Using AV Receiver such as Denon, Marantz for Audiolense measurements

[2234rew]

I have one HQP license. Switching endpoints thus requires logging into my Roon core server, which also is my HQP host, and changing the HQP settings accordingly. That’s more of a hassle than just switching endpoints in Roon, so I’ve used HQP only for the three endpoints that I can scale to DSD. Of course, all my HDMI endpoints are 24/192. Maybe I am missing out on some sonic benefit to HQP at 24/192. JCR

Thanks JCR, yes that makes sense.

Changing HQP endpoints with HQP Desktop version isn't as quick as changing Roon endpoints from an iPad for example.

I use HQP Embedded version though, so i can change endpoints from iPad or iPhone (anything with web browser).

 

[juicehifi]

^^Yes, that should work as long as you have 8 channels "activated" in Windows.

And check the connections by using a 3s sweep at low level. You can cancel the session once you hear audio from the wrong channel.

 

[2234rew]

^^Yes, that should work as long as you have 8 channels "activated" in Windows.

And check the connections by using a 3s sweep at low level. You can cancel the session once you hear audio from the wrong channel.

Hi @juicehifi

Some good progress running AVR HDMI measurement today.

So even though the 'speaker connection test' was wrong again, the measurement itself sounded correct (correct sweeps coming from correction drivers).

In terms of measurement quality, Audiolense reported both Good and Excellent dynamic ranges (I repeated the measurement numerous times at differing volumes - see screenshots below.)

However in AL itself, I can't view the measurement or even filter it at all.




What can be the cause of this? Measurement quality reported as Excellent but I can't view anything? And can't filter the measurements and can't create corrections.

The measurement zip file is attached also.

I got one step closer and then hit a wall again.

 

[juicehifi]

I will take a look and get back to you, 2234rew.

 

[2234rew]

I will take a look and get back to you, 2234rew.

Thanks Bernt. Hoping there's something in the measurement file that can point me in the right direction.

I got one step closer with sweeps working well and then saw the desired Excellent measurement screen and smiled thinking I had solved the HDMI case - and then hit this brick wall.

If I can get this to work I will write up a bit of a guide to help others but an interesting observation was: I still need needed to set the AL output channel numbers to the correct Windows spec, shown below.

Even though the Audiolense 'speaker connection test' didn't follow it correctly, the measurement does respect the Windows spect (as you said it would in emails and earlier in the thread).

 

[2234rew]

2234rew,

You should only specify used channels in Audiolense. If you measure something that isn't producing audio you will get a measurement error.

Hi @juicehifi - I got it working! I'm currently in COVID lockdown so I spent a bit of time on this. I had the spare time :-)

I will write up how I got there shortly but wanted to let you know.

Simulations are different to previous measurements using Focusrite 18i20 Gen3 interface as the DAC.

I'm still playing around with simulations to see why there are differences and optimisations . Both measurements are attached.

Delays are a bit different too.

Both measurements use Focusrite + Earthworks mic for ADC microphone measurements (Focusrite ASIO driver).

8th April measurement uses the Focusrite as the DAC, feeding Denon's analogue inputs

26th April measurement uses Denon as the DAC, feeding Denon AVR's HDMI input

Both measurements use the Denon for amplification of each driver

 

[2234rew]

Also a question @juicehifi - when using clock drift correction like I did with the HDMI connection, is the sweep supposed to sound a bit different compared to when I use only Focusrite interface mic/DAC WITHOUT clock drift correction?

I heard some 'tapping' sounds during the proper measurement sweeps when using clock drift correction, that I hadn't heard before. The tapping sounds are a bit similar to the 'speaker connection test' sounds.

The filltered frequency response measurements look pretty similar though, even though simulation looks different, strangely.

Probably related to time domain more than frequency domain (since delays are different).

New (HDMI using ASIO4ALL) - with clock drift correction:




Old (Focusrite for both DAC and mic ADC using ASIO driver) - without clock drift correction:




You can see below, left speaker tweeter and woofer delays are different

 

[2234rew]

And trying to optimise the new HDMI measurements (dated 26th April) I need to increase the 10Hz time domain subwindow to 10 cycles / 1000 ms to get nice simulations. Anything less than 9 cycles isn't nice.

Is 1000ms at 10 Hz too large? Not recommended?

The simulations do look pretty though LOL

 

[juicehifi]

10 TTD cycles is a lot, but if it works just try it out. But the general recommendation is to keep the TTD window shorter than the correction window.

However, those clicking sounds you reported shouldn't be there. THe clock drift correction is post processing, the seeps are the same.

 

[juicehifi]

The delay differences are due to glitches in the data stream.... at least in one of the measurements.

 

[juicehifi]

Also a question @juicehifi - when using clock drift correction like I did with the HDMI connection, is the sweep supposed to sound a bit different compared to when I use only Focusrite interface mic/DAC WITHOUT clock drift correction?

I heard some 'tapping' sounds during the proper measurement sweeps when using clock drift correction, that I hadn't heard before. The tapping sounds are a bit similar to the 'speaker connection test' sounds.

The filltered frequency response measurements look pretty similar though, even though simulation looks different, strangely.

Probably related to time domain more than frequency domain (since delays are different).

New (HDMI using ASIO4ALL) - with clock drift correction:

View attachment 41499


Old (Focusrite for both DAC and mic ADC using ASIO driver) - without clock drift correction:

View attachment 41498


You can see below, left speaker tweeter and woofer delays are different

View attachment 41500

I have taken a look at 4 measurements that I have here from your rig. Two are simular to the one above right,

The question is which has the right timing or if both has the wrong timing?

I would have put my money on the hdmi measurement if you didn't mention the clicking. It looks most similar to a measurement here:

Heil AMT (Air Motion Transformer) Tweeter – AmpsLab

However, for all I know the little wrinkles we see could be highly position dependent ... and it may not even be the same model as you are using.

I suggest that you examine the sweepmeasurement file. You can open it in Audacity and look for discontinuities in the recorded sweeps. It will typically look like a part of the sweep has ben cut out ... or several parts.

PS you are wery close here. You could even be close enough that the time domain correction will work. The tweeter has practically the correct shape. There is perhaps a 4 sample difference for the midrange delay between the two measurement.

PPS: I did not see a blank measurement in you archive, but one labelled Mac something was corrupt and would not open.

 

[juicehifi]

And I checked the "Check speaker connection" from your measurement on my Lynx Hilo sound card. All channels checked as they are supposed to.

 

[2234rew]

Firstly @juicehifi - thanks so much for your help looking into this. Your help has helped motivate me to persist with trying AVR HDMI. I was enjoying things perfectly fine with my Focusrite. Lots of progress made.

And I checked the "Check speaker connection" from your measurement on my Lynx Hilo sound card. All channels checked as they are supposed to.

Do you mean for my HDMI measurement (26th April filename) that worked, not the previous days failed attempt? If so yes that's correct, I noticed that speaker connection test worked for the first time over HDMI when I changed from:



... and changed to ASIO4ALL for the first time ever which worked:




I never had any success with ASIO4ALL - I could never get ASIO4ALL popping up in the icon tray area . Yesterday I now realised it only pops up when you either start speaker connection test or the measurement. Huge lesson learnt.

So using ASIO4ALL resulted in correct AL channel mapping with speaker connection test.

So when I eventually write up a 'how to guide' for using AVR + HDMI, this will be an important note. Use ASIO4ALL and do the speaker connection test in AL to see the ASIO4ALL tray icon come and then configure ASIO4ALL.

Below are the measurement advanced settings that 'worked' for HDMI measurement although since you mentioned there should be no clicking during sweeps, I need to experiment more here:




When using ASIO4ALL 'playback problems workaround' this time it didn't seem to help and on my multiple trials, a couple times made things worse.

Notice also that I was able to get measurement done WITHOUT any of the WASAPI, Direct Sound, MME driver options enabled... instead I configured in ASIO4ALL. Is this the correct approach? Or could be one reason for the clicks I heard during sweep?

10 TTD cycles is a lot, but if it works just try it out. But the general recommendation is to keep the TTD window shorter than the correction window.

However, those clicking sounds you reported shouldn't be there. THe clock drift correction is post processing, the seeps are the same.

Noted, will re-try and play with different measurement advanced settings

I would have put my money on the hdmi measurement if you didn't mention the clicking. It looks most similar to a measurement here:
Heil AMT (Air Motion Transformer) Tweeter – AmpsLab

However, for all I know the little wrinkles we see could be highly position dependent ... and it may not even be the same model as you are using.

Agreed, there's multiple 'Heil AMTs' out there.

The tapping/clicking sound (which sounds similar to the speaker connection test sound) is a good reason to try different things in the measurements advanced settings screen and re-measure.

I suggest that you examine the sweepmeasurement file. You can open it in Audacity and look for discontinuities in the recorded sweeps. It will typically look like a part of the sweep has ben cut out ... or several parts.

Will do, thanks for this advice.

PS you are wery close here. You could even be close enough that the time domain correction will work. The tweeter has practically the correct shape. There is perhaps a 4 sample difference for the midrange delay between the two measurement.

Nice ! This was my first attempt since getting ASIO4ALL working. Still some things I can play around with in the measurements 'advanced settings' menu.

PPS: I did not see a blank measurement in you archive, but one labelled Mac something was corrupt and would not open.

I will email you this measurement file (measurement dated 25th April) which was not using ASIO4ALL - it was using the below settings.



Audiolense reported a Excellent measurement quality but had the blank measurement screen.

If you can spot something that would allow me to re-try without ASIO4ALL I am happy to try too because there were no clicks during these sweeps.









Thanks heaps again

Things are very close as you say :-)

If I can get this working I think a good 4 ohm capable Denon AVR + miniDSP UMIK 1 or UMIK-2, is a superb combination for DSP active speaker setup.

In my case, 9 independent DACs channels and 9 independent amp channels in one box for decent price with decent power.

I'd love to write a guide with tips of things to do / try to help others in future.

 

[juicehifi]

Yes, when you use Asio4All you don't need wasapi etc.

I have no answer to why you got this "empty" impulses in the measurement with the good quality reports. i have never seen this happen without an error message in the post proccessing. It may have been a freak accident. Maybe it won't happen next time.

Was this btw the measurement with "mac" in the file name that you included in the download zip above?

 

[2234rew]

I have no answer to why you got this "empty" impulses in the measurement with the good quality reports. i have never seen this happen without an error message in the post proccessing. It may have been a freak accident. Maybe it won't happen next time.

Ok will retry, thanks again

Was this btw the measurement with "mac" in the file name that you included in the download zip above?

Just to be clear, you mean post #28 of this thread?

Yes this is normal when you create a zip file on macOS and open it in Windows, you will always see this macOS folder below but you don't need to open it.

I created this zip file on macOS.

Just ignore this macOS folder in all cases.

 

[2234rew]

I have no answer to why you got this "empty" impulses in the measurement with the good quality reports. i have never seen this happen without an error message in the post proccessing. It may have been a freak accident. Maybe it won't happen next time.

I retried multiple times today and get the same empty measurements each time. And measurement quality still reported as 'Excellent' by Audiolense. I will explore the sweep in Audacity this weekend as you suggested.

But I have some good news - using ASIO4ALL I finally got a click free sweep over HDMI using these below settings. I don't know why the 'recording problems workaround' option helped here but it did.




New measurement file attached in .zip file.

I'm quite pleased with these initial results. I managed to make the HDMI input sound as good (subjectively) as Focusrite interface handling measurements and doing playback D to A :-)

I think I could replace the 10 channel Focusrite and just using a miniDSP UMIK-2 for mic measurements.

I think I can do some optimising but straight away it sounds as good and initial simulations look kind of 'respectable'.

I took your advice of keeping TTD subwindow less than frequency response window.

I also took your earlier advice to make all drivers positive polarity. I used the impulse response measurement to verify. Does the sub look like positive polarity to you? Its hard to tell because there is such a huge dip after.

Also is it normal for tweeter impulse peaks to be different magnitudes? See below.

Seems like the case of HDMI input of AVR has been mostly solved - it can definitely be done and with seemingly half decent results so far.

What are your overall thoughts on new HDMI measurements? Any other tips you can give would be appreciated.

New HDMI delays (ASIO4ALL, no clicking in sweeps and now all drivers in positive polarity I think):

Yesterdays HDMI delays (ASIO4ALL with clicking sounds and tweeters and sub were possibly in negative polarity):

Previous Focusrite interface USB delays (Focusrite ASIO driver handling everything, tweeters and sub were possibly in negative polarity):

 

[2234rew]

I would have put my money on the hdmi measurement if you didn't mention the clicking. It looks most similar to a measurement here:
Heil AMT (Air Motion Transformer) Tweeter – AmpsLab

However, for all I know the little wrinkles we see could be highly position dependent ... and it may not even be the same model as you are using.

This tweeter measurement shown is negative polarity yes?

Mine was initally like the above but since you mentioned we should aim for all drivers to be in positive polarity I now have this below after changing tweeter wiring: