TOPIC:

Hi :   
   I use a SHD in my "stereo" system.  As I learn from Dirac user manual, the microphone should pointing between two speakers in a  "stereo" system when doing measurements.
   But the Dirac user manual page 26 section 4.10  (Volume Calibration) says the microphone should pointing toward ceiling without specifying "Stereo" or "Multichannel".
   I wonder this means that the microphone should point to ceiling when doing volume calibration no matter in a  "Stereo" or "Multichannel" system ?  
   Thank you !

Hank 
   

The recommendation from Dirac is that the microphone should be pointing vertically for all systems, for both volume calibration and the actual measurements. You'll find it won't make a huge difference either way though, provided you use the correct microphone calibration file for the orientation you choose.

As clear documentation is hard to come by, here is some info. from Flavio Fellah of Dirac Live posted on another forum:

www.hifiwigwam.com/forum/topic/138704-mi...ents#comment-2620420

I read (forget exactly where) to take measurements with the mic pointing forward towards the speakers for 2-channel setups and at the ceiling for surround setups.

Stephen300 wrote:

I read (forget exactly where) to take measurements with the mic pointing forward towards the speakers for 2-channel setups and at the ceiling for surround setups.

I used to do so myself and loads of others do too. I changed on the basis of the information I shared with you above - make up your own mind on it. There is no 100% correct answer here since the microphone will receive sound from all directions but you can't somehow simulataneously use different calibration factors.

As I said above, you really aren't going to notice a fat lot of difference either way so long as you make sure you're using the correct calibration file for the orientation you choose.

You could contact Dirac Live yourself of course. This question is very much one of what works best for Dirac Live rather than anything unique to an SHD.

Stephen300 wrote:

I read (forget exactly where) to take measurements with the mic pointing forward towards the speakers for 2-channel setups and at the ceiling for surround setups.
 

That is what it says on the umik-1 and umik-2 product pages.

This is one of those unexplained things. Dirac Live says in no uncertain terms to use a vertical position and a 90 degree calibration file. It’s published online, but Flavio did the same in a private email to me when I asked. However, miniDSP says to use a horizontal position and 0 degree file. Considering Dirac Live is very much in the position to know how to best use their software, it’s up to miniDSP to explain why they tell people to do something different.  

To play devil's advocate, I would suggest its up to the microphone supplier to decide how to use it.

For a nominally omnidirectional capsule it won't make much difference, the on and off axis responses seem to differ primarily at high frequencies, probably not at key frequencies where Dirac will do the most important things and possibly only audible to younger ears. For a cardiod capsule etc. pointing up would be necessary.  Dirac Live obviously works on many different systems with potentially different microphones so perhaps they have just simplified the advice so it works with any mic capsule.

Fortunately the SHD has multiple presets so anybody can do their own experiments and try both ways. I will likely do that myself next time I do a calibration,

First up for context, below is what the two calibration curves look like for my UMIK-1. As you can see there only starts to be a difference above about 1 kHz so it makes zero difference for those just using Dirac Live in the bass region, whilst there will be some influence for those interested in higher frequencies.



For measuring the direct signal from a loudspeaker, the horizontal calibration only truly applies when the end of the microphone is directly pointing at the speaker. For Dirac measurements this won't actually be the case. By contrast, the vertical caibration is actually valid for all directions in the horizontal plane and so is arguably more accurate to use here.

In terms of the Dirac algorithm I believe the argument goes that there will generally be more reflected signals from the sides than the floor and ceiling (in part since the speakers will radiate more in the horizontal plane than up/down) and so it's better to bias the accuracy slightly to the horizontal plane.

As I've said I really don't think it makes a fat lot of difference either way but for me I think pointing upwards makes the most sense overall (making sure to use the corresponding calibration file).

After reflecting (pun intended) on the on and off axis graphs I wonder whether pointing the mic at the floor might be better than pointing at the ceiling.

The 0 degree response is around 3dB higher than the 90 degree response at 10kHz, a difference that will be maintained after the calibration file is used to normalise the response.
If the mic is pointed at the ceiling the front and side sensitivity will certainly be balanced  and nominally flat but the ceiling will be 3dB more sensitive. Given that ceilings are often hard flat surfaces with significant reflection that may not be the best direction for 3dB extra sensitivity. In contrast floors are often covered with carpet and likely provide significant absorption at 10kHz so may cancel the 3dB extra sensitivity.

It is probably moot, the apparently minor differences between front and ceiling directions suggests in practice it may not have a significant effect pointing it at the floor. I would try it myself but I have rather extensive ceiling absorption and diffusion treatment as well as only being able to hear up to 11kHz on a good day, and only around 7kHz in my left ear due to an infection at the moment.
 

We don't have data for the response from behind the front (above 90 degrees) but I'd guess the sensitivity drops off. I think an argument could therefore be made that pointing up is better since there are more significant reflections from the ceiling than the floor, at least in a room with a hard ceiling and a carpeted floor.

Interestingly I could make the opposite argument, that pointing at the ceiling with significant reflections from there with an excess sensitivity of 3dB at 10kHz could lead Dirac to believe the room is brighter than it really is. It could then make the compensation filters duller than they ought to be.

Like you I would suspect that the 180 degree response is duller than the 90 degree response. Pointing at the floor could make Dirac less sensitive to ceiling reflections than it ought to be which could lead to the opposite result of Dirac making its compensation filters brighter than they ought to be but the reduced contributions from the ceiling should mean that has a lesser effect relative to the front and sides than when pointing at the ceiling. My gut feeling is that pointing at the floor could be slightly less wrong than pointing at the ceiling but that either would be better than pointing forwards.

I am avoiding considering impulse response effects and concentrating primarily on potential frequency response effects. Also, if the only slight effect was on high frequency response you could presumably obtain similar results by modifying the target curve.

My thinking is partly that the response getting towards 180 degrees would be dramatically down, vs the much smaller difference between 0 and 90 discussed above.

Impulse response corrections are based exclusively on the early arriving signal I believe, and so should be largely unaffected by what we're discussing.