3 Attachment(s)
Summing at crossover - LR24db/BU24db
As a continuation from this thread i want to discuss my efforts in integrating sub to midbass and midbass to tweeters.
As my phase cohesion wasn't that good, clearly to see in the aforementioned thread by not summing of 6db in my crossover regions, i decided to start from scratch. Yes, from scratch, means EQing, level matching drivers, and then adjusting time delay from my baseline (through tapemeasure) to have better summing at the crossover.
For the speakers see my sig.
This time the crossovers were chosen as follows:
Sub: Lowpass 63Hz LR24db, Level in DSP: -4.5 db
Midbass: Highpass 63Hz LR24db, Lowpass 2500Hz LR24db, Level in DSP: left 0db, right 0db
Tweeters: Highpass 2500 LR24db. Level in DSP: left -8.5db, right -6db
DSP used: Mosconi D2 100.4 DSP (DSP amplifier).
I use a custom curve made with Justins Spreadsheet.
Results from EQing and level matching are in the following screenshot.
Attachment 11117
You can see the slopes follow the LR24db alignment almost perfectly. Almost, as there is a 0.5-1db difference in some areas.
Problem: when entering the time delay values from Erins calculator (and Jazzis, but result is the same) i only had 3db summation at all crossover points for all drivers (using a screenshot where my sub-midbass xover was at 80Hz, but result is the same between both. It's just for demonstration purposes. Sub is missing in the screenshot, but it had only 3 db of summation just like the midbass to tweeters).
Attachment 11118
Now i did the following.
I entered the initial delay values for my midbass and tweeters:
Left Tweeter: 2.31ms
Left Midbass: 1.17ms
Right Tweeter: 0.83ms
Right Midbass: 0ms
Sub: 0ms
After that i added 10ms of delay to all drivers, muted the tweeters (have to do the fine adjustment for the tweeters still) and removed delay from the subs in 0.4ms steps and measured with periodic pink noise.
At one point i had a big dip in the response. At that point i stopped and added an amount of delay that equals half a wavelength of 63Hz in time (think it was around 15.87ms, so i added half of that: 7.94ms) and then i lowered the delay of all drivers uniform until the delay on the subs was at 0ms again.
This brought me to the following time delay settings and the following summation in the sub-xover:
Left Tweeter: 13,82ms
Left Midbass: 12.64ms
Right Tweeter: 12.36ms
Right Midbass: 11.56ms
Sub: 0ms
Attachment 11119
Looks a lot better, but still missing around 1db of summation.
I haven't yet measured both midbass drivers together, nor tweeters together, as i want to get each side done first as good as i can.
Now, would you go for that last db, or would you call it a day and live with it? Is this missing 1db really due to phase or can i savely eq this by minor adjustments in the xover region?
Is my way of doing all this like i'm doing it right now valid at all?
Next on my list is the midbass/tweeter crossover integration. I'll follow up with that in another post in this thread, but right now the plan is to do it like with the subwoofer, i.e. add some delay and then substract delay from midbass and sub until i have cancellation and then flip phase on the midbass back to 0db.
Re: Summing at crossover - LR24db/BU24db
The only time I’ve ever got a full 6db of summation is while using smaart to actually set delays and phase by a mic positioned between the ears and then adjusted the delays to suit, of I do this I can sum my front midbass and get a full 6db between them, that means they are perfectly in phase
the tape measure is off by +/- 2cm and the resulting phase mismatch causes all manner of issues through the passband of the speaker to its other side counterpart and also the higher and/or lower drivers also
seeing phase live while adjusting delays is a game changer for me for tuning
1 Attachment(s)
Re: Summing at crossover - LR24db/BU24db
On to session two.
I muted the subs and one side midbass and tweeter and linked left and right tweeter together.
Measured in 0.2ms steps first, but not matter what, i wasn't able to get the mids and tweeters to sum up to 6db when played together.
Somewhere i read that if two drivers only sum up to 3db one of the drivers is 90° out of phase. Might this be something that only an allpass filter is able to solve?
What i managed to get though was a more uniform sum between left and right.
Attachment 11120
All in all i arrived at around 0.34ms more delay for both left and right tweeter.
Might it be the reason for not fully summing that the midbass drivers are low in the doors and the tweeters up in the A-pillars?
Maybe i should think about adding a dedicated midrange to the system. As i had the 12 speaker shaker system before i have the cutouts in the door panel already.
Overall the setup sound really good right now (to my standards at least). Drums hit hard, no harshness anywhere, guitars sound wonderfull... don't know if i should hunt the missing few decibels in the mids/highs xover for now.
Re: Summing at crossover - LR24db/BU24db
I'll post some measurement of my graphs for comparison shortly. For what it's worth, I've found that I only get about 3dB of summation when playing my L+R 3.5" dash speakers together, but get up to 6dB summation when I play my L+R 6x9 midbass speakers together. I say "up to 6dB" because in certain areas of the midbass frequencies, I get less (and allpass filters didn't seem to change that whatsoever). I've been told that it's rare to get 6dB summation on the higher frequencies in a car (due to waveform length or something like that).
Great thread though - I think this is a great topic that often goes unlooked at by us "novices" (not sure about your experience level, but I consider myself a novice at this).
I know one thing, I must like my midbass more than most, because my midbass speakers go up on a pretty significant slope as the frequency gets lower. :-) I also overlap my midbass and "sub" (just a small 8" underseat powered sub). Graphs coming shortly...
1 Attachment(s)
Re: Summing at crossover - LR24db/BU24db
Oh... i'm definitely a novice, but trying to learn (to my wifes dismay i may add).
So i had my third session today...
I guess i nailed the bass now at least that's what the measurements show me.
I used "nulling" for the bass/midbass integration and left the mid/tweet integration alone for now.
Lowered the delays all around for mids and tweets though. Looks like they were a bit too high.
Attachment 11121
Tweeters will still follow. Tried a short round of nulling for them, but the dip was really hard to see in the graph.
Used bandwith limited periodic pink noise to find the null.
The sweeps afterwards showed that the Macbook Pro i use for tuning/measuring has some weird frequency response on it's headphone jack.
When using the tone generator the results were different from the periodic pink noise (same settings exported from REW) that i played via CarPlay.
So i stick to the ppn from my phone when doing frequency measurements and switch to the Macbook only for nulling/impulse responses and stuff.
Re: Summing at crossover - LR24db/BU24db
For what it's worth, here is where I'm currently at after replacing my dash and door speakers. I now have Illusion Audio C3CX 3.5" coaxials in the dash and CDT Audio CL-69S 6x9" midbass speakers in the doors. Sub is just a JBL BassPro SL under-seat powered sub. Crossing over door/dash speakers at 450hz. Currently have sub low-passed at 55hz and door speakers high-passed at 50hz/55hz.
FIrst pic is of all of my front speakers (L=Blue, R=Red, Combined=Purple) and sub (Yellow/ish) - as well as entire front+sub (Grey):
https://uploads.tapatalk-cdn.com/202...ae2f3440b1.jpg
Second pic is of my dash and door speakers and the response of them playing together (so you can see how the crossover sums):
https://uploads.tapatalk-cdn.com/202...8dd27cfa2d.jpg
Third pic os of my door midbass speakers and the response of them playing together (so you can see how the crossover sums):
https://uploads.tapatalk-cdn.com/202...c88b2671db.jpg
Probably not done with this tune yet, but it sounds pretty good so far. Crossovers are definitely not summing perfectly - and I definitely don't get 6dB of summing on front 3.5" dash speaekrs. Only get some 6dB summing on 6x9 door speakers. You can see how I had to "shape" the individual door speakers to get the actual shape that I wanted when they play together (individual response and combined response differ due to lack of full summing in certain areas).
Not saying any of this is "good" or "bad" - just what I'm seeing - right or wrong. :-)
Re: Summing at crossover - LR24db/BU24db
Quote:
Originally Posted by
jtrosky
I'll post some measurement of my graphs for comparison shortly. For what it's worth, I've found that I only get about 3dB of summation when playing my L+R 3.5" dash speakers together, but get up to 6dB summation when I play my L+R 6x9 midbass speakers together. I say "up to 6dB" because in certain areas of the midbass frequencies, I get less (and allpass filters didn't seem to change that whatsoever). I've been told that it's rare to get 6dB summation on the higher frequencies in a car (due to waveform length or something like that).
Great thread though - I think this is a great topic that often goes unlooked at by us "novices" (not sure about your experience level, but I consider myself a novice at this).
I know one thing, I must like my midbass more than most, because my midbass speakers go up on a pretty significant slope as the frequency gets lower. :-) I also overlap my midbass and "sub" (just a small 8" underseat powered sub). Graphs coming shortly...
All pass filters won’t work Likely because it’s a room mode, and it’s a result of the direct sound and a reflection cancelling, an all pass can’t do direct without effecting the reflection also, so the cancellation remains if that makes sense
if your dash speakers don’t sum 6db lower down then they aren’t timed correctly or you have massive reflection issues if above the dash with glass around them
Re: Summing at crossover - LR24db/BU24db
Quote:
Originally Posted by
Cathul
I use a custom curve made with Justins Spreadsheet.
Can you please share the target curve text files you are using?
Re: Summing at crossover - LR24db/BU24db
remember, you're almost never going to get a 6db sum in the car. reflections will show up as phase discrepancies in crossover regions. Also, why delay a half wavelength when you could just flip polarity?
1 Attachment(s)
Re: Summing at crossover - LR24db/BU24db
Quote:
Originally Posted by
Justin Zazzi
Can you please share the target curve text files you are using?
Sure...Attachment 11161