Printer Friendly Forum Posts - focal tweeter tuning

I have heard many horror stories about the Focal TN-52 tweets sounding harsh or spitty up high, but after some reading on ECA.com, I have also learned that in the hands of the right tuner, they can sound great. I got mine fairly cheap, and they came bundled with a set of midbasses I wanted, so I think I will try and get as much out of them as possible before I trade them for something else.

Has anyone gotten good results from using these, and if so would you be willing to share some of your tuning knowledge?

Currently, I have them paired with DLS Iridium 6.3 midbass's, running active. While I know this is not ideal(IMO, the DLS midbass does not do too well playing up high - and it was not designed to), it is what I have until I can build A-pillar panels for my TG9's. I want to maximize the tweeters and keep them from sounding harsh, as the DLS are quite the opposite and are very natural sounding. I currently have the x-overs set at 50-2500hz @ 12db for mid bass, and 2500hz + @ 12db for tweeters.

Any takers?

In my opinion you have one of the best tweeters ever engineered specifically for mainstream car audio. The trick to avoiding that harsh sound is to always be aware of comb filtering and take steps to avoid it. Comb filtering is when a change in the speakers phase occurs at a lower frequency that causes a ripple in the response. This ripple will repeat its self over and over again as frequency rises. In other words a simple alteration like poor equalization or reflective surfaces that effect 3khs can have horrendous effects at 10khz.

posted_image

Here you can see the effects of comb filtering on what would otherwise be a flat smooth response. A shift in phase at 1khz causes problems throughout the pass band. This can be caused by tweeter location relative to the mid range, incorrect crossover frequency, incorrect equalization, and other things.

Incorrect crossover frequency selection is a common cause of comb filtering. Your tweeters have an Fs of 1955hz. The accepted rule of thumb is to always crossover tweeters at a minimum of twice the resonant frequency using a 12db crossover. This would be 3910hz in your case.

After you have established a good crossover point for your application, the next step is to equalize if you have the equipment. Always start by equalizing lower frequencies first and working your way up. This actually applies to all speakers of all types. If you randomly equalize you run the risk of inadvertently causing comb filtering. Then you will find your self trying to equalize a problem that you actually caused. Comb filtering can be masked by the way that an RTA displays its information. This is why many people don’t trust these machines. They don’t realize that they are equalizing improperly, and that the RTA lacks the resolution to display that. Like I said comb filtering can be avoided largely by taking steps to avoid it. The correct tool to use to view it in the event that you think it is the source of harsh sounding highs is an FFT analyzer.

All car audio systems suffer from this and most suffer a lot. I stress this because this is the absolute largest cause of bad sounding highs.
I hope this helps with your decisions.

^^^ This is awsome info. It looks like this system will really be able to shine as a 3-way, and until I can get that in, I am going to have to compromise and settle for less. Without putting in a dedicated midrange or changing to a higher-playing midbass(or much lower playing tweet), I am not going to get the best FR...

Maybe I can try raising the tweeter's HP to 4khz w/ a 6db slope, and raising the midbass's LP to 2800hz w/ a 6db slope to see if that helps at all. I jsut don't want to raise the midbass's LP any more to avoid beaming - these DLS's already do not sound that great up high.

I just wanted to avoid using 6 & 18 db slopes b/c I thought I read something in my research that said using these slopes could cause weird phasing issues that would be difficult to work around. Is this true?

You’re on the right track. Making the most of what you have is what its all about. As far as the crossover slopes go I would not count on any specific slopes theoretical phase response. The information that you read is correct but applies to home audio applications. It of course is also valid in a car but in a car there are several other factors that will modify the phase response anyways. For instance the location and angle of your speakers. In a home audio application all of the drivers are typically mounted on a flat surface so phase characteristics are much more predictable. I could go on and on here but the skinny of the situation is always keep the theoretical phase information in mind (like using 12db slopes) but experiment and trust your ears, then make an educated decision. 12db crossover slopes will typically invert the phase of your tweeter but not the mid. So experiment with inverting the phase, and listening. Also in the event that you have equipment capable of time delay you can manually modify the phase of any speaker incrementally. You might seriously want to look into purchasing Auto sound 2000’s test discs part # CD-102 and CD-103. With these two disks you can definitively find the best phase, gain, time alignment settings.

What type of processor are you using for your crossovers?

I am using the PPI DCX-730 - it has T/A, 1-4th order slopes, 3-way capable(6 seperate channels of processing), crossover selection anywhere from 20-20000hz, butterworth/LR filters(selectable) - basically anything I need to do as far as processing is concerned I can.

After I got off of work last night, I set my LP on the mids at 3khz, and the HP on the tweets at 4khz - I think I like that alot better. I also left the slopes at 12db for now. I think where most of my muddiness is coming from is my midbass's - I cut 1500hz/1000hz/750hz down about -5db and it sounds much better. I have not began EQ'ing the tweeters. I think I have my cart a little ahead of my horse here, as I need to get my mechanical phasing for all the drivers down first - I think that's where some issues lie as well.

Another question - If the slopes were to change the phase(say a 12db slope for a 180' phase shift(?) ), wouldn't just flipping the mechanical phase(+ & - wires) be different the taking the x-over away?

I am not sure that I understand the last part of your question, But yes. You must first make sure that electrical phase is correct, and yes inverting the phase at the tweeters terminals is the best way of compensating for a 12db hi pass with inverted phase characteristics. All processing should be done last.

speakermakers wrote:

I am not sure that I understand the last part of your question, But yes. You must first make sure that electrical phase is correct, and yes inverting the phase at the tweeters terminals is the best way of compensating for a 12db hi pass with inverted phase characteristics. All processing should be done last.

While your last response answered my question, I'll reitterate my comment. for clarity...

Doesn't introducing a 12db/ octave crossover into a drivers' signal path effectively create a 180 degree phase shift?

Also, wouldn't a 6db/octave slope cause a 90 degree shift - this is one that would make it difficult to tune around since reversing the mechanical phase(+ & -) would cause a 180 degree shift and not be able to compensate for the it - correct?

Now - isn't it true that flipping the + & - wires (electrical phase) is NOT the same as changing the acoustical phase( introducing a phase shift from a crossover slope)? In my last question I listed the terms mechanical phase and electrical phase - this was an error on my part. I should have used electrical and acoustical phase. My apologies...

Pat your self on the back. It is evident from your questions that you get all of this so much more than the typical enthusiast. It’s nice to see that once in a while. Theoretically a 6db crossover will have no phase shift. In the event that you want to account for phase shifts that are greater or less than 180 degrees then time alignment or physical relocation is your only practical option.

I have often wondered if crossover phase off sets the time alignment of that driver by one half cycle (in the case of a 12db), or if there is no time delay involved in a 12db crossover and re-inverting the phase simply sets things 100% rite. I honestly don’t know. I have also wondered, if the signal is delayed by the crossover (inverted with delay) wouldn’t the effects be frequency dependant and dependant on the type of filter (active or passive).
This last bit of pondering has led me to believe that there must not be a predictable amount of delay involved with any crossover design (but rather an instantaneous shift in phase with no delay). Active at least. If any one has information and references to the source of this information (not just opinions) I would be interested. Please post. Another thought That I have is that, I have read dozens of books on this topic and most of them written by ultra anal mathematicians and physicists. Never once have I come across any author describing the phase shift as an actual delay.

As I said before, even if you could calculate the exact effects of electronic phase you are still faced with the difficulties of overcoming acoustical phase shifts caused by speaker placement in the car environment. All of this knowledge plus time alignment capable processors is our saving grace!

Well I have the ability to time align; is there a rule or formula I would use to see how much time alignment I would need (in addition to what I would use to account for path length differences) to account for the phase shift caused by the crossover? Also, what would I need to consider if I already had the electrical phase flipped on certian drivers, like my driver's side mid bass?

And thanks for the compliment - audio and electronics has always interested me ever since I was about 12-13, but ever since I've had a car I have not been able to get enough info or read enough about it :)

Also, another question comes to mind: say you have a bandpass filter set on a driver. Each filter ( LP & HP) will have its own slope, so if they both had 12db/octave, would you combine them to view it as a 24db/octave slope, would they subtract from each other and have no phase shift whatsoever, or would they do something completely different?

.5ms = 6.8” and as further reference .1ms = 1.4”. Use this formula to estimate the delay that you will need in order to compensate for distance.
To account for phase at a specific crossover frequency you would want to add or subtract delay by the length of one half of a cycle for 180 degrees. For example if the wave length of 1khz is 13.57” then a change in time alignment of about .5ms will cause a 180 degree phase shift. For a 90 degree phase shift you would want to calculate ¼ the wave length of that frequency. Google “wave length calculator” to find an applet that will do this math for you. Because wave lengths become larger as frequency gets lower the time delay amount will depend on the frequency. Remember that in real life several variables exist that will change this slightly. Use the math to get your self in the ballpark and then conduct listening tests with the cds that I recommended to finalize your settings. Read the information included with the CDs carefully.
My understanding is that only hi-pass 12db crossovers invert phase. I would assume that a band pass would exhibit the behavior of a high pass crossover with inverted phase. I have searched extensively for an answer to this. I have yet to find any definitive answers. I have however found a truck load of opinions. This is a difficult one because acoustic phase can not be directly measured.

When calculating time alignment always start with the second farthest speaker from you head. Make the time arrival for this speaker match the time arrival of the farthest speaker(usually your sub). Then work your way to the closest speaker. Never break this rule, or all efforts will be in vain. Many people get caught up in the high frequency speakers and forget that phase is a collective phenomenon. Start with the subs. Check out Steve Brown’s university. Google it.

If I may interject, here...

speakermakers has you WAY on the right track, but if your TD device processes in the digital domain, there will be no phase issues... Input will equal output on all outputs. Keep this in mind as you relocate, "re-phase", and adjust delays.

-------------
It all reminds me of something that Molière once said to Guy de Maupassant at a café in Vienna: "That's nice. You should write it down."

haemphyst wrote:

If I may interject, here...

speakermakers has you WAY on the right track, but if your TD device processes in the digital domain, there will be no phase issues... Input will equal output on all outputs. Keep this in mind as you relocate, "re-phase", and adjust delays.

BY TD, do you mean Time Delay? If so, I understand - my processor is digital, so I should have no phasing issues...

How would you know if the TD is digital? Does my AlpineCDA-9830 have digital TD?

-------------
2013 Kia Rio -90a alternator
DDX470HD GTO14001 GTO1014D (x3)
Big3 in 1/0G
1/0G to GTO14001

jmelton86 wrote:

How would you know if the TD is digital? Does my AlpineCDA-9830 have digital TD?

I would assume if you have a digital processor or not - if it's in a headunit, probably not.

(Hope I'm right on this)

jmelton86 wrote:

AlpineCDA-9830

My 9833 has TD processing - greenbroncoguy's recent thread's acctually got me playing with it.

greenbroncoguy wrote:

if it's in a headunit, probably not.

Since the function is contained on the integrated circuits and microchips in the headunit, it would sort of have to be 'digital.' This is why all CD players have a "DAC" - digital-to-analog converter that finishes the signal for the RCA's.

-------------
"I'm finished!" - Daniel Plainview

sedate wrote:

jmelton86 wrote:

AlpineCDA-9830

My 9833 has TD processing - greenbroncoguy's recent thread's acctually got me playing with it.
greenbroncoguy wrote:

if it's in a headunit, probably not.

?
Since the function is contained on the integrated circuits and microchips in the headunit, it would sort of have to be 'digital.' This is why all CD players have a "DAC" - digital-to-analog converter that finishes the signal for the RCA's.

https://i-personalize.alpine.com/nam/index.html#

After you register, you can input the distances from your speakers/subs to you head and it'll give you approximate delay times. -I did it and it was off by 2 clicks, but still pretty close.

Plus, if your unit is compatible, you can burn the settings (x-oxers, TD, EQ) to a disc and when you put the disc in your unit, it'll load the settings to your unit.

If you want to, measure the distances from your speakers to head, post them here, and i'll do it for you and post the approx. delay times...

-------------
2013 Kia Rio -90a alternator
DDX470HD GTO14001 GTO1014D (x3)
Big3 in 1/0G
1/0G to GTO14001

What haemphyst says is quite possibly true. It depends on how the software was written. It can be done both ways, and is. Typically I have found that if your software includes features like selectable crossover types (Butterworth, Linkwitz-Riley, Bessel, etc.) there will be phase characteristics that mimic the analogue design. This can be of great use in the right hands when dealing with asymmetrical configurations and uneven path lengths. But it can cause mass confusion in the wrong hands. There are no hard rules on this though as far as I know. And I have found that its nearly impossible to get this information from the manufacturers. When you think about it, who’s answering the phone. Its not likely to be the guy that designed the processor, or anyone close to his pay grade. So how reliable could any information that you get be?

If you get your hands on those CDs though, it’s easy to determine proper time alignment by ear (after calculations). And like I said after all the calculations are done, and done correctly you will still most likely be off by some amount on a couple if not several of your speakers. The fine tuning here must be done by ear using material specifically recorded for this type of tuning.

One good way I set my TD with is a track from a BMW Logic7 audio test CD (highly recommended). It is just a 'click' noise that repeats over and over. It seems to be at all frequencies, so setting the door speakers to the sub is easy.

-------------
2013 Kia Rio -90a alternator
DDX470HD GTO14001 GTO1014D (x3)
Big3 in 1/0G
1/0G to GTO14001

sedate wrote:

Since the function is contained on the integrated circuits and microchips in the headunit, it would sort of have to be 'digital.' This is why all CD players have a "DAC" - digital-to-analog converter that finishes the signal for the RCA's.

I think just the reading of the CD part would be digital. If you were to listen to the radio, it would not have a digital signal I wouldn't think...

speakermakers wrote:

What haemphyst says is quite possibly true. It depends on how the software was written. It can be done both ways, and is. Typically I have found that if your software includes features like selectable crossover types (Butterworth, Linkwitz-Riley, Bessel, etc.) there will be phase characteristics that mimic the analogue design. This can be of great use in the right hands when dealing with asymmetrical configurations and uneven path lengths. But it can cause mass confusion in the wrong hands. There are no hard rules on this though as far as I know. And I have found that its nearly impossible to get this information from the manufacturers. When you think about it, who’s answering the phone. Its not likely to be the guy that designed the processor, or anyone close to his pay grade. So how reliable could any information that you get be?

If you get your hands on those CDs though, it’s easy to determine proper time alignment by ear (after calculations). And like I said after all the calculations are done, and done correctly you will still most likely be off by some amount on a couple if not several of your speakers. The fine tuning here must be done by ear using material specifically recorded for this type of tuning.

Maybe it's not such a good thing that my processor has selectable filter types(Butterworth, Linkwitz-Riley) - I may not be "the right hands" posted_image

greenbroncoguy wrote:

sedate wrote:

Since the function is contained on the integrated circuits and microchips in the headunit, it would sort of have to be 'digital.' This is why all CD players have a "DAC" - digital-to-analog converter that finishes the signal for the RCA's.

I think just the reading of the CD part would be digital. If you were to listen to the radio, it would not have a digital signal I wouldn't think...

If you're controlling time delay, filters, etc. with the headunit it must use an A/D converter on the radio and any other analog input signals first. If the filters were analog then I don't think you could change them by pushing buttons. Of course the question of whether the filters were programmed to behave exactly as analog filters (i.e. phase shift) remains.

Don’t worry your self too much about this. On the surface that sounds quite irresponsible but, let me explain. In a perfect world where everything works out just the same in the car as it does on paper everybody would use 24db slopes with no phase altering characteristics and all would be jolly. But this is not the case. In the real world we don’t even know for sure where the acoustic centers (the point which sound starts) of our speakers are. This point is theoretically at or near the dust cap but varies greatly by speaker design. There fore even if you mounted all drivers on the same plane and the same distance from your ear and used phase coherent crossovers you would still not likely achieve a phase coherent acoustic wave front. Now don’t get me wrong. Calculate everything you can. Twice, three times. Account for every possible variable. In the end you should be within one wave length or less of your target on all speakers in the system. After that you can easily determine precise phasing, time alignment, and amplitude per speaker using gated pink noise (no equipment necessary). Using gated pink noise you can quickly achieve elements of music reproduction that most people don’t even know are possible.
At this point and only at this point you should also experiment with alternative slopes and types of filters (Butterworth, Linkwitz-Riley, Bessel, etc.) if they are available in your software. In almost all but, not all cases there will be considerable advantages. This is why I encourage you to master your abilities with 24db Butterworth (or what ever your software defaults to) slopes and then move on to experiment with uneven and odd order slopes. You can always go back to a previous setting. Ya gota love digital!

First post time, sorry for the noob question.

When measureing the distance for a rear firing sub should I measure the distance to the rear of the trunk from the drivers head, plus the distance from the rear of the trunk to the woofer itself. Or should I just measure to the woofer from the drivers head.

Thanks