REW: time sync issue for impulse response

**z_accoustics** · **03-06-07, 01:39 AM**

I have an edirol UA-5 (http://www.roland.com/products/en/UA-5/index.html) and it works great except that the impulse timming is different everytime i take a measurement even if i don't change delay settings. The impulse graph will show the arrival up to +/- 2ms different with the same measurement repeated. I heard this is some sort of soundcard time sync issue with the recording start not being synched with the play start. Is this correctable with the REW software or is there some other approach to fixing this in general or with the UA-5 in particular?

Thanks a lot.

**JohnM** · **03-06-07, 05:02 AM**

REW places the impulse peak at time 0 (it has no absolute time reference), the measurements would usually be very consistent unless there is something changing in the room being measured. One thing to be wary of with the Edirol cards is to make sure that monitoring is disabled, any monitoring causes a feedback loop that creates inconsistent (and invalid) measurement results.

**z_accoustics** · **03-06-07, 11:49 AM**

Thanks for the tips on monitoring, i usually keep it off but i'll make sure. Also good to know how REW does impulse response. If it always puts the peaks at 0 then how do I go about measuring the delay from one driver placement to another? If i want my right speaker output to arrive at the listening position at the same time as my left speaker output, how do I use REW to assist in assessing the proper delay to set on my HU time alignment?

**JohnM** · **03-06-07, 04:58 PM**

You could make a measurement with both speakers running and adjust the delay until you get a single peak rather than a pair of peaks.

**z_accoustics** · **03-06-07, 06:31 PM**

That's a good idea, except that when I do the sound sweep measurement there isn't always a clear "beginning" to the impulse graph. Meaning, there's a bunch of small ups and downs before the major peak and there are a few major peaks back to back and that's just taking the measurement with 1 driver. Seems like doing it with two at a time I would not be able to interpret what i need to from the graph. Perhaps if the sound measurement consisted of a very brief interval of pink noise instead of the sound sweet... then maybe i'd be able to interpret things but of course with a 2-way system that has 2 tweeters 2 mids and bass, this method becomes overly complicated. hmmm. Any other ideas?

**JohnM** · **03-07-07, 08:29 AM**

Give it a try, it is easier than it sounds - you are only interested in the largest peak. Make sure the sweep extends high enough, at least to 5kHz, that way you will get a narrower peak. If you only want to adjust for arrival time differences between the speakers a tape measure also works very well

**PeterSt** · **03-27-07, 06:33 AM**

Hi John,

I signed up to this forum, after which a search link emerged, so I exactly found what I was going to post about ... this thread.

I have been trying what you suggest here, but IMHO it just can't work like that. Not when all first largest peaks are mapped onto the 0 point ... Think of this :

When there is two drivers in a cabinet, and you want them perfectly aligned (physically shifting) including (filter) electronics behind them, I'd need to know the relative arrival times of the first stroke.
There is IMO no way of letting the both (etc.) drivers play at the same time at measuring, because the picture can't be interpreted because of adding (etc.) waves then. The only thing which -with a long shot- seems a bit possible, is first measuring one driver, and then add a second, where the first one is to be interpreted as a reference (plot). But then too, the impulse plots of either driver change because of the both interfering with eachother.

I know, this is ROOM equalizing software, not speaker alignment software. But I thought it would be so easy to just let the plots start where they do (as an option), instead of mapping all to one 0 point.

Thank you, and thank you very much for great software !
Peter

**JohnM** · **03-28-07, 05:47 AM**

Playing a full range sweep through the speakers still gives the desired result Peter, the effect of the individual crossovers and their time alignment will be evident in the impulse response derived from that full sweep. If everything is perfectly time aligned there will be a single peak (followed by the various room influences).

**PeterSt** · **03-28-07, 06:26 AM**

Thank you John.
But are you saying that I then should look before the 0 point ? I mean, *at* the 0 point it's always equal. Hehe. And then, which will be the relieable peak to look for ? What I'd obviously want is -when two drivers are 2 cm displaced (all other stuff equal)- the 2cm can be seen on the screen ... and that I need to shift 2cm ...

I think I just don't understand ...
When I have two measurements of two drivers (say bas and mid), the first highest peak will be at the 0 point for both. I don't see a reason that one of them would behave differently before the 0 point. Obviously, the bass will start slower, but since you re-align everything to the 0 point ... how to see that ?

Sorry to be stupid

(feel like that anyway).
Peter

**brucek** · **03-28-07, 09:22 AM**

Quote:

But are you saying that I then should look before the 0 point

Left of the main peak zero are distortion components. Every peak is due to an increasing order of distortion component (2nd harmonic, 3rd harmonic, 4th harmonic .....). The further to the left you go, the higher the distortion order.

Quote:

how to see that ?

If the energy arrived at the mic at two different times (i.e. phase shift in one driver), then there would be two main peaks after zero (although it's hard to distinquish from reflections). Perhaps set up in the middle of the room with the mic somewhat nearfield so you can determine the reflections better. Or even better still, do it outside.

brucek

Advertisement