My first REW Measurement - DIY Speakers - Did I get speaker polarity wrong?

[Keith_W]

May I suggest that you connect your power amp directly to the driver with no passive XO in between? You have to be VERY CAREFUL not to accidentally send low freq signal to the tweeter because you might blow it up. If you want, you can leave a capacitor to protect the tweeter. This will tell you if it's the driver or the XO which is at fault. If you have enough DAC channels, you can even make a DSP crossover and forget about passive XO's. Digital XO's are simply better.

Re: longer sweeps for lower SNR, it's not always necessary. As my eBook says, as long as your measurement is about 20dB above the noise floor, it's sufficient. Maybe even 10dB, if all you want to do is look at the graph (and not DSP it!). Note that the noise floor typically rises at low frequencies - this is a problem because most speakers typically roll off the bottom end and it can be difficult/impossible to tell what is signal and what is noise at about 20-50Hz. The eBook also tells you how to look at the noise floor, see the section on waterfall.

 

[nikb]

So, I finally found and fixed the midrange problem with the right side speaker!

There were two bare leads, both from large, heavy, capacitors, right next to each other. When the crossover was flat on my bench they would not touch, but mounted on the side inside the cabinet, the weight of the caps made the leads touch (they were originally glued to the board, but must have worked themselves loose). Fixed it by prying them further apart and applying lots of electrical tape.

The image below is how the speakers currently measure. This measurement was taken at listening volume with the mic at tweeter height, 1m from the speaker, with a long
sweep (1M). As you can see, there is still a 1.5db difference between the speakers. I would really like to fix that as well, but currently I have no idea where to look.

The distortion issue that some of you reported seems to be OK now as well. Only the impulse response in REW really messy compared to examples I see online, but that could be a user problem - I just do not know how to properly zoom in the IR window.

The speakers are nice to listen to now, giving a good stereo image, without any eq.

Do you see any faults that still need fixing? Should I stop here? I would like to know your thoughts.

 

[Keith_W]

Well there's the fact that your left tweeter is about 2dB louder than your right. This could be a tweeter performance issue, an XO issue, or an amplifier issue. As before, try reversing the channels on your amp, then try swapping your tweeters left and right. This could be easily fixed with some DSP, but it's better to make sure that the tweeter/amp/XO isn't faulty.

The rest of the issues I see are room related.

RT60 is pretty high.

You can see that the decay is very long in the spectro.

And if you switch to the wavelet view, you can see the reflections.

Step response is pretty poor. You can see two peaks after the main impulse which are higher in amplitude than the main impulse itself. I presume that this is at the listening position, because the 1m measurement does not look like this (this is "L fixed crossover").

Anyway it's a considerable improvement compared to your earlier effort. Your speaker is close to sorted. Now you need to do something about your room.

 

[MaxwellsEq]

There were two bare leads, both from large, heavy, capacitors, right next to each other. When the crossover was flat on my bench they would not touch, but mounted on the side inside the cabinet, the weight of the caps made the leads touch

Excellent sleuthing! Well discovered.

 

[nikb]

Well there's the fact that your left tweeter is about 2dB louder than your right. This could be a tweeter performance issue, an XO issue, or an amplifier issue. As before, try reversing the channels on your amp, then try swapping your tweeters left and right. This could be easily fixed with some DSP, but it's better to make sure that the tweeter/amp/XO isn't faulty.

Yes, I will continue with the tweeter issue. I want to fix as much as possible in the crossover and only the rest by eq.

I am quite sure that the problem is still in the crossover - I have switched amplifier channels between measurements without any effekt. Also I have switched tweeters between the enclosures already. Must be the crossover. I have to come up with an idea/hypothesys of what could be the cause.
I'll post an update when I find anything.

The rest of the issues I see are room related.

View attachment 493556

RT60 is pretty high.

View attachment 493557

You can see that the decay is very long in the spectro.

View attachment 493558

And if you switch to the wavelet view, you can see the reflections.

View attachment 493559

Step response is pretty poor. You can see two peaks after the main impulse which are higher in amplitude than the main impulse itself. I presume that this is at the listening position, because the 1m measurement does not look like this (this is "L fixed crossover").

Anyway it's a considerable improvement compared to your earlier effort. Your speaker is close to sorted. Now you need to do something about your room.

Thanks for showing me the room effects, specially the step resonse. Can I ask you how you get there?
For me, the step response for "R fixed crossover Nov27" looks like this:

Anyway, I will fix the rest of the speaker issues when I have come up with a plan. Then I will go on to look at what can be done with my room (with acceptable effort) and eq.
Cheers, Nik.

 

[Keith_W]

Thanks for showing me the room effects, specially the step resonse. Can I ask you how you get there?
For me, the step response for "R fixed crossover Nov27" looks like this:

All in the REW eBook (you want book 2, see the section on step response). To get that view, change from dBFS to % in the drop down box in the top left. It disappears until you hover your mouse over it.

 

[nikb]

All in the REW eBook (you want book 2, see the section on step response). To get that view, change from dBFS to % in the drop down box in the top left. It disappears until you hover your mouse over it.

Thank you so much for the handholding and the extensive ebook - I had started in it a couple of days ago but had not come to this section yet. Now I have a couple of days off where I can hopefully read and prepare for using it when optimizing my room.

I will update you when I have the last problem with the speaker sorted out.