Ideally the sum should be flat, however, quite often the measured dip at crossover isn't audible.
It's where the drivers are producing the same frequency (you know that right?
). Anyway, there could be a number of reasons, for example the crossover itself is misaligned, but let's presume that's all ok. It could simply be that where you have the mic there is subtractive interference between the drivers due to differences in distance between them and the mic. In another position it might be flat or even a peak
). Anyway, there could be a number of reasons, for example the crossover itself is misaligned, but let's presume that's all ok. It could simply be that where you have the mic there is subtractive interference between the drivers due to differences in distance between them and the mic. In another position it might be flat or even a peak
I think the problem is that you do not have a true electroacoustic 4th order on the sub and second order Bessel on the fullrange. You may have implemented electrical filters of that nature - but it’s the overall response that counts. If you had well-behaved drivers that are flat - then applying the electrical filter is one and the same. What you have is non flat driver response convolved with your filter which is then higher order. So what you need to do is apply custom filters of lower orders and apply PEQ to get your individual drivers to match the textbook response. Only then will the sum work out as theoretically prescribed.
Visually it looks like you have two 4th order filters. Even order filters will cancel and make a dip At xO.
If you do it right according to the textbook curves, and get the delay just right, the mild dip occurs just to the left of the XO like predicted. Here is my attempt at the Harsch, note mild 1dB dip to left of XO. M
Thanks for the input. Your first paragraph is over my head. The Bessel filter available in the Mini DSP is what it is and there are no options as far as an order to choose from. If I'm not mistaken the Bessel is offered @ 30 dB so, I went in a different direction and use LR. These measurements have no PEQ added whatsoever. I really appreciate the help, thanks
Attachments
Mkane, please you must use PEQ to tune in (straighten) on-axis response at least 1 preferably 2 octaves past the xo frequency!
The recommended procedure is to do PEQ first (by measuring at roughly 3' distance) and after that set xo slopes. Then fiddle with spl and timing to get desired matching! Delays are very important when frequency rises above 200Hz and even more critical when F goes up.
Asymmetric slopes do not match perfectly neither do they null well when poalrity is changed - this makes evaluation of success a bit vaque. Keep eye on step response instead.
The recommended procedure is to do PEQ first (by measuring at roughly 3' distance) and after that set xo slopes. Then fiddle with spl and timing to get desired matching! Delays are very important when frequency rises above 200Hz and even more critical when F goes up.
Asymmetric slopes do not match perfectly neither do they null well when poalrity is changed - this makes evaluation of success a bit vaque. Keep eye on step response instead.
Last edited:
Nice device!
Yes, use the inbuilt PEGs and not the inbuilt crossover - Later on you maybe can try Rephase (still ONLY the pegs), but better to keep things simple till you get it right.
Do you have the textbook filters already? By overlaying those curves with what you actually measure, you can see which PEGs need to be applied.
The main point here is to understand that it may be the case that if you apply a LR24 filter (the theoretical crossoverfilters in the miniDSP), the actual response of the speaker (what you measure) may be a LR12 filter. Or something else. I think this is a very important point to grasp.
You have to unzip it before you can import the files to REW.
To import the textFRD responses to REW you can either drag and drop it to REW or use the import function. See attachment.
To offset the textbookfilters, hit the All SPL button and then hit the controls button. Pick the masurement and the offset it the dBs section. It may be a lot of dBs you have to offset it. Doesnt matter!. See attachment.
To import the textFRD responses to REW you can either drag and drop it to REW or use the import function. See attachment.
To offset the textbookfilters, hit the All SPL button and then hit the controls button. Pick the masurement and the offset it the dBs section. It may be a lot of dBs you have to offset it. Doesnt matter!. See attachment.
Attachments
Okay guess you saw post 385 where in red trace it was manufactures response curve times simulated baffle step and diffraction for sizes of your OB, now if overlaid with your raw mid-tweet orange trace we seems not far and therefor suggest up XO point one octave to 160Hz. Attached below is two new target files @160Hz and sorry but have to zip because txt-files is limited to 193kb sizes.
Attachments
Last edited:
