Sound card & microphone fixes

[juicehifi]

Getting valid measurements is the biggest challenge for us who support new users. Usually we find a way, but we struggle to keep up with what works with which gear.

If you had problems getting a good measurement and found a solution, please share.

 

[whoareyou]

Exasound + UMik1 -
ASIO is only option for Exasound DAC
UMik1 WASAPI driver does not work with this combo. AL accepts the Exasound ASIO / UMik1 WASAPI combination and measurements seem ok, but timings are consistently out of whack.

For my configuration, only thing that works is direct mode for UMIK1. I always disable WASAPI.
This provides consistent results.

 

[2234rew]

My DAC is actually my old Denon AVR 4520. I run it in "Pure Direct" model using it as 8 channel DAC and 8 channel amp - with a remote control for volume.

It is 2 x DIY 3-way speakers plus 2 subs.

Audiolense handling digital crossover and speaker/room correction.

AVR on Windows can be a nightmare with getting channel mapping to work correctly with Audiolense

I use a Focusrite Solo 3rd Gen + Earthworks M23 mic, so I use ASIO4ALL for measurements

With previous Audiolense 6.17 I used the clock drift compensation without problems.

But with recent AL 6.19 update (and ASIO4ALL update to 2.15) I was getting a 0.7 second delay on my right side drivers and sub.

Measurements looked fine by I lost my phantom center completely. Sound was coming from left and right speakers only. I repeated measurements multiple times and even rolled back ASIO4ALL version.

So finally i disabled clock drift compensation this time and now left and right were under 0.1s difference and phantom center returned.

I lost a couple days trying to solve why the sq was negatively affected, but now am very happy with how it all sounds.

If using a pro audio multichannel interface (Mic input and DAC outputs) and it's ASIO driver you won have any of these issues,

But using an AVR like this is cool - all in 1 box and remote control for volume.

The Denon volume control is actually analogue (IC), even thought it shows volume on digital display

 

[Marcus DiBenedetto]

Exasound + UMik1 -
ASIO is only option for Exasound DAC
UMik1 WASAPI driver does not work with this combo. AL accepts the Exasound ASIO / UMik1 WASAPI combination and measurements seem ok, but timings are consistently out of whack.

For my configuration, only thing that works is direct mode for UMIK1. I always disable WASAPI.
This provides consistent results.

Hi. I will be using Audiolense XO to analyze my surround system (5.2) with my ExaSound S88 DAC. I have the UMIK1. I would like to draw on your experience to save me from similar issues.
-- Music server to S88 via USB.
-- S88 to Parasound A52+ via XLR.
-- Front three speakers are full range and each includes a powered subwoofer.
-- UMIK1 to music server via USB.

Screenshots:
1. Audiolense XO Setup
2. Current Measurement Setup. I get sweeps through speakers.
3. Measurement Setup Choices. I do not see "direct sound" as an option.

Your Thoughts?

Thanks in advance.

 

[whoareyou]

Take a look under advanced settings, and you should see an option to enable the direct sound option.

 

[juicehifi]

What he said, and it is sometimes a matter of trial and error, especially with the two components you are using. Do several measurements and look for consistent timing differences between tweeters. Ideally these differences should only vary by max 2 samples.

 

[mccarty350]

Heads up, this is all on my side as theorized, the minidsp UMIK only supports 48khz recording. This is NOT an audiolense issue.

 

[juicehifi]

I am just about to delete those two posts, mc350. I want this thread to contain solutions only. No pun intended. And soon this post will be gone too.

 

[2234rew]

Do several measurements and look for consistent timing differences between tweeters. Ideally these differences should only vary by max 2 samples.

So for 48kHz measurement:

2 samples = 1000 * (2/48000) = 0.05 ms ?

So tweeters should not vary by more than 0.05 ms?

0.1 ms is too much?

 

[juicehifi]

Ideally, yes. But while a larger variation than this is indicative on some inconsistency in the measurement process, the drift is likely to be inaudible.

 

[2234rew]

Ideally, yes. But while a larger variation than this is indicative on some inconsistency in the measurement process, the drift is likely to be inaudible.

Thanks Bernt - but any drift "should" be able to be corrected by AL with the drift compensation enabled?

 

[juicehifi]

Yes, but it requires glitch-free streams during the measurement to work. This should be a given, but unfortunately that isn’t always the case.

 

[2234rew]

Yes, but it requires glitch-free streams during the measurement to work. This should be a given, but unfortunately that isn’t always the case.

Yes noted. I've experienced such glitches during measurements but it's hard to know if its the driver issue or ASIO4ALL issue or something else.

I have to play with some of the measurement options like Exclusive Mode on/off and some of AL's other options.

Obviously the ideal is have one interface for both ADC and DAC during measurement but sometimes not possible.

For example I use 8 channel HDMI input DAC and amp in my AVR - I use DSP crossover.

The AVR solution is really neat but this AVR doesn't have 8-channel RCA input (older Denons did) so I can't simply use an 8-channel pro interface for the measurement for its single clock because I don't have 8channel power amp.

 

[pl1jn]

I have spent a lot of time trying to get going but i cant get the "test signal" out of my speakers so i cant take any measurements. I am using Steinberg UR12 and Behringer ECM8000 microphone . I have downloaded the correct drivers for the
UR12. I attach a pic which details the full system. I can get the test signal out my laptop which i am using as a roon core.

 

[juicehifi]

hi @pl1jn,

I want this thread to be packed with solutions to sound card problems, not the problem solving itself. I recommend that you start a separate thread searching for yours. And when you find the solution please post it here.

 

[mccarty350]

Here are some solutions I've found to contribute Bernt.

Getting measurements in a stable manner with an ASUS Xonar U7 MkII and Umik.

Using ASIO as the playback device works fine on the Xonar U7 MKII. I use WASAPI as the input device. These are probably evident suggestions here are the ones that are not so much evident:
1. Lowering your sweep duration from 10 to 5 seconds definitely increases your chances of success.
2. Go into advanced settings and then turn WASAPI exclusive mode to enabled.
3. EVERY TIME YOU TAKE A MEASUREMENT turn clock drift correction on and off. It resets something in the software, I'm not sure what but when you turn it on and off the entire UI seems to flash for a second like something major has happened. I don't know what it is but I was able to take over 10 measurements without audiolense crashing by toggling this between measurements. I then tried a few measurements without toggling it and got an immediate 'file open' error so I believe this is a method of increasing stability.

Another thing I've noticed:

Under speaker setup as you go through the tabs from left to right from speakers to cross over configuration there are two places where you can set octave width. Make sure that when you change it in the cross over configuration tab that you return to the speakers tab to verify that it 'stuck' because often it will return to it's default 2 octave value so MAKE SURE that you check it before saving your speaker setup configuration.

 

[juicehifi]

^Excellent contribution

 

[mccarty350]

If I find out any more tips and tricks I'll post them. I will also say that using this sound card and configuration when I A/B the exact same correction file and measurement with the only change being using clock drift correction versus not having clock drift correction I find that turning it off sounds a bit 'clearer' in my scenario and I prefer it. Bernt if I'm off target on any of this or anything is just perceptual on my part let me know, i don't want to spread misinformation. I must say that I have made tweaking with this quite a passion I spend at least an hour a day playing with audiolense. I also want to share (and I know maybe I shouldn't add clutter to this thread but I'm proud) my active modular testbed:

Basically I have dual subs in the room, the 8" bass module above (still have to add round overs for cosmetics) on the bottom, a midrange module that uses screw inserts and machine screws so that I can exchange baffles and drivers on the front and back. On the back I can remove the rear baffle so that I have an 'open midrange' configuration. I have at least half a dozen tweeters that I can place on top of the box each with their own baffle and a plethora of waveguides. I have Bridgestone ISO pond liner between each module for isolation and the construction techniques of the mid enclosure adheres to BBC learned techniques (bitumen damping and 1/2 inch birch ply) while the bass module is constructed of far sterner stuff. This is why I never stop playing with audiolense, I'm always switching drivers, crossover designs, etc. and ever optimizing and documenting. So much fun. If anyone is wondering why the center to center distance on the mid module is so great vs the woofer it's because I built it to be able to accommodate up to 6 1/2 mid drivers and the one shown is only 4" so I chose to optimize it's center to center distance to the tweeter rather than to the bass module. Learning is a blast and audiolense has made this endless tinkering off the charts gratifying.

 

[Cornelis Huizinga]

An interesting post, mccarty350. I played with some of your targets for a while.
Your measurements show a textbook step response. I'm curious how you managed that. Mine look a lot messier.
Has the room been acoustically treated? Your linear step response just looks more like a minimum phase. Is it maybe a mix of linear and minimum phase? Your step response target will also continue horizontally after the peak instead of falling.
Looks like you are a very experienced user and maybe you have some tips to get a better step response.
So far I haven't really been able to get a better step response.
Despite that, I am very satisfied with the sound quality that I have achieved with Audiolense.

 

[mccarty350]

Good questions, let me give the answers with things as far as I know (and I would love to be told I'm wrong and taught better lol):
1. I'm using the button in each target that designates linear phase and then in my corrections I'm doing total time domain filters without checking minimum phase which I think results in linear phase correction. Where did I post measurements BTW? I think I only posted room curves.

2. The room has bass traps where I've been working and has some owens corning 703 panels so it is treated to a degree although probably not perfectly.
3. I don't know if I'm an experienced user but I've put a lot of time in. one thing I've done is to use the full 130k tap filters and then go into the filters and designed some with longer cycles times for both bass and treble. I THINK that basically if you increase your cycle times that audiolense has more time to create corrections but the negative is taking up more taps/making things more granular and also as a result increasing latency but I leave that to Bernt to confirm, it's just my suspicion as a laymen that has spent a lot of time playing.
4. Audiolense's sound quality is just staggering, figuring out how to wrench every last bit of improvement out of it is the challenge but I love learning through trial and error.

Bernt, let me know if we're murdering this thread and we can take this out of here.

 

[mccarty350]

I just had another thought Cornelis. Are you correcting speakers that have passive crossovers or are you using multichannel with an amp channel per driver and doing correction on a per driver basis? If you're correcting a set of speakers with passive crossovers you won't be able to do 'per driver' correction. I am fully active and am doing correction per driver and that might be what you're seeing. Does that make sense or am I speaking gibberish?

 

[Cornelis Huizinga]

You certainly don't speak gibberish, but I was confused with the measurement of 2234rew (May2, 2021) I opened it assuming it was your measurement :-(. His step response looks perfect by the way.
I too have an active system. To be precise; diy three way speakers.
My room is the living room without special acoustic treatments.
I hadn't tried that many taps before, but I did see that the step response target changes compared to 65k taps.

 

[Badbruno]

Exasound + UMik1 -
ASIO is only option for Exasound DAC
UMik1 WASAPI driver does not work with this combo. AL accepts the Exasound ASIO / UMik1 WASAPI combination and measurements seem ok, but timings are consistently out of whack.

For my configuration, only thing that works is direct mode for UMIK1. I always disable WASAPI.
This provides consistent results.

Thank you for this insight with ExaSound, it’s my next tuning project
cheers… thanks for sharing

 

[whoareyou]

Thank you for this insight with ExaSound, it’s my next tuning project
cheers… thanks for sharing

Interestingly enough, I just received new computer and measuring with WASAPI now works (same UMIK-1 and same exasound). When I measure, I still use direct sound since it's given very consistent results, but I was a surprised that WASAPI now works for me. Bottom line is, I'm sure you'll get a good measurement, but every bit of hardware/drivers in the chain seems to impact process.

 

[Badbruno]

This is good to know! I would also say that when there is a mismatch ( delay) the first thought is the signal chain in computer is the problem. Not always the case I’ve had two systems where Audiolense found a hardware problem.
I had just rebuilt passive 4 way JBL L250 xover and the software pointed out an mis wire on left channel. Second issues found was a amplifier inverted polarization one the channels.

I wish Audiolense would add a POP test for polarity testing speakers