Is This Considered A Decent Step Response After Audiolense Correction

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Hello,

I am trying to interpret Step responses and even after reading Mitches book on Acourate that explains how to interpret a step response to figure out if the drivers are time aligned I am still struggling.

I can post this also on the REW forum but firstly want to know if there is an anomaly or if the step response is okay considering it has been corrected with Audiolense? This is a measurement with REW from the listening position with the DSP on.

?
56687
 

Omid

Member
Joined
May 28, 2017
Messages
131
Hello,

I am trying to interpret Step responses and even after reading Mitches book on Acourate that explains how to interpret a step response to figure out if the drivers are time aligned I am still struggling.

I can post this also on the REW forum but firstly want to know if there is an anomaly or if the step response is okay considering it has been corrected with Audiolense? This is a measurement with REW from the listening position with the DSP on.

? View attachment 56687

Unfortunately that does not look great. Here's what mine looks like for example. It looks pretty close to the target step AL showed. The target step response is the ideal response you should get given your target curve.
56688
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Okay that is bad news. My questions then are

1. I didn't measure with loopback does that make a difference?

2. What do I do in Audiolense to try and get that response? If its automatic correction then how do I or to what do I make a change to ensure that I get a proper step response with REW?

It seems my target step in AL looks more like yours but not as good. Unfortunately, the filter is old so I just did another generation to see how the simulated step reponse will look like. Here it is, but it is approximately, same crossover points, same filter and measurement.


56689
 

Omid

Member
Joined
May 28, 2017
Messages
131
The timing reference in REW (either acoustic or loopback) shouldn't make much difference (it makes the estimated time delay between signal and impulse more accurate)
The pic above doesn't show your ideal target, but yes it usually looks more like mine.
You need to read through the various instructions on the forum and elsewhere on how to change the variables until you get a good frequency, and step response. The target response is key (low and frequency ends),
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
The timing reference in REW (either acoustic or loopback) shouldn't make much difference (it makes the estimated time delay between signal and impulse more accurate)
The pic above doesn't show your ideal target, but yes it usually looks more like mine.
You need to read through the various instructions on the forum and elsewhere on how to change the variables until you get a good frequency, and step response. The target response is key (low and frequency ends),

Can you give me an example of those variables so I can get an idea? It will just point me in the right direction.

And is that what you did play with those variables till you got a decent step response in REW or did the automatic correction give you a good result first go? Or do you play around with it till you get a decent simulated correction in Audiolense?
 

Omid

Member
Joined
May 28, 2017
Messages
131
I did not bother checking in REW until AL predicted a good response. The predictions are very accurate. This allows you to go through numerous iterations rapidly until you land on the right one. Look at Mitch's articles in audiophile styles and his youtube video.

You can attach a zip file with your measurement, speaker setup, target and see what people can do.
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
I did not bother checking in REW until AL predicted a good response. The predictions are very accurate. This allows you to go through numerous iterations rapidly until you land on the right one. Look at Mitch's articles in audiophile styles and his youtube video.

You can attach a zip file with your measurement, speaker setup, target and see what people can do.

Okay got it, so I need to play around with it till the prediction looks good.

I have read all of Mitches articles, meticulously, but not from the point of view of what needs to be tweaked to get a decent step response, I will take another look.

The main things that can be tweaked in AL is

1. The target curve
2. Filter Length
3. Time Domain Sub windows which I think is the main thing that is tweaked but does that effect the step response? I definitely do not know how this works yet but Mitch has something on it in his articles.

I mean do any of them influence step response?
 

Omid

Member
Joined
May 28, 2017
Messages
131
filter length I would play with last.
Time subdomain is usually 1-2 cycles short of window length. The longer the window length the higher the resolution of the correction.
Crossover frequency and target can have a big impact.
The suppress pre-ringing box can really affect your step (not always in a good way).

I have gone through hours of various iterations before getting a really good response. It takes time, don't get discouraged.
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
filter length I would play with last.
Time subdomain is usually 1-2 cycles short of window length. The longer the window length the higher the resolution of the correction.
Crossover frequency and target can have a big impact.
The suppress pre-ringing box can really affect your step (not always in a good way).

I have gone through hours of various iterations before getting a really good response. It takes time, don't get discouraged.

Thats brilliant Omid, that helps a lot. Thanks.

And now that you explain it in this way, I am presuming the XO widths also play a role in time alignement.
 

Omid

Member
Joined
May 28, 2017
Messages
131
Thats brilliant Omid, that helps a lot. Thanks.

And now that you explain it in this way, I am presuming the XO widths also play a role in time alignment.
Yes for sure. There are so many variables to play with. That's why it takes time. It'll be worth it.
 

Omid

Member
Joined
May 28, 2017
Messages
131
PS. Make sure your microphone stays in the exact same place that you performed the initial measurement with audiolense, when crosschecking with REW. If the placement is not the same, the step response will look different than predicted.

Also make sure that your microphone is exactly centred, meaning that they should be 0 ms delay between right and left tweeter.
 

kevinzoe

Member
Joined
Jun 10, 2022
Messages
39
More  
Preamp, Processor or Receiver
PrimaLuna DiaLogue Premium preamp
Main Amp
PS Audio BHK Signature 250 stereo amp
Additional Amp
Valve Amplification Company 30/30 Renaissance
Computer Audio
JRiver
DAC
exaSound e48 (8 channels)
Front Speakers
Acoustic Zen Crescendos
Subwoofers
Pair of Rythmik F15's
Other Equipment
Cables by: TEO Audio, Cardas and AudioSensibility
Trdat,
As the step response and frequency response and mathematical derivations of the impulse response, maybe we should take a step back and consider some early step fundamentals first . . .

> when you take a measurement are the tweeter delay amounts (right most column) equal or within 0.04ms of each other if you're bi- or tri-amping? If using a single amp then is each speaker's delay the same or nearly the same?
> Correction Procedure Designer - try 10 and 8 cycles in the upper window for 10hz and 24khz respectively and try 7 and 6cycles in the lower window as a starting point. You'll need to experiment here with longer windows to take more of the room into affect. Keep the top window's cycles 2-3 more than the lower window which I've seen Bernt mention. You might want to start with 4-6dB Max Correction Boost and filter length of 32K or higher. I also click the boxes for "Prevent treble and bass boost"
> how have you got our crossover settings laid out?
> for target curves make sure the high end mimics your speaker's natural roll off and take the bass on the target curve right to 1hz at say -80dB. I've seen Bernt mention this somewhere so it must help the software in some way,
Try these things to see if there is room for improvement
 

Mitchco

Member
Joined
Oct 1, 2017
Messages
79
@Trdat a good place to start is having a look at your measured step response: Analysis->Measurement->Step Response Measurement.
Two views/screen shots. We want to look at the first 100ms in one screen shot and the first 5ms in the 2nd. Thanks.
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
@Trdat a good place to start is having a look at your measured step response: Analysis->Measurement->Step Response Measurement.
Two views/screen shots. We want to look at the first 100ms in one screen shot and the first 5ms in the 2nd. Thanks.

Hey Mitch there is a step response for tweeter, midwoofer and sub do you need them seperately or all in one screen because I can't manage to get them all within 5ms or let alone 100ms in the one screen shot. Unless I post them indivually or there is a way to set the size somewhere?

I am presuming the way the measured step response looks there is a certain ways you can tweak AL to get it right, anyway I'll post it and any tips will be appreaciated.
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Trdat,
As the step response and frequency response and mathematical derivations of the impulse response, maybe we should take a step back and consider some early step fundamentals first . . .

> when you take a measurement are the tweeter delay amounts (right most column) equal or within 0.04ms of each other if you're bi- or tri-amping? If using a single amp then is each speaker's delay the same or nearly the same?
> Correction Procedure Designer - try 10 and 8 cycles in the upper window for 10hz and 24khz respectively and try 7 and 6cycles in the lower window as a starting point. You'll need to experiment here with longer windows to take more of the room into affect. Keep the top window's cycles 2-3 more than the lower window which I've seen Bernt mention. You might want to start with 4-6dB Max Correction Boost and filter length of 32K or higher. I also click the boxes for "Prevent treble and bass boost"
> how have you got our crossover settings laid out?
> for target curves make sure the high end mimics your speaker's natural roll off and take the bass on the target curve right to 1hz at say -80dB. I've seen Bernt mention this somewhere so it must help the software in some way,
Try these things to see if there is room for improvement

This is great info thanks. I think just the first recomendation I don't know much about.

1. I do have different amps for each frequency range and have absolutely no idea to figure this out? When you say right most column you mean in AL there is somewhere which I can check if each frequency range is withn .04 sec of each other?


2. I read Mitches book for the 4th time and no I am slowly understanding about the FDW and started experimenting but still not sure about the FDW in the timedomain so I will experiment with your recommendations. When you say lower window I presume you mean the timedomain subwindow?
3. Before I had different widths and different crossovers with different widths to each crossover but I experimented with the same crossover point with at least the same width at that crossover junction and I am getting better simulated step responses.
4. I have heard about mimicing the nartural roll off of both the low end and high end, I will also try this for the high end and see what I get with -80db at 1hz for the low end.

So far it seems setting the preringing off always gives better results probably at the cost of preringing but I am yet to see where I read if I am getting preringing in the graphs.
 

Mitchco

Member
Joined
Oct 1, 2017
Messages
79
Hey Mitch there is a step response for tweeter, midwoofer and sub do you need them seperately or all in one screen because I can't manage to get them all within 5ms or let alone 100ms in the one screen shot. Unless I post them indivually or there is a way to set the size somewhere?

I am presuming the way the measured step response looks there is a certain ways you can tweak AL to get it right, anyway I'll post it and any tips will be appreaciated.

Have a look at the measured step response graphs in this article: https://audiophilestyle.com/ca/ca-a...fers-with-stereo-mains-using-audiolense-r712/ One is showing the first 20ms and the other about 45ms.

Use AL zoom to show us the first 5 or 10 ms and yes it will cut off everything to the right, that is fine. Then zoom out and show the first 200ms.

If it is easier, DM your measurement file or upload here on the forum. Cheers.
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Have a look at the measured step response graphs in this article: https://audiophilestyle.com/ca/ca-academy/ integrating-subwoofers-with-stereo-mains-using-audiolense-r712/ One is showing the first 20ms and the other about 45ms.

Use AL zoom to show us the first 5 or 10 ms and yes it will cut off everything to the right, that is fine. Then zoom out and show the first 200ms.

If it is easier, DM your measurement file or upload here on the forum. Cheers.


I can't attach the measurement file here for some reason. Any ideas?

I don't really know where the measure\ment is starting from so I will analyse it more and see if I can get a better shot and upload it again, I think I got it.
 

Attachments

  • Step1.jpg
    Step1.jpg
    299.7 KB · Views: 34
  • Step2.jpg
    Step2.jpg
    280.8 KB · Views: 32
Last edited:

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
I think this zooms in better.
 

Attachments

  • Step3.jpg
    Step3.jpg
    303.6 KB · Views: 25

Ofer

Member
Joined
Mar 15, 2021
Messages
244
More  
Preamp, Processor or Receiver
miniDSP 4x10hd
Main Amp
Emotiva XPR200 midrange amp
Additional Amp
Crest audio 2001A bass amp, Crest audio 8002 sub
Other Amp
Rotel RA930ax twitter amp
Computer Audio
Sony Bravia android TV
DAC
RME FF400
Universal / Blu-ray / CD Player
Marantz original 5E CD
Front Speakers
Andromeda MkII
Subwoofers
18" OEM powered subwoofer, 18" Martycube Dayton A.
Screen
Sony bravia 65XF9005
Streaming Subscriptions
Deezer HiFi
Other Equipment
HP i5 running W10, HLC convolver for Audio Lense filters
I can't attach the measurement file here for some reason. Any ideas?

I don't really know where the measure\ment is starting from so I will analyse it more and see if I can get a better shot and upload it again, I think I got it.
You need to zip the files to attache them
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Thanks.
 

Attachments

  • 2.0 High End PA 200Hz 950hz wider high pass 12 Sep 21_12 52Speakcloserexcdyn.zip
    2.9 MB · Views: 10

Mitchco

Member
Joined
Oct 1, 2017
Messages
79
Cheers, Ofer! @Trdat I am afraid the timing response of this measurement is not good to start with:

56806


Once a measurement has been taken and saved, you will want to check the timing by selecting new measurement again and then looking at the delay column.

Assuming the mic is centered and equal distance from both left and right speakers, we should be seeing a much tighter tolerance between left and right channels. For example, the left and right midrange drivers. The left is reading 0ms and the right 6.71 ms. Given sound travels roughly 1 foot per millisecond...

What I recommend is taking 3 measurements back to back, changing nothing. Then inspect the delays, if each of the measurements vary by milliseconds, (i.e. not consistent) then you will need to spend some time trying various combinations of playback and microphone input devices (like WASAPI or Direct Sound) or use ASIO4ALL and play with some of the Advanced Settings. Assuming you are using a USB mic or separate interface for the mic, ensure clock drift correction is enabled. Try enabling/disabling separate play and recording streams. Try Playback and/or recording problems workaround. One of these combinations should give you consistent/repeatable results. Good luck!
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Cheers, Ofer! @Trdat I am afraid the timing response of this measurement is not good to start with:

Once a measurement has been taken and saved, you will want to check the timing by selecting new measurement again and then looking at the delay column.

Assuming the mic is centered and equal distance from both left and right speakers, we should be seeing a much tighter tolerance between left and right channels. For example, the left and right midrange drivers. The left is reading 0ms and the right 6.71 ms. Given sound travels roughly 1 foot per millisecond...

What I recommend is taking 3 measurements back to back, changing nothing. Then inspect the delays, if each of the measurements vary by milliseconds, (i.e. not consistent) then you will need to spend some time trying various combinations of playback and microphone input devices (like WASAPI or Direct Sound) or use ASIO4ALL and play with some of the Advanced Settings. Assuming you are using a USB mic or separate interface for the mic, ensure clock drift correction is enabled. Try enabling/disabling separate play and recording streams. Try Playback and/or recording problems workaround. One of these combinations should give you consistent/repeatable results. Good luck!

You learn something new everyday hey...

I had no idea that we needed to check the delays during a measurement and ensure they were within a decent threshold. I had read it in your article but presumed Audiolesne would take care of the timing. So it seems we need decent delays to start of with which is also what kevinzoe mentioned in his list of things to look out for as the rest of the list I knew about except that particular tip.

ASIO never works, ever, although I am sure the focusrite solo is ASIO compatible. I do use clock drift correction but I will have to try disabling and enabling random things in the Advance settings and see. When you say try playback or recording problems workaround is this a tab in the settings?

So I am presuming a better measurement will give me better results. I will measure up again, thanks so much for this I absolutely had no idea.

I did though try different combinations of widths, crossovers and FFD windows and I did get much better simualated step response. For now, a better measurement with decent delays and more teaking to get a better step repsonse. I also need to read in your e-book on how to get a decent group delay.

Appreciate everyones help.
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
So I have been trying to remeasure.

And I have had this problem since day one, what happens is that I never can use the focusrite ASIO or even with my previous presonus ASIO driver to measure with an ASIO driver.

I use a focusrite solo and a minidsp U-DIO8 as a digital to digital converter for my 8 channels.

The only way for me to choose the focusrite ASIO driver is for me to switch the output from the minidspASIO to the USBstreamer mutlichannel either directsound or wasapi. But then that leaves weird speaker check connections with the output with the USBstreamer, meaning that either some speakers wont play or they wil play in the wrong order when with the minidspASIO chosen the speaker connection is absolutely fine. It might even be that I only measured this way previously with each channel with wrong speaker connections but I should have noticed that.

When choosing minidspASIo for output and focusrite ASIO for input it says that the coundcard is supported but I get an error message when running measurement.

But there should be no reason for me not to use the focusrite as its driver is an ASIO driver right?

On the occasion I do measure with the above output and input option all the delays are totally out of whack by huge proportions, it could be from not measuring ASIO it also could be from the weird speaker connections when the USBstreamer for output is chosen, to be honest I have no idea what the difference between the USBstreamer and the minidspASIO is.

Anyway, the point is why would the focusrite chosen as the input not work? Do I need a universal ASIO driver? Is there something I may have missed?

In REW from the dropdown list the option is Analogue 1 +2 (focusrite USB audio) which I am presuming is not the ASIO driver because in Audiolense there is an option as well from the drop down list Analogue 1 + 2(Focusrite USB audio) Wasapi/Directsound
 
Last edited:

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Out of curiousity if out by 1 millisecond is okay? I got 4.5 and 4.6 between tweeters?

And I chose miniphase measurument to achieve this result what exactly is miniphase...?

I did manage exactly the same sec between tweeters so it is possible, but it was a bad measurement so I need to try again but at least it tells me it is possible.
 

juicehifi

Audiolense
Staff member
Joined
Feb 5, 2018
Messages
698
Re the asio error: You can not use asio for both when you have separate input and output devices. But you can e.g. use asio in one place and wasapi in the other.
 
Top Bottom