Question on Waterfall and Transient response

[wintermute]

Hi,

I have a basic understanding on how to interpret waterfall plots but I haven't been able to find anything which explains how to tell whether the transient response of a driver is good or not. I think I'm ok with being able to spot resonances etc but what for instance is a good transient response for a mid base driver. I've attached a waterfall plot here for a 5" mid bass driver. Can someone comment on whether the transient response is good bad average

And what would be good, bad, average (relative to what this one is).

Sorry about the image quality, I think the highest time shown is about 1.95ms It's definitely under 2ms.

I guess the other thing is, Is it good that this driver seems to be pretty consitent from low Hz up to about 4000Hz (I'm assuming this is a desirable thing).

Also any links to sites that give a good explanation of how to interpret waterfall plots would be great.

Regards,

Tony.

[SY]

I think you'll get a better handle on transient response by looking at impulses or steps in the time domain. Waterfalls are more useful for spotting resonances and stored energy (though a high-resolution plot of impedance in the complex plane is even better. imo).

[wintermute]

Hi Sy,

My beleif was that transient response was basically how quickly the driver can react to a sound input, and how quickly it stops producing the sound once the input is removed. Is that right? So basically a speaker with poor transient response will take a while to get up to speed so to speak and then also take a while to stop again, resulting in a muddy (ie unclear/ not detailed) sound.

I thought maybe the cumulative spectral decay graph would indicate at least the quickness of decay and provide some insight (but I don't have a baseline as to what is quick and what is slow). I'm just going of published stuff here not my own measurements.

I've heard of impulse and step response but haven't really read up on them, looks like it's time for that now too ..... Not sure where I can get that data from though, is it something you can measure yourself? Bit hard if trying to find out before buying something though

I think I'm ok with spotting resonances, although I must admit the plot I attached does confuse me somewhat in the way it slopes to the right, most graphs I have seen the resonances come straight out at 90 degrees to the freq axis not cut across it at an angle like the one I showed.

Regards,

Tony.

[planet10]

Originally posted by SY
Waterfalls are more useful for spotting resonances and stored energy

And they are even more useful if they use a axis in units of period instead of time.

dave

[SY]

Tony, normally waterfall plots are derived from the impulse response. And an impulse is about as pure a transient as you'll find. Impulses can be gathered directly or by taking an autocorrelation of an MLS excitation. In English, you hit a bell with a hammer and see how it rings.

Interpreting the waterfalls is difficult- understanding how transform parameters affect the waterfall display is even tougher (the "slope" you're having trouble with is a perfect example). D'Appolito's book gives a fairly good explanation of what the plots are and aren't and how to interpret them.

[wintermute]

Thanks SY. I've seen Dappolitos book mentioned a bit, I'll have to keep an eye out for it!

I'm also going to have to get off my backside and buy some drivers, before I go nuts with too much theory

Regards,

Tony.

[stokessd]

Here's some waterfalls with time on an axis...

Check that transient response:

http://www.quadesl.com/madisound/sds_esl_impulse.gif


Sheldon

[Brian]

"My belief was that transient response was basically how quickly the driver can react to a sound input, and how quickly it stops producing the sound once the input is removed. Is that right?"

I believe this is exactly right.
Like You, I also do not know how quickly the graph ought to decay to indicate a musical speaker.
E-SPEAKERS has both step response and impulse response graphs of the Josef Manger driver at the bottom of this page http://www.e-speakers.com/products/manger.htm also a waterfall plot which You can use for comparison. This is an 8" Titanium membrane with 1.3Teslas of flux density. I think any mid cone or woofer which decays as quickly is apt to be a good one. Perhaps other catalogs have step or impulse plots for cones but I only remember seeing them for tweeters.
Another way to consider speed before buying a driver is to read the construction details. High power woofers with their huge heavy 3 and 4" voice coils tend to be slow to respond. These drivers also have rugged therefore heavy cones. Paper cones seem to be the fastest. Also stiff paper surrounds are said to make a faster speaker with foam coming next then neoprene rubber. I do not know where treated cloth surrounds fit in this. Higher flux density, like a Lowther, increases the response speed.

"I think I'm ok with spotting resonances, although I must admit the plot I attached does confuse me somewhat in the way it slopes to the right, most graphs I have seen the resonances come straight out at 90 degrees to the freq axis not cut across it at an angle like the one I showed."

I had a piano like that. When sustained one of the strings said wooowww. With the pitch rising with the passage of time. On this speaker I wonder if perhaps it is a vibration being excited in the metal basket.

[wintermute]

Thanks Brian,

Thats a very interesting driver (and at the price I guess it would want to be!)

I think I have a tendency to over research before making a purchase decision (at least when there is a reasonable amount of money involved).

I think you are probably right about comparing the decay on a known fast driver with another, I think I should stop worrying and just get on an buy it

You might be onto something with your piano analogy too, perhaps the freq changes somewhat as the sound decays, hadn't thought of it like that.

Regards,

Tony.

[SY]

Honestly, one of the major influences in the choice of drivers for me is hearing a speaker I really like, then noting what the drivers are. There's a lot that the data sheets don't and can't tell you.