Absolutely love the discussion! I was hoping for these types of challenges earlier on, but glad that now that it's starting to become more AVS mainstream, these discussions are happening. Love it, and appreciate the thoughtful insight.
Originally Posted by neutro
I have a few comments about the score by the way. I totally understand that the score is not perfect and is only a quick figure that help us compare setups.
The main question I have is that the score is normalized by room volume. It's intuitive that the same gear in a smaller room will result in more SPL. But is the effect directly proportional to volume? I.e. is the same setup placed in a room half the volume automatically twice louder (i.e. +6 dB)? I understand specifics like materials, furniture and room shape can also affect levels at the main listening position, so let's assume an empty rectangular room.
The part about normalizing to a popular sub (the sealed SI) is a good idea too, yet it's still a bit abstract as I'm pretty sure the majority of us here have never seen nor heard that idealized sub. So while it gives us an idea (wow, didn't now that it would take 3 sealed SI's to hit the 16 Hz level I get with my two ported 12"), the score could also be expressed as cu. ft. per max burst dB SPL at a given frequency. Or, if we want a score that goes *up* when it's better, simply the inverse, i.e. max burst dB SPL (or equivalent SI's) per cu ft. Anyway, the stars is the final score and those can be computed in cu. ft. per equivalent SI, or in max burst dB SPL per cu. ft. ... the conversion is direct anyway.
Also there is no reason to limit this to the ULF portion, unless perhaps the room size has much less importance for mid-bass and there would be no reason to normalize room size?
Great point Neutro, and you're right. I could have done just volume per max burst. The reason why I decided to use SI, was what you've already eluded to above: Physical 18in drivers where we have pictures, people own them, etc. are much more 'relatable' and intuitive compared to decibels. When comparing from room to room, I feel we can better relate to physical objects than sound pressure levels. It's much more meaningful and comparable to the average joe that in his 1000cf room he has 1 18in sub compared to the same 18in sub in a 3000cf room. That intuitively makes more sense that saying his sub in a 1000cf room can reach 89.2db compared to that same sub in a 3000cf room can reach 89.2db.
Either metric could have been used, I just felt converting it to a sub would help visualize things better, and ultimately easier to understand.
As for using the SI? Not going to lie...the acronym really worked with its brevity. Additionally though, it is popular and some of the most well know HT systems use them. So, by using the SI as the baseline, again folks could ultimately compare to those uber AVS HT systems (yes, I'm talking about you pop). The point is, it doesn't matter what sub is used to convert to because all it is doing is setting a frame of reference or baseline to compare with others.
Originally Posted by ironhead1230
I was just working on posting something very similar. This thread is being referenced a lot now and some are using the scores and output estimates as absolutes. Comparing subs down to the db. Until now, there has really been no testing to see if these methods are valid and if room volume alone makes as large an impact as reflected in the ratings. I think its time we attempt to verify these scores using speclab captures. We could try and pick a clip or two that highlight the different frequency scores and see what people's setups actually do during playback. I started a thread awhile ago to try and put a procedure together to take mic'd speclab captures, but nobody else seemed very interested. I recently found my first attempts with speclab from the PA sub GTG last year. Looking through them again, I think it is worthwhile to revisit.
I definitely think it's worth a revisit! I too want to verify these scores...
Originally Posted by neutro
Ultimately the score can be replaced by max sweep frequency measurements at your listening position. Only then will you have the whole story, and if you already have the gear to take accurate speclab measurements, then I guess you can also load REW and perform max output sweeps.
However, there are drawbacks to that. First, it's pretty involved, and close-to-max output FR measurements at the MLP are most often already posted by the guys with the 5-star setups
It's pretty far from the quick back of the envelope type estimate that the score provides. Second, some people (me included) can't really do a max sweep. Third, if the goal is to condense your setup in a single metric, you still have to convert the FR into a score / stars. Fourth, the most interesting / funny part of the ULF score cannot be achieved by measurements: it's the what-if! What if I had four JTR Captivators in my 1680 cu ft room?
That spreadsheet gives us a quick glance at what can be achieved for us who are contemplating an upgrade. (Disclaimer: no I'm not.)
Still, I totally agree with you about at least verifying the effect of room volume. It is the major driver in the current ULF score. What if SPL varies with the square root of room volume instead? That changes everything. I for one would love experimental confirmation but I'd settle for a theoretical argument from acoustics and physics.
It's just that subwoofer don't act like pistons. I agree that for a given piston displacement, in a static situation, half the room volume will yield double the pressure level induced by the piston displacement inside the room. But our rooms are not perfectly sealed, and the sub's piston displacement is actually negligible compared to the total room volume, as evidenced by the fact that all our subs have pretty poor performances at 0 Hz. (Unlike rotary subwoofers
by the way...
) So pressure level achieved by our subs is a dynamic phenomenon and linked to the acoustical energy pumped by the driver. How that is affected by room volume is not straightforward for me.
Yes, that is the challenge. We need some test subjects in each star category...folks that are willing to take some max sweeps and ensuring we all follow a repeatable process to limit variation. Again, not an easy ask IMO...I am personally hesitant to run a max sweep because I don't know REW, and I don't want to blow up my subs...
Originally Posted by neutro
Of course the room has a tremendous importance. We are just arguing about the specifics. I don't know enough about acoustics to have a clear idea. As I see it, there are two factors in play. The first is distance: sound pressure drops as the inverse of the distance. If you're twice farther, SPL drops by half (-6 dB). But that's not all, since if that were the case, the distance to MLP would be the main driver. The other factor is reflections on the wall, ceiling and floor surfaces which has a reinforcing effect. This is complicated by the resonances and room modes a bit, but if a wall is farther, then the reflection will also come from further. Perhaps this is the basis for the volume normalization.
Then again, you could have a room of infinite volume (e.g. parking lot) and still have some great ULF, whereas this ULF score would simply be infinite cu. ft per SI equivalent (0 star). This is just to show how things break down if you push the idea to the limit, while still acknowledging that this is not an interesting case (unless you have an outdoor home theater I guess). On the other side, if the volume is sufficiently small, you could get a very large ULF score. Get in a 200 cu. ft. closet with an SB12 and there you go, 4.5-star setup. I wonder if that would really give me reference levels @ 12.5 Hz? Again, this is not a very interesting scenario, so the direct proportionality with volume might still be warranted for most home theater rooms.
Another great point...certainly outlier situations in practice, but I do realize your point was illustrative. The Star rating should ideally be bell shaped curves and HT systems would fit within certain standard deviations....meaning if your score lands with a given star range, we should have some confidence level that you're in a certain SPL range.
Originally Posted by neutro
Well if we neglect standing waves (room modes) for now, the effect of boundaries in a room is to reflect sound waves. As such, if the reflection is perfect (no absorption), then it is as if there were mirror images of your sub in imaginary rooms on each side (and up and down) of an hypothetical rectangular room. The first reflection then adds 6 subs at various distances, and SPL decreases as the inverse of the distance. Double reflections will add more subs at greater distances, etc. The sum of all possible reflections (which are infinite), weighted down by the path distances and any reflection loss, is the level provided by the sub inside the room.
So right there we see that room size has an impact, as the "images" will be further apart in a larger room. However, one would expect the effect to be proportional to the inverse of the average room length (i.e. cube root of the room volume). So perhaps the ULF score should be cubic root of room volume / SI equivalent. I think it just doesn't matter too much, because most rooms are 7 to 9 ft high anyway (i.e. not much variations in height), and overall in a rather small range of volume that is mainly affected by the length of the average dimension. Also, the focus is more on the number of stars than the actual ULF score. The root cube of the volume would not change much -- just the limits of the bins for the various stars.
IMO, this is what we need to hash out more.
The score itself is factual. It is
That is a metric, and it can be compared with others. It's the interpretation and categorization where we need to hone in on.
The above chart is what we need to figure out and potentially change. IMO, the ULF scores can stay the same. What should be the ranges and how does that relate to reference output?
The output estimate started based some very good data that Bosso posted here. He basically stated that in his experience, the recipe for reference output was a sealed design where you would have .01 Liters of displacement per cubic foot. So for 2000 cubic feet, you would need 20 liters of displacement. 4 SI is equal to 21.2 liters of displacement. 2000cf / 4 SI = 500 ULF score....around the middle of the range. I created the categories before relating his post, so it confirmed I was in the general ballpark.
If we can keep the score the way it is, it will keep it simple (the goal of this metric). I think we can just put more science around the above chart to determine how close it is, and what needs to change.