Step Response

[Robh3606]

Just wondering if any of you making measurements have been looking the the step responses of systems you have measured?? Some seem to place a great amount of emphasis on it. With the larger systems particuarly the horn systems they can be a bit spread out in time yet they clearly are the most dynamic sounding systems.

Any thoughts??

Rob:)

[grumpy]

Vs. impulse? (which has a long airplay history). That there is a smooth
handoff (in time) from one driver to the next -seems- to be a common
quality of many well-liked multi-driver speaker systems... although I'm
sure there are many contrary-measuring systems that give a listener joy.

[Oldmics]

A quick read thru explained what it means.

This immeadently differentiates pro rigs from hi fi measurements.

In setting up a pro rig,once the measurement tool is calibrated,we usually measure box /individual componet delay times and then go right to coherence.

None of that messy passive stuff to worry about.

Just the hell of processer latency,slope wormholes,:blah::blah::blah: and the list goes on.

Oldmics

[Ian Mackenzie]

Just wondering if any of you making measurements have been looking the the step responses of systems you have measured?? Some seem to place a great amount of emphasis on it. With the larger systems particuarly the horn systems they can be a bit spread out in time yet they clearly are the most dynamic sounding systems.

Any thoughts??

Rob:)

Agreed,

I think it depends on how hard you want to be on yourself with your design aims.

Direct radiators look much cleaner (less ringing..aberations) but dont have the dynamics of horns as you say.

Perhaps a better question is the audibility of measured aberations?

I think its more of an issue in the transition from the cone to the horn in the overall system step function ie a square wave where there are large phase shifts.

Depending on the genre the perception of the problem varies but I hear it as imaging, timbre and depth when comparing a Trad horn/cone system crossing over 800- 1300 htz comnpared to a cone / wave-guide time aligned system where the crossover point up around 2.5 k hertz

I guess it also depends on the driver in that the break up modes dias get mingled with other time related issues of the horn and if you are asking the horn operate down near its cut off where you might take rising distortion as a step response problem?

The ear appears to be most sensitive to garbage in the 800-2000 region what ever the cause.

As with all this stuff there are lots of compromises but I like the LS80 design where they run the woofers up to 2.5K..bring on the 2118s

Rob,

Have you tried crossing over the PT wave guides at 2.5 kertz and giving it a whirl?

[Robh3606]

Hello Ian


Have you tried crossing over the PT wave guides at 2.5 kertz and giving it a whirl?

No but I can certainly try it. I ended up doing a passive x-over at 1.4k with 2108 and the PTH1010 2435 combo. It looks pretty good and actually can be made better with a spacer to bring the horn forward a bit like on the 4430/4435.


Perhaps a better question is the audibility of measured aberrations

Yes I wonder. I noticed when taking 2 meter measurements on the Jubilee clones the step response had a sweet spot as did the previous DIY system. Looking at an older measurement on the 4344 there obviously is quite a difference.

In your travels have you ever listened to a pair of L250ti's?? The Jubilees are a really nice sounding speaker. I have not even powered up the CC networks yet and they are simply a pleasure to listen too.

Rob:)

[Mr. Widget]

In setting up a pro rig,once the measurement tool is calibrated,we usually measure box /individual componet delay times and then go right to coherence.How do you use coherence? I understand that if the system isn't linear this will show up as an obvious deviation, but let's say you want to improve your gain structure to reduce system noise... how do you track down the problem area?


Widget

[Ian Mackenzie]

Hello Ian




In your travels have you ever listened to a pair of L250ti's?? The Jubilees are a really nice sounding speaker. I have not even powered up the CC networks yet and they are simply a pleasure to listen too.

Rob:)

Only in a shop a while back and Ti dome had some L250's as i recaall

What you say about distance makes sense when you have an array of drivers vertically. Unless they are in a curves concave arch the path lenghts will vary with chnages in distance.

[Oldmics]

How do you use coherence? I understand that if the system isn't linear this will show up as an obvious deviation, but let's say you want to improve your gain structure to reduce system noise... how do you track down the problem area?


Widget


If I told you-I would have to kill you. :(

[Mr. Widget]

If I told you-I would have to kill you. :(You old tease.:D


Widget

[4313B]

Yes I wonder. I noticed when taking 2 meter measurements on the Jubilee clones the step response had a sweet spot as did the previous DIY system. Looking at an older measurement on the 4344 there obviously is quite a difference.Do you still have the plots?

It looks pretty good and actually can be made better with a spacer to bring the horn forward a bit like on the 4430/4435.This is one of the reasons why Greg used the 1200FE for his personal system (with the deeper 4338 horn).

[Mr. Widget]

This is one of the reasons why Greg used the 1200FE for his personal system.Any of the 43XX systems with the baffle mounted slot tweeter have positively terrible impulse responses... as do my own speakers. In my own system I can fix it for the most part with digital delay, but only for one location... if you do not have coincident drivers, you can never have perfect impulse response everywhere.

How important is it? In conversations with merlin and others, it seems to be a big deal for some and far less for others. Most horn based systems do come up pretty short in this regard. I don't know if that old Fostex system with the radical offset was really time aligned or just had time align styling like so many others.


Widget

[Ian Mackenzie]

There is a good article on a private web page I will try and find regard .step response.

Its one thing to have a smooth step response and another for a system to pass a square wave:blink:.

http://customanalogue.com/elsinore/elsinore_2.htm

http://www.customanalogue.com/elsinore/elsinore_8.htm
http://www.customanalogue.com/elsinore/elsinore_mk2.htm

[4313B]

How important is it? In conversations with merlin and others, it seems to be a big deal for some and far less for others.That pretty much sums it up. Some people care, some people don't.

Some people go looking for trouble where there isn't any and they get caught up in tons of measurements. Some guys only hear via amplitude plots so if the plot doesn't look good the system can't possibly sound good. :rotfl: That's why corporations drill down the measurements and start taking subplots and then piece it all together so it looks real nice and pretty (but not too pretty), you know, for the guys who hear with their eyes. ;)

Some guys only use measurements to help pinpoint anomalies they are hearing.

Any of the 43XX systems with the baffle mounted slot tweeter have positively terrible impulse responses...Yes, that's true and it may even be why some guys can't stand to listen to them. :dont-know Given all that is against them they still manage to entertain alot of people and can sound pretty damn good while doing it. :)

[Robh3606]

Do you still have the plots?

Sure for some of it. The 250 plots were done on the fly so I have a response I can post later but I wouldn't say it is correct for the real deal. My DIY stuff is the real deal so I feel a bit more comfortable posting that. The 4344 is all over the place in comparison. Sure sounds good though.

The multi driver arrays change because of the geometry. As Widget says coincident makes the most sense. It became obvious you could optimze the driver layouts to have the best response in the listening window which makes sense.

Rob:)

[4313B]

Sure sounds good though.Yeah... goofy eh?

[Ian Mackenzie]

That pretty much sums it up. Some people care, some people don't.

Some people go looking for trouble where there isn't any and they get caught up in tons of measurements.

Some guys only use measurements to help pinpoint anomalies they are hearing.


Best post this year.

As far as diy goes I dare say people in many cases aren't familiar with how to measure what they might be hearing.

ie a resonance below a crossover point that needs notching, mismatch in off axis polar of the woofer/horn and crossing over a horn way to low (distortion), early reflections.

While I dont go with everything Earl Geddes says he knows what & where the perception thresholds problems are that irritate listeners and how to design around the problem areas.

I mean if the early reflections off a side wall swamp all the crude happening inside the speaker box and that is most audible / irritating and you start to control that then its worth looking at the other problems.

[boputnam]

In my own system I can fix it for the most part with digital delay, but only for one location... if you do not have coincident drivers, you can never have perfect impulse response everywhere.Hey, Widget...

I don't think it's as bad as you suggest? Once you "remove" the combined effects of the electronic timing differences and driver offsets, the coherence is much improved pretty much everywhere in the listening field. Or, are you suggesting that baffle reflections vary by listening position...?


Some guys only use measurements to help pinpoint anomalies they are hearing.I live in this space - every minute of every show.

I run my reference signal into Smaart from the plf (solo, or monitor) output. That way, the frequency response of whatever input I solo can be seen on an RTA. It really helps me be efficient and effective in applying accurate (high Q) notching to specific anomalies using that input's filters. When not solo'ing, the rest of the time I watch the Magnitude (Transfer Function) trace, ever on the hunt for room / system anomalies...

[Mr. Widget]

Hey, Widget...

I don't think it's as bad as you suggest? Once you "remove" the combined effects of the electronic timing differences and driver offsets, the coherence is much improved pretty much everywhere in the listening field. Or, are you suggesting that baffle reflections vary by listening position...?What I am saying is that even using digital delay to "time align" a system, you can only have it perfectly aligned for a single point in space. If the drivers were coincident then they would be time aligned throughout the room.

However as I have posted... I am not really certain that it matters all that much. I am sure that it is not a black and whit issue. As the time smear increases I am sure at some point it becomes increasingly problematic.


Widget