Left and right summation is off from target curve. Adjust Jazzis curve? Raise? Lower?

When I set my curves in Jazzis program. And match my speakers to the individual targets in a 3 way system. The drivers all hit the curves nice and all have good phase at the crossover points. Their independent left and right channel total curves match the overall target house curve and each other. But when I take a measurement of the left and right side together the curve is way off. All the drivers don't sum the same amount. The midbass sum much higher than the rest about 5 db. The subs are run mono so they have no summation or db rise especially at the lowest frequencies. How should this be adjusted to reach the final curve? Should I for example raise the bottom end of the sub to get the overall curve to happen again? Should I lower the midbass and try to adjust the curve for the final curve which will make the individual left and right curves different then the final curve of both sides? How about shape and db discrepancies for the mids and highs? Here is a pic of the sub and midbass. That's the curve I am going for, again the left and right side individually hit that. But the drivers in the pic are left and right midbass and the sub together which is creating a different shape.

Are the subs on the same channel of the amplifier or do they have independant channels? If only one channel you should raise the sub about 6dB for the overall measurement of all drivers.
If they have independant channels mute one of the channels while measuring and enable it again for the overall measurement.

For Mids and Tweeters it's not that easy and i basically have the same "problem".
Due to all the reflections in the car mids and tweets will never sum up to 6dB, even if timing is totally correct. Sometimes even with mids you recognize 6dB in the lower range of the mids and max. 3-4 dB of summation in the upper parts of the mids. Mids to tweeters crossover the same, very rarely i've seen full 6dB summation, even if impulse response are perfectly aligned. Measuring phase in that region is hard as hell, too, so i wasn't ever able to do phase alignment like between midbass/sub or midbass/mids (when xover was low enough).
For this i don't have an answer myself. Raise? Leave as be as it's not really only 3-4dB summation? I don't really know.

This is a really interesting question! I've not considered how the summation of the speakers means that the overall curve will be a different shape than a textbook-perfect example.

For your subwoofer not summing together any hotter than your midbass speakers, try tuning your subwoofer with only one of the channels getting signal. If you are using (for example) channels 7 and 8 on your DSP to send to the subwoofer, try tuning with only channel 7 enabled but with channel 8 muted. Then when you measure everything together, you can unmute channel 8 so your are sending twice as much signal to the subwoofer and it will increase output to keep up with the summed output of your two midbass drivers.

It's nice the your midbass drivers sum together at +5dB'ish. That's not easy to do so congrats.

The midrange and higher frequencies will be hard to sum together at +5dB though, like Cathul mentions. I also do not put too much emphasis on getting my midrange and treble to sum together perfectly because the moment I move the microphone a tiny bit, it changes.

Actually, if you are using the moving-mic technique to make your acoustic measurements, then the midrange and treble should not be able to sum together greater than about +3dB even if you get it perfect. This is because as you are moving the mic around and taking a spatial average, the path length differences are constantly changing and the midrange and treble frequencies will be either summing together or cancelling eachother out as you move the microphone around. So expecting to get a +5dB summation in the middle and higher frequencies is going to be difficult if not impossible.

Fascinating though. I wonder if this is why people prefer a treble response that tapers off? In a car, the summed treble response will appear to measure with a falling taper because of how the mid and higher frequencies sum together like I mentioned above. I think this phenomenon will be more common in a small enclosed car than in a larger room with a higher direct-to-reflected sound ratio. Hmm this is something to think about.

Sorry, I don't think I answered your question.

The "final" curve that my spreadsheet creates is going to be the sum of all speakers in one quadrant of your car, plus the subwoofer playing the equivalent of "one speaker worth" of signal. So for example that would be the front left midbass and tweeter, and the subwoofer also playing as if you were sending it only the "left" channel's signal.

To make this measurement, mute every channel except the front midbass and tweeter. This is the equivalent of the "left" channel playing in your system. To get the correct amount of bass to play from the subwoofer, you need to only send it the bass from the "left" channel. So either fade your radio all the way to the left, oe use only one RCA cable instead of the "left" and "right" inputs on the subwoofer. Or if you are sending the subwoofer signal from (for example) channels 7 and 8 from your DSP, then mute ch8 and only send signal to the subwoofer from ch7.

If you get it right, then the total acoustic energy played by the "left" channel in your car will match the overall target curve that my spreadsheet creates, and you will not have the midbass louder than the subwoofer like in the screenshot you posted.

If you want to play all of the speakers at the same time and measure them, then shift the overall target curve up by +6dB. Your system, if tuned absolutely perfectly, should reach this new +6dB level at all frequencies. In practice, as you have discovered, the lower frequencies will sum together at +5db'ish and the midrange might sum together at +3dB'ish and the treble might sum together or it might not. I would consider that a success for a car, even though it appears like a failure because the performance doesn't match the curve when you play all the speakers at the same time.

A bit counterintuitive!

When timing between tweeters is correct they will almost always sum up to 3dB at least, even in the reflective environment that the car is. So if you got 3dB summation when playing both tweeters i wouldn't worry too much, but rather think "nice, i got it (almost) perfect". ;)
Same for higher midrange frequencies.

Quote:

---

Originally Posted by Justin Zazzi

To make this measurement, mute every channel except the front midbass and tweeter. This is the equivalent of the "left" channel playing in your system. To get the correct amount of bass to play from the subwoofer, you need to only send it the bass from the "left" channel. So either fade your radio all the way to the left, oe use only one RCA cable instead of the "left" and "right" inputs on the subwoofer. Or if you are sending the subwoofer signal from (for example) channels 7 and 8 from your DSP, then mute ch8 and only send signal to the subwoofer from ch7.

---

Man that's an interesting point. At the moment I am only using one channel of the helix split to the left and the right channel of the sub amp. In the i/o of the helix I have it 50% right and 50% left. Do you think its better to run two channels to the sub from the helix and do 100% right and 100% left to the sub amp?

That's up to you. If you run as is, you need to raise your sub level by about 6dB when playing everything to get it in level with the rest of the system.
When running left and right to your subamp, you need to mute one of the channels when measuring one side only. When playing together you get double the signal --> 6dB more level.
You just have to know this and adjust your level/measuring when measuring all drivers together.

A single input to the sub amp is fine, turn it down 6db when tuning individual and setting levels, this is far easier than finding 6db to add on at the end, we tune to the lowest driver, normally midbass in my experience with the adjustments and headroom we use to get a nice response… if you can nail midbass and get 6db of summation then you’ve cracked it… you don’t need to move the mic for midbass measurements, in fact for me you’re better with the mic stationary as it lets you adjust timing better (the top end of any driver always gives the best timing indications… 6db at the upper crossover is what you should be looking for… this is because at higher freqs the phase changes more with timing changes so it’s a better indicator of timing being good)

Cursedbat, I tried altering my target curve generator to take this summing behavior into account. It totally works! I want to update my spreadsheet with this new concept thanks to your really insightful questions.

I thought (and I am a total newbie to this stuff) that if you measure with crossovers not set (or set an octave or 2 above below desired level) and EQ response flat an octave above/below your crossover points that when you sum the signals you will get a near flat result?

Hmm, Maybe i misunderstood the conversation. I was thinking about summing drivers on each channel to get a flat response not summing L+R.

I think you're right. Both concepts are related here in this conversation especially.

I also have a single mono sub (summed in the Helix left+right @50% each) and for the longest time I used to just cut the sub channel by ~6db and tune each side to the curve. This works, but I found that I have better results just tuning the MB, MR and TW on each side to the curve (set in REW as speaker driver xxHP to None,) tuning the sub to it's desired curve and then measure the summed response of the front and adjusting sub level to to match the front.

The summing of the two sides exhibits the same summing issues OP mentioned, and always has. I actually just tried a tune where I added EQ to the summed MR to get it to match the curve and raised the TW ~2.5DB and got the total response to match the curve, but it sounds terrible for me lol.

Quote:

---

Originally Posted by swrocket

The summing of the two sides exhibits the same summing issues OP mentioned, and always has. I actually just tried a tune where I added EQ to the summed MR to get it to match the curve and raised the TW ~2.5DB and got the total response to match the curve, but it sounds terrible for me lol.

---

Sure it sounds terrible, as you're basically pushing the tweeters 2.5 dB above the desired target curve. ;)
Just use the Jazzis curves for each side and not for combined left/right response.

I don't think the spreadsheet can be altered anyway to account for this as every different car will be different and have another frequency where you start to see less than 6dB summation due to different distances between drivers (need to be less than half a wavelength apart in distance to always get 6dB summation. If you cross mids to tweeters at 3kHz you need < 5.5cm distance between drivers centers).