Tuning Help

I have only tuned 1 other time a few years ago and it has taken me a week to get to a point where I feel like I can kinda tune something.

I have the following response curve with target curve. I intentionally tuned bass with only 1 channel enabled so I could get a stronger signal iiin the end.
View attachment 17821

Most music still feels like it is missing bass and I have to admit, I have no clue what I am doing. Could it be the sub enclosure is just garbage, not sure what to expect.

System details if it matters: Alpine SWS12D2 (Punch 60), Focal TA-246 Tweeters, NVX XSP65 (SB Aoustics SB17MFC35-4) powered by Alpine PDX-F4 Mosconi 6to8 v1.

If it helps to see any of my progress here are my response files.
 
That sub response looks ok to me overall, but you could be having phase issues between the subs and midwoofers. Hard to tell from those graphs. Could you show us an individual of each driver on the same graph?
 
Thank you for sharing your measurement data file. Below is the best I can understand what Jacob is asking for. Left tweeter, mid, and sub all with eq ... plus an overall measurement of left.

It seems like your mid level was decreased some when you made the overall left measurement.

You might feel like there is more bass if you make the transition smoother from 60Hz to 300Hz (ish) like I drew with the green line (super non specific please don't aim for that exact shape).

Looks like you have some cancellation in the midbass region around 140Hz between the subwoofer and the mid. The cancellation is not huge, but it should ideally sum together better in that area. You might be able to improve this with a combination of time delay and polarity changes on the mid and/or subwoofer.

View attachment 17827
 
Looking at the measurements in the file, definitely some phase issues (get dips when different speakers are played together) that weren't there in individual measurements.
Look at the crosshair part. That is a dip from about 160hz to about 300hz. That is a pretty crucial zone to have a dip.
You can look at this chart and see how many instruments have their fundamental frequency in the specific range: https://alexiy.nl/eq_chart/
View attachment 17828
 
That dip at 160Hz on the left side measurement is typical for a door mounted midbass driver on the drivers side. There is absolutely nothing you can do about that.
 
Right, but the dip in blue/green (when both left and right are playing together) is not there in the individual sides are played, so there is something you can do about that.
 
Last edited:
Right, but the dip in blue/green (when both left and right are playing together) is not there in the individual sides are played, so there is something you can do about that.

I would always listen first and check if it's an audible problem. Sure you could try an allpass filter to fix this, but in my experience allpass filters almost always destroyed the staging for me. Therefor listen first and if it isn't an audible problem leave it as is.
 
Why does it look like the midbass are playing to 40hz? Is that pre tuning? I don’t have my laptop here so I can’t look, I’m basing this on Justin’s picture and the cancellation at the sub/midbass that’s cirocked and the huge response peak below it
 
Let me make sure I understand how to track down the dip in 160-300hz.

Flip phase on 1 miidbass and remeasure response of system
flip phase on subwoofer only and measure system
Flip phase on 1 midbass and subwoofer and measure system
Analyze and see if any combination helps
 
Why does it look like the midbass are playing to 40hz? Is that pre tuning? I don’t have my laptop here so I can’t look, I’m basing this on Justin’s picture and the cancellation at the sub/midbass that’s cirocked and the huge response peak below it

Not sure, I assume it was some kind of resonation in the car. I will def confirm i though.
 
Possibly. It could be a polarity issue, it could be a time issue, it could be some other issue.

I would check the following in order

1) The polarity are all wired the same and the dsp settings are all the same
2) the time alignment settings are correct for each
3) You have followed a house curve and have 24db/oct slopes on the mids and the subs at the same frequency. (it doesn't appear this is followed very well yet)

If all of those are correct and you still have that phase issue I would then try the following

1) Depending on the dsp, you could do input FR viewing and see what is happening on the input side of things. May be a factory all pass filter in place.
2) Try flipping polarity of one driver at a time and seeing if it improves
3) try changing time alignment ever so slightly (dangerous game here without being able to measure phase)
4) Try adding an all pass filter on one of the drivers at 230hz and see what happens there.
 
I have only tuned 1 other time a few years ago and it has taken me a week to get to a point where I feel like I can kinda tune something.

I have the following response curve with target curve. I intentionally tuned bass with only 1 channel enabled so I could get a stronger signal iiin the end.
View attachment 17821

Most music still feels like it is missing bass and I have to admit, I have no clue what I am doing. Could it be the sub enclosure is just garbage, not sure what to expect.

System details if it matters: Alpine SWS12D2 (Punch 60), Focal TA-246 Tweeters, NVX XSP65 (SB Aoustics SB17MFC35-4) powered by Alpine PDX-F4 Mosconi 6to8 v1.

If it helps to see any of my progress here are my response files.

In addition to what jdunk54nl said... what car is it? Stock headunit? Input to your aftermarket system from the original soundsystem or low-level?
Some people may already know if there is a factory allpass filter set if they know the car and stock soundsystem.
 
Vehicle: 2013 Jetta Sedan TDI
Headunit: Kenwood DMX905 refurb

I checked polarity with test tone and all show green.
Re-measured driver distances and nothing wrong found.
Checked crossovers and everything was on butterworh (WTF), changed to LR 24db
I re-measured the system and while checking the right channel I noticed some strange cancelations around 100hz (mid/sub is set at 80hz) so I inverted the right mid and now see a big gain at 100hz.
Measured both channels together and no more strange cancelations at 160/300 etc (I think)

I have been trying to tune to a house curve, clearly I have been failing
I think my tweets are still a little high but I have them set -9db in DSP
I don't know how much of the irregularities in the response to attack.
View attachment 17854
 
Last edited:
Set a high shelf filter to bring the tweeters down a bit. My estimation would be a high shelf at ~7kHz with -2dB to start with.
Same for a low shelf at the bass. low shelf at ~60Hz with -4 dB to start with.
Remeasure afterwards and check if the response is better aligned to your target.
 
Set a high shelf filter to bring the tweeters down a bit. My estimation would be a high shelf at ~7kHz with -2dB to start with.
Same for a low shelf at the bass. low shelf at ~60Hz with -4 dB to start with.
Remeasure afterwards and check if the response is better aligned to your target.

OMG, thanks so much for that response - I had no idea what a shelf filter was nor the fact that my DSP supports it!

I can't wait to measure again and see how close to the target curve I can get now.
 
Shelf filters are amazing!

Also, to help with what to focus your energy on, try using more smoothing. I see you're using 1/48th octave smoothing in the last chart you posted. Try using 1/3rd smoothing and you'll end up chasing your tail a lot less. When you get a good feel for how to make adjustments at that resolution (1/3rd-octave), and you can get the response to match your target, then you can try less smoothing like 1/6th-octave. Any more than 1/6th-octave is going to stress you out more than help, but higher resolution can be useful for chasing down really specific anomalies.

For tonality, I use 1/3rd or 1/6th-octave only now. It's so much faster and so much less stress and it sounds great. I'll often squint my eyes a bit and look for the overall big trends and correct those instead of focusing on individual peaks and valleys. One way to enforce this concept is to use a parametric equalizer with a Q value between 0.5 and 1.5, but not greater than 1.5. You can also use up to one high shelf and up to one low shelf filter if you like, but not more than that. Then you won't be able to chase tiny wiggles anyways.

Oh! It looks like you're using the RTA mode to capture the measurements. If that's the case, then use the highest resolution in the RTA mode (probabbly 1/48th-octave) then apply your smoothing in the main REW window. This way you can change to higher resolution if you want to later on to investigate stuff.
 
Shelf filters are amazing!

Also, to help with what to focus your energy on, try using more smoothing. I see you're using 1/48th octave smoothing in the last chart you posted. Try using 1/3rd smoothing and you'll end up chasing your tail a lot less. When you get a good feel for how to make adjustments at that resolution (1/3rd-octave), and you can get the response to match your target, then you can try less smoothing like 1/6th-octave. Any more than 1/6th-octave is going to stress you out more than help, but higher resolution can be useful for chasing down really specific anomalies.

For tonality, I use 1/3rd or 1/6th-octave only now. It's so much faster and so much less stress and it sounds great. I'll often squint my eyes a bit and look for the overall big trends and correct those instead of focusing on individual peaks and valleys. One way to enforce this concept is to use a parametric equalizer with a Q value between 0.5 and 1.5, but not greater than 1.5. You can also use up to one high shelf and up to one low shelf filter if you like, but not more than that. Then you won't be able to chase tiny wiggles anyways.

Oh! It looks like you're using the RTA mode to capture the measurements. If that's the case, then use the highest resolution in the RTA mode (probabbly 1/48th-octave) then apply your smoothing in the main REW window. This way you can change to higher resolution if you want to later on to investigate stuff.
I will always have a look at 1/48 first and see if there’s any obvious big peaks or dips, make note and tackle if needs be, then swap to 1/3 octave and ignore any sections with big suckouts that looked ok at 1/48 octave aside from the dip, it makes more efficient use of limited tuning time if it’s a tune for someone else
 
Last edited:
Shelf filters are amazing!

Also, to help with what to focus your energy on, try using more smoothing. I see you're using 1/48th octave smoothing in the last chart you posted. Try using 1/3rd smoothing and you'll end up chasing your tail a lot less. When you get a good feel for how to make adjustments at that resolution (1/3rd-octave), and you can get the response to match your target, then you can try less smoothing like 1/6th-octave. Any more than 1/6th-octave is going to stress you out more than help, but higher resolution can be useful for chasing down really specific anomalies.

For tonality, I use 1/3rd or 1/6th-octave only now. It's so much faster and so much less stress and it sounds great. I'll often squint my eyes a bit and look for the overall big trends and correct those instead of focusing on individual peaks and valleys. One way to enforce this concept is to use a parametric equalizer with a Q value between 0.5 and 1.5, but not greater than 1.5. You can also use up to one high shelf and up to one low shelf filter if you like, but not more than that. Then you won't be able to chase tiny wiggles anyways.

Oh! It looks like you're using the RTA mode to capture the measurements. If that's the case, then use the highest resolution in the RTA mode (probabbly 1/48th-octave) then apply your smoothing in the main REW window. This way you can change to higher resolution if you want to later on to investigate stuff.

I am using RTA mode to make measurements, I wasn't aware there was another way.

When tuning, I am under the impression I should measure left, eq, then make the right match the left. This sounds simple but the only thing I have been able to kinda do is match to a target response. Are there any tricks for matching left/right response?
 
I use my microphone to set my levels (gain) on the amps to be as close as possible to the target curve first. Then fine tune using the gain on the dsp to get a good baseline for levels. Set Time alignment based on distance or use one of the built in time alignment tools depending on your dsp.

Then I measure each speaker individually and adjust the crossovers and the eq to best match the target curve. I don't worry about matching spl levels at this point. The dsp settings only matter in that they get the driver to match the curve. Don't pay too much attention to what you are using, like butterworth, linkwitz, shelf, peq, etc. Use the tools that you have to get the acoustic (measured by microphone) curve to match the wanted curve. Like Justin, I limit the q that I am willing to use. I go up to a q of 5, but that is the highest. I do this based on Floyd Toole's Research that anything beyond a q of 5 is really really hard to hear.

Then I measure each speaker post crossovers and eq to see if it matches my target curve and predicted response in REW. If so I am done with that driver's eq. If not, I fine tune on the spot with the RTA set to like 8 averages so it updates when I make a change.

Once the driver matches the curve to my expectations, I then fine tune the levels of each drive to match each other and the curve.

After all of that. I put away the microphone and computer and pull out test tracks of bandwith limited pink noise and songs that I really know and listen to what needs to change or needs adjusting (something pulling left/right, backwards, etc.) I don't make any changes at this point as my ears are probably dead but write them down so I know what I thougth. I usually do this for a like a week or so without making a bunch of changes but keep writing them down. If it still is an issue a week later, then I will make the changes. I do this so I make sure I am not changing something that is just a result of my ears being tired that day.
 
I am using RTA mode to make measurements, I wasn't aware there was another way.

When tuning, I am under the impression I should measure left, eq, then make the right match the left. This sounds simple but the only thing I have been able to kinda do is match to a target response. Are there any tricks for matching left/right response?

There are many, MANY different ways to make a system sound good. I prefer to match all the speakers to a target because the target never changes, but that's just my preference.

If you like to match the left and right together, it's the same process as trying to match to a target. I assume you're having trouble getting the left and right to mach eachother because one is not responding to your efforts and being a but more stubborn? I haven't tried this method, but my best guess is the car is a non-symmetrical environment so whatever abnormalities you have on the left channel might be difficult to reproduce with the right channel, because the right channel will have some unique abnormalities of it's own. If the left and the right fundamentally behave different because you are measuring from the driver's seat, then it might not be your fault.

As an experiment, you could try putting the microphone in the center of the car like on the arm rest and then see if you can get the left woofer to match the right woofer any easier than before. This is one way to maybe confirm this total guess that I'm making right now.
 
Back
Top