Timing Measurements

[Chris A]

I assume that this is the product line that you are referring to?  I actually found more than one "Dirac" product, unfortunately.

 

The only compliant that I've heard about miniDSP is that it might be noisier than desired in a small room due to the DSP noise itself when used with very efficient loudspeakers.  I've not heard one before, so I can't tell you personally that there are any issues or whether it might sound very good.

 

Chris

[DrWho]

I wonder how audible a 4.8 ms delay is?

It's a real miracle how once something is confirmed through measurements it becomes "painfully" audible to some folks..

I would agree that people tend to over emphasis principals they've come to understand, but are you saying you can't hear the loss of definition in transient sounds (like a snare or piano)? Or you saying you don't care?

I personally find this effect to be an order of magnitude more audible than the difference in sound between amplifiers.

[DrWho]

BTW: I believe that the timbre shift heard when time misalignments are present are due to "zero lobing error".

Mark, I believe that the deconvolution integral to produce an impulse response from an upsweep within REW (the "impulse" view) will give you a time delay - and if the sweeps are done separately for each driver, a characteristic time delay is determined, as long as the microphone position and room temperature both don't change.

I calculate the path length difference from the bass bin to the midrange, and the midrange to tweeter (for a 3-way loudspeaker), then plunk those into the channel delays (midrange and tweeter channels). When I run REW upsweeps using the FR and phase view, I can see the remaining phase misalignment between the two horn/drivers. I then fine tune the delay by varying the delay and looking at upsweep FR and phase at the crossover points to eliminate the FR dip at crossover and verify by looking at the un-smoothed phase curve: works like a charm.

Wow.......you use a microphone to try and set a time delay on driver path lengths? I have learned by using the same stuff in my rooms that a large % of the sound picked up by a microphone in your room is reflected sound.

I would not do what you are doing to set driver path length delays........

What you seem to be attempting is to do is acoustically align drivers..........which seems based at least partly on reflected sound. You must have some interesting results.

Do your delays mirror Roy's or come close?

I use a much higher technology method............its called a tape measure. My processor allows you to enter the path lengths in "feet". I can get it down to inches with "tenths". I just measure the length of each driver from voice coil to horn entry and the processor creates a time delay in "seconds" (milliseconds in my case).

Without an oscilloscope everyone is approximating anyhow. But "close enough" works very well. When I switch off the delay the room clouds up and snaps into cripsy clarity when I re-engage. Very noticable........Works well.

The big thing Heyser introduced was the idea of gating the calculated impulse response. You lose frequency resolution in the process, but it allows you to ignore reflections.....Or if you prefer, you could look at just the reflection instead.

It's really cool to see a frequency response dip caused by a reflection go away as you bring in the gating window.

[Chris A]

I wonder how audible a 4.8 ms delay is?

It's a real miracle how once something is confirmed through measurements it becomes "painfully" audible to some folks..

...I personally find this effect to be an order of magnitude more audible than the difference in sound between amplifiers...

 

I agree with Mike's assessment here--if you've not heard the difference in A-B demo using an active crossover, it's a bit difficult to describe just how audible it is.

 

I've sometimes wondered why some people like the sound of the Cornwall - and whether or not some of it could be due to less bass bin to midrange time misalignment.  Note that the Khorn midrange horn is also the same basic horn and therefore the delay as in Cornwall midrange horn - woofer is actually going the opposite direction with the woofer leading the midrange. But the 6 millisecond delay on the Khorn bass bin relative to the midrange is a very large time misalignment, IMO.

 

Note also that I find the 1.1 millisecond delay of the Belle or La Scala to be very audible in terms of overall timbre shift and in harsh sounds in the tuning fork A-440 Hz region.]

[NOSValves]

1.1 millisecond..... gosh I wish I had your ears

[Don Richard]

1.1 millisecond..... gosh I wish I had your ears

 

It's not like a person can hear the two sounds 1 ms apart... it's the comb filtering generated by the two signals  displaced in time by 1 ms that is easily audible. 

Edited August 22, 2014 by Don Richard

[Chris A]

Actually the most audible artifacts includes a timbre shift due to "zero lobe error" (affecting all frequencies), followed closely by an even more audible and sensitive-to-error distortion of pure tones in the crossover region - such as a synthesizer, flute, clarinet, or saxophone, oboe or trumpet might make (or sine/square/triangular/sawtooth wave tones).  The 1.1 millisecond delay error is at the 500-650 Hz region, and that's very audible - above tuning fork frequency.  It's also just as audible around the 400 Hz region.  That's about a 180-degrees-out-of-phase dual source distortion of the signal (i.e., a full wave period at 500 Hz is 2 ms). Actually, I found that delays less than 1/10th of a wavelength were clearly audible, and I could easily find the correct delays (starting within +/-180 degrees of in-phase) by ear.

 

Not to be impolite, but if you've not heard the effects using an active crossover, I don't think that you will believe it.  However, it's there for anyone's ears to hear.  Try it first before pooh-poohing it.

Edited August 22, 2014 by Chris A

[djk]

A 1/4" is audible at higher frequencies, that's why I said "Depends on the frequency." back in post #16.

[Chris A]

One quarter of an inch is ~18 microseconds--I assume at ~6 KHz, about where Heritage loudspeakers cross over from midrange to tweeter.

 

Shifting gears to 600 Hz, about where I crossed the Belle from bass bin to midrange, that'd be equivalent to 0.18 milliseconds, about 2.5 inches of alignment at 600 Hz.

 

That's about 1/10th of a wavelength in both cases: we're in violent agreement I believe.

 

Chris

[Rudy81]

Can anyone suggest an experiment we can run with an active XO?  Although I align my drivers, I would like to 'hear' what happens when they are not.  

 

In particular, can REW send a tone or help in such an experiment?

[Chris A]

REW or TrueRTA can - you can hear when you begin to dial the midrange off of alignment while playing a square wave at the crossover mid-frequency.

 

BUT, you mentioned that you're crossing at 200 Hz, which is getting fairly low.  The 1/10 of a wavelength rule of thumb at 200 Hz is 0.5 milliseconds.  You'd have to be dialed off of alignment by more than half a millisecond to begin to hear it, if that rule of thumb holds.

 

Chris

Edited August 23, 2014 by Chris A

[Rudy81]

Chris,

 

Thank you.  I'll give it a try when I get a chance.

[jcmusic]

Rudy,

I am pretty sure you are gonna hear your sound stage shrink up alot smaller than what it is now...Also you will detect smearing of the sound, less clear...

Edited August 23, 2014 by canyonman

[Bofkont]

I assume that this is the product line that you are referring to?  I actually found more than one "Dirac" product, unfortunately.

 

The only compliant that I've heard about miniDSP is that it might be noisier than desired in a small room due to the DSP noise itself when used with very efficient loudspeakers.  I've not heard one before, so I can't tell you personally that there are any issues or whether it might sound very good.

 

Chris

Indeed these are the ones with DIRAC Live incorporated. It's a variation on a theme: AA is in the analog part of your chain, DD in the digital domain or DA as a complete converter/preamp. I have not yet heard them but like the theory and from experience know that they make good boards without anything audible added like noise. The little board with D in/out on the miniDSP i.e. works nice and sounds good. Needless to say I have no connection, but simply like their products and a Khorn in a small(er) room needs stuff like this.  Have a great weekend, Dirk

[kodomo]

For Khorns, I physically aligned my mids and tweeters and it helped. This is the lowest they can be with a slight tilt without having effects of hitting the top of tophats. The drivers are now horizontally and vertically aligned on top. They are VoltiT1 tweeters. We have made blindtests and everytime we ended up prefering this to tweeters being in the tophats to the sides (volti upgrade fits inside tophats like that). Here are two photos:

 

[Chris A]

1.1 millisecond..... gosh I wish I had your ears

It's not like a person can hear the two sounds 1 ms apart... it's the comb filtering generated by the two signals  displaced in time by 1 ms that is easily audible. 

 

Right: classical comb filtering clearly occurs in the crossover region.  The precedence effect masks the  delayed sound in the human hearing system when we are talking about 1 ms.

 

Lobe shifting affects a much broader range of frequencies and unfortunately it's frequency dependent. If lobe shifting were not a function of frequency, perhaps it wouldn't be as troublesome as it is.

Edited August 23, 2014 by Chris A

[jcmusic]

Here are the last graphs....

Edited August 23, 2014 by canyonman

[oldtimer]

Thanks for those.

[jcmusic]

Your welcome I hope they helped in some way.

[Bofkont]

The quest went on. Past year have brought a MiniDSP with DIRAC Life software to replace my DIY "time aligner" and a new DAC: CHORD Hugo that also doubles as a more than excellent HP amp. Both made me very happy but that is of lesser interest. Let me show you a few graphs to proof my point. They reflect on the room measurements that one needs to do to.

 

First a before and target frequency response

 

Now it looks more like the ideal curve from B&K

 

Through some IT magic also the impulse curve was improved:

 

Music is very crisp and detailed. Spot on purcussive sounds with a lot of energy. Every horn in a smaller room benefits from this I would reckon.

 

Anyway just to let you know and maybe provoke some interest.