Crossover Design Trouble(Acoustical measurements fixed!)

[Sdiver2489]

Hello all,

I have been experimenting with different setups to see which gives me the most accurate acoustical results for my drivers. I measured the Seas tweeter on a 11 x 17 baffle flush mounted. The mic(New version of radioshack SPL Meter) was set up .5 M away directly on axis with the tweeter. My "gated" frequency measurements are always fairly consistent:



As you can see, there is a pretty large spike at around 8000 Hz. I do not know why the driver is doing this. I have also measured it off the baffle and got very similar results. Just to be sure it was nothing in the room I measured the gated frequency response then disconnected the speaker and measured it again. This way any room noise would be apparent. After subtracting the results of the two measurements, the results were still nearly the same.

Could this be a bad tweeter?

*****PROBLEM FIXED BY BEHRINGER MIC*****

[crackyflipside]

From what I've heard around different audio forums, the Rat-shack SPL meter is very unreliable in the high frequencies.

[---k---]

I was going to say the same thing, but figured I would let the measurement experts talk first.

[JonMarsh]

Hello all,

I have been experimenting with different setups to see which gives me the most accurate acoustical results for my drivers. I measured the Seas tweeter on a 11 x 17 baffle flush mounted. The mic(New version of radioshack SPL Meter) was set up .5 M away directly on axis with the tweeter. My "gated" frequency measurements are always fairly consistent:



As you can see, there is a pretty large spike at around 8000 Hz. I do not know why the driver is doing this. I have also measured it off the baffle and got very similar results. Just to be sure it was nothing in the room I measured the gated frequency response then disconnected the speaker and measured it again. This way any room noise would be apparent. After subtracting the results of the two measurements, the results were still nearly the same.

Could this be a bad tweeter?

Using an RS SPL meter mic for tweeter measurements?


:rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl:


Please, PLEASE, get a Behringer ECM8000 if you can't afford a calibrated electret or other low dollar solution. They're soft above 15kHz, but in a very predictable, compensatable kind of way....

And DON'T forget that an 11 X 17 baffle is NOT an IEC infinite baffle, and has diffraction effects also- try modeling it in BDS or the EDGE if you want to understand what I mean. 8 kHz is a very common diffraction frequency for small baffles, depending on where you've mounted it in relation to the edges.

[Sdiver2489]

Using an RS SPL meter mic for tweeter measurements?


:rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl:


Please, PLEASE, get a Behringer ECM8000 if you can't afford a calibrated electret or other low dollar solution. They're soft above 15kHz, but in a very predictable, compensatable kind of way....

And DON'T forget that an 11 X 17 baffle is NOT an IEC infinite baffle, and has diffraction effects also- try modeling it in BDS or the EDGE if you want to understand what I mean. 8 kHz is a very common diffraction frequency for small baffles, depending on where you've mounted it in relation to the edges.

Seas own tests of the 27TDFC/TV are not done on an infinite baffle per their datasheets. Thanks for the info though.

What would be a good preamp/phantom power supply to go with that Mic? Trying to stay as cheap as possible with still getting a fairly linear response.

[---k---]

MAudio PrePro sound card is a quality sound card with phantom power. Last I saw, it runs about $140.

There is also a Behringer Mixer board that I've seen recommended several times. It runs about $50. I think it was this one: http://www.behringer.com/MX602A/index.cfm?lang=ENG
I did a search for it at Parts Express and Musician's Friend, but neither had it. It may have been replaced.

[Sdiver2489]

MAudio PrePro sound card is a quality sound card with phantom power. Last I saw, it runs about $140.

There is also a Behringer Mixer board that I've seen recommended several times. It runs about $50. I think it was this one: http://www.behringer.com/MX602A/index.cfm?lang=ENG
I did a search for it at Parts Express and Musician's Friend, but neither had it. It may have been replaced.

So lets say I pick up the ECM8000. Which would be the better approach. The analog outs on my current sound card suck. The frequency response starts to roll off at about 9KHz. So I have been using the digital outs which is what the card is designed primarily to do. This works great as it avoids a bad D/A conversion by the sound card and uses my receiver(HK 435) to do it instead. This results in very low noise output...but there is always a delay that speaker workshop seems to have trouble with as it can no be compensated for.

Soo....I could just continue to deal with it and kinda guess and check until I find the right gate times. Or I could get the M-audio Delta 1010LT which has mic preamps built in.

Another option is to keep the sound card and get a mixer such as the

http://www.zzounds.com/item--BEHXENTX502

with the ECM8000.

Or I could get a basic sound card with better analog outs, the preamp shown above and the mic. Anyone have any opinions they could offer?

[dlneubec]

I don't believe that the 502 includes a phantom power supply. The one most people get is the 802:

http://www.partsexpress.com/pe/showdetl.cfm?&Partnumber=248-576

[Sdiver2489]

I don't believe that the 502 includes a phantom power supply. The one most people get is the 802:

http://www.partsexpress.com/pe/showd...number=248-576

Very nice catch. The manual is for the 502 and 802 and so on and they only specify it is not for the 502 in one very small section. :T

[Sdiver2489]

I don't believe that the 502 includes a phantom power supply. The one most people get is the 802:

http://www.partsexpress.com/pe/showd...number=248-576

ok I ordered the mic, the 802 and the needed cables. Lets hope this fixes the problems with my measurements...I suppose at worst I get more accurate readings. I dount baffle defraction would cause a 15dB gain in output at 8KHz. But the again...I am not THAT experienced with acoustical measurements. ;x(

[JonMarsh]

EDGE and BDS (Baffle Diffraction Simulator) are both free, you need Excel to run BDS, though.

Try simulating your test setup. You might be surprised... peaks and dips of several dB in a panel without treatment are very common, and if you have edge distances stack up (same distance to two or more edges), then peaks can be even higher.

Try doing you measurements with a heavy felt sheet surrounding your tweeter on the baffle. Or do your measurment ultra-nearfield- say, 1". That will give you a good idea of the performance of the driver, but it won't clue you in to how your driver performs on the baffle design, which is critical to an actual speaker design.

[JonP]

Very nice catch. The manual is for the 502 and 802 and so on and they only specify it is not for the 502 in one very small section. :T

Heh... I had to catch it by buying the 502... at least I was able to exchange it and get the next one up! I believe that family line has been discontinued for the newer "Xenyx" line, pretty much the same thing tho the mic amps are supposed to be somewhat quieter.

Another little "gotcha" when using these mixers. Check the EQ pots periodically. Every now and then I find one has gotten turned a bit off the center detent... there goes the flat response! Kinda wish they didn't have that, but if I ever use it for a actual mic mixer, would be nice to have.

And yes, as one who started using SW with a RS meter.... you just might be looking at funky response plus "interesting" resonances in that big plastic mic holder. There's actually a little electret mic inside the big silver enclosure. It did suck, though I didn't appriciate how much till I got the ECM8000.

Alternative to the ECM8000 (which I also have) would be to find contact info for Kim Girarden (sp?) who does mic calibration, he also sells the small electret capsules with calibration info for about $40 if I remember right. If you don't mind a little soldering and hand manufacturing, you can mount it in a thin tube, add a mic stand adapter, and have a very nice and calibrated mic.

[JonP]

ok I ordered the mic, the 802 and the needed cables. Lets hope this fixes the problems with my measurements...I suppose at worst I get more accurate readings. I dount baffle defraction would cause a 15dB gain in output at 8KHz. But the again...I am not THAT experienced with acoustical measurements. ;x(

Congrats... always good to know you can (mostly) trust your test equipment.

You also might want to look at where you're placing your start cursor, as well as your windowing... That can have large impact on the interpereted frequency response.

It's been many moons since I've worked with it, but I think there are ways to compensate for delay, search around for it.

I'd also scale the graph to 500-20K or so, won't change the FR, but will give you more info in the important range. At least SW lets you do that pretty easily.

And lastly, if you have any other tweeters or small drivers that should be fairly flat in that region, give them a try to see if that separates the problem from that driver...

[Sdiver2489]

Congrats... always good to know you can (mostly) trust your test equipment.

You also might want to look at where you're placing your start cursor, as well as your windowing... That can have large impact on the interpereted frequency response.

It's been many moons since I've worked with it, but I think there are ways to compensate for delay, search around for it.

I'd also scale the graph to 500-20K or so, won't change the FR, but will give you more info in the important range. At least SW lets you do that pretty easily.

And lastly, if you have any other tweeters or small drivers that should be fairly flat in that region, give them a try to see if that separates the problem from that driver...

I plan on testing the other Seas Tweeter I have shortly. My equipment comes tomorrow(yes...fast shipping from PE) so I will hopefully be getting some good measurements. I may borrow a card with a better analog section than my card has if the digital->receiver path still leads to the same results.

I did try using the time delay but got odd results. Perhaps I should play with it some more but when I adjusted the delay in what I believe was the correct direction I heard the tone play...stop...and then play again.

I will have to experiment more. I wish I had more time.

[JonMarsh]

One other comment- look at all the phase turn-overs in your tweeter response measurement- that's a clue that the MLS window setting is probably off a skosh... means the phase data won't be useufl for doind a design in software.

Give that a close look and checkout in your next measurement effort.

[Sdiver2489]

One other comment- look at all the phase turn-overs in your tweeter response measurement- that's a clue that the MLS window setting is probably off a skosh... means the phase data won't be useufl for doind a design in software.

Give that a close look and checkout in your next measurement effort.

Thanks for the tip. I made a VERY quick and rough measurement today and it seems like the problem is gone. My only issue now is that the ECM8000+XENYX 802 seem to pick up a large amount of ambient noise compared to the rat shack meter. Should I be changing something on the mixer or in software to adjust the sensitivity? While I can see what my mic is picking up in the graphs...the noise surrounding the data is rather annoying.

[JonMarsh]

Yeah, those anechoic chambers are for keeping noise out as well as for controlling reflections within. Doing measurements on a street corner will be very frustrating, for example. That's why CLIO and Praxis have averaging modes- to average out the non repeitive noise from the actual data.

Maybe threre's a setting for that in Speaker Workshop.

[Sdiver2489]

Yeah, those anechoic chambers are for keeping noise out as well as for controlling reflections within. Doing measurements on a street corner will be very frustrating, for example. That's why CLIO and Praxis have averaging modes- to average out the non repeitive noise from the actual data.

Maybe threre's a setting for that in Speaker Workshop.

Ok. The mic+preamp solved the odd measurement problem. Now I am having a slightly less weird issue.

I need to measure my woofer both nearfield and gated farfield to be able to get a curve which accurately follows the FR of the speaker. Unfortunately, my phase measurements for both the near field and far field don't match up precisely so I can't easily splice them together. I eliminated the time up to the MLS impulse in my gates which made my gating phase look good. The nearfield measurements were taken with the mic less than a quarter inch from the driver. Any tips on why they aren't matching up.

Another question I have is that my gated far field measurements exhibit low frequency roll off starting at 500 Hz for my woofer. This is not the case with my near field measurements. Is this the typical case when measuring both drivers just on the baffle with no surrounding box or is there something wrong?

[KJP]

Don't worry about your near field phase measurements. Phase plots in SpeakerWorkshop are relative to the test signals "time of flight" from speaker to mic position compared to the reference input channel. Change mic position and this relationship changes becoming more or less meaningless to compare to one another. What matters to the crossover simulations is your gated plots of the woofer and tweeter at a fixed mic position. Phase is then relative between drivers and contains offset information relieving the need for offset guessing in crossover modeling.

Hope that makes sense.....

Measuring a speaker indoors becomes problematic below 300-500hz. What you are seeing is probably a combination of this and the open baffle effects that you alluded to. Sounds like nothing is wrong to me.

[Sdiver2489]

Don't worry about your near field phase measurements. Phase plots in SpeakerWorkshop are relative to the test signals "time of flight" from speaker to mic position compared to the reference input channel. Change mic position and this relationship changes becoming more or less meaningless to compare to one another. What matters to the crossover simulations is your gated plots of the woofer and tweeter at a fixed mic position. Phase is then relative between drivers and contains offset information relieving the need for offset guessing in crossover modeling.

Hope that makes sense.....

Measuring a speaker indoors becomes problematic below 300-500hz. What you are seeing is probably a combination of this and the open baffle effects that you alluded to. Sounds like nothing is wrong to me.

Thanks for the response! That does make sense. However, I don't really get meaningful response due to the limitations of my room and its reflections in the gated response below 1000 Hz. Should I sacrifice having some reflections in my chart to get more accuracy? This doesn't make sense to me as I am getting that roll off I described before so below 1000 hz the acoustic data does not reflect the actual driver as I believe it should. I will add a graph shortly to show my results.

EDIT: Here is the picture

[KJP]

It looks to me like your opening marker is set too close to the pulse. Setting it a little earlier in time, .5ms or so, should improve your high frequency response.

I like to get a measure pulse to use for playing with the calculate FFT time range settings. This way I can quickly see the effects of marker settings and choose their placement knowing a little bit more about the effects that they have. Seems to help in choosing my evils.

That said your low end roll off does look suspicious. Is your baffle centrally located in your room and away from objects that may be causing reflections?

[JonMarsh]

Don't forget, if your first goal is to verify the integrity of the tweeter/mic combo, an ultra nearfield is the way to go, say 1" from the dome. Turn down the levels of course, and be sure to enter the mic cal data in Speaker Workshop if you can, ECM8000's are soft above 15 kHz. A generic cal file (web search) for ECM8000 is better than no cal file.

Don't worry, you're getting there...


Here's the ECM8000 setup I used to use (still have it) with a DMP preamp from MAudio.




Here's one of my bookshelf projects from the early part of this century (2002), and the measured response on and off axis using the ECM8000 in my living room, with CLIO measurement system.

The diffraction control on the front panel is an integral part of the design with the standard Woodstyle grille panel, to avoid the effects it would usually have in the presence and treble region.




Measured plot- sorry I don't have a smaller version hosted online I can refer to. This measurement was only intended to verify the 1 meter repsonse from ~300 Hz up at on axis and 15 degree increments off axis. Bass response was verified with nearfield measurements of woofer and port. Crossover is an elliptic 8th order equivalent at 1250 Hz, which is why the presence region holds up very well up to 45 degrees off axis. This system has also been built with SS98000 tweeters and RS28A.





Something to ponder, and why baffle and room behavior are so important, you ears hear what the mic hears...


~Jon

[Sdiver2489]

It looks to me like your opening marker is set too close to the pulse. Setting it a little earlier in time, .5ms or so, should improve your high frequency response.

I like to get a measure pulse to use for playing with the calculate FFT time range settings. This way I can quickly see the effects of marker settings and choose their placement knowing a little bit more about the effects that they have. Seems to help in choosing my evils.

That said your low end roll off does look suspicious. Is your baffle centrally located in your room and away from objects that may be causing reflections?

As much as it can be. The room is pretty small so theres not much I can do to avoid reflections except reduce the gating time.

I get this no matter how long I set my gating time. Same result just much messier as gating time increases. Low frequency always starts to roll off. If I measure nearfield this does not occur.

Always another problem after one is solved... :banghead:

[Sdiver2489]

Don't forget, if your first goal is to verify the integrity of the tweeter/mic combo, an ultra nearfield is the way to go, say 1" from the dome. Turn down the levels of course, and be sure to enter the mic cal data in Speaker Workshop if you can, ECM8000's are soft above 15 kHz. A generic cal file (web search) for ECM8000 is better than no cal file.

~Jon

Ok, I got a generic calibration file that shows the same softness about 15KHz that you were talking about. I will use it in my next measurements. In the meantime, any idea why I am getting the low frequency rolloff on my gated measurements(does not appear in my nearfield measurements when measuring my speakers on an open baffle? (The same baffle they will be mounted on when completed.

After researching a bit on open baffles it appears that bass cancellation is not unusual for this type of mounting. So I would say I am not surprised I am getting the rolloff I am getting on a relatively small baffle(11x17). I suppose I should measure it as I have been and not worry about the low freq response as I am just concerned about the crossover at this point.

[JonMarsh]

The shorter your gating, the less low frequency data you get in an MLS meausrement. Inherent in the critter. Different systems display that short coming in different ways- some just using a soft interpolation (little data), some showing a graph fall off representative of the loss of data.

Plus the tweeter does roll off in the bottem end!

If you do an ultra nearfield measurement, the direct arrival swamps out the influence of the room reflections- START HERE!!! Then you can use a longer gating window, etc. You may find in a small room that the practical measuring distance you'll want to do is 18" or 24" instead of one meter. That may be a valid trade-off for you.

~Jon

[Sdiver2489]

The shorter your gating, the less low frequency data you get in an MLS meausrement. Inherent in the critter. Different systems display that short coming in different ways- some just using a soft interpolation (little data), some showing a graph fall off representative of the loss of data.

Plus the tweeter does roll off in the bottem end!

If you do an ultra nearfield measurement, the direct arrival swamps out the influence of the room reflections- START HERE!!! Then you can use a longer gating window, etc. You may find in a small room that the practical measuring distance you'll want to do is 18" or 24" instead of one meter. That may be a valid trade-off for you.

~Jon

I was referring to the woofer. Sorry I was not clear. I will try the ultra nearfield measurements for the tweeter and see how those turn out. Still gotta try them with the new calibration file anyway.

[Sdiver2489]

Also, JonMarsh, would you happen to know if the large terminal on a Hivi woofer is the positive or negative terminal? I have assumed it was positive but I suppose I shouldn't be making assumptions like that. I don't see any polarity marking on the woofer itself.

[JonMarsh]

large terminal is the positive- thank heavens for a few consistencies in this world...

[Sdiver2489]

large terminal is the positive- thank heavens for a few consistencies in this world...

I used this calibration file:

http://www.claudionegro.com/download.html

I can never seem to get a good summing of both my high pass and low pass filters. I think its because I have bad phase measurements as my woofer and tweeter responses seperately look decent...but when summed...everything falls apart.

(IMAGES REMOVED AND UPDATED BELOW)

[KJP]

Your measurements are looking pretty good.

Try reversing polarity on your tweeter in your crossover and then resum.

Some screenies of your crossovers could shed some light as well....

[Sdiver2489]

Your measurements are looking pretty good.

Try reversing polarity on your tweeter in your crossover and then resum.

Some screenies of your crossovers could shed some light as well....

I will try to get some screens...but the filters were 4th order linkwitz-riley on both sides(to start out)....I don't think I should have to reverse the polarity.

I actually did try that. When I did things just got worse I believe.

Maybe I will make another attempt at some measurements and make sure that phase difference is staying consistent.

[Dennis H]

Neither of your graphs looks like an LR4 rolloff and they don't look like they're tuned for the same XO frequency. Try rescaling them so your total dB range is 50dB or so and you'll be able to see what's going on better in the XO region. Both should be -6dB at the same Fc and roll off at 24dB/octave beyond there.

[KJP]

Nice catch Dennis.

Sdiver2489, try creating a target response and run the crossover optimizer.

[Sdiver2489]

Nice catch Dennis.

Sdiver2489, try creating a target response and run the crossover optimizer.

I did just that this morning and the results were initially roughly the same. I flipped the polarity on the tweeter and then they improved a great deal. I do not understand why I would have to do this as the crossovers placed and the goals targeted were all 4th order.

I updated my pictures to reflect the changes. I was in a bit of a hurry so I set all the levels equal manually. I have a little bit of trouble modeling my crossover and keeping the zobel networks effects on the impedance so I removed that for this modeling. I will have to figure out why later today.

Do a hard reload(hold shift and click reload) if the images don't make sense(I replaced them on my webspace do your browser could be using its cache)

Tweeter:



Woofer:



Total(Polarity reversed):



Total(Polarity not reversed):



Crossover:



I think I should take some acoustic measurements again to make sure that there wasn't a error that occurred in my phase measurements.

[KJP]

When you take those measurements, keep an eye on your Measurement.in for your reference channel to make sure that the pulse is arriving at the same time for both woofer and tweeter measurements. Phase data relevancy between drivers depends on this.

P.S. You could also post some of this at the SW forums...

[Sdiver2489]

When you take those measurements, keep an eye on your Measurement.in for your reference channel to make sure that the pulse is arriving at the same time for both woofer and tweeter measurements. Phase data relevancy between drivers depends on this.

P.S. You could also post some of this at the SW forums...

Shouldn't the pulse change slightly just because of the acoustic offset? I thought that is where the phase change originates from. The pulse response of each speaker are very close and only vary by maybe .1ms or so. I set the start gate to the earlier of the two and do the gated measurements with the gates constant.

P.S. This started as a measurement problem and has now progressed into more of a speaker workshop issue :-p. I will post over at audua.com as well. I have been making some posts there but due to the low traffic they don't receive much of a response.

[KJP]

The pulse recorded from your mic input changes from driver to driver, is why I said watch the reference input channel. It should not change.

[Sdiver2489]

The pulse recorded from your mic input changes from driver to driver, is why I said watch the reference input channel. It should not change.

I thought measurement.in was the same thing(just temporary) as the file that is stored(like the pulse response measurement.

So you are saying I should run a pulse response and the spike in the pulse response file may change but it shouldn't in the measurement.in file?

[KJP]

You have a Measurement.in.r and a Measurement.in.l
One of them is comes straight from the output of your sound card. This one should not move between measurements. The other travels through your device under test, be it a voice coil, resistor, or a speaker and microphone. This one can and will move.

[Sdiver2489]

I realize this is a very old thread...but I just want to say thank you for everyone that helped me out here. I apologize for how dense I was on some of these things. When I reread this I realize how confused I was at the time. I keep rereading these threads though and each time I understand more and more of what I was doing wrong. I am still working on my project and the construction is well underway. I am glad that I took the time to perfect my measurements before I moved on. I still need to improve my acoustic measurements but I am building the cabinet so I can get individual driver measurements which more accurately reflect the final design as opposed to just using a baffle.

Just for kicks I'll attach some of my final imdedance measurements(these are the free air measurements without any smoothing). They have come a long way from my first measurements(there is a thread where that was the subject). My goal is to get much better acoustic measurements now.

Anyway, I'll quit blabbering, but that you guys for all the help.