• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

SVS Ultra Bookshelf Speaker Review

Dennis Murphy

Major Contributor
Technical Expert
Joined
Mar 17, 2020
Messages
1,071
Likes
4,547
He did not recomend them, and that is his choice. Yes he is clear about that. And yes I do not think that detracts anything from subjectivity except to say that at least this subjectivists is honest even when it hurts his subjectivity itself as a concept.

I just finished modding the Ultra's to see if I could get at what was bothering Amir. There are two issues that may have contributed to the perceived brightness. One, which doesn't show up on Amir's measurements due, I think, to the protective device he uses on the mic, is that the tweeter takes off like a rocket above about 14 kHz. I don't think I can hear that, but maybe Amir's ears are less battered than mine. The other is that SVS kind of fudged the tweeter circuit. There is a big hump in the raw tweeter response at the low end (which is very common in baffles this size), and the crossover doesn't attempt to achieve an accurate 4th order acoustic roll-off. The tweeter response instead levels off on the way down, but still sums fairly flat with the woofer. Maybe it's the greater-than-normal tweeter contribution in the 2-3 kHz range that Amir is picking up on.

Frankly, I didn't find the Ultra's particularly bright. I was just bothered by a lack of focus in the mid-treble due to the U-shaped response and perhaps the less-than-precise roll-offs. Here are the before and after on-axis plots. I deliberately ramped the tweeter response down a little just to avoid any remaining brightness that people might hear.
SVS Stock On Axis.png
SVS Mod On Axis.png
 

KaiserSoze

Addicted to Fun and Learning
Joined
Jun 8, 2020
Messages
699
Likes
592
The previous broad/shallow dip centered around 5 kHz is now much less conspicuous due to your having lowered the tweeter's response at 2 kHz and lower. 2 kHz is probably high enough in frequency for this to be a reasonable hypothesis for the exaggerated treble Amir heard, but there is also another good hypothesis which has been brought up, i.e., the ringing centered at, oddly, the same region where the dip is evident in the mid-treble, 5 kHz. This is evident in Amir's CSD graph, and I think it raises a couple of interesting questions. I've observed people say in essence that unless you see a localized peak in the response, there shouldn't be any ringing. The rationale for this idea is that ringing implies a localized high Q, which ought to show up as a peak in the response. While this seems to sort make sense, the numerous CSD plots I've looked at over the years say otherwise. Amir's CSD plot for this speaker is a perfect example: there is clearly ringing in the region of 5 kHz, where there is also a dip in the response.

In the CSD plot, we see the indication of persistent energy emission at 5 kHz, indicating a moderately strong, highly undamped resonance of unknown origin. It seems reasonable to hypothesize that this is the manifestation of breakup of the woofer cone. If so, it would likely occur to a moderate extent even when the woofer is being stimulated by frequencies much lower than 5 kHz. According to SVS the woofer is made of fiberglass, but this doesn't mean that cone breakup wouldn't occur, or that if it did, the intrinsic damping of the fiberglass would be strong enough to suppress the resonance. What you did is unquestionably useful, but it would be nice if someone could identify the source of that anomaly in the CSD at 5 kHz and make a determination as to whether it is likely to be audible.
 

escott82

Member
Joined
Aug 26, 2020
Messages
36
Likes
4
The main advantage of the midrange is that it keeps the two woofers from operating so high in frequency that they will interfere with each other through much of their operating range. This matters only for listeners who are closer to one woofer than to the other one. When used as a center channel, it matters for listeners who are located off to one side. This is less important when the speaker is stood on one end and used as a stereo speaker, because then it would matter mainly for listeners sitting on the floor close to the speakers, or standing above the height of the speaker and right up close. Thus, the benefit of the midrange is not as strong in this configuration as in the center-channel configuration. But if you buy one with a midrange (so that when used as a center channel the interference between the two woofers will be avoided for listeners horizontally off-axis) and you have any expectation of using it as a stereo speaker with it stood on one end, then in this case I would very much recommend choosing one where the tweeter and midrange are mounted on a plate that can be rotated. Otherwise, when you stand it on end, the midrange and tweeter will be side-by-side, which means that the interference between them, at and near the crossover frequency (typically around 2.5 kHz), will affect the sound quality for any listener who is laterally off-axis by more than a small amount. If you toe them in so that they both point to the center of the room and if everyone sits close to the center of the room it probably won't matter much, because in this case the tweeter and midrange will be in phase notwithstanding that they are side-by-side.
I’ve been thinking about building them with the peerless drivers. Using two 6.5” for the center and the left/right still using that same 3 way design with the 5.25” mid but only 1 6.5” woofer.
To me the tweeter in the svs and the crossover design are both the weak links. Looking into getting ribbon tweeters for the other drivers but not sure which.
thanks for your explanation I wasn’t sure not being able to flip them would have a bad effect on frequency response.
 

Dennis Murphy

Major Contributor
Technical Expert
Joined
Mar 17, 2020
Messages
1,071
Likes
4,547
The previous broad/shallow dip centered around 5 kHz is now much less conspicuous due to your having lowered the tweeter's response at 2 kHz and lower. 2 kHz is probably high enough in frequency for this to be a reasonable hypothesis for the exaggerated treble Amir heard, but there is also another good hypothesis which has been brought up, i.e., the ringing centered at, oddly, the same region where the dip is evident in the mid-treble, 5 kHz. This is evident in Amir's CSD graph, and I think it raises a couple of interesting questions. I've observed people say in essence that unless you see a localized peak in the response, there shouldn't be any ringing. The rationale for this idea is that ringing implies a localized high Q, which ought to show up as a peak in the response. While this seems to sort make sense, the numerous CSD plots I've looked at over the years say otherwise. Amir's CSD plot for this speaker is a perfect example: there is clearly ringing in the region of 5 kHz, where there is also a dip in the response.

In the CSD plot, we see the indication of persistent energy emission at 5 kHz, indicating a moderately strong, highly undamped resonance of unknown origin. It seems reasonable to hypothesize that this is the manifestation of breakup of the woofer cone. If so, it would likely occur to a moderate extent even when the woofer is being stimulated by frequencies much lower than 5 kHz. According to SVS the woofer is made of fiberglass, but this doesn't mean that cone breakup wouldn't occur, or that if it did, the intrinsic damping of the fiberglass would be strong enough to suppress the resonance. What you did is unquestionably useful, but it would be nice if someone could identify the source of that anomaly in the CSD at 5 kHz and make a determination as to whether it is likely to be audible.

I think you're on to something there. See graphs below. The first is the unfiltered response of the Ultra woofer as measured on the tweeter axis. The breakup peak is quite evident around 5 kHz. The second plot shows the individual woofer and tweeter responses with the stock crossover. Although the 5k peak is no longer visible as such, the woofer response in that area isn't suppressed very far. In the last plot, it can be seen that the problem area near 5k is further down with the modded crossover.

SVS Woofer Raw Response.png
Stock SVS Driver Response.png
SVS Mod Driver Response.png
 

KaiserSoze

Addicted to Fun and Learning
Joined
Jun 8, 2020
Messages
699
Likes
592
I think you're on to something there. See graphs below. The first is the unfiltered response of the Ultra woofer as measured on the tweeter axis. The breakup peak is quite evident around 5 kHz. The second plot shows the individual woofer and tweeter responses with the stock crossover. Although the 5k peak is no longer visible as such, the woofer response in that area isn't suppressed very far. In the last plot, it can be seen that the problem area near 5k is further down with the modded crossover.

View attachment 82407View attachment 82408View attachment 82409

These are very nice. The first one clearly reveals the cone breakup at 5 kHz, so there can be little doubt that this is the cause of the ringing seen in Amir's CSD graph. We still don't know whether or to what extent this was the source of the high frequency sound that annoyed him, but chances are good, I think, that this is at least a strong contributor to that problem. Your third graph looks much more like a proper crossover, especially with respect to the tweeter, compared to the second graph which is the individual responses with the original crossover. It looks to me like you did a very good job there. I'm betting that if Amir gets a chance to listen to your modded version that he'll like it a lot better. One thing, though, when I look at that unfiltered response of the woofer, that thing is butt-ugly. Even if all that stuff below 300 Hz happens to be measurement artifacts or maybe room effects that you couldn't clean from the measurements, it still looks like it isn't very flat at all, with that big dromedary thing going on in the vicinity of 1 kHz. I expect that to get rid of that, the cutover point in the LPF itself was originally fairly low, although it looks like you moved it a bit higher in conjunction with a steeper slope, or at least a slope that is steeper once above 3 kHz. I doubt if that woofer is a woofer that you would consider using in a speaker you were building from scratch. (And the same is probably true for the tweeter.)
 

Dennis Murphy

Major Contributor
Technical Expert
Joined
Mar 17, 2020
Messages
1,071
Likes
4,547
These are very nice. The first one clearly reveals the cone breakup at 5 kHz, so there can be little doubt that this is the cause of the ringing seen in Amir's CSD graph. We still don't know whether or to what extent this was the source of the high frequency sound that annoyed him, but chances are good, I think, that this is at least a strong contributor to that problem. Your third graph looks much more like a proper crossover, especially with respect to the tweeter, compared to the second graph which is the individual responses with the original crossover. It looks to me like you did a very good job there. I'm betting that if Amir gets a chance to listen to your modded version that he'll like it a lot better. One thing, though, when I look at that unfiltered response of the woofer, that thing is butt-ugly. Even if all that stuff below 300 Hz happens to be measurement artifacts or maybe room effects that you couldn't clean from the measurements, it still looks like it isn't very flat at all, with that big dromedary thing going on in the vicinity of 1 kHz. I expect that to get rid of that, the cutover point in the LPF itself was originally fairly low, although it looks like you moved it a bit higher in conjunction with a steeper slope, or at least a slope that is steeper once above 3 kHz. I doubt if that woofer is a woofer that you would consider using in a speaker you were building from scratch. (And the same is probably true for the tweeter.)

Actually, the Peerless woofer response is very nice up until around 3 kHz. There will be a 1 kHz hump on any woofer measured anechoically in the intended cabinet. That's just the 6 dB baffle step, and it's very smooth on this unit. Usually there's some kind of nonsense going on in the 1K area that's due to interference between the cone and surround. This is a very well engineered woofer with lots of power handling, and I'm pretty sure it could go a lot lower in a larger cabinet.
 

KaiserSoze

Addicted to Fun and Learning
Joined
Jun 8, 2020
Messages
699
Likes
592
Actually, the Peerless woofer response is very nice up until around 3 kHz. There will be a 1 kHz hump on any woofer measured anechoically in the intended cabinet. That's just the 6 dB baffle step, and it's very smooth on this unit. Usually there's some kind of nonsense going on in the 1K area that's due to interference between the cone and surround. This is a very well engineered woofer with lots of power handling, and I'm pretty sure it could go a lot lower in a larger cabinet.

Which Peerless woofer is it?
 

Beave

Major Contributor
Joined
May 10, 2020
Messages
1,396
Likes
3,016

Do you think they cut corners on the crossover purely for $$ reasons?

I haven't seen schematics for their crossover, but I did notice from the pics I posted earlier that the board has placeholders for other components that aren't installed. I can't help but wonder if they started with a more complex (more $$) crossover but went cheaper in the end, deciding the cheaper version was 'good enough.'

On that note, can you post a schematic or tell us what their crossover topology is?

And how much more would yours cost compared to theirs?
 

Dennis Murphy

Major Contributor
Technical Expert
Joined
Mar 17, 2020
Messages
1,071
Likes
4,547
Do you think they cut corners on the crossover purely for $$ reasons?

I haven't seen schematics for their crossover, but I did notice from the pics I posted earlier that the board has placeholders for other components that aren't installed. I can't help but wonder if they started with a more complex (more $$) crossover but went cheaper in the end, deciding the cheaper version was 'good enough.'

On that note, can you post a schematic or tell us what their crossover topology is?

And how much more would yours cost compared to theirs?

Perish forbid. I'm sure they would never cut corners to increase profits. At least they put a little more into it than DynAudio did for their $800 Emit M10. We're talking one tiny iron core (not even steel laminate) inductor and one capacitor. Here's the stock schematic for the Ultra. The inductor values may be off slightly because I didn't want to disconnect anything on the board to make an accurate measurement. The board does have spaces reserved for two more resistors, another cap, and an inductor. I doubt that those were intended for the Ultra, however.
SVS Ultra Stock Net 1.png
SVS Ultra Stock Net 2.png
I'll have to check what my new crossover parts came to, but it wasn't all that much.
 

Beave

Major Contributor
Joined
May 10, 2020
Messages
1,396
Likes
3,016
Thanks Dennis!

More questions: Do you ever sleep? Have you cloned yourself? How do you have time do post on so many forums while at the same time doing so much design work? It's truly impressive.
 

justcheeze

Member
Joined
Aug 3, 2020
Messages
10
Likes
6

KaiserSoze

Addicted to Fun and Learning
Joined
Jun 8, 2020
Messages
699
Likes
592

Thanks. I had glanced at this woofer last night because it was the only one I found on PE that used a fiberglass cone, but I got confused because I was comparing it to pics of the wrong woofer, from the less expensive Prime Bookshelf, on Noaudiophile.

But now my question is why the frequency response you showed above for the woofer's unfiltered response looks so different from the graph published by Tymphany. Their graph shows an essentially flat response from 150 Hz to 2 kHz. Your graph clearly indicates a peak at 900 Hz and sharp declines to either side. Is the difference simply the difference between reality and what the manufacturer publishes? Even if the circumstances of your measurement renders the results inaccurate below, say, 500 Hz, this still would not explain the stark differences between 500 Hz and 2 kHz.
 

Dennis Murphy

Major Contributor
Technical Expert
Joined
Mar 17, 2020
Messages
1,071
Likes
4,547
Thanks. I had glanced at this woofer last night because it was the only one I found on PE that used a fiberglass cone, but I got confused because I was comparing it to pics of the wrong woofer, from the less expensive Prime Bookshelf, on Noaudiophile.

But now my question is why the frequency response you showed above for the woofer's unfiltered response looks so different from the graph published by Tymphany. Their graph shows an essentially flat response from 150 Hz to 2 kHz. Your graph clearly indicates a peak at 900 Hz and sharp declines to either side. Is the difference simply the difference between reality and what the manufacturer publishes? Even if the circumstances of your measurement renders the results inaccurate below, say, 500 Hz, this still would not explain the stark differences between 500 Hz and 2 kHz.

You might want to spend a little time reading about the baffle step if you haven't already. There is no standard among manufacturers for measuring woofer frequency response. Tymphany apparently mounts their woofers in a very large "infinite" baffle that largely eliminates the baffle step. Other manufacturers use a smaller baffle typical of real-world cabinets. Those curves will show a 6 dB baffle step beginning in the lower midrange and culminating in a 1 kHz peak, followed by a dip.
 

KaiserSoze

Addicted to Fun and Learning
Joined
Jun 8, 2020
Messages
699
Likes
592
You might want to spend a little time reading about the baffle step if you haven't already. There is no standard among manufacturers for measuring woofer frequency response. Tymphany apparently mounts their woofers in a very large "infinite" baffle that largely eliminates the baffle step. Other manufacturers use a smaller baffle typical of real-world cabinets. Those curves will show a 6 dB baffle step beginning in the lower midrange and culminating in a 1 kHz peak, followed by a dip.

I'm reasonably familiar with baffle step and with the closely related diffraction effect. In fact in the thread on the GR-Research kit speaker I had point out that both that speaker and this speaker are 8.5" wide and that the dip in the on-axis response and the peak in the off-axis response at the same frequency, 3.2 kHz, is a diffraction effect. The first, prominent on-axis peak caused by baffle diffraction occurs for this speaker at 1.6 kHz and thus is not a viable hypothesis for explaining the difference between the manufacturer's FR curve for the woofer vs. your unfiltered curve.
 

Dennis Murphy

Major Contributor
Technical Expert
Joined
Mar 17, 2020
Messages
1,071
Likes
4,547
I'm reasonably familiar with baffle step and with the closely related diffraction effect. In fact in the thread on the GR-Research kit speaker I had point out that both that speaker and this speaker are 8.5" wide and that the dip in the on-axis response and the peak in the off-axis response at the same frequency, 3.2 kHz, is a diffraction effect. The first, prominent on-axis peak caused by baffle diffraction occurs for this speaker at 1.6 kHz and thus is not a viable hypothesis for explaining the difference between the manufacturer's FR curve for the woofer vs. your unfiltered curve.

I don't think very many people are reading this thread, or else someone would have jumped in and explained the difference between ordinary diffraction effects and the special case of the baffle step. The diffraction effects you're referencing are caused by sound waves diffracting back from the baffle edges and causing cancellations and summations depending on the baffle width and the sound wave lengths. You obviously understand that aspect of diffraction. But we're not communicating on the baffle step

It's a gradual process caused by the sound wave lengths increasing as you go down in frequency from around 1 kHz. As the pulsating sphere of sound pressure widens with decreasing frequency, the sound pressure starts to slip behind the baffle, thus lessening the forward sound pressure and decreasing output as perceived from the front of the speaker. This process ends when the sound waves are so long that they radiate completely omni-directionally. The difference between the sound pressure above around 1 kHz and below 200 - 300 Hz is 6 dB if measured anechoically. That's what is shown in my plot of the Peerless woofer. (My measuring system (Praxis) is only anechoic down to about 300 Hz, at which point it transitions to a room mode, which is why my plot is so filled with room effects down there.) As I stated in an earlier post, any woofer will exhibit this change in sound pressure when mounted on a typical baffle. Here's my measurement of the excellent Purifi 6" woofer. The response above 1000 Hz is almost entirely a function of the physics of the woofer itself. It's quite different from the Peerless, but not necessarily "better."

Purifi Woofer Raw Response.png
 
Last edited:

CBM

Active Member
Joined
Aug 7, 2020
Messages
109
Likes
49
I have had the SVS Ultras for a couple of years. These were the first "premium" speakers I have owned so I thought they had to be good. I spent a lot of time positioning them and running Audyssey and listening. Then repeating the process. Using REW, I was finally able to locate them so they had a wonderful soundstage and overall good sound (so I thought). Then I read this review and thought maybe I can do better. I purchased Focal 906's and Elac DBR-62's for a trial. I placed them in the same position as the Ultras and immediately noticed the difference with both speakers. Now I know what Amir meant by brightness on female vocals. It's really there. Now they sound so smooth. I'm going to buy a replacement. I'm not sure if it will be the Elac or Focals but I will make a change. I have a few more weeks left to make up my mind.
 

justcheeze

Member
Joined
Aug 3, 2020
Messages
10
Likes
6
I have had the SVS Ultras for a couple of years. These were the first "premium" speakers I have owned so I thought they had to be good. I spent a lot of time positioning them and running Audyssey and listening. Then repeating the process. Using REW, I was finally able to locate them so they had a wonderful soundstage and overall good sound (so I thought). Then I read this review and thought maybe I can do better. I purchased Focal 906's and Elac DBR-62's for a trial. I placed them in the same position as the Ultras and immediately noticed the difference with both speakers. Now I know what Amir meant by brightness on female vocals. It's really there. Now they sound so smooth. I'm going to buy a replacement. I'm not sure if it will be the Elac or Focals but I will make a change. I have a few more weeks left to make up my mind.

Give the Aperion Verus Grand 3 if you haven't yet. They have free in-home trials too. They to me strike the right balance between just bright enough for music that requires some bit of brightness (hello rock and metal) but not so much where it would start to become fatiguing on the ears (which SVS has done on ALL their speakers, the Primes being the worst offender).
 

CBM

Active Member
Joined
Aug 7, 2020
Messages
109
Likes
49
Give the Aperion Verus Grand 3 if you haven't yet. They have free in-home trials too. They to me strike the right balance between just bright enough for music that requires some bit of brightness (hello rock and metal) but not so much where it would start to become fatiguing on the ears (which SVS has done on ALL their speakers, the Primes being the worst offender).
Thank you for the recommendation. I looked at their website and they seem to be very innovative. I'm actually thinking about trying them out. I wish Amir would measure them.
 
Top Bottom