REW and System Baseline Tune Help

Today was my first real time getting into REW, attempting to get my system ready for tuning. My crossover points and time alignment are set, and I've had sound all week and the speakers are starting to break in nicely. My goal was to set some levels but either my ears are broken or I'm using REW incorrectly.

It's a 3-way system controlled with a Helix DSP, with xover points at 2.2kHz (12dB Butterworth), 250Hz (24dB), and 80Hz (24dB). Using a UMIK-1 with the calibration file loaded. I'm still reading a ton on level setting and trying to figure out how to set gains in the DSP (can't/don't know how to set input gain, output set at -3dB on each channel) and amps (gains are down, bumped up a bit on the midbass and sub). Hopefully this doesn't set off any immediate red flags, but noise floor is low and noise from the factory head unit is now at least acceptable.

Part of the way through trying to achieve a flat-ish curve I checked to see how it was going with an actual song, and it was miles off. Highs were overly bright and the bass was almost nonexistent. I ended up giving up and just going by my ear and it's not too bad, but the RTA results aren't what they should be:

https://i.postimg.cc/XNdQtV61/Untitled.jpg

The yellow line is by ear and the green is when I started going for a flat curve. What am I doing wrong?

REW is a great program but certainly takes some getting used to. I personally found the Audiofrog Tuning Guide to be an awesome starting point for tuning and REW setup. The entire guide is worth a read, but the REW settings and suggestions were on page 38. They break down some simple and repeatable starting points to get consistent results, complete with pictures and explanations.

Check it out! https://testgear.audiofrog.com/wp-co...y-it-Works.pdf

How was the mic oriented? Pointing up with a 90 degree calibration file and moving for a spatial average? Or aimed forwards and still?

according to those graphs you have you have 60db less treble than you have bass, at most that should be 25db and that would represent a fairly bass heavy tune... for example the audiofrog house curve is approx 20db between 20khz and 40hz

so either your aren’t actually using the umik as the mic or the calibration file is very wrong!

if I do a quick knock up by ear with 1/3 octave tones I can normally get 20db between lowest and highest... does it recognise your umik when you open rew? Are you using the latest version of rew? (I use version 5.20 beta 30 from AV nirvana forums which host all the rew updates and beta versions :) )

Quote:

---

Originally Posted by blockrocker

REW is a great program but certainly takes some getting used to. I personally found the Audiofrog Tuning Guide to be an awesome starting point for tuning and REW setup. The entire guide is worth a read, but the REW settings and suggestions were on page 38. They break down some simple and repeatable starting points to get consistent results, complete with pictures and explanations.

Check it out! https://testgear.audiofrog.com/wp-co...y-it-Works.pdf

---

Thanks, I will definitely check this out tonight.

Quote:

---

Originally Posted by dumdum

How was the mic oriented? Pointing up with a 90 degree calibration file and moving for a spatial average? Or aimed forwards and still?

according to those graphs you have you have 60db less treble than you have bass, at most that should be 25db and that would represent a fairly bass heavy tune... for example the audiofrog house curve is approx 20db between 20khz and 40hz

so either your aren’t actually using the umik as the mic or the calibration file is very wrong!

if I do a quick knock up by ear with 1/3 octave tones I can normally get 20db between lowest and highest... does it recognise your umik when you open rew? Are you using the latest version of rew? (I use version 5.20 beta 30 from AV nirvana forums which host all the rew updates and beta versions :) )

---

Yeah, I know something is wrong and that it's almost certainly operator error. The mic was pointed forward and still using the regular calibration file. I had my laptop outside of the car so I do know that the mic was being used (I did discover early on that I was originally using my laptop's mic). My REW version is 5.16 so I should update regardless, but that shouldn't be the cause of my issue.

I'll have a read of the Audiofrog guide and try to give it a go by tomorrow night and report back here. Any other suggestions are always welcome as well!

I went out and took some quick measurements in the car just now. Sitting in the driver's seat with the mic pointing up using the 90 degree calibration file and the results are as expected.

I'm not exactly sure what it was I was doing before, but the graph looks different (more jagged) and it was showing responses differently than on Sunday. So the good news is that my mic and ears are OK, and that it was operator error. dumdum, I didn't realize that's what the 90deg cal file was for, so thanks for indirectly telling me what it was!

blockrocker, thanks again a ton for the link to the tuning guide. I've only had time to look at some parts of it but before I do anything more I'll be reading this from front to back. And playing with REW a bit more maybe.

https://i.postimg.cc/7ZMDbwsP/Untitled.jpg

Glad to hear that you are making progress. I also have a Helix DSP (DSP.3) - here are what my responses look like (1/3rd smoothing), just for something to compare it to. I have the following setup:

Dash speakers: 3.5" coaxials (Kenwood Excelon) - mids/highs
Door speakers: 6x9" midbass (Kenwood Excelon) - midbass
Rear Deck speakers: 6.5" coaxials (Kenwood Excelon) - rear fill - bandpassed
Sub: 8" under-seat powered sub (JBL BassPro SL)

I'm not saying any of this is right or wrong - just what I've found that *I* like. The sub response is not pictured here separately (but is included in the "All" light brown'ish response. I used quite a few PEQ filters to get it to this point.

https://uploads.tapatalk-cdn.com/202...4715abf349.jpg

Here is what my original pre-EQ responses looked like(!):
https://uploads.tapatalk-cdn.com/202...46e2b3be8e.jpg

As you can see - big difference! :-)

Quote:

---

Originally Posted by ejeffrey

I went out and took some quick measurements in the car just now. Sitting in the driver's seat with the mic pointing up using the 90 degree calibration file and the results are as expected.

---

When taking measurements, most people will set REW to average the responses "Forever" and let REW keep taking measurements and averaging while the move the MIC from ear to ear somewhat slowly (some people move the MIC slowly and others more quickly) - until you tell REW to stop. I usually let it take (and average) about 85-100 measurements for each measurement.

Also - you may want to try taking measurements of each speaker individually. That will go a long way to showing why the overall response is the way it is. I like to match the response of left/right speakers very closely - really helps with keeping the imaging "centered".

With a Helix DSP, you can really fine tune things. Personally, I've found that getting each speaker EQ'd really well just causes the overall response to work out without needing any "global" EQ. I just EQ each individual speaker really well.

Tuning a DSP like a Helix is really fun and rewarding. I *love* being able to "shape" the overall sound to my specific preferences. It does take some time to learn the tools and to understand how to tune, but it's worth it.

Also - Helix DSP's have an "auto-EQ" feature where it will do the EQ for you! Definitely take a look into that feature - might give you a great tune to use while you learn REW and tuning. Nick posted a how-to video on the Helix auto-EQ feature.

EDIT: One more thing. When setting up an active-crossover system, most recommend using 24dB Linkwitz Riley slopes (instead of the 12dB you are using). 24dB slopes will help avoid dips or peaks at the crossovers.

Quote:

---

Originally Posted by jtrosky

When taking measurements, most people will set REW to average the responses "Forever" and let REW keep taking measurements and averaging while the move the MIC from ear to ear somewhat slowly (some people move the MIC slowly and others more quickly) - until you tell REW to stop. I usually let it take (and average) about 85-100 measurements for each measurement.

Also - you may want to try taking measurements of each speaker individually. That will go a long way to showing why the overall response is the way it is. I like to match the response of left/right speakers very closely - really helps with keeping the imaging "centered".

With a Helix DSP, you can really fine tune things. Personally, I've found that getting each speaker EQ'd really well just causes the overall response to work out without needing any "global" EQ. I just EQ each individual speaker really well.

Tuning a DSP like a Helix is really fun and rewarding. I *love* being able to "shape" the overall sound to my specific preferences. It does take some time to learn the tools and to understand how to tune, but it's worth it.

Also - Helix DSP's have an "auto-EQ" feature where it will do the EQ for you! Definitely take a look into that feature - might give you a great tune to use while you learn REW and tuning. Nick posted a how-to video on the Helix auto-EQ feature.

EDIT: One more thing. When setting up an active-crossover system, most recommend using 24dB Linkwitz Riley slopes (instead of the 12dB you are using). 24dB slopes will help avoid dips or peaks at the crossovers.

---

jtrosky some really helpful info here. I've had a chance to read the Audiofrog tuning guide posted above and have made the decision to go to the 24dB slopes as you suggested. And interesting that there's a ~5dB dip very close to my crossover point... Also I've been studying Nick's guide on the auto-EQ function and will be using that to get a solid tune that I can enjoy for a while.

I like your idea of measuring each speaker individually. It seems like a great way to let the auto-EQ do its thing. My USB extender cable came today so that means I'm ready to make more progress this weekend and will report back here on how things are going.

I am looking forward to playing with the tune quite a bit afterwards to dial in the sound better, but also learn. Thanks again for everyone's help on this!

Quote:

---

Originally Posted by ejeffrey

I like your idea of measuring each speaker individually. It seems like a great way to let the auto-EQ do its thing. My USB extender cable came today so that means I'm ready to make more progress this weekend and will report back here on how things are going.

---

Just to clarify - the Helix auto-EQ doesn't require that you connect your MIC to the Helix itself. The MIC just connects to your laptop. When you mentioned the "USB extender cable", I just kind of got the impression that you were looking to connect your MIC to the Helix itself for the auto-EQ (you would not be the first person to think that is what you need to do!!).

Also, just FYI - I personally haven't used the Helix auto-EQ yet (my tune was pretty much done when the thread was posted about it) - so I'm not sure if you actually want to use it on each speaker. The stuff I've seen says to do whole sides at a time with the auto-EQ (you "link" all channels for the side being EQ'd and the Helix software adjusts the appropriate speaker for the frequency it's adjusting). Not sure if auto-EQ'ing the individual speakers with the Helix auto-EQ is beneficial or if it even works correctly that way.

I need to test out the Helix auto-EQ... Not sure why I haven't yet.

Quote:

---

Originally Posted by ejeffrey

jtrosky some really helpful info here. I've had a chance to read the Audiofrog tuning guide posted above and have made the decision to go to the 24dB slopes as you suggested. And interesting that there's a ~5dB dip very close to my crossover point... Also I've been studying Nick's guide on the auto-EQ function and will be using that to get a solid tune that I can enjoy for a while.

I like your idea of measuring each speaker individually. It seems like a great way to let the auto-EQ do its thing. My USB extender cable came today so that means I'm ready to make more progress this weekend and will report back here on how things are going.

I am looking forward to playing with the tune quite a bit afterwards to dial in the sound better, but also learn. Thanks again for everyone's help on this!

---

Auto eq is automated and literally applies a global eq over the channels I believe, but watch nicks video for more info

measuring individual speakers is handy so you can see they are in phase at the crossover points and sum correctly (you will get six dbs of summation throughout, so a pair of lr24 crossovers in phase will give a flat response as the crossover point is the -6db point for both sides so it sums as flat, butterworth is the -3db crossover point so it sums with a 3db hump, so you can then part the crossover to flatten the hump a little, it’s handy if one side or the other has a dip at the side of the crossover freq to flatten the response when summed

also you can play one side mid for example, then the other and see how they sum when played together, again if phase and timing is good you will get 6db of summation, this can reveal phase differences between sides also and if timing isn’t perfect as it will put equally spaced dips (comb filters) into the summed response :)

You can use rew with its auto eq feature on individual drivers and that can then be imported into the helix software also

there's lots to learn and take in, but give it time and it does get clearer and easier to do

Quote:

---

Originally Posted by dumdum

also you can play one side mid for example, then the other and see how they sum when played together, again if phase and timing is good you will get 6db of summation, this can reveal phase differences between sides also and if timing isn’t perfect as it will put equally spaced dips (comb filters) into the summed response :)

---

I'm sure 2 speakers playing together would be a 3dB boost not 6?

On the helix auto EQ, you can do each speaker individually. You'd have to create your acoustical EQ curve for each speaker set (woofer, mid, tweet, etc) to get the auto EQ to EQ the speaker to its correct curve. Then auto EQ as a speaker set, and then all left speakers and right speakers, and then finally, all speakers. That's a lot of steps and a lot of house curves you have to switch around. But it would get you the best results and follow the same steps you'd take with a manual EQ tune. Nick's point on the auto EQ is you can get a good enough tune by doing all left, then all right and then globally. And that most people wouldn't notice the difference from that and a more advanced tune.

Also a point to make that I noticed using auto EQ. Once you start the process of auto EQing sets of speakers, which should have a summed 3dB increase, you need to increase the reference curve to match up to the new increase in output. Otherwise, you'll tell the auto EQ to EQ out the 3dB increase it is getting from the summed output.

Quote:

---

Originally Posted by jrwalte

I'm sure 2 speakers playing together would be a 3dB boost not 6?

---

I think it should be 6...you double the speakers for 3db and you doubled the power for 3db.

Quote:

---

Originally Posted by Jdunk54nl

I think it should be 6...you double the speakers for 3db and you doubled the power for 3db.

---

That^ is correct.

I could have sworn that I only saw ~3dB increase in level when measuring left+right compared to just left or right with REW?

Someone please correct me if I am wrong.

But if you see only a 3db increase and not 6db with double the power and double the speakers, that should mean that the speakers aren't 100% in phase with each other. They would be somewhere between 0 and 180 degrees out of phase. In other words, not completely in phase, but also not completely out of phase.

Quote:

---

Originally Posted by Jdunk54nl

Someone please correct me if I am wrong.

But if you see only a 3db increase and not 6db with double the power and double the speakers, that should mean that the speakers aren't 100% in phase with each other. They would be somewhere between 0 and 180 degrees out of phase. In other words, not completely in phase, but also not completely out of phase.

---

Yup. Probably caused by reflections

Sent from my SM-G975U using Tapatalk

I'll have to double check the measurements with REW. I don't remember seeing a 6dB increase when playing both speakers - could be wrong though - didn't pay close enough attention to exactly how much higher the measurements were when playing both speakers - just checked to see that response stayed the same across the frequency range - and was higher. I can't imagine that the speakers (all three sets - dash, door and rear-deck) could be out of phase the same amount - I mean the physical polarity is definitely the same on them and time alignment is set properly, so just can't see how they would all be out of phase by the same amount like that.

I'll take some new measurements today and confirm if they are ~3dB or ~6dB higher when playing both left and right and go from there. Who knows, maybe there is some issue with my system that I'm unaware of.

Have you checked measurements on your system and can confirm that you see +6dB of gain when both left and right are playing?

And just to be clear, you are saying that the frequency response should be basically the same with both playing, just 6dB higher than with only one playing, correct?

EDIT: Just saw Nicks post - maybe the phase issue is reflection related - if really only 3dB higher.

Ya, what is causing it and can you fix it are two important questions that are semi-hard to answer....

Quote:

---

Originally Posted by jtrosky

I'll have to double check the measurements with REW. I don't remember seeing a 6dB increase when playing both speakers - could be wrong though - didn't pay close enough attention to exactly how much higher the measurements were when playing both speakers - just checked to see that response stayed the same across the frequency range - and was higher. I can't imagine that the speakers (all three sets - dash, door and rear-deck) could be out of phase the same amount - I mean the physical polarity is definitely the same on them and time alignment is set properly, so just can't see how they would all be out of phase by the same amount like that.

I'll take some new measurements today and confirm if they are ~3dB or ~6dB higher when playing both left and right and go from there. Who knows, maybe there is some issue with my system that I'm unaware of.

Have you checked measurements on your system and can confirm that you see +6dB of gain when both left and right are playing?

And just to be clear, you are saying that the frequency response should be basically the same with both playing, just 6dB higher than with only one playing, correct?

EDIT: Just saw Nicks post - maybe the phase issue is reflection related - if really only 3dB higher.

---

Yeah, you'll be hard-pressed to get that 6 db increase from speakers a car width apart, they need to be within a quarter wavelength of each other to get the full coupling effect. Subs and midbass are fairly easy to couple like that, mids and tweets are not. And the reflections absolutely figure into that problem, and all the other nastiness our cars provide.

Quote:

---

Originally Posted by ckirocz28

Yeah, you'll be hard-pressed to get that 6 db increase from speakers a car width apart, they need to be within a quarter wavelength of each other to get the full coupling effect. Subs and midbass are fairly easy to couple like that, mids and tweets are not. And the reflections absolutely figure into that problem, and all the other nastiness our cars provide.

---

It seems that you are absolutely right! I found some old measurements and this is what I see:

3.5" coaxial dash speakers (play 500hz->20khz) - about a 3dB increase for the entire range of frequencies they play - slightly more at bottom end of frequency range
6x9" midbass door speakers (play 65hz->500hz) - Mostly ~6dB increase - slightly less at upper range of frequencies
6.5" coaxial rear-deck speakers (play 75hz->4khz) - ~6dB increase as lower end and ~3db at upper end.

So it looks like speaker position, frequencies played and possibly reflections and/or phase can all play into this. But generally speaking, at least in my car, the midbass freqs seem to add closer to 6dB and the mid/treble freqs only closer to 3dB.

In the end, I guess it's really not even a big deal as long as you tune for the difference appropriately. At this point, I'm not going to change anything because of it since I had already taken the relative levels between midbass/mid/treble into account when I originally tuned, so as they say "it is what it is". :-)

I played around more with both REW and the auto-EQ function today and was able to get a good baseline tune that I'll be happy with for a while. The car sounds really good and I'm sure there's a decent amount of improvement to be made by someone who knows what they are doing, so I'm really encouraged.

Also, I found out what my original issue on this thread was caused by. I had the mode in REW set to "Spectrum" and not one of the RTA settings. So, yeah, don't use that...

A massive thanks again to those who helped out with this!

After more listening (actually just jumping in the car yesterday morning) I noticed that the most of the image was skewed over to the right, which was mostly corrected by 2 clicks to the left on the balance. Because of this I figured that I may have set the levels wrong before tuning, so I went through the whole process again tonight. Interestingly enough, I ended up with pretty much the same result.

Is this common? Or more user error?

You are on to the tuning by ear part now. Play correlated pink noise and adjust levels on speaker pars until they are centered. Then play band limited pink noise with speaker pairs to center the frequencies even more.

Quote:

---

Originally Posted by ejeffrey

After more listening (actually just jumping in the car yesterday morning) I noticed that the most of the image was skewed over to the right, which was mostly corrected by 2 clicks to the left on the balance. Because of this I figured that I may have set the levels wrong before tuning, so I went through the whole process again tonight. Interestingly enough, I ended up with pretty much the same result.

Is this common? Or more user error?

---

That happens when you use REW's level (calculate target level from response) for each speaker instead of setting them all to the same level (in the eq window, target settings tab, target level). I usually pick the lowest level from the calculated levels and set them all to that. You can just manually adjust the gains in the dsp accordingly.

So while investigation the whole "how much of an increase do you see when playing both left/right speakers at the same time?" thing, I noticed what appear to be some phase-related issues with my current setup (at least I think they are phase-related issues).

The picture below shows the individual left/right door speaker and individual left/right dash speaker responses - along with the purple "both speakers playing" response. I'm guessing that the circled areas are caused by phase issues? The one dip also causes my crossover area to go from 500hz down to 435hz when both left and right speakers are playing.

So my question is... Should I attempt to "fix" the dips with EQ by adding equal amounts to both the left and right speakers in order to "fix' the dip when both speakers are playing (which would also bring my crossover freq back to 500hz)? Or should I just leave them be?

Thoughts? Opinions? Also, note the small dB increase with both dash speakers playing compared to the larger dB increase with both midbass speakers playing (which was the topic originally being discusse here).

https://uploads.tapatalk-cdn.com/202...a7a4a92a03.jpg

Also, with the dash speakers, the areas to the left and the right of the red circle actually seem to be *higher* that normal (instead of a dip) - would I also decrease those areas with EQ (again, to help bring the crossover point back to 500hz)?

Thank you!

You seem to have a 6db summation (ideal) in some areas below the crossover on the dash speakers, and just under above that frequency also... personally as long as the frequencys around that point are centred when played on speaker pairs I wouldn’t adjust it, sometimes the mic picks up stuff that the ear doesn’t... sometimes the mic doesn’t pick up what the ear does... for fine tuning I prefer to check with test tones and the ears

So I have a stupid question about tuning to a target curve... I've always noticed that my measured frequency response always came out different than the target curve I was tuning to and never understood why. However, when I found that I was getting ~6dB increase when both midbass speakers play and only about a ~3dB increase when both mids/tweeters play (2-way dash speakers), then it made sense. :-)

So if I EQ each speaker towards a target curve in REW using the same "Target Level" value for each speaker, the midbass freqs will always end up about 3dB higher than the mid/high freqs, thereby making the resulting frequency response not match the curve I was tuning to.

I'm thinking that the easiest way would probably be to set the "Target Levels" to different values in REW depending on whether I'm tuning a midbass speaker or a mid/highs speaker? For example, if my "Target Level" is 65dB for my mid/highs speaker, maybe I should set the "Target Level" to 62dB when EQ's the midbass speakers? That way, when both left/right speakers play together, the result should match my target curve correctly. At least that is my expectation... I think I'll give that a shot.

Just curious how others handle this since the level of summation for left/right speaker pairs seems to be different depending on the specific speaker and frequencies it plays (at least in my car).

You should get 6db of summation when two speakers play together, if not they are not in phase with one another (this is why LR 24db crossovers work, the crossover point is 6db down and in phase so the average level through the crossover becomes flat when summed

watch some sound design videos from Nathan lively for more info, good guides on phase and levels (even though it’s more commercial sound and not automotive, the rules don’t change... we just get more phase anomalies to deal with :( )

Quote:

---

Originally Posted by dumdum

You should get 6db of summation when two speakers play together, if not they are not in phase with one another (this is why LR 24db crossovers work, the crossover point is 6db down and in phase so the average level through the crossover becomes flat when summed

---

Hmm... Well, if they are not completely in-phase, I'm not so sure I can really do anything about it. Wiring polarity is correct, time alignment is correct and I'm using LR 24dB xovers - what else can I do?

Above, @ckirocz28 mentioned this:

"Yeah, you'll be hard-pressed to get that 6 db increase from speakers a car width apart, they need to be within a quarter wavelength of each other to get the full coupling effect. Subs and midbass are fairly easy to couple like that, mids and tweets are not. And the reflections absolutely figure into that problem, and all the other nastiness our cars provide."

and @SkizeR confirmed that:

"Yup. Probably caused by reflections"

I took those comments to mean that "it is what it is" and that there really isn't much I can do about it with the way the speakers are installed (factory locations - aiming striaght up into the windshield).

If that is the case (that there is not much I can do about it), probably the best I can do is to add 1.5dB to "Target Level" when EQ'ing the dash speakers in REW and subtract 1.5dB from the "Target Level" when EQ'ing the door midbass speakers in REW. That would give me my 3dB difference and make it so the resulting frequency response would pretty much match my target curve. I think. :-)

Or am I missing something here?

This sounds like you are trying to overcomplicate something...If the phase issues are caused by reflections that you do not have control over, then just do the following:

Tune to your curve, then measure the speakers in pairs. If you need to adjust levels, do so in pairs. If your midbass are 3db higher than your mids/tweets, adjust both mids/tweets equally +3db (to keep their side to side level the same difference). Similar if you adjust your midbass. If it is tuned flat, it is tuned flat...it doesn't really matter what target level.

Plenty of people don't even care about target level and just tune each speaker the best, then adjust levels to match after they tune. That order doesn't really matter.

Quote:

---

Originally Posted by jtrosky

So I have a stupid question about tuning to a target curve... I've always noticed that my measured frequency response always came out different than the target curve I was tuning to and never understood why. However, when I found that I was getting ~6dB increase when both midbass speakers play and only about a ~3dB increase when both mids/tweeters play (2-way dash speakers), then it made sense. :-)

---

Well that's interesting .... I'm going to look for that next time I tune something. Neat observation.

Quote:

---

Originally Posted by jtrosky

Just curious how others handle this since the level of summation for left/right speaker pairs seems to be different depending on the specific speaker and frequencies it plays (at least in my car).

---

I was thinking this might be why. When you make a measurement by moving the microphone around a bit, the phase between two speakers is relatively stable since the lower frequencies have longer wavelengths. The higher frequencies have shorter wavelengths and so the phase interaction between two speakers as you move the microphone around becomes more random.

Two coherent sources sum together at +6dB (lower frequencies in this example)
Two random sources sum together at +3dB (higher frequencies in this example)

At least, it sounds solid in my mind's eye :)