Any guide available somewhere to do this right? Because I'm pretty sure that the tweeter and mid measurement were made the right way. The tweeter driver was the reference in all cases, but the sweep started at 200 Hz in the mid measurement and at 1000 Hz for the tweeter.
Ok, let's say that's ok. What about the physical distances, do the differences match your listening position differences?
Wired loopback on one channel, but the timing sound comes from the tweeter (high Hz). I run a sweep from 200 Hz (out of range for tweeter) with either bass or mid muted. Then I run a sweep from 1000 Hz with only tweeter unmuted. This results in a delay of the mid only of 180 us relative to the bass and tweeter. It seems like a confusing result, but I cannot find an understandable method online.
With tweeter delay the sound seems a little better, but it can easily be my mind playing tricks.
Last edited:
yes loopback right channel output to right channel input, will work.Left channel out to amplifier. If you have suitable attenuation then the loopback can be from the amplifier output rather than just a loop of the sound card. This is the most accurate as it takes into account any delay in the amplifier as well. BUT make sure you have a suitable attenuation circuit if you do this or you can easily fry your sound card.
For a paper on doing measurements check out kimmo's guide for REW for getting polar measurements https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_Preparations_REW.pdf
I seem to remember that there are some subtle gotchas when doing 2 channel measurements in REW, but it's over a year ago since I did it, and can't remember what they were.
Tony.
For a paper on doing measurements check out kimmo's guide for REW for getting polar measurements https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_Preparations_REW.pdf
I seem to remember that there are some subtle gotchas when doing 2 channel measurements in REW, but it's over a year ago since I did it, and can't remember what they were.
Tony.
So the mic is at tweeter level so the tweeter is closer than the others? Maybe it's the others you should be delaying?
It sounds as though he wants to simulate the theoretical result to make a point about how good the step could appear. This is, however just another way of presenting the same information.
Last edited:
This can also help you get better S/N. I have a little rig with 0, -20, -40dB and RCA or crocodile clips to RCA for amp outputs or loopbacks.
The idea is if the delay of two drivers is matched and one driver is inversed, the two sine waves are 180 degree out of phase and thus fully cancel each other. At least in theory, in practice a bit less as TNT already explained...
Fedde
I actually learned why the 6 db at crossover is so important.
Just one more question.
Where do you measure what.
I flatten the drivers measuring at driver level (20-30 cm away) and the rest between drivers (bass mid or mid tweet) for the crossover. Timing then farther away. What makes sense. Using the reversed polarity technique, it seems at the physical crossover is the right place.
Or measure everything at one place?
I suspect that you are going to get very differing oppinions on this! 
My personal take is that you should do it in as big a space as you can (outside is ideal) with gated measurements (for higher frequency crossovers). This will give you the cleanest data and phase and you are only dealing with the drivers responses and phase. (even then it is still difficult to get good data for low crossovers).
This data will allow you to get good integration of the drivers. Dealing with room response issues is IMO something that should be done separately.
I've tried doing a 270Hz crossover with in room measurments but it was not good. I ended up using good clean outdoor measurements not gated but 1/8th octave smoothed and that gave me a much better result. Other people will likely say they get the best results measuring at the listening position
Tony.
My personal take is that you should do it in as big a space as you can (outside is ideal) with gated measurements (for higher frequency crossovers). This will give you the cleanest data and phase and you are only dealing with the drivers responses and phase. (even then it is still difficult to get good data for low crossovers). This data will allow you to get good integration of the drivers. Dealing with room response issues is IMO something that should be done separately.
I've tried doing a 270Hz crossover with in room measurments but it was not good. I ended up using good clean outdoor measurements not gated but 1/8th octave smoothed and that gave me a much better result. Other people will likely say they get the best results measuring at the listening position
Tony.
I suspect that you are going to get very differing oppinions on this!
This data will allow you to get good integration of the drivers. Dealing with room response issues is IMO something that should be done separately.
I've tried doing a 270Hz crossover with in room measurments but it was not good. I ended up using good clean outdoor measurements not gated but 1/8th octave smoothed and that gave me a much better result. Other people will likely say they get the best results measuring at the listening position
Tony.
If it's dry we may measure outside some day. However, when measuring 30 cm from a driver you'd say that is okey as well.
My questions specifically:
1. What are gated measurements.
2. Bass driver measurements at ear level or at bass driver level?
3. Time alignment of drivers also close to drivers at ear level?
I have Dirac Live on my NAD C658 so optimization at listening position is of no concern.
Thanks,
Geert