TOPIC:

Wanderer wrote:

The response based on combined L+R measurements may have nothing in common with individual channel filters.

Yes but I'm not sure what a combined left and right measurement would have to do with the current discussion? The aim here is to measure each of the Dirac filters individually. You'd want to know the highest boost in either filter.

I essentially never make a combined measurement actually as they're of no use to me, since I use a mono bass signal below about 100 Hz.

Tony_J wrote:

On the convenience issue, I think it is a case of six of one and half a dozen of the other. My original approach (do a sweep and spot the peaks/values on the DDRC peak meters) is actually pretty easy and quick to do, and my subsequent more critical measurements seem to validate that approach as long as you then round up the biggest peak value (see my most recent post above). To me, that is much easier and les prone to misinterpretation than doing two sweeps in REW and dividing one by the other, and then worrying about which parts of the resultant curve you need to ignore because of measurement artefacts. It also works for the whole of the audio range.

I hadn't appreciated that you were able to use a slow sweep to be able to look for peaks by looking at the meters. This does make your approach more usable.

If people are just interested in the sub-200 Hz region (as I am currently) I still think the measurement approach is the easiest, most informative (gives the full picture and phase info.), and I really don't think is limited on the accuracy front by any meaningful extent. I would be interested in your comparison if you left the room for the measurements as this may be a factor in the remaining differences shown in the table above. I may try to have a play with this myself at some point as I'm struggling with why the results don't match better if not.

It may just be me but I always like to make a comparison measurement of results with and without the filter anyway to see what is actually going on.

I still suspect the max. boost is always 10 dB though and so this can just be used as a basis of any attenuation applied to prevent digital clipping.

Wanderer wrote:

I don't have specific knowledge about room non-linearity, but guess you are right that it should be a minor factor when the dynamic range is small. (Although, even ±10 dB is actually a huge variation if you convert it from dB to linear scale.)

True. The whole premise of Dirac Live working would fall apart if there were significant non-linearities over such a range though, as the applied corrections then wouldn't get close to doing what the calculations had assumed/predicted.

Thinking about it I have done some compression tests on an old subwoofer which show pretty good linearity actually. Below is an example, where the shape of the curves really is pretty consistent over a 35 dB range of drive signal.

Ultrasonic wrote:

Yes but I'm not sure what a combined left and right measurement would have to do with the current discussion?

Nothing at all... Sorry, I wasn't thinking clearly. Need to sleep a bit more

Ultrasonic wrote:

Below is an example, where the shape of the curves really is pretty consistent over a 35 dB range of drive signal.

That is a comforting result, thanks for testing.

Thanks @Tony_J for the detailed research. I've recently acquired an SHD, which I guess has more or less the same signal processing architecture, and this is the best description I've found so far about how it all fits together, and has clarified a couple of things I was confused about.

Something I learnt about recently is that you can use REW to directly measure what the SHD is doing, via USB. So the chain is:

REW -> USB Output -> SHD processing -> USB Input -> REW .

Obviously you don't see what's happening directly in the analogue domain, but I assume any clipping would show up in REW anyway.

Attached is a graph from this method using a measurement sweep that shows the frequency response of the Dirac filters in my setup. It looks like the maximum boost applied for me is around 5db.

Hi,
I’d be interested in understanding the procedure you followed to produce that graph - can you link us to it? Also, what does it look like without the smoothing applied?
Thanks

Thanks @Tw99 - that is a very useful idea indeed as it takes out all of the guesswork. Brilliant stuff!

asx77 wrote:

Hi,
I’d be interested in understanding the procedure you followed to produce that graph - can you link us to it? Also, what does it look like without the smoothing applied?
Thanks

The way I did it was to plug my SHD's USB port into a Windows laptop with all the MiniDSP software installed. I then used Windows Sound config to set both the default input and output devices to be the SHD.

Then I just ran REW, toldl it to use the default devices, and ran a measurement sweep. That's it...

The graph is unsmoothed.

Awesome thanks. I might even manage that!

So for me it was a little more complex than that. First attempt gave a bit of an odd result where nothing was shown below the sub crossover.



I tried again muting the sub and spare outputs (3 and 4) and then removed the crossovers from outputs 1 and 2 (left and right). I then ran the sweep on the left only and then on the right only and got something more helpful.



But question... where is the zero level/starting point for Dirac? I ran the same measurements with Dirac off then got REW to divide one by the other which gave the following which I presume is the info I’m looking for(?) but what on Earth is going on above 4kHz?

I also had that "fuzzy tail" in the higher frequencies (it's off the RH edge of the graph I posted). It's there even with no processing in the SHD enabled, so I'm assuming it's some measurement artefact, and could be explained or worked around if I knew what the hell I was doing
It looks better at least if you apply smoothing....

I've just had a play with this...

First up I didn't need to change any windows settings as I could simply select 'Line (SHD)' as an input device in REW, on the Soundcard tab of Preferences.

The two inputs that are measured are Outputs 1 and 2 for the SHD, so as @asx77 found it was necessary for me to disable the crossovers on these two channels to get full-range results. This does actually give a way of checking the crossover filter applied if anyone wanted to though. It is note necessary to mute or do anything else with Outputs 3 and 4.

I did all measurements with my power amp and sub switched off, and used a normal frequency sweep rather than pink noise.

I don't though get the high-frequency noise issues shown above. Here's an example:

Here's what I get with Dirac Off (blue) and On (red):




One thing this shows is some imperfect behaviour below 10 Hz, although whether that is the SHD itself or another factor I have no idea.

Dividing the Dirac On by the Dirac Off measurement gives the applied Dirac filter. The amplitude response looks like this



And the phase response like this:



These results were with no timing reference applied but repeats show identical phase behaviour so I think this is correct.

I forgot to say, thank you very much @Tw99 for sharing this idea .

For anyone who had already read it I have now updated my post above to include phase response data. I thought it best to keep all the graphs together rather than post it separately here.

I've had a bit of a play looking at higher sample rates and frequency ranges and they show some odd behaviour which I'm not sure if it's down to REW or the SHD...

My results above were for a sweep from 1 Hz to 20 kHz using 48 kHz sample frequency and a 256 k long signal (which is what I use for acoustic measurements).

If I run a sweep from 1 Hz to 96 kHz using 192 kHz sample rate (with the SHD spec. says the USB connection supports) and a 4 M long signal, I get the following result:



The rapid fall off above 20 kHz is consistent for all settings I've tested, but the behaviour before this varies, with shorter measurements causing issues at lower frequencies rather than just the little overshoot in the graph above. I may ask about this on the REW forum but does anyone here have any thoughts? Might there be a setting on my laptop that is having an effect for example?