• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

step response is a important part to show speed of speaker that is good enough for ITD. See measures

Status
Not open for further replies.

Wesayso

Member
Joined
Dec 30, 2021
Messages
74
Likes
144
Location
The Netherlands
How could this be ? Your cabinets are closed and cones of TC9 are treated. Are you sure that it happened over one winter period ?

When I started to research this, i found that more drivers of the former Vifa/Scan Speak stable had this problem. It might not have been the storage in my garage, it could have been age related due to the glue that was used in these drivers.

I have contacted Tymphany, current owner of the Vifa brand speakers without result. Scan Speak have acknowledged the problem and have changed their glue to prevent this from happening quite a few years ago.
Yes, I know... pretty crazy huh? But hey, we only live once...
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
611
Likes
613
Are there any drawbacks? Sure, you see those horn shaped curves in the plot? They are the "dreaded" side lobes of an array consisting of multiple drivers. That is because it isn't a continuous array, but rather a series of 3.5"drivers. Use a smaller driver and those horn like lobes would go up in frequency. One drawback of doing just that, smaller drivers wouldn't be able to play the bottom end, so you'd need a support system down low.
I always wanted to do a true line array with the Tectonic BMR's but never had the time. Did you look at them? https://www.tectonicaudiolabs.com/audio-components/bmr-speakers/
 

Wesayso

Member
Joined
Dec 30, 2021
Messages
74
Likes
144
Location
The Netherlands
Yes, I did look at the Tectonic BMR's with interest. It should work out pretty good, even if their sensitivity is a bit low to start with.
Not too much info on them, except for the round models though. I'd love to see a full set of measurements on one of those square types and run some Vituix simulations with that.
 

Zvu

Addicted to Fun and Learning
Joined
May 1, 2020
Messages
700
Likes
1,185
Location
Serbia
Your system is one of the best executed DIY systems/concepts i know of. Peerless TC9 is great little transducer and multiplied by 25 makes its performance high-end. Going from TC9 to 10F is worth, if nothing else, for durability.

About that glue issue, you've got me worried there for a second. I have 5 pairs of TC9's bought at different times. Here are 2016 and 2019 batch:

IMG_20220103_142709.jpg

It doesn't seem all that different from one another. Slightly darker color of 2016 but i might be imagining. We'll see what happens in future.
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
611
Likes
613
Yes, I did look at the Tectonic BMR's with interest. It should work out pretty good, even if their sensitivity is a bit low to start with.
Not too much info on them, except for the round models though. I'd love to see a full set of measurements on one of those square types and run some Vituix simulations with that.
I just did some crude measurements back then. The small one looked promising:

TEBM36S12-4A.png
 

Wesayso

Member
Joined
Dec 30, 2021
Messages
74
Likes
144
Location
The Netherlands
Your system is one of the best executed DIY systems/concepts i know of. Peerless TC9 is great little transducer and multiplied by 25 makes its performance high-end. Going from TC9 to 10F is worth, if nothing else, for durability.
Thanks for that huge compliment :).
I'm hoping Tymphany changed glue also, I just couldn't get any type of response from them. I tried online though their own web channels and trough their mayor European distributor. No answer from the first and that last one could not give me a conclusive answer.

If I didn't have 5 drivers fail within a 2-3 month period I wouldn't have given up on those drivers. They have done all I expected them to do and more. Even though I've always been curious about the Vifa TG series and the Scan Speak 10F, without these failures I wouldn't have had a cause to upgrade. The responsiveness from Scan Speak and their acknowledgement of the problem swayed me that way.
I've always said: You've got to like the driver you use in an array, as you'll get it back on steroids. I do know I like the 10F 8424G00 and expect the 8414G10 to be a hair better suited for the line array job. And it's a bit more affordable. The 8424G00 has a closed basket construction below the spider, that's what worried me for full range use.

In the mean time I have replaced the faulty drivers and have had no failures since... All has been well for about half a year...

About that glue issue, you've got me worried there for a second. I have 5 pairs of TC9's bought at different times. Here are 2016 and 2019 batch:

View attachment 176386

It doesn't seem all that different from one another. Slightly darker color of 2016 but i might be imagining. We'll see what happens in future.
I can only hope I received a bad batch. I 'think' my project may have sold quite a few extra TC9's :D. I just hope they will hold up fine over time.
My batch was bought in September/October 2011, delivered in December that same year. Plenty of time for Tymphany to do some upgrades ;).
.
I just did some crude measurements back then. The small one looked promising:

View attachment 176387
This is not the first time I've seen that strange wavy pattern control in the highest sensitivity area (2-4 KHz) in a plot like this.
The round one as tested by Erin didn't show this though....

For me the TC9 was a no-brainer, all of the measurements I could find were exemplary for such a driver. It's low price point was a big bonus.
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
611
Likes
613
This is not the first time I've seen that strange wavy pattern control in the highest sensitivity area (2-4 KHz) in a plot like this.
The round one as tested by Erin didn't show this though....

For me the TC9 was a no-brainer, all of the measurements I could find were exemplary for such a driver. It's low price point was a big bonus.
If memory serves well I just slapped the unit into a cardboard (!) box, so don't read too much into the graphs.
 
OP
B

bennybbbx

Addicted to Fun and Learning
Joined
May 27, 2020
Messages
568
Likes
105
Location
germany
Well, your jumping pretty fast. I've ran my arrays as full range array from the end of 2004 until september 2020. Over the winter of 2020-2021 I have converted my speakers, out of curiosity, to frequency shaded arrays. That doesn't change anything I said, you just had a misunderstanding of my post where I tried to explain that your idea of a single driver above 100 Hz wasn't an array in any form, way or shape.


The picture I've shown you demonstrates that after DSP the frequency curve of an array shows less bumps and dips due to room effects than a single driver speaker would exhibit. True for both the full range array and the frequency shaded example by the way. Meaning it varies less in room than a single driver speaker would. That doesn't mean it will make it sound more dense. Quite the opposite.
Actually, if it weren't for the physical senses you get you'd be fine to compare it to giant headphones, but in a good way. Better than headphones because of the imaging/stage and the physical senses. It does sound dryer in-room than the single driver with DSP applied and absorbing the early reflections. Strange huh? How a limited vertical pattern can clean up the results. I hope you didn't miss the tight vertical pattern an array exhibits. It covers seated and standing positions but not much more. It "seems" like it misses the floor and ceiling, while in reality they are virtual extensions of line length. Can you imagine that?

So no! You can't mimic an array with your speakers. In reality I stand a better chance of mimicking your speakers with my arrays. So You'd be better of finding those sound clips if you're really curious.

sure it is not perfect but most of the array sound and room reflections are record.

I have done a video
that show when 2 delays are add to simulate the diffrent ear distance to line array speaker. you notice transient lost and comb filter effects. later i do delay of 2 ms and 500 ms so you can hear how it sound with more and that i set the delays correct. I can also such a video do when the line array play only upto 1 khz, when below 100 hz all speakers play then it is a bass array ;) Now i try to do the step response from thissituation

EDIT: there can see clear the position of the 2. and 3. rd speaker in step response and it look slow. thats correct transients are much reduce. the measure is at same microphone postion wirth and without delay

step response with delay.jpg


step response without delay.jpg
 
Last edited:

Wesayso

Member
Joined
Dec 30, 2021
Messages
74
Likes
144
Location
The Netherlands
So you missed the point entirely... well done. If you do feel like learning something instead of assuming you already know it all, let us know.

If you make the model above have about a 3 ms delay, you could learn why your speaker sound like it does ;). You just don't get it, how can I blame you.

Look at your waterfall plot, look at my waterfall plot... what do you see? Can't make it more basic than that. Which of those two looks the most like your experiment here?

I'll give you a hint, what you model is 3 sources with a certain distance between them.

What I have is a series of sources with a fixed distance between them, but that distance (to each speaker) varies as seen from the listening spot, the output has been modified to fit the frequency it needs to play (as it is frequency shaded) which causes the beam forming that avoids most ceiling and floor reflections, and it is knocked back into shape by using DSP (FIR) correction that will do it's best to make it into an ideal impulse again. The beauty of it being that it holds true for quite a generous listening area. As can be seen in the IR, the STEP and Frequency plots, wavelets etc.

In other words: it isn't what you want to make of it, however hard you try. Read it again, try to start learning and stop assuming.
zappa-parashute.jpg
 
Last edited:
OP
B

bennybbbx

Addicted to Fun and Learning
Joined
May 27, 2020
Messages
568
Likes
105
Location
germany
So you missed the point entirely... well done. If you do feel like learning something instead of assuming you already know it all, let us know.

If you make the model above have about a 3 ms delay, you could learn why your speaker sound like it does ;). You just don't get it, how can I blame you.

maybe you understand me complete wrong. I dont want say the JBL ist the best speaker on earth i want only show that with a shorter step response stereo width and transients are better and so it is usefull to see step response. and when there is a speaker below 1000 eur with rise time less than 300 µsec and 6.5 inch woofer i buy it and test. kali is 6.5 inch have much slower step response, less transients and stereo width. and i want show that the FR have not much influence to step response.

should i do a video that show how 3 ms delay really sound ?. with 3 ms delay you hear double attack transients only shorter as when do delay of 500 ms
 

Wesayso

Member
Joined
Dec 30, 2021
Messages
74
Likes
144
Location
The Netherlands
maybe you understand me complete wrong. I dont want say the JBL ist the best speaker on earth i want only show that with a shorter step response stereo width and transients are better and so it is usefull to see step response. and when there is a speaker below 1000 eur with rise time less than 300 µsec and 6.5 inch woofer i buy it and test. kali is 6.5 inch have much slower step response, less transients and stereo width. and i want show that the FR have not much influence to step response.
As can be seen in the STEP of an ideal band-passed impulse, your story is garbage (wanted to type something else but I'll keep it civilized)
STEP.jpg

Here's the STEP of my speaker (with a non flat response, making the shape a tiny bit different) all the way down to 17 Hz when it was still running full range.
actualSTEP.jpg

Had it been equalized flat beyond all boundaries it would have looked more like my direct SP-Dif pass trough shown here:
S-Pdif.jpg

Just because the bandwidth of the SP-Diff signal is wider (both at high and low frequency) than the band passed model, it
stays above that center line for longer.

The SP-Dif as seen in the program APL_TDA:
dac.jpg

Instant signal at all frequencies. Transient perfection.

My speaker (not yet frequency shaded) as seen with the same APL_TDA program:
stereo.jpg

Near transient perfect over its range, reflections can be seen after the impulse. Clearly seen is the band-passed behavior at low frequencies.

should i do a video that show how 3 ms delay really sound ?. with 3 ms delay you hear double attack transients only shorter as when do delay of 500 ms
No thanks... as I do not think anyone would be interested in viewing it. I could be wrong though.
You could be the only one that has it right in this whole thread, and all other posters, me included could be delusional.

Just magically a transient perfect response should look like your STEP, and not like a perfect STEP response of a band passed signal like I've shown more than once now. :(

Why do I keep trying....
 
Last edited:
OP
B

bennybbbx

Addicted to Fun and Learning
Joined
May 27, 2020
Messages
568
Likes
105
Location
germany
As can be seen in the STEP of an ideal impulse, your story is garbage (wanted to type something else but I'll keep it civilized)
View attachment 176455
Here's the STEP of my speaker (with a non flat response, making the shape a tiny bit different) all the way down to 17 Hz when it was still running full range.

this is 1 speaker or all in the line ?. as i told before i dont want say the JBL is perfect so wy you compare so often with your speaker. your speaker have no LP crossover and is fullrange. so there is much much shorter rise time. if you have 2 way speakers then you maybe can show for compare,
 

Wesayso

Member
Joined
Dec 30, 2021
Messages
74
Likes
144
Location
The Netherlands
25 drivers at once.

I show it as an example because I have it available. My speaker technically is a two way I guess, as I'm running subs nowadays that come in at 160 Hz. But... it has a linear phase crossover to not loose anything from that transient nature. It's what I wanted, its what I have.

What you are showing with the JBL is a tweeter, firing in front of a woofer, etc... The bandwidth of the tweeter will determine what the first part of the STEP looks like. This has nothing to do with transient response or speed. It's just the tweeter's output that dominates the impulse/step.
If it had a linear phase crossover and hardly any reflections within the room, the STEP would stay flying above the center line for longer, and it could be called a good transient. But as is, it just isn't.

If it had a different type of crossover, it would look different too. Look up Dunlavy, check out his speakers and his STEP response.
Stereophile Dunlavy SC-IV

99DAL4fig4.jpg

He had first order crossovers and used a single tweeter, 2 mids, 2 woofers in one big speaker. His speaker is fast.
1) Dunlavy Audio Labs SC-IV Signature Collection Speakers – SkyFi Audio

(plucked this from the net, for information purpose only, for the original look here: link)
His STEP is longer above the center line than yours. This isn't done with DSP, just with common sense and engineering.
I cheat with DSP, why? Because it is available. John Dunlavy was working on a DSP model speaker and didn't get to finish it.
Look what he wrote about it in his paper, you can find it online. Read and learn. He knew his stuff....
An interview on Stereophile might help: Interview John Dunlavy but his paper: Loudspeaker Accuracy holds way more information.
 
Last edited:

Wesayso

Member
Joined
Dec 30, 2021
Messages
74
Likes
144
Location
The Netherlands
To share two relevant pages of his paper:
CCF03052009_00008.jpg

CCF03052009_00020.jpg

This last page shows the differences within STEP responses. Guess which one is fast....
 
OP
B

bennybbbx

Addicted to Fun and Learning
Joined
May 27, 2020
Messages
568
Likes
105
Location
germany
25 drivers at once.

I show it as an example because I have it available. My speaker technically is a two way I guess, as I'm running subs nowadays that come in at 160 Hz.

and what happen when you measure without subwoofer and without phase linear ?. how does step response then look and does it sound much diffrence ?. I have the conclusion that with phase correction you can fake the measure impulse that it give nice results. but it sound not really better. i test dirac on kali. maybe there should test multi tone burst and look what the speaker bring as results. on my measure there is near nothing change with and without sub. https://www.audiosciencereview.com/...r-itd-see-measures.28585/page-11#post-1028997 how much delay your linear phase system have ?

But... it has a linear phase crossover to not loose anything from that transient nature. It's what I wanted, its what I have.

Linear phase is not possible for low frequency when play music realtime because the delay is too much and hearable

What you are showing with the JBL is a tweeter, firing in front of a woofer, etc... The bandwidth of the tweeter will determine what the first part of the STEP looks like. This has nothing to do with transient response or speed. It's just the tweeter's output that dominates the impulse/step.

see my 1.5 cm measure of the woofer/mid. the kali have rise time of 500 µsec. the jbl is not so good to compare because it have a coaxial system and so the raise time of tweeter is always in front
 

Wesayso

Member
Joined
Dec 30, 2021
Messages
74
Likes
144
Location
The Netherlands
this is 1 speaker or all in the line ?. as i told before i dont want say the JBL is perfect so wy you compare so often with your speaker. your speaker have no LP crossover and is fullrange. so there is much much shorter rise time. if you have 2 way speakers then you maybe can show for compare,
Do you want a more than excellent example of a 2 way speaker with the same APL_TDA plot style?

524831d1452732108-group-delay-questions-analysis-apl-tda-35ms-3d.png

This is what a 2 way speaker with subs looks like in a Studio grade room with a passive Haas Kicker at 25 ms.
Nothing in that graph is left to chance, the room is planned completely to get this, as measured at the listening spot.
Pure dedication and lots of research, the speaker is a Troels Gravesen build with second order crossovers.
More info on room + speakers on Gearslutz here: Jim's excellent room

I think what is shown here surpasses many studio rooms and literary was the work of many years of dedication.
The only thing you might want to do different in a Studio is getting rid of the Haas Kicker. But for a listening room
it is highly welcome.
 
OP
B

bennybbbx

Addicted to Fun and Learning
Joined
May 27, 2020
Messages
568
Likes
105
Location
germany
To share two relevant pages of his paper:
View attachment 176470
This last page shows the differences within STEP responses. Guess which one is fast....

the first step response have faster rise time as 2. on the 3. can nothing see about rise time. but look the speakers that are out in the world. on very much speakers see delay between tweeter and woofer. When all speakers look as 3 then i also agree step response is not usefull. but currently it can good see if a speaker have good time align and fast mid
 

Wesayso

Member
Joined
Dec 30, 2021
Messages
74
Likes
144
Location
The Netherlands
and what happen when you measure without subwoofer and without phase linear ?. how does step response then look and does it sound much diffrence ?. I have the conclusion that with phase correction you can fake the measure impulse that it give nice results. but it sound not really better. i test dirac on kali. maybe there should test multi tone burst and look what the speaker bring as results. on my measure there is near nothing change with and without sub. https://www.audiosciencereview.com/...r-itd-see-measures.28585/page-11#post-1028997 how much delay your linear phase system have ?
About 740 ms delay. So if I press play I need to wait 0.74 seconds for sound, no big deal as I am a listener, not a producer.
I've done the whole song and dance of comparing non phase linear to phase linear and aside from not wanting to correct the systems natural roll off I highly prefer a corrected phase. Which isn't easy to achieve without warts and stuff... It took me long enough to get it right, really right.

Is it a big difference? That depends, it can be. For all sounds higher than ~1 Khz, not so much. For the lower notes it can bring a difference.
But that would take a room + speaker that works toward that goal of keeping the STEP in-tact at the listening spot. Then it can be fun.
As said: I prefer it.

You can't just EQ phase and call it perfect. You need the room and speakers to work with you (read: room treatment, see Jim's room for an excellent example of a treated room, no DSP, and no linear phase though)
Linear phase is not possible for low frequency when play music realtime because the delay is too much and hearable
For realtime, nope, I wouldn't use it. Though for entertainment, where all I do is press play, a delay is nothing to worry about.
For video: I use a media player that automatically adjust video sync, so no trouble there. I don't do live shows.

I can't figure out what you mean by hearable, you mean it stays silent before it starts? Yes. For the whole 740 ms. I use many taps in my FIR correction, why? Because I can. Not limited by hardware.
see my 1.5 cm measure of the woofer/mid. the kali have rise time of 500 µsec. the jbl is not so good to compare because it have a coaxial system and so the raise time of tweeter is always in front
Not this again.... it will tell you nothing without looking at your tweeter's band width. You're making up things again.
Use science, that will work way better.
 
Last edited:
Status
Not open for further replies.
Top Bottom