Differences between SVS and Rythmik

I'm trying to get a good understanding of why many reviewers think Rythmik subs sound better, cleaner, or better with music than other competitors. One such competitor is SVS. I am just comparing two subs that are very close in performance and price. Like the SVS PC 13 Ultra and the Rythmik FV15 HP. Both subs were extensively measured on data-bass by Josh Ricci. (Technically Josh measured the PB 13 Ultra, but according to SVS the performance of the PC 13 Ultra is basically identical.)

In most regards their performance is very similar. They both have multiple tuning options. They both have very similar long term output compression curves, keeping flat frequency response as the output level is increased until the sub reaches the limit and eventually it just stops playing louder (in the case of SVS) or it stops putting out the lower frequencies (in the case of Rythmik). The transient response of both subs are very close to one another looking at the waterfall and group delay plots. But then we come to distortion. Both subs don't seem to have any higher order harmonic components of distortion, when the subs are pushed hard. It is only 2nd, 3rd and 4th order harmonics present in the plots. But, the SVS sub has very low distortion. Even when pushed to about 118 dB output level (at 50 Hz) it still manages to keep less than 10% THD from about 22 Hz and up. However the Rythmik sub on the 115 dB sweep from 35 Hz and below has quite large THD in comparison. The only slight weakness that I can see of the SVS sub is that the 3rd order harmonic is the largest component of the distortion rather than the more pleasant sounding 2nd order harmonic. However, this doesn't seem like it would matter since the SVS sub has such low THD to begin with, even when pushed hard and then it just stops playing louder rather than letting THD get larger than about 10%.

I'm not trolling trying to pick a fight. I just want to understand what I am missing. One of the reasons I gave my Sierra 2s an audition in the first place was because the measurements seem to indicate that it truly is an exceptional speaker. After listening to them compared to a couple other close competitors, they did in fact sound more realistic and lifelike in many ways. However, with subwoofers of this size there are more substantial costs involved with demoing them in my own room. I'm hoping that I just purchase a pair of subwoofers and be done with it.

So why are Rythmik subs praised so much for their sound quality, in comparison to most of the other competitors, like SVS for example? Is there more than all of the aforementioned quantities that drastically influence the "sound" of a subwoofer? Is it that some things just can't be measured?

Quote:

---

Originally Posted by N Boros

I'm trying to get a good understanding of why many reviewers think Rythmik subs sound better, cleaner, or better with music than other competitors. One such competitor is SVS.

---

I can add one more here from Ed who worked at Apple as a hardware team director. Before he purchased our subs, he called me asking why our servo is better. I explain everything and he understand those principles. But the final test is still in listening test and here is his feedback.

Hi Brian,

Thanks for checking in. Yes! More than meets, it *exceeds* my expectations. Nicely done!!

The F15HP is remarkably cleaner and more satisfying than anything else I've heard!

I can hear more detail & texture to the low bass, plus, I've never experience LF extension like this in a home system before. Also I feel as if, somehow, my system has much greater overall dynamic range. :-D

FYI, I did pick up a SBxx-U for comparo... it was too easy not to, since there's a Magnolia center just a couple of blocks from me. Good sub, though it lacked LF extension in most scenarios, and was far more distorted too (both audibly & measured with spectrum analyzer + cal'd mic). The SBxx-U has since gone back: you win!

I spent an awful lot of time optimizing setup for each sub: I tried countless placement & tuning options. After I get a new pre-pro (or maybe receiver), I'll decide whether I want a 2nd instance of F15HP. The 2nd sub doesn't seem warranted strictly for the sake of SPL; it'd be more for the purpose of calming the room modes.

Some constructive feedback for you...
The 2nd instance of PEQ on the SBxx-U was quite useful. On the other hand, I appreciate being better able to control the frequency & Q with higher resolution on the F15HP: narrow peaks (at e.g., 42Hz or 56Hz) that don't coincide with their predetermined set of center frequencies (40/50/63Hz) were hard to tame without negatively impacting adjacent frequencies.
Another feature I'd recommend to enhance your offering: a better way to remotely control "modes" for the sub: it's quite frequent that I'd switch between "movie mode" (20 or 28Hz + low damp + rumble filt) and "audio mode" (14/high/off). I found that I just leave it in music mode now for all but the most exciting, effects-laden movies, since I find myself forgetting to switch or to switch back... honestly it's the last thing I want to get up for (or even think about, for that matter) when it finally comes time to relax & enjoy.

I've been meaning to write you & give some kudos, though I have been extremely busy growing a new team at Apple. I do prototyping there (hardware systems, though not audio). I find a whole lot of joy coming home to enjoy my audio system after a long day: that joy has been increased quite a lot with the addition of your sub.


Thanks for your efforts... keep up the great work!

The only change I made is the model number from SVS.

Quote:

---

But, the SVS sub has very low distortion. Even when pushed to about 118 dB output level (at 50 Hz) it still manages to keep less than 10% THD from about 22 Hz and up. However the Rythmik sub on the 115 dB sweep from 35 Hz and below has quite large THD in comparison. The only slight weakness that I can see of the SVS sub is that the 3rd order harmonic is the largest component of the distortion rather than the more pleasant sounding 2nd order harmonic. However, this doesn't seem like it would matter since the SVS sub has such low THD to begin with, even when pushed hard and then it just stops playing louder rather than letting THD get larger than about 10%.

---

PB13U above 22hz can never hit 118db if you read the max output SPL numbers. They are at least 3db lower than ours above 30hz. So where do you get the conclusion that they can make the harmonic distortion test at 118db, but not at the max SPL output, won't you think so? The fact is you have read the chart wrong and the better part is you are not the only one (and that unfortunately is bad for us, wouldn't you agree?). Many other people also think the same as you do. But are they right? The answer is even though the curve is labelled as 118db, it is not 118db anywhere. Second, do you know about amplitude modulation (AM)? AM does not generate harmonic distortion, but it generates intermodulation distortion. Is intermodulation distortion better or worse than harmonic distortion? Compression causes AM distortion. At 118db, PB13U has 6db compression. That is like 50% amplitude modulation. So the correct way to read the chart is harmonic distortion is only relevant when the compression (AM distortion) is low. If AM distorton is high, the harmonic distortion is less meaningful.

Quote:

---

Originally Posted by RythmikAudio

PB13U above 22hz can never hit 118db if you read the max output SPL numbers. They are at least 3db lower than ours above 30hz. So where do you get the conclusion that they can make the harmonic distortion test at 118db, but not at the max SPL output, won't you think so?

---

You are absolutely right, I did misread it, since Josh labeled them in kind of a strange way on the SVS sub. I think what we see that just slightly under 115 dB the SVS was designed to just stop playing much louder at all. I think Josh is trying to indicate that if more output is asked for (say 3dB), then it just doesn't really output much more. Fortunately on the long term output compression curves you can move the cursor of the mouse over it to get the actual number.

Okay so let me say this again, comparing 110 dB sweeps (in 20 Hz modes for both subs) we can see that the SVS sub has about half the THD from 20 to 35 Hz than the Rythmik. However, both are (basically) still under 10%. Who knows if such a difference would be audible. As the input signal asks for more 5dB output from the two subs, the Rythmik rises to the challenge except for at about 25 Hz and below where it doesn't have much more to give. The SVS can just about do it at 45 Hz, but going lower or higher in frequencies it starts to fall off and not really be able to output much more. It isn't that it can't output much more, but it can't do it with the distortion levels that the designer wanted, so it just doesn't output what was called for. The price that was paid for asking extra output in the Rythmik sub is that the distortion on this sweep rises dramatically from 30 Hz and below. I see, so the two are actually quite closer than I was thinking AND the Rythmik is a lower price.

Now I just need to wrap my head around AM distortion. I've never learned out this. I wonder why Josh doesn't measure this as well? So are you saying that this can likely give us a difference in how they sound and even one maybe having more detail than the other?

Quote:

---

Originally Posted by RythmikAudio

Second, do you know about amplitude modulation (AM)? AM does not generate harmonic distortion, but it generates intermodulation distortion. Is intermodulation distortion better or worse than harmonic distortion? Compression causes AM distortion. At 118db, PB13U has 6db compression. That is like 50% amplitude modulation. So the correct way to read the chart is harmonic distortion is only relevant when the compression (AM distortion) is low. If AM distortion is high, the harmonic distortion is less meaningful.

---

I understand what you are saying now. The SVS sub looks so neat and clean as far as THD is concerned, even when it is asked to push 118 dB. However, as it is being pushed like this there is a quite a bit of compression going on, to where the THD becomes less important. The AM distortion is going to dominate what we hear. If we look at the 115 dB sweep (even though the SVS isn't actually putting out 115 dB everywhere, whereas the Rythmik is from 30 Hz and up) the SVS is actually compressing quite a bit, looking at the long term output compression curves. So it is in fact the Rythmik that is cleaner from 30 Hz and up, even though the SVS looks like it is from the THD curves. From about 23 Hz to 30 Hz and up the Rythmik sub is actually outputting more on the 115 dB and it shows slightly more output compression than the SVS. Actually, if we were to level match the two I bet they would be nearly equal in output compression below 30 Hz. Summing up, it seems that when the AM distortion sets and becomes dominant below 30 Hz, both subs perform equally as well. But, above 23 Hz the Rythmik seems to have more output and above 30 Hz the Rythmik has much less AM distortion, which will likely lead to this "cleaner" sound. Add on the fact that the Rythmik sub is cheaper than the PC 13 Ultra, makes it even more exciting.

The long term output compression curves I didn't really think had much meaning before, except for showing that the subwoofer is running up close to (or past) its limits in either driver excursion or amplifier power for the box it was built in. Thanks for helping me understand this new layer of meaning in these curves!

I know I am in the minority. But when I look at speaker drivers, I think of their mathematical modeling (even without hearing the sound). In math, an important concept is continuity which means you cannot go from point A to point B without visiting the points in between. So you cannot go to 115db without going to 70db first, then 80db, then 90db, and etc. We often call the opposite, or discontinuity as singularity. Another concept relevant to modeling is curve fitting. One can do a polynomial fitting of any mathematical curve. When the number of terms expressed in polynomial is small, it is relatively easy for us to understand how it behaves. This is understandable. We understand rational numbers before we understand irrational numbers. For instance, 1/2 is a rational number, but square root of 2 is an irrational number and so is pi. These irrational numbers can be expressed as sum of infinite sequence and that is why you cannot clearly express these numbers in finite numbers of digits.

We can apply the same concept to distortion modeling. If the speaker has linear distortion, the 2nd harmonic should double in percentage (2x), when the output increases by 2x (or 6db). In other words, the strength of distortion quadruples when the output is only doubled. The 2nd harmonic should quadruple in percentage when the output has increased by 4x. In other words, the strength of distortion increases by 16x when the output only increases by 4x. So you follow the idea. 2nd harmonic distortion is represented by the lowest order term in polynomial modeling. And for 3rd harmonic distortion, it increases by 4x when the output has increased by 2x (or 6db). In other words, the strength of distortion increases by 8x when the output only increases by 2x. The 3rd harmonic is represented by the second lowest order term in polynomial modeling and it grows rapidly. You can infer the rest. Granted, the higher order the harmonic, the higher order of polynomial it represents and fast it grows when output increases.

The distortion at a single SPL point is not as relevant as how we connect the distortion numbers at these various SPL points and compare them to the above modeling. So for instance, if the 2nd harmonic distortion is very very small and suddenly increases by 4x at 100db, it is not as good as another case where it strictly follows what I describe above on how 2nd harmonic distortion should behave. This is because in the former case, when we connect all these various SPL points, the distortion model cannot fit a polynomial with only few terms. It is almost as if it has hit a singular point. You won't find this singular point until you do a "curve fitting" connecting the harmonic distortion at various SPL level together. Now note, all the study on audibillity of harmonic distortion were done with linear model because the researchers in the old days do not differentiate linear distortion vs non-linear distortion. So all of synthesized harmonic distortions can be well curve fitted into very simple polymonials. In other words, they don't know how to synthesize the type of distortion the drivers can generate.

Now you can extend this idea to compression, where no polynomial models exist, and realize how important it is to maintain zero compression. It is again through the connection of how speaker drivers behavior at all these various SPL output points and relate them. But here we need to introduce the concept of time. So why is "time" important in the understanding of compression (which is a form of distortion)? I will borrow a knowledge that we as human offer proof for the existence of dark matters: by looking at series of photos of the night sky over different time instances and found the location of other galaxies wobble (because invisible gravity effect between us and the galaxies). The same concept can apply to compression. It is not the static compression that is important or interesting. It is how the compression changes over time (which becomes a memory effect) that is important and interesting. If we just look at a particular time instance by itself, we cannot even define compression. We want the least amount of change in distortion characteristic over time (like we wish there were no dark matters). If you ever hear a steady 20hz tone, you know it is annoying. It is not something we will hear as music. Music is dynamic. It is how they are connected over time that is just as important as, if not more important than, the harmonic contents at individual time instance.

Now we can also apply this time changing behavior to distortion that is time-variant and time-invariant. The latter is much benign than the former. So the final note. The way we currently measure distortion is like our understanding the cosmos very very long time ago (when no concept of time is introduced). We really haven't advance as much. But why we won't spend time and money to advance that? It is not where the research fund or money is and also because it is more complicated than most of us want to understand. For instance, knowing the distortion at various SPL level is only the first dimension. How to connect them at various SPL level (how distortion can be modeled) is the 2nd dimension. And change over time is the 3rd dimenson. In the least, we should understand the distortion number is a composite of distortion from various sources (for instance, the nonlinearity of spider has a different distortion characteristic than the one from motor nonlinearity). Some are more benign than others. But our current distortion measurement procedure cannot even separate them. So let us not read the number at face value.

Wow, that was an incredibly interesting reading. Lots to digest, good thing is the beginning of the weekend, thanks for sharing.

Later today I am most likely going to place and order for an L12. :-)

I also need to add that when you do curve fitting, the more SPL snapshots points we have, the more clear picture we have for the distortion charactersitc. I think that is easy to understand.

Time-coherent response is very important. This is because what we hear is a mix of direct wave and wall reflection, small signals mixed with large signals, ..etc. Therefore it is a mix of energy generated at various time instances and various signal strength. If sub outputs don't follow time coherent characteristic, it is not easy for our brain to connect the dots. The impression customers have comparing our subs to other subs is our subs are more dynamic and also cleaner. This is because with other subs, it sounds like there are lots of busy background noises. But in our subs, those background noises seem to disappear. But in reality, the energy is still there. But becasue our subs are more time-coherent, each energy component falls into the right place when our brain connects them. As a result, we don't hear them as background noise.