Post by: George Friedman-Jimenez on July 06, 2008, 02:53:18 PM
According to Bassbox Pro 6, the unprocessed box and driver would be 3 dB down at 39 Hz (11 dB down at 20 Hz), with Qtc of 0.9. The original plans call for building a Linkwitz Transform circuit, essentially a fixed analog line level EQ that boosts the low frequencies so that the low end FR rolls off more gradually and is down 3 dB at 20 Hz. Preliminary calculations suggest an 8 dB boost at 20 Hz, tapering down to 0 dB boost at 45 Hz, will be needed. I have a software realtime analyzer (TrueRTA 1/24th octave) I can use with an Earthworks mic in the nearfield (1/4" from driver) to adjust the shape of the PEQ to get the response as flat as possible. Other configurations in the future are likely, as I am now just starting to experiment with this. What I want are:
1) LF boost as described above to extend and flatten LF response
2) Sub to Woofer Xover at around 120 Hz
3) Woofer to tweeter Xover at around 1.6 kHz
4) Delay/phase adjustment of Sub to full range monitors
5) Flexibility in the future for other drivers / boxes / configurations / live sound applications
6) Steep high pass filter at about 10 Hz for speaker protection, if possible
What would be the best way to do this using outboard hardware? I am looking for excellent sound quality on a limited budget (under $1k excluding speaker). The 3 major options I have considered are:
a) DSP device such as Driverack 260 or Xilica XP3060, used as multiple PEQs, crossovers, delay, and filters.
b) Get an analog PEQ and keep the analog crossover for the tops.
c) Build a dedicated Linkwitz Transform, rely on the sealed box and careful use for LF speaker protection, keep my Ashly X1001 Stereo 2 way analog crossover, forget about delay/phase adjustment.
What advice / cautions can you give me?
Post by: Ivan Beaver on July 06, 2008, 04:06:21 PM
I do not know if you actually want a box that has a flat response down to 20Hz. You said this was for a small recording studio. In a small room you are going to get room gain-due to the size of the space being in the pressure zone as far as low freq are concerned.
How much will really depend on your particular space.
So the real question is are you looking for flat "in room" response or a loudspeaker that is flat all by itself-in a non enclosed space-say outdoors? That can really affect what the final result of your recording/mix is.
Your RTA will only be useful up close to the loudspeaker, so trying to measure the "in room" response is going to be a bit tricky as you really need to be in the listening position to properly align your loudspeakers. The RTA will then be subject to all of the reflections in the room. You really need a time based system such as Smaart-Systune-Praxis-TEF etc. But there goes your budget.
There is no way to see any phase information with the RTA, so you have no idea what is happening in the time domain (the most critical for accurate reproduction-in my opinion), only the amplitude response-which is easily fooled.
Having a DSP is a great thing, but without the tools to properly set it up/align, it is pure guess work.
You may have a hard time getting a 10Hz highpass in most DSP's-there are a very few exceptions. Most are limited to 20Hz. So whatever filter you choose, you are already going to be down a couple of dB at 20Hz.
Post by: George Friedman-Jimenez on July 06, 2008, 05:36:16 PM
Room gain is an issue I thought about and I think I have addressed most of the main issues with having a subwoofer flat down to around 20 Hz. The room is 13x15 feet, so the lowest mode is approximately 40 Hz. I have done a lot of acoustic treatment in the room, there are at least 50 cubic feet of rockwool in bass traps in the corners and ceiling. The room is pretty dead sounding, which is good for playing and recording percussion. Most of the worst modes are reasonably attenuated at my mixing position according to some measurements I did last year using ETF / RplusD using my current sub. My current sub (Event) goes low but has a lot of distortion and vent noise. Variability of the LF loudness walking around the room is not bad, probably due to the bass traps. I would prefer to have the sub as flat as possible and deal with the room gain with bass traps, rather than attenuating the sub and depending on room reflections to augment the bass.
I have been using the RTA as sort of a poor man's loudspeaker measurement tool, for nearfield measurements only (per D'Appolito's book). It is quick and seems to work pretty well for a single LF speaker, although I have no gold standard to compare. Anything beyond real nearfield (mic 1/4" from cone), or with multiple drivers, or above about 800 Hz gets complicated due to wave interactions and room reflections. I have used different methods to try to get the room response flat at the mixing position, including ETF / RplusD measurements with MLS/swept sine waves and waterfall plots, a Radio Shack SPL meter, and Bob Katz' lay method for setting up a subwoofer in a listening room. Nothing is perfect, but with some effort it sounds pretty good at low levels. The bass distortion at higher levels is audible and annoys me, so I am looking to upgrade the sub.
The DSP approach seems like the most flexible, and I think I can use RplusD to get the room response right at the mixing position. If it takes too much EQ I may have to put in more bass traps or maybe add some diffusors, but I think with the current bass traps the room response will homogeneous enough that we can have rehearsals there and use a flat sub for listening back.
Your observation that most DSP PEQs don't go down to 10 Hz is helpful, I didn't know that. I am not too worried though, the box is tiny and sealed, Xmax is 12.5 mm, no unknown engineers will be using it, and the SPL I need is not real high.
My real questions are:
1) Whether a DSP like the Driverack 260 or the Xilica XP3060 will have enough PEQs to extend and flatten the sub.
2) Will the DSP introduce phase or delay issues that could affect sound quality or recording?
If I can adapt the circuit for a balanced in/out, I may opt for building the Linkwitz Transform, it is potentially the cheapest way to go if I can figure it out and get it to work.
Any further thoughts you or anyone else has would be most welcome. Thanks.
Post by: Ivan Beaver on July 06, 2008, 08:09:51 PM
Assuming an 8' ceiling (and your other room dimensions), the freq below around 37Hz are in the pressure zone (where actual gain occurs). The modal zone is from around 37hz to 425Hz. Above that you are in the diffusion zone.
The Xilica has 6 PEQ per output channel. Is that enough? It depends on your particular needs. Probably. If you are using an RTA to measure with, that gives you one band per 1/3rd octave interval.
I would not be worried about the audio quality of todays DSP's, except possibly the lower end models.
What is the rest of your equipment? That might be your limiting factor in terms of quality.
Post by: Silas Pradetto on July 06, 2008, 08:38:01 PM
Post by: Antone Atmarama Bajor on July 07, 2008, 02:11:09 AM
For a subwoofer. At minimum wouldn't you want SPL capability at least equal to that of the mains across the subwoofers intended operational range.
So the desired -3dB @ 20Hz should be in relation ship to the mains spl capabilities. Probably not possible with 1 "tiny" sealed subwoofer so probably not to useful.
This may ruffle some sound peoples feathers Isn't it ideal to have a subwoofer that has what I choose to call Equal Loudness Headroom or say up to 30 or 40dB more available headroom in the 20 or 30Hz range? Of course this is normally impractical.
Regarding the optimal sealed subwoofer configuration for linkwitz transform, isn't a Q of .707 typical.
I think the box you are proposing for the peerless is very Low Q thats a Qtc of .352 , and I don't think the 830450 exists I think you mean the 830500. They don't rate the 830500 power I guess close to the lab12. You will not get -3dB at 39Hz with that driver or any other 12 that I know of without perhaps a stiff SPL penalty.
I think both the Lab 12 driver and the peerless will be closer to -3dB at 50Hz, and the lab is a little more linear in the desired range. The peerless seems to want a much smaller box for a .707 Qtc. and its -3 is in the up mid 70's.
With the peerless in the 1.6cu' box you will exceed xmax at 400 watts at around 35Hz with 108dB output. Your max output at 20Hz will be 99.2dB @ 20Hz @ 170watts, more power than that and you exceed xmax. Which is the same as you can get from a lab 12 in a smaller box at 400watts but you don't have to worry about distortion from gross over excursion.
I guess thats apples and oranges similar output for a box 1.5 times the size, but if someone abuses it could damage the driver mechanically. I'd rather go for 2 Lab 12's in a 2cu' box. Same output + 3dB power headroom. Less easy to dislodge.
I'm convinced that the Lab 12 driver is one of the better options for a sealed subwoofer driver.
I'm building a couple of sealed Quad lab 12's right now 1 standard one push pull. Using 2kW of ICE amps. They will not exceed rated Xmax at 2kW at 1Hz so no worries of mechanical damage only thermal. These cabs need to be ~4cu' for a Qtc of .707 heavy fill.
They should be capable of 111dB @ 20Hz and 117dB @ 30Hz, so with in room gain perhaps have some nice usable headroom.
I'm really excited to see what 99dB @ 10Hz can be like. Probably inaudible but good fun.
Anyhow enough craziness.
How much SPL are you using at mix position? Clean bass headroom good! But who actually has it??? Not many.
Good luck
T-
Post by: Mike {AB} Butler on July 07, 2008, 07:38:59 AM
You say you want 120 Hz Crossover point. In my experience, I would suggest that you may not like the results. You will probably be much happier with 80-90 Hz. At 120 Hz, the midbass "boominess" tends to affect your mix results, so you would be compensating for this effect.. a lot!
Also, be careful when setting up the amplitude response of your system.. Ivan is indeed correct on all that he says.
A $1K budget might be quite a bit light if you don't already have power amps - what were you thinking for these? Maybe buying your DSP and amps used will help the tight budget.
Provided you don't expect super output levels from the system at 20 Hz, you should be in great shape.
Regards,
Post by: George Friedman-Jimenez on July 07, 2008, 09:01:20 AM
Ivan, your point about the pressure zone is well taken. Thank you for reminding me about this concept. The f=c/2L equation for deriving the pressure zone frequency is not very accurate when the walls are nonrigid. Newell's book Recording Studio Design has a good discussion of pressure zones on p. 327-330. The single sheetrock walls that I have are pretty transparent to LF and I think more meaningful dimensions to use in that equation would be the inside dimensions of the cinder block foundation of the house. The studio is in a basement. I will have to measure those dimensions and get back to you. In any case, your point is essentially in favor of the DSP route, with measurement of the room response at mixing position. Direct empirical measurement would be the only way to deal accurately with the complexities of misspecification of mathematical models due to nonrigid walls and ceiling, absorption, etc. Also, I could have multiple settings for flat room response at mixing position, flat speaker response, and best sounding responses for high SPLs, and for different locations in the room (for group listening). With imperfect bass absorption these settings could be significantly different.
Antone, you are right, the Peerless 830450 doesn't exist. I meant to write 830845. The TS parameters for the 830500 (the original driver specified by Linkwitz for his 14x14x22 box) are quite different, and do give a Qtc of 0.35 and f3 of 72 Hz. I chose the 830845 because it gives a much lower f3 in a sealed box, and the Qtc is closer to .707 without stuffing. The SPLs I use for monitoring when editing recordings are generally around 85 dB or less. Looking at the Fletcher Munson curves motivates me to check the mix at higher SPLs as well since the tendency is to overhype the bass when mixing at low SPLs to compensate for the ear's lower sensitivity at low SPL and low frequency. At any rate, I don't expect to use even 1/4 of the power of the PL236 driving the sub, because the power handling of the 830845 is only 175 W RMS @ 8 ohms. Without calculations, I am sure it will be loud enough in this small room.
Mike, regarding crossover, I have Dynaudio BM6 monitors that really don't go much below 100 Hz. Currently I am using BM15s as midbass speakers in addition to the BM6s to fill the gap to my current Event 20/20 subwoofer, and the 120 Hz crossover from the BM6 to the BM15 gives me the flattest response at mix position. I was hoping to be able to dispense with (ie, sell) the BM15s if this sub can go high enough. I have learned not to try to predict things like this, and again the DSP / empirical measurement approach seems the best for optimizing the system.
Silas, why do you say 4 PEQs is not enough? Those are in addition to crossover filters, right? Do you think I will need more than those 4 to flatten the sub response?
Thanks again for all your input, it has been very helpful.
Post by: John Chiara on July 07, 2008, 11:52:12 PM
Post by: George Friedman-Jimenez on July 08, 2008, 01:01:32 AM
Ivan, I measured the inside foundation dimensions of my basement and it is 25' x 28', with a 7' ceiling to the bottom of the 1st floor floor above. That would mean the pressure zone would be below f=c/2L=20Hz, approximately, which is good news for our room gain discussion. Following this logic of the sheetrock walls being transparent to LFs, I might move 1 or 2 of the corner bass traps out to the actual cinderblock corners. This would free up some floor space and allow a slightly more live sound in the room but still attenuate the room modes from the foundation walls. I don't know how the partially rigid sheetrock walls and 2 layer wood floor above will affect the transition frequencies between the pressure zone and the diffraction zone. I will have to play around with the bass traps and an MLS Sweep or other room acoustics program to answer this. I have used ETF/RplusD in the past with an Earthworks TC30 mic and it seems to work well, generates nice gated waterfall plots and LF room response graphcs, but I have no gold standard to compare. Do you think this is adequate or do I really need to get a program like SMAART or SYSTUNE?
Post by: Antone Atmarama Bajor on July 08, 2008, 02:56:23 AM
Of course the explanation is probably as simple as, the average consumer doesn't have accurate sub bass reproduction capability, reasonable listening environment, or even have an awareness of what it really is. So why waste precious head room on something that isn't going to be heard by the average listener.
85dB is right at the threshold of hearing for 20Hz.
Food for thought for Full scale cinema (exhibition houses) the surrounds and mains are mixed at 85dB with 20dB of Headroom and the LFE is mixed at 95dB with 20dB of headroom.(Movies actually take advantage of a large dynamic range compared to modern recorded music.)
Although Cinema X curve is Far from Flat due to perf screen attenuation. They shoot for SPL down to 35Hz and ignore things bellow, and have a 3dB per octave roll of from 2kHz to 8Khz, Khz mix against.
Bag End:
I heard that Ron Wickersham (Wickersham Alembic Basses, and Greatful dead wall of sound fame)designed the ELF circuit, but had a falling out with Bag End, so they can't use his design anymore.
So now Bag End have the Infra system.
I was shocked when I read about the ELF system being a Dual Integrator circuit. An integrator Circuit is an elementary filter circuit with a 6dB per octave slope. A dual integrator is nothing more than 2 of these circuits in series. Probably the most complicated part of the circuit is the limiter, changing where the slope lines up with the box.
I made an op amp circuit that sums a "dual integrator" with a High pass filter. I think the Linkwitz transform is actually a little simpler to build. You may just want to build a simple Linkwitz transform for your sub, using the linkwitz lab spread sheet. And put it in between the sub amp and the xover. You may not have enough gain available in DSP unless you attenuate everything above the sub knee. Hopefully you can work everything out in the wash.
Ahh Bass.
Good luck.
Post by: George Friedman-Jimenez on July 08, 2008, 09:05:12 AM
The effect of reduced perception of bass, especially at low SPLs, is huge. The Fletcher Munson curves are actually conservative in predicting the perceived effect. More recent curves, the ISO Equal Loudness Curves, are apparently more accurate than the Fletcher Munson curves and are more extreme in their predictions. Wikipedia has an internet accessible graph that compares the two, here: http://en.wikipedia.org/wiki/Fletcher%E2%80%93Munson_curves
As an example, for a 1000 Hz tone, 100 phon is the perceived loudness of a 100 dB SPL tone. The perceived loudness of a 40 Hz, 100 dB SPL tone is around 70 phon. At 20 Hz, your 100 dB SPL tone is perceived as having equal loudness as a 40 phon tone at 1000 Hz.
This has big implications for sound production. To produce a 40 Hz tone that is perceived as 100 phon requires about 117 dB SPL, interpolating from the graph. For a 20 Hz tone to be perceived at 100 phon requires 128 dB SPL. The situation gets even worse at lower SPLs. For a 40 Hz tone to be perceived at 80 phon requires about 106 dB SPL. For a 20 Hz tone to be perceived at 80 phon requires about 120 db SPL. The threshold of hearing for a 1000 Hz tone is at 0 phon but for a 20 Hz tone, the threshold of hearing is at 75 phon and at 50 phon for 40 Hz!
These perceived loudness curves are averages over small groups of study subjects that were included in the research studies, and I don't know how much variability of these parameters exists in the general population.
For this reason, seeking a subwoofer that is "flat" in its dB SPL frequency response seems very arbitrary, and it does not come close to ensuring that the perceived loudness is equal at different frequencies. A "flat" frequency response based on perceived loudness would sound massively different than what we are used to, aside from being technically unfeasible for bass frequencies. For mixing music, however, a "flat" frequency response in SPL gives us some basis to judge when the bass notes would sound softer or louder than the real life sound that was recorded, assuming the frequency response of the listener's stereo system or PA is also flat. To the extent that most people's stereos are not flat, our mixes will sound different on those systems, but there is nothing we can do about that. Also, some people choose to crank the bass and others cut it down.
So do you think that for purposes of monitoring recordings striving for a natural sound similar to the real acoustic sound, the small sealed sub I am proposing will be adequate?
If I need another 6 dB, I could build 2. I even thought of building 2 boxes 14" x 14" x 34" long, putting feet on the driver end and using them vertically as combined downfiring subs and speaker stands! Also, I already have the Peerless driver but the LAB12 driver has similar TS parameters and would give a similar response in the same box, with slightly higher sensitivity and double the power handling capability. I just don't know how clean (distortion free) the sound of the LAB12 would be between 20-40 Hz compared with the Peerless.
Post by: Antone Atmarama Bajor on July 08, 2008, 04:56:16 PM
I have heard from an old engineer friend of mine that there is a trend almost a convention in the recording studios to EQ the LF in the monitors about 6dB Hot starting at either 80-100Hz. I think some of this convention was started to help make sure tracks were recorded bass light to prevent to much bass to be put on a vinyl. I also heard that it was sort of held on to because of the popularity of "Bass Boost" Circuits in consumer audio. Because that slope is closer to what the average boom box owner is listening too.
I'm sure you've taken an FFT of your average music track and seen the spectral density in the LF I think on the average recording there is probably hotter 20dB+ in the 80Hz-100Hz range vs anywhere else in the spectrum, so if your 85dB spl mixing (That would be your spectral sum) You are probably using most of your headroom in the bass area.
Since you are trying to preserve the natural characteristics of the percussion perhaps take a profile of the drums spectrum with fft. Then you can determine what sort of headroom in the range you are operating is required. Not that Natural is often the desired in produced music. If you have a huge 15Hz transient from a particular piece of percussion, which eats massive amounts of headroom, are you going to go to extraordinary lengths to faithfully reproduce it?
I have built a Quad Lab12 and pushed them with about 3kW with some Car sub stress test tracks like "Bass I love you" it took quite a bit of Hype to make the tracks sound pleasing to me. But they were pretty clean. I may have been getting a bit of extra 2nd harmonic On the 6Hz tone I think maybe I heard a little mechanical noise. My top speakers were little Blue sky monitors so for my subjective taste the sub bass balance was just about right.
I think the Lab 12 in a smaller box would have less distortion due to better cone control. You could have similar response in the Peerless, in a similarly small box, but your RMS power handling is less than 1/2. And again in ~2cu' you could have a pair in push pull.
Flat bass low is possible with all of the modern tools but normally only with a limited power bandwidth.
Any engineers here feel free to chime in, I really am just a tourist in this world of sound reproduction.
Good luck-
Antone-
Post by: John Chiara on July 08, 2008, 11:40:36 PM
| George Friedman-Jimenez wrote on Tue, 08 July 2008 01:01 |
Thanks, John for the tip. I looked at the Bag End website and the Infra MXB is their balanced I/O subwoofer processor that has replaced the ELF Integrator. It seems to be limited to a 12 dB per octave slope, which would work better for some sealed driver/box combinations than others, depending on the TS parameters, box size, stuffing, etc. Also, the MSRP for the Infra MXB is over $1k (I didn't find a street price on the web), while the Driverack 260 on the web is around $800, so I don't think the cost would be substantially different. The Bag End processor also seems much less flexible and adaptable than a DSP unit like the Driverack 260 or the Xilica 3060, which have crossovers, many different filter options, and delays. The disclaimer on the Bag End website even says not to use with subs other than the Bag End Infra series, although that is probably CYA language for folks who aren't paying attention to Xmax. |
The thing is..the ELF unit is designed to do what you want to do. I have a homemade sealed box and it works fine. Getting enough level for rock shows is gonna be tough. It sounds like you want to reinvent the wheel. They already figured it out..and it works. Unless you have a lot of free time...
Post by: George Friedman-Jimenez on July 09, 2008, 01:05:15 AM
I am not clear on how adjustable the filters on the ELF are, were you able to adjust your homemade sub for a flat frequency response? What are your sub driver, box dimensions, and stuffing?
I am definitely not in favor of reinventing the wheel, but for a homebuilt sub, I would have to do all the frequency response measurements, filter adjustments, crossover adjustments, optimization of physical placement, and level adjustments from scratch with the Bag End ELF anyway. Would it really take that much more time and effort to do it with a generic DSP like the Xilica or the Driverack?
I don't think there will be much cost difference, the new model Bag End Integrator looks like it sells for almost $1k. Do they still sell the original ELF unit? I couldn't find it on the web.
Post by: John Chiara on July 09, 2008, 01:23:36 AM
Good Luck whatever you settle on.
Post by: Silas Pradetto on July 09, 2008, 09:17:43 AM
Post by: Ivan Beaver on July 09, 2008, 09:55:01 PM
| Silas Pradetto wrote on Wed, 09 July 2008 09:17 |
Ivan stated that the Xilica's 6 PEQ's would probably be enough. Based on that, 4 wouldn't be enough, so I wanted to mention that. Since the PEQ's can also be shelving EQ's, I'm sure you could make do with 4 or 6. Actually, the less you actually use the better. |
I didn't say it would take 6, only that 6 were available IF NEEDED.
I do agree that less is often more in audio.