sanders Sound Systems 10e Hybrid Electrostatic Speaker

[sleepysurf]

Cleaning the Sanders 10e

... You don't want to use a vacuum cleaner on an electrostatic panel since this will certainly unduly stress the thin film membrane. You definitely don't want the membrane to touch the stators as it might if the vacuum has enough suction...

Actually, if Sanders electrostatic panels use a similarly durable membrane like MartinLogan, it's probably safe to gently vacuum them. I've owned three different ML speakers over the past 25 years, and vacuumed the panels every 2-3 months, with no degradation in fidelity. In fact, ML even has a page on their website addressing this... https://www.martinlogan.com/en/support/faqs/q/197

I suggest calling Roger and asking him if gentle vacuuming is OK.

 

[atmer]

Sound Lab does indicate the panels can left plugged in 24/7. When not in use my own Sound Lab speakers are turned off via an inline switch on the AC cord. The panels charge fully in less than a minute according to Roger West so there's no penalty for turning them off.

Most of the SL models have no issue with dust as the wood frame with panel attached is connected directly to the metal back plate that is grounded so any charge that could accumulate is bled off through that pathway.

It might be worth pursuing a means of grounding the woofer box to AC ground, to dissipate the charge as the wood is somewhat conductive.

 

[tmallin]

Experiments With Equalization and Crossover Settings

Why Turn Up the Bass?

Frequency measurement graphs of speakers made in anechoic rooms ideally should look flat. Such rooms do not have "room gain" in the lower frequencies.

However, in real-world listening rooms, playing most commercial recordings, most people prefer the sound of speakers which measure a bit up in the bass frequencies and a bit rolled off in the highs. Thus, we talk about equalizing the speaker frequency response to one's preferred "target curve."

Ordinary rooms tend to boost the low bass response a few dB ("room gain"), moreso the closer to room walls and corners you place the speakers. Speaker manufacturers know about this effect, so that is one reason they tend not to engineer speakers to have elevated bass response when measured in an anechoic room.

The bass frequency room gain of most ordinary listening rooms helps listeners achieve a bass target curve which is up a bit in the bass without too much, if any, electronic bass boost from an equalizer as long as the speaker has been engineered to measure flat in an anechoic room. However, some speaker manufacturers want their speakers to measure flat-ish in the bass in real listening rooms. Thus, such speakers tend to need more electronic bass boost to sound natural in real-world listening rooms. You can tell such speakers from the NRC and Stereophile measurements. In NRC graphs, the bass response will be a bit lower than flat. In Stereophile's real-room-but-extremely-near-field bass measurements, such speakers will look like they have flat bass response, whereas speakers which have anechoically flat response (e.g., some BBC-influenced designs like the Stirling LS3/6) will show a bit of excess bass in Stereophile's graphs. (John Atkinson has stated time and again that his in-room-near-field bass response measurements tend to exaggerate the measured bass response.)

As far as I can tell, the Sanders speakers are pretty flat as far as the panel goes and the transmission line woofer does not inherently produce any excess bass warmth. If you want a somewhat warm, generous low end as I and most listeners prefer, you have to add electronic equalization, at least as long as the cabinets are positioned well away from walls and corners as I have them deployed.

Equalization

Roger Sanders videos and hard copy instructions recommend that when you run the Automatic Equalization (AEQ) Wizard via the LMS, you should select the "Recommended Curve" for the equalization result. That is the way I've been doing it up to now. The Recommended Curve inserts a parametric low shelf boost of the low end starting above 500 Hz and topping out at about 7 dB at 100 Hz and below. It also uses an additional 10 bands of parametric EQ to smooth the full-range response. Watching the real-time results of further EQ using my OmniMic v2 measuring system on my computer while I adjust the EQ through the LMS via the dbx VENU360 app on my iPad, I then manually tweak the AEQ result with other available parametric filters: four more are available within the AEQ function. Eight more are available through the PEQ function controlling the panel and eight more through the PEQ function controlling the woofer. There are plenty of parametric filters available, in other words, to smooth the response to taste.

This method sounds excellent. However, I've sometimes found the "Recommended Curve" low frequency boost to be a bit heavy handed, especially on male voices which gain a bit too much chest tone if you don't use other parametric EQ bands to reduce the amount of boost from 250 Hz and higher. For the low end boost amount, if you think 7 dB at 100 Hz and below is a bit too much of a good thing, you can, as I have, reduce the overall level of sound put out by the bass amp/woofer by reducing the level of all sound below 172 Hz (the crossover between woofer and panel) by a subjectively ideal amount. This is what Roger Sanders recommends to tweak the low end response, and is what I was doing, reducing the overall level of the woofer by 0.8 dB. That sounds just about right in terms of mid and deep bass.

But recently it occurred to me that there is another, perhaps more accurate and flexible means to smooth the 10e response. What if I could equalize the response of the speaker to flat while still using the AEQ function? Then, I could hold one of the AEQ parametric filters in reserve to design and apply a low frequency parametric shelf boost more to my liking.

That is what I have now done. The AEQ Wizard allows you to automatically smooth the response to "Flat" rather than to the "Recommended Curve." It then uses 10 parametric filters to flatten the overall response.

By looking at the graph of the "'Recommended Curve" boost, which is displayed on my iPad screen within the dbx app as the "Target" curve, I was then easily able to design a parametric curve of similar shape and amplitude but with less boost at 250 Hz and above. The parameters of my chosen bass low shelf boost curve are: Frequency = 172 Hz, Q = 5, and Amplitude = 7 dB or less.

This creates a situation where I can better eyeball the graphed frequency response using my OmniMic v2 frequency response measuring program on my computer since I am now judging the deviation of the graphed frequency response from a horizontal grid line on the screen rather than trying to judge the graphed response deviation from the "recommended curve" which is not a flat line and is not displayed on the OmniMic v2 computer screen.

This works extremely well. I have expanded the graph scale of the OmniMic v2 system so that there is only one dB between horizontal graph lines. With the exception of the 50 Hz area where my room has a persistent null, I am now able to get the measured response to fall within plus or minus one dB from 20 Hz up past 15 kHz, and within plus or minus 0.5 dB over much of this range, above which the measured response falls off a dB or two by 20 kHz. This is using either 1/6-octave or 1/12-octave smoothing.

The 50 Hz null shows being 3 dB down with 1/6 octave smoothing. With no smoothing at all this null maxes out at about 10 dB deep over a range of just a couple of Hz. I make no attempt to fill in this null other than what the AEQ system does automatically.

Crossover Tweaking

As delivered by Sanders, the dbx VENU360 is set up to have a 48 dB/octave high pass crossover to the woofer at 23 Hz. However, one of PEQ bands is also set to deliver a low-frequency shelf EQ to the woofer of plus 10 dB with a frequency of 30 Hz, thus boosting the response at 23 Hz and below by about 10 dB.

With these parameters, for some reason when the AEQ is used to equalize the speaker to "Flat," a strong bass boost occurs below 30 Hz. Apparently the AEQ is not flattening the PEQ's 10 dB low frequency shelf filter. To get rid of this measured boost, I first tried a countervailing PEQ bell-shaped parametric filter set to 27 Hz at minus 7 dB. This flattened the bottom half-octave of bass nicely, with 20 Hz then about 3 dB below the reference level.

But by playing around with these two crossover parameters--the 23 Hz 48 dB/octave high-pass on the woofer and the 30 Hz low frequency shelf boost, I determined that if each of these is removed, the speaker's measured low frequency remains remarkably flat in room down to about 15 Hz! So that is what I'm now doing. I turn off the 23 Hz 48 dB/octave high-pass crossover on the woofer and turn off the PEQ for the woofer, thus eliminating both the factory's 10 dB low frequency shelf boost at 30 Hz and the countervailing PEQ bell-shaped parametric at 27 Hz at minus 7 dB.

With these crossover changes, the sound gains considerable openness/spaciousness (it was already phenomenal at that, or so I thought). Imaging is more specific, yet rounded. Staging depth gains vastness. The bass has new-found punch and definition. There is added clarity and definition throughout the spectrum without any additional brightness. Dynamic contrasts seem wider. Generally the sound seems better at lower volume levels than before. The phrase "overall transformation" keeps coming to mind, but the sound was so wonderful before, how can this be? Perhaps it has something to do with improved phase relationships between the bass and upper ranges. Removing high-pass filtering--such as extending the bass of a system down into the subsonic region--has been experimentally shown to provide sonic benefits not only in the entire bass range but also far above.

I think the reason Sanders ships the VENU360 with the bass settings I mentioned is to avoid overdriving and possibly damaging the woofer with subsonics. His settings roll off the bass below 25 Hz. But if you're a bit careful and don't relish extremely high SPL levels, my revised settings should be safe enough and certainly demonstrate many sonic advantages. Watching the woofer cones move on heavy bass material does not reveal excessive excursion at the SPLs I favor.

Fo some reasons Sanders also ships the dbx unit for his speakers with a 48 dB/octave low pass crossover for the electrostatic panel set at 20 kHz. This is also easily turned off, which I did. The highs gain additional openness and clarity and about a half dB in measured level above 15 kHz, another sonic improvement.

I'm still zeroing in on the exact level of the bass shelf boost beyond flat bass which sounds best. So far, with the substantial subsonic bass extension I now have, I am finding that a parametric bass shelf filter with the parameters of frequency = 172 Hz, Slope = 5, and amplitude = +4.8 dB seems subjectively about right.

 

[tmallin]

Foundation Equalizing for Speakers Generally

The basic methods for foundation equalizing apply to all speakers. The two things you need are (1) a way of measuring the response of your speakers, preferably in real time--usually a computer program--and (2) a way of adjusting the measured response of your speakers, preferably in real time--an equalizer program or box. You definitely should not trust your ears for this task.

Really advanced audiophiles use Acourate or other other very sophisticated programs to achieve unbelievably precise results. This technology is complex to implement and, I think, beyond the abilities of most here.

At the other end, you could buy a used analog 31-band graphic equalizer like the sonically transparent but pain-in-the-butt-to-adjust Audient ASP231, or the not as transparent but easier to work with digital Rane DEQ 60L (don't use the Perfect Q function and set it up so that you are adjusting both channels with a single set of sliders).

You could use such an equalizer in conjunction with the OmniMic v2 measuring program which costs about $300, including the microphone, microphone clip, and measurement program, and you supply the boom mic stand. Then you run the mike's USB cable (perhaps with an added USB extension cord) to your computer outside the listening room. Follow the OmniMic v2 instructions for setting up the program on your computer and measuring frequency response.

You should not be in the room when measuring; your physical presence anywhere in the room will skew the measurements; I know, I've tried it in the room watching the OmniMic screen as I moved from place to place in the room. The measurements are only reliable and stable if you are outside the room with the microphone between where your ears would be when listening.

Starting at the bass end, adjust each slider so that the response on the OmniMic display on your computer smooths out. Keep going back and forth between the listening room where the equalizer is and where your computer screen is, iteratively adjusting each slider. I estimate that this method with the Audient ASP231will take you a full eight hour day to get the response as smooth as possible due to all the interactions among adjacent filters. It would take about four hours with the Rane since you only need to adjust one set of sliders, not one set for each channel.

If you are willing to invest in two sets of long balanced XLR terminated cables, you can probably cut the adjustment time down to an hour for the Rane and two hours for the Audient by having the EQ box set up right next to your computer screen while still being wired into your system. You would eliminate all the walking back and forth time and all the exercise that would provide and could observe in real time the effect of what you are doing. I think it's worth the investment if you choose this method. Just be very careful not to touch the EQ sliders when you move the EQ box back into the listening room! Attaching the proper sized clear plastic tamper guard (these are sold for pro-audio installations) would be a good idea until you have the equalizer installed back in your listening room.

In between the manual graphic equalizer and Acourate methods, you could try REW. Here is a link to how to use REW to measure and correct your system using the equalization built into Roon. Yes, it really is that complex, but many people seem to master it. It's one of those things which is at least a bit more complicated to explain than to do. Roon is well-integrated with the Dutch & Dutch 8c speakers, so if you use it with those speakers, it's much easier, but it still may be a bit challenging.

What I suggest if you are willing to invest $1,500 or so, is the combination of the OmniMic v2 measuring system with the dbx VENU360 equalizer. The dbx unit, as far as I can tell with the Sanders 10e speakers, is as transparent as they come. You will need to connect the equalizer to your home network with ethernet cable from your router. If you don't have ethernet available and don't want to install it, you can use this device in Client Mode to pick up Wi-Fi signals from your router and run an ethernet cable from there to the equalizer. (You are not running your audio signal through the Wi-Fi; you are just using your home network to access the equalizer to adjust it.)

First download the dbx VENU 360 app onto your Windows computer. This app is a marvel of ease of use and quality of the display. The only problem is that you have to use a Windows computer since neither the dbx nor the OmniMic programs currently work with Mac or iOs. Once your dbx unit is connected to your home network with an ethernet cable, the app should find and connect to the dbx unit. If you are only equalizing a pair of stereo speakers like the Quads, use the system set up wizard on the app to set up the app for stereo speakers, rather than any more complex option.

Second, set up the microphone that comes with the dbx unit on your mike stand so that the capsule is between where your ears would be sitting in the sweet spot. Aim the mike straight ahead. Plug the supplied mike cord into XLR jack on the front of the dbx unit.

Third, on the dbx app, open the AEQ screen and choose Wizard. Select the AutoEQ only wizard. Choose "Recommended Curve." Run four sweeps, not moving the measuring microphone at all. This will take less than five minutes. The AEQ function will then use 10 parametric EQ bands to flatten the response.

That's it. Unless you're really picky about flattening the response, this will take care of everything that ails the frequency response of your speakers quite nicely. The only thing to check is that no EQ band is boosting the response by more than 6 dB and that the AEQ has not tried to extend the lower limit of your speakers below their natural roll off frequency. Those are no-no's for any speaker.

Now, if you're really picky like me, you can now use the dbx app in combination with the OmniMic v2 program to flatten the response even more. I suggest using split screen mode on your Window's computer monitor. Run the dbx app on one side of the screen and OmniMic on the other.

If you choose this route, when you run the AEQ Wizard on the dbx app, pick "Flat" rather than "Recommended Curve."

After you run the AEQ test sweeps for the dbx equalizer, then switch out the microphone to the one which came with OmniMic v2, arranging it the same way on the mike stand. Connect it to a USB input of your computer and follow the OmniMic instructions for finding and applying the correction curve for that individual microphone.

On the OmniMic screen, choose Frequency Response, sine sweep test tone, All, and set the vertical graph division so that each line represents a 2 dB difference in SPL. Choose 1/6-octave smoothing, with the horizontal graphed frequency response running from 20 Hz to 20 kHz. Play track 6 on the OmniMic CD (left channel short sine sweep) or track 12 (right channel short sine sweep) on continuous repeat and adjust your system volume so that 1 kHz displays somewhere between 70 and 80 dB; louder is better at ignoring ambient noise such as traffic, planes, trains, etc.

After the dbx unit auto corrects the frequency reponse, there are four more parametric bands available on the dbx app within the AEQ function. There are at least 8 additional parametric EQ bands available in the PEQ function, perhaps as many as 16. Hold band 14 within the AEQ bands in reserve for now. You may also want to reserve one of the PEQ bands to insert a BBC dip (F = 3kHz, Q = 3, A = minus 4.2 dB, for example) for some material.

Now the manual frequency response adjustment fun can begin. Examine the displayed frequency response on the OmniMic screen. Starting again at the bass end, using the dbx app, insert any additional parametric filters you think will help flatten any bumps or dips in the response line you see. You can watch the result of what you do via the dbx app in real time on the OmniMic screen. Adjust the filter parameters until the bump or dip is minimized, keeping in mind the 6 dB limit on boosting--you can cut a peak as much as you like.

Do the same for each significant departure from flat as you move up the frequency response range. This will take some time, but the constant visual feedback speeds the process. I suspect that with time, patience, and the limits on available bands of parametric EQ, you may be able to flatten the speaker response to plus or minus 1 dB or better from its low frequency rolloff up to about 15 kHz.

When you are done flattening, you can set up band 14 of the AEQ bands as a low shelf filter (not a bell-shaped filter) with slope of 5 and frequency of 172 Hz. Adjust the boost of this filter to your target curve taste. Most folks will probably choose to boost this filter by 2 to 6 dB for the most pleasing frequency balance.

If you saved one of the PEQ filters for the BBC dip, you can insert that, too, modifying the parameters I stated to taste.

And then you are "really done" with foundation equalizing for your system. Except that over time you may decide to tweak this or that parametric filter based on what you hear. Never forget that your ears are the final arbiter. The amount of boost or other parameters of that band 14 filter are the most likely things you will fiddle with at first, but eventually you will lock into what sounds best on most material.

Now you can add that Schitt Loki Max to EQ individual recordings to taste. And so it goes....

 

[tmallin]

Getting the power range of 100 to 300 Hz fully fleshed out is very important to musical realism. See Robert Greene's short essay on this here. It is also important not to have too much treble. A flat response in the top two octaves, while easily achievable with the Sanders as discussed above in my posts #163 and #164, is not necessarily the most musical sounding on commercial recordings.

I'm fortunate that the Sanders 10e and the Dutch & Dutch 8c before them in my room do not have problems in the 100 - 300 Hz power range. The built-in equalization of these speakers allows flat response in this area. My room's problem area is a sharp null around 50 Hz just a few Hz wide. I suspect this comes from setting up the speakers symmetrically in the room. The only time this null went away was with the AudioKinesis Swarm plus Stirling LS3/6 mains run full range where bass in this area was generated by six different sources, four of which were not symmetrically located in the room.

Using the OmniMic v2 measuring system with the microphone at the listening position, except for that 50 Hz null, I can equalize the Sanders flat to about plus or minus 0.5 dB from 17 Hz up to about 15 kHz. (I'm not boosting the bottom octave at all; this is just the natural mechanical roll off of the woofers in my room.) But actually the high frequency end sounds better--as in with all traces of "too much" or "aggressive" removed--if I let the response "relax" or slope down a bit above 3 kHz so that it is down 2 dB by 10 kHz and about 4 dB by 20 kHz.

The Auto EQ function of the dbx VENU360 box which equalizes the Sanders automatically provides this sloped down high end if I don't fiddle with it manually. And even if I command it to equalize the low end to "Flat" rather than "Recommended Curve," the low frequencies gradually rise about 4 dB from 200 Hz down to 20 Hz. This sounds quite nice and takes less than five minutes to accomplish once I set up the measuring microphone. Letting the Auto EQ function equalize the response to the "Recommended Curve" sounds warmer and bassier, perhaps a bit too much of a good thing on male voices, but still quite acceptable.

Manually tweaking the response further, lately I run the Auto EQ to produce "Flat" response, then I manually really flatten the range below 3 kHz to that plus or minus 0.5 dB standard from 17 Hz up to 3 kHz (except for the 50 Hz null), not fiddling with the Auto EQ results above 3 kHz. Then I use one parametric EQ band to dial in a subjectively satisfying low frequency boost taking the form of a low frequency shelf filter with the parameters Slope = 5, Frequency = 172 Hz, and Amplitude = +4.8 dB. The parameters of that boost filter are all easily adjustable in real time via my iPad from the listening seat, but those settings are about what I think sounds "right" at this point. The result is increased bass below 250 Hz, flat from 250 Hz up to 3 kHz, with gradually decreasing treble above 3 kHz.

 

[Ron Resnick]

Getting the power range of 100 to 300 Hz fully fleshed out is very important to musical realism. See Robert Greene's short essay on this here. It is also important not to have too much treble. A flat response in the top two octaves, while easily achievable with the Sanders as discussed above in my posts #163 and #164, is not necessarily the most musical sounding on commercial recordings.

I'm fortunate that the Sanders 10e and the Dutch & Dutch 8c before them in my room do not have problems in the 100 - 300 Hz power range. The built-in equalization of these speakers allows flat response in this area. My room's problem area is a sharp null around 50 Hz just a few Hz wide. I suspect this comes from setting up the speakers symmetrically in the room. The only time this null went away was with the AudioKinesis Swarm plus Stirling LS3/6 mains run full range where bass in this area was generated by six different sources, four of which were not symmetrically located in the room.

Using the OmniMic v2 measuring system with the microphone at the listening position, except for that 50 Hz null, I can equalize the Sanders flat to about plus or minus 0.5 dB from 17 Hz up to about 15 kHz. (I'm not boosting the bottom octave at all; this is just the natural mechanical roll off of the woofers in my room.) But actually the high frequency end sounds better--as in with all traces of "too much" or "aggressive" removed--if I let the response "relax" or slope down a bit above 3 kHz so that it is down 2 dB by 10 kHz and about 4 dB by 20 kHz.

The Auto EQ function of the dbx VENU360 box which equalizes the Sanders automatically provides this sloped down high end if I don't fiddle with it manually. And even if I command it to equalize the low end to "Flat" rather than "Recommended Curve," the low frequencies gradually rise about 4 dB from 200 Hz down to 20 Hz. This sounds quite nice and takes less than five minutes to accomplish once I set up the measuring microphone. Letting the Auto EQ function equalize the response to the "Recommended Curve" sounds warmer and bassier, perhaps a bit too much of a good thing on male voices, but still quite acceptable.

Manually tweaking the response further, lately I run the Auto EQ to produce "Flat" response, then I manually really flatten the range below 3 kHz to that plus or minus 0.5 dB standard from 17 Hz up to 3 kHz (except for the 50 Hz null), not fiddling with the Auto EQ results above 3 kHz. Then I use one parametric EQ band to dial in a subjectively satisfying low frequency boost taking the form of a low frequency shelf filter with the parameters Slope = 5, Frequency = 172 Hz, and Amplitude = +4.8 dB. The parameters of that boost filter are all easily adjustable in real time via my iPad from the listening seat, but those settings are about what I think sounds "right" at this point. The result is increased bass below 250 Hz, flat from 250 Hz up to 3 kHz, with gradually decreasing treble above 3 kHz.

This is very interesting, Tim. Two weeks ago I started using a spectrum analyzer and an external professional quality microphone and mic stand to make objective measurements for the first time ever in connection with audio.

It would be great and interesting to see your curve on a thread I started precisely for this purpose: https://www.whatsbestforum.com/threads/post-your-frequency-response-curve.36361/#post-855991

It would be better, still, if you were so inclined, to download the same Spectrum Analyzer RTA app we all are using, and with the same settings, for an even more apples-to-apples visual comparison.

 

[Ron Resnick]

Getting the power range of 100 to 300 Hz fully fleshed out is very important to musical realism.

I completely agree with the importance to musical realism of the power range of 100 Hz to 300 Hz. I might actually extend the upper range of that important region to 500 Hz.

But actually the high frequency end sounds better--as in with all traces of "too much" or "aggressive" removed--if I let the response "relax" or slope down a bit above 3 kHz so that it is down 2 dB by 10 kHz and about 4 dB by 20 kHz.

I also know that my ears prefer a downward sloping treble range. Interestingly, compared to your preference, I think I prefer a much greater roll-off than 2 dB by 10 kHz and 4 dB by 20 kHz.

PS: I often wonder if declining high frequency response due to age, or tinnitus, or both, has the unexpected effect of making me extremely sensitive to, and making me find edgy and irritating, almost any musical information in the treble range.

 

[tmallin]

Part of the amount of preferred treble roll off is surely related to the high frequency directivity of the speakers in question. With the back wave of my Sanders 10e damped with lots of foam in the wall corners behind them, the resulting front radiation does not bounce off walls, floor, or ceiling much at all. The wall behind the listening seat is also highly diffusive. See the pictures at post #7. In my experience, the less obnoxious room surface reflections of high frequencies you have, the less the treble will need to be rolled off for best subjective sound quality from commercial recordings.

I have not been capturing screen shots lately of my system response. Much of the way a graph looks is determined by the amount of smoothing you apply and the number of decibels between vertical divisions. The vertical scale of the graphs in the thread you referred to seem to be between 5 and 10 dB per division. Most any speaker on earth will look fairly flat and smooth with that kind of vertical divisions. Stereophile usually uses 5 dB per vertical divsion, also. No offense meant, but the fact that many of the posted responses on the thread you linked to look quite bumpy even at 5 to 10 dB per vertical resolution shows that much work remains to be done in terms of smoothing the system response.

The graphs I work with mainly are 1/6-octave or 1/12- octave smoothing and 1 or 2 dB per vertical division. It is only by working with a graph with 1 dB per vertical division that I can be assured that the equalized smoothness is plus or minus 0.5 dB. If I used 5 dB per vertical division, I guarantee you that except for my 50 Hz null, the graphed line would look flat indeed and that even the narrow 50 Hz null, with the 1/6-octave or 1/12-octave smoothing would appear as a very minor notch in the line, not at all a major deviation from flat.

As far as the OmniMic v2 program I use, I've used it and its Liberty Instruments (Bill Waslo) predecessors SynRTA and Praxis for decades now. I know this program well and it provides the flexibility I need. I'm thus reluctant to abandon it.

 

[Ron Resnick]

As far as the OmniMic v2 program I use, I've used it and its Liberty Instruments (Bill Waslo) predecessors SynRTA and Praxis for decades now. I know this program well and it provides the flexibility I need. I'm thus reluctant to abandon it.

Forgive me for being confusing. I was not suggesting that you abandon your long used and liked program. The one we have been using commonly on the frequency response curve thread is either free or downloadable for $20.

 

[Fred Crane]

An aside...apologies where applicable.

Regarding the post's name, for most it will recall Sinatra doing his minted version of the classic song by Sammy Cahn and Jimmy Van Heusen (Chet Babcock originally...he changed his name at 16, while finishing a butt, looking out of his apartment window onto a 'Van Heusen Shirts' billboard)

For me, it pulls up the dulcet tones of Joe Mooney. Joe was many things; a wit, a marvelous bebop accordion player, a fine pianist and organ player, and a vocalist with some phrasing capacity. He was a Sinatra favorite. I came up with stories about Joe from my folks and other local musicians in New Orleans who had made the pilgrimage from NY to Miami in the early 50's as was the trend. Joe was left legally blind after a serious car accident. It was during 18 months of recovery that he came up with the sound of the group he was to form...mostly a trio with an occasional guest. Anyway, here's Joe's version of The Second Time Around.

 

[picears]

You should not be in the room when measuring; your physical presence anywhere in the room will skew the measurements

Well... if one is going to be in the room when listening (we hope), should we not be concerned about the in-room response when we are in our listening chair?

 

[tmallin]

There is no way to both physically sit in the listening chair and place the measuring microphone between where your ears would be when you sit in the listening chair. Fractions of an inch of microphone placement make a difference in measured response.

And even putting the microphone anywhere near your head produces a reflection which skews the measured response.

Also, by holding the OmniMic display (I use a Microsoft Surface Pro tablet for this) in my hand and walking crawling around the room, I can tell that the measured response varies depending on where my body is in the room.

If you must be in the room for taking measurements, in my room, the most stable measuring location seems to be kneeling down midway between the speakers, behind the speakers and directly in front of my equipment rack. If you do not use the dbx VENU360 app on an iPad or portable computer, that would be the way you'd have to do it if you didn't want to be constantly running back and forth between a computer display of frequency response outside the room and the dbx adjustments on the dbx unit in the audio room. Ugh. That would be very tedious and tiring and ultimately less accurate because of the lack of constant visual feedback in real time between trial adjustments to the filters and the effect of those adjustments on the measured frequency response.

Using a long USB cable from the microphone to a location outside the listening room and leaving the door to the listening room shut (as it is for all serious listening), results in very consistent frequency response readings from moment to moment and even session to session days, weeks, or months apart as long as the system set up is not changed.

 

[picears]

And even putting the microphone anywhere near your head produces a reflection which skews the measured response.

Exactly.

This is why I find Prof. Choueiri's device so fascinating. He's attempting to correct for the effects of head shape, ear pinnae shape, etc.

So this gets me back to my question: if our presence alters the response, and we correct (via DSP) the system for when we are not in the room, there is a disconnect in what we are attempting to do. This again is why I find Choueiri's ideas something that will be interesting to follow going forward.

Long ago the audiophile community had a contingent that cringed at tone controls. Yet here we are in 2023 and DSP is accepted by many.

 

[tmallin]

You are talking about the audible effects of using the BACCH program to process stereo playback as heard from the listening seat. See the detailed TAS review and discussion of how BACCH works here. That device uses a pair of small microphones which you place in your ear canals to set up the system by measuring the sound impinging on your two ears at the listening position and "correcting" what the speakers sound like to compensate for a number of effects related to headshadowing and the fact that stereo imaging is not "real" but "virtual" in the sense that the brain operates on what your two ears hear to create phantom virtual images from the two separate sources (two speakers) generating the sound rather than the single "real" source we hear in nature.

The goal of BACCH is not so much to create a flat frequency response than to create a more (MUCH more) three dimensional sound field. From the demos I've heard of it so far (at the 2022 AXPONA event), the system goes far beyond what I hear at a live event in creating three-dimensionality. If normal stereo soundstaging is not three dimensional enough, BACCH sounds surreal in a laughable fun-house way. Others seem to disagree (see the linked review), but I think they are just fascinated with the rather obviously exaggerated stereo effects BACCH produces compared to what one hears in a concert hall. Interesting and fascinating perhaps, but hardly realistic. And BACCH effects are hardly new to the audio scene. Ralph Glasgal's Audiophonics, the Carver Sonic Hologram, and TacT's XTC system are all eariler attempts. Yes, BACCH removes the "phasey" tugging at your ears that prior systems tend to produce, but it also further exaggerates the stereo effect far beyond what is natural.

Having measured the frequency response of my audio systems when the Carver Sonic Holgram and the TacT XTC were employed, I know that such systems drastically alter the measured response one gets from a measuring microphone positioned between where your ears would be in the listening position, especially when measuring both channel outputs at once. The idea of such systems is not to produce flatter frequency response, but to alter the way our brain perceives the soundfield sound generated by two separate loudspeaker sources. Frequency response manipulation is part of the processing involved in that effort.

I think you will find (at least I do) that setting up the Sanders 10e speakers as I have to eliminate room reflections to a great extent, you will hear much more three-dimensional stereo than with most other speakers. You can improve the stereo effect with all speakers using acoustical room treatment, but you get further with the Sanders since, other than the back wave, it is so very directional in the first place. What I hear in terms of soundstaging is quite natural sounding to me, much moreso than what I've heard from any of these systems, including BACCH. That's my current opinion, obviously, subject to alteration on further experience with BACCH at the coming AXPONA.

 

[picears]

See the detailed TAS review and discussion of how BACCH works here.

Thanks. Yes, I know what the BACCH does.

And I agree with you that people can be immediately enamoured with the gee-whiz capabilities.

Still, one can use such system to not do all the gimmicks possible (and the professor has many more gimmicks in development), but instead to simply correct for the soundfield at one's head/ears.

And I think that could be quite useful if there is a setup in the home which is used by more than one person. Having processing intended for each person, that can be easily called up, will allow a system to compensate for each person's hearing idiosyncrasies.

It's my belief that is the goal of what I and many others wish out of their audio system - not a flat response curve but a sound that is good to us individually.

 

[tmallin]

Hearing idiosyncrasies among individuals usually don't need to be compensated for in an audio system since each person hears both natural sounds and audio system sounds with their same set of ears and same brain.

But it seems clear that listeners do have individual preferences as to how they would like their home audio systems balanced in terms of frequency response. For this, you don't need BACCH, only an equalizer with recallable, stable presets.

I've discussed such equalizers earlier in this thread as well as elsewhere. One is the dbx VENU360 which is the LMS (Loudspeaker Management System) at the core of the Sanders 10e.

Ideally one would daisy chain what I've called a "foundation" equalizer with another equalizer meant to adjust individual recordings on the fly. Usually, one would set what I've called the "foundation" equalization to be flat or according to a preferred target curve, which most people prefer to be rising at the low frequencies and rolling off gently at the high frequencies. That is what I have done via the dbx unit in my own system. On can also add individually preferred enhancements to such a foundation equalization curve, such as a "BBC dip" which adds a bit of response "relaxation" centered around 3 kHz to back off the soundstage and tame aggressive string sound from orchestral recordings a bit. The dbx unit has about 75 presets available which can be recalled at will.

Then, for equalizing individual recordings to individual taste, one can use a unit with just a few bands of EQ to make on-the-fly adjustments to the response. An ideal unit for this, and one that doesn't even add any DSP to your system, is the Schiitt Loki Max. It is remote controllable and has many presets available as well. It uses six broad, overlapping bands of equalization to adjust the overall frequency response in a manual way strongly reminiscent of the old Cello Audio Palette and Palette Preamp units with equal or better all-analog transparency.

The dbx unit can also be used to mimic the action of the Schitt Loki unit in addition to the foundation EQ chores, but I doubt that it can match the Loki's ease of on-the-fly equalization. Six parametric bands of the dbx unit could be reserved to provide the type of low-Q equalization provided by the Loki. The frequency response graphs of the Loki filters could be studied to construct dbx filters of similar Q and center frequencies. I suspect the Q of each filter will be 1 or less. See the graphs of the Loki Max filters on the pages labelled 2 of 2 in the test report here.

 

[picears]

I mention the BACCH because it does in-ear measurements (with a microphone for each ear) and the results will show how different people really are.

Long ago when I first got interested in stereo I used tone controls a bit but then went in the opposite direction to eschew them.

With digital processing one can add as many filters as one likes, such as the ones you mentioned. I get that some people are opposed to DSP but I do think it is going to be ubiquitous in our lives.

 

[tmallin]

More on speaker and listener positioning

We all like shortcuts and rules of thumb.

Let's assume you are dealing with only two stereo speakers, no subs. Yes, any positioning of two stereo speakers is a complex trade off among bass extension/power, bass smoothness, and imaging/staging. You have to decide how relatively important each is to you.

For me, that decision is easy. I have found that I can fix bass smoothness quite well with electronic equalization.

However, with the exception of a few cardioid-radiating speakers, most speakers sound much better in terms of imaging and staging if they are kept far away from the walls of your listening room. If bass is the only thing that matters to you, put your speakers in the room corners, and equalize the bass for maximum smoothness from the listening position. You will get the deepest possible bass for your speaker/room combination, as well as the highest bass SPL without obnoxious distortion. But you will then just have to live with the unnatural imaging and staging because, believe me, padding the walls is only a partial solution for speakers so positioned. Even with cardioid speakers like the Dutch & Dutch, there is at least a bit of imaging/staging penalty for keeping the speakers near the wall behind them.

Thus, I tend to heavily favor imaging/staging when setting up my speakers, fixing the bass with electronic equalization. In my current small (132 inch x 161 inch) room, even with near-field listening (currently about 55 inches from speaker drivers to ears), it is difficult to get the speakers more than five feet from the wall behind them and much more than three feet from the side walls, while maintaining a 60-degree subtended angle between the speakers and a listening position more than five feet from the wall behind the listening position.

Yes, the only way to be sure of what will happen in terms of bass response when you move a speaker in your listening room is to try something, measure, and listen. Then try another position and another until you find the smoothest bass response position.

This iterative measurement method seems simple enough, but is really tedious in practice. The problem is that the measured response at the listening position depends heavily not only on the distance of the speaker from the side wall and wall behind it, but also on the position of the listener with respect to the front and back walls.

So, you set up the measuring microphone where you think the listening position might be. Then you move the speaker around, watching your measurement graph for the smoothest response. You only need to do this with one speaker if your room is fairly symmetrical acoustically and you plan to set up the speakers symmetrically in your room and most listeners do.

Now you find the speaker position where the bass measures as smoothly as possible. BUT: you must now position the other speaker symmetrically in the room, measure the distance between the two speakers and move your listening position to create a 60-degree subtended angle between the two speakers. Chances are, this means moving your listening position and thus your measuring microphone forward or back several inches or more. In a small room like mine, even a few inches changes bass response considerably. Not only does bass smoothness vary, but in every listening room I've ever experienced, the general trend is that the closer to the center of the room you move the listening position, the less overall weight the bass has and the closer you move the listening position to the wall behind the listening seat, the more overall bass level there is.

Thus, once you move the listening position to create your 60-degree subtended angle, iterative cycles of moving the speaker again, locating the new smoothest response spot for the speakers, then again moving the listening position to re-establish your 60-degree subtended angle must be undertaken.

Yes, you can eventually zero in on the best speaker and listening positions in terms of smoothest bass, but it is tedious. If you don't want to use electronic equalization to smooth the bass, this is the only way to get the smoothest bass response while still giving you a proper stereo spread and depth.

But if you ARE willing to use electronic equalization to smooth the bass, and you have a dedicated listening room which allows you to put the speakers and listening chair most anywhere in the room that you want, then I suggest trying my "lazy man's approach." Use the Cardas calculators for placing your speakers and listening position. Once you do that, smooth the bass with electronic equalization and you will have fine bass, plus outstanding imaging and staging, probably about as good as you will get from your speakers in your room.

Yes, the Cardas calculators are just another rule of thumb, but they seem to work even in very small rooms like mine to yield truly exceptional stereo imaging and staging, with decent bass thrown in. Why this "golden ratio" method works, I don't know, but it seems to not mess up the bass beyond what can easily be corrected electronically and gives you a spatially fabulous presentation. See the Cardas calculator on this page. Remember that the length of the "main wall" in the calculator is the wall behind your speakers, not necessarily the longest wall of the room. That calculator will also tell you where to put the listening position with respect to the wall behind the speakers so as to automatically create a 60-degree subtended angle between your two stereo speakers.

Also, if you are placing pure dipole speakers like the Quads, the proper Cardas calculator is here. For my room measurements, the only difference between dipole and monopole Cardas speaker placement recommendations is that dipoles are to be placed about five inches further from the wall behind them.

Actually, this is one excellent method of arranging your main stereo speakers even if you have a Swarm or other multi-sub arrangement, or corner woofers like the TacT. You put the separate woofers/subwoofers either in corners or some asymmetrical arrangement which smooths the bass. With the Swarm, in my room I did not need electronic EQ for the low bass beyond what is provided in the amp which is part of the Swarm package. Then, for the main speakers, you could arrange them according to the Cardas calculator and electronically equalize them for smoothest lower ranges.

 

[tmallin]

I agree that there is nothing magical about 60-degree stereo separation. Minor deviations from 60 degrees probably are not that important to the overall effect.

However, from what I've read, 60 degrees seems to be the established standard for studio mixing, whether or not it is often not observed in real-world studio monitoring. Here's a quote from one pro-audio source:'

"The established standard in the studio mixing world is a 60-degree angle between the two loudspeakers as measured from the listening position. In other words, an equilateral triangle with the two loudspeakers and the listener’s head located at the points. You’ll also see the same 60 degrees show up as part of the ITU-R BS.775-3 recommendation for multichannel stereophonic sound systems, which is often referenced in home theater work. Thus, if you want to hear the stereo part of recordings as intended, it is best to set your speakers up with a 60-degree subtended angle between them."

I suspect that studio mixed recordings make up 95+ percent of all commercial recordings, even classical. The percentage of our "best" recordings done with Blumlein or some other quasi-coincident miking array for which a wider separation--90 degrees or even more--is ideal is quite small.

A decade or more ago, for at least a year I set up my speakers for 90-degree separation. Yes, a few purist recordings definitely sounded best that way--the Sheffield Labs were some. But most were at best just different or interesting that way. For many recordings, there were rather obvious "holes" in the imaging/staging with "pools" of sound separated by clearly audible gaps in the smooth spread of instruments on the stage. I eventually concluded that since I listened mostly to recordings for which 60-degree separation sounded best, that set up offered the most rewarding overall experience and I've used that separation ever since.

Many audiophiles, stereo shops, and show demos seem to set up their speakers with a combination of 45-degree-or-less separation and listen in the far field (10 feet or more away from the speakers). I think this results from room size/shape considerations, a desire to keep the speakers away from the side walls (good idea!), a resistance to toeing in the speakers to face the listener's ears, as well as the fact that most large, multi-driver speakers really do not "gel" when listened to in the near field. With most large speakers with drivers spread out vertically and/or horizontally, it is easy to hear out the positions of the individual drivers when you listen from five feet or less away from the speakers. I try to avoid such speakers. But listening in the near field makes me much more aware of the proper geometry where everything gels. The proper listening height for the LS3/6 type of speaker is one such example; listen above the height of the lower tweeter and the treble sticks out. With the Sanders 10e, the only rule to avoid hearing out the woofer is to place your ears at least a foot above the bottom of the electrostatic panel. Once that is done, I never hear the woofer as a source even though my listening position is only 55.5 inches away from the large panels.

 

[tmallin]

The Limits of Automatic Equalization Software

Frequency response needs to be manually adjusted using a program different from any automatic EQ present in the system. This is so even in the bass regions (below 200 Hz, say) where frequency response measurements taken from the listening position should fairly closely match what your ears hear.

Back when I was using the Stirling LS3/6 speakers, I, too, liked what RoomPerfect, as implemented in Lyngdorf electronics, did for bass smoothness and extension. But I also found its automatic setting to emphasize the highs in a subjectively unpleasant way which made to overall impression of the automatic RoomPerfect settings intolerable for more than a few minutes. Checking the frequency response from the listening position with OmniMic v2 before and after implementing the automatic RoomPerfect adjustments indicated that the automatic RoomPerfect introduced a measured broad peak of a few dB centered around 4 kHz. Certainly that was why I subjectively felt that the automatic RoomPerfect made the sound too bright.

On the other hand, the Automatic EQ function of the dbx VENU360 device which is part and parcel of my Sanders 10e system seems to do just the opposite when measured in the same way with OmniMic v2. It emphasizes the bass end a bit (a shelf rising to about 4 dB at the very bottom bass) and introduces a small amount of pleasant roll off in the top octaves above 3 kHz even when it is set to automatically produce a flat response. While I find this subjectively much more pleasant, if I really want measured flat bass response, I manually adjust the EQ settings so as to produce flat bass response as measured by OmniMic v2 from the listening position.

Then, since, like most listeners, I subjectively prefer a target frequency response curve which boosts the bass below 200 Hz a bit, I use one of the dbx device's parametric filters to introduce a controlled bass shelf which provides a bit more bass lift than the dbx does when it is set to automatically produce flat response, but less bass boost than the dbx introduces when it is set to automatically produce what it calls the "recommended curve" in the bass. This is all very subjective and I tweak it by ear for best bass balance on a wide variety of program material. You can't rely on measurements alone. You must ultimately listen to determine what sounds "right."

Looking back as my list of equalizers (other than plain vanilla bass and treble tone controls) I've owned:

• Cello Palette Preamp
• Z-Systems rdp-1
• Legacy Steradian (for Legacy Whisper speakers)
• Rives PARC
• Rane DEQ-60L
• TacT RCS 2.2XP AAA stock
• TacT RCS 2.2XP AAA with fully Maui Mods
• Audient ASP231
• DSPeaker Anti-Mode 2.0 Dual Core 2012 model
• DSPeaker Anti-Mode 2.0 Dual Core 2013 model with both stock and two different after-market power supplies
• Behringer DCX2496 + DEQ2496
• ART EQ355
• RoomPerfect (in Lyngdorf TDAI-2170)
• Z-Systems rdq-1
• DSPeaker X4
• REW (as integrated with the Dutch & Dutch 8c)
• dbx VENU360

most had no automatic functions. Some might quibble, but I wouldn't call the excellent sounding REW integrated with the Dutch & Dutch 8c speakers "automatic" since a lot of set up and choices are involved before the EQ runs.

Of the ones which have automatic functions, the RoomPerfect in the Lyngdorf TDAI-2170 was by far the least sonically satisfactory to my ears. That unit had no parametric bands which you could use to fix up the automatic EQ's problems, so I had to add the ART EQ355 31-band graphic EQ to fix the broad peak centered around 4 kHz which RoomPerfect introduced.

The automatic Behringer EQ functions resulted in too much top octave, but since it was just adding extra "air," it was not as objectionable as what RoomPerfect did in my system. The Behringer was used with the Sanders 10c speakers I had a decade ago.

The automatic TacT functions worked okay. The DSPeaker Anti-Mode 2.0 models had fine sounding auto-EQ functions and the DSPeaker X4 is even better in that respect. The auto-EQ function of the dbx VENU360 is about on par with the X4 in terms of personal listenability, at least with the Sanders speakers.

All these auto-EQ programs could be improved upon with manual tinkering guided by listening. My point is that, in my experience with auto-EQ functions, RoomPerfect was--at least as implemented in that Lyngdorf unit and when used with the Stirling LS3/6 speakers--the least satisfactory of the bunch.