problems with acoustic measurements

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
I've been doing acoustic measurements of my prototype MTM speaker and have been having some problems.

I'm using speaker workshop and a Wallin II preamp with a panasonic WM60AY capsule (not linkwitz moddified).

I just discovered one what one of the problems is. the phase plots I have been getting seemed wrong compared to a couple that I had (where I must have accidentaly reversed the polarity). I've attached the two plots below.

the top one is with the speaker wired with correct polarity tested with 1.5V cell (when connected +ve battery terminal to +ve speaker get an outward movement of the cone).

the bottom one is with the speaker wired with reverse polarity.....

so:

1. Is the bottom graph how my phase should look??
2. Is it likely that my amp or mic preamp is reversing the phase??
3. should I connect my speakers up with the polarity reversed at the amp terminals?


I have another question regarding the freq response. but I'll leave that for a while so as not to confuse the issue :)

Tony.
 

Attachments

  • phase_compare.gif
    phase_compare.gif
    63.5 KB · Views: 698
The phase plot doesn't really become important until you start measuring multiple drivers and attempt to concoct a crossover for them. Your nearfield phase measurement is just showing you that if you reverse the polarity of the signal to the driver the phase will shift...theoretically 180 degrees. Nearfield response is usefull for looking at how a driver behaves in an enclosure at lower frequencies, or for examining frequency response aberations the driver may have, but you won't be using it to develop a crossover, so phase isn't important at this point.
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
Hi Jbateman,

Probably not important right now, but I will in future be crossing (actively) over to stereo subs (and was planning to do it at about 300Hz).

What concerned me with these plots was the phase doing a reversal at exactly that point (ie 300Hz) when the driver is wired up with correct polarity.... I haven't built new cabinets for the subs yet so haven't done any measurements on them, they are currently the bass drivers in my existing 3 Ways :), but because of the low freq's involved I figured that I could only use near field (or possibly ground plane) measurements for this part of the project. Sorry I didn't include that detail :) I moved this from the other thread I had going because I figured it wasn't on the topic of that thread, and I guess I forgot to put in some details :angel:

So in that context would a phase reversal at 300Hz be an issue?? (my feeling is that it is a big one) and if it is should I be running the midbass and subs in reverse polarity?? I guess this question is a little pointless without measurements from the subs.....

I also have a concern about the phase on my gated measurements... it just doesn't look right, its inverting all the time and is very jagged, like a saw tooth......

Attached is an example. I'm not really happy with this. The resolution is too low due to not being able to get better than about 7ms gate time. I will try and do some outside measurements on a still day, but this is what I can get at the moment.

The measurements were taken at a distance of 1M with mic 1.25M off the floor (1.5M to ceiling), egg crate foam on the floor, and no other nearby walls or objects. Gate time was set at 6.8ms (much higher than floor or ceiling bounce would occur at but based on the pulse response I thought it would be ok. Changing to a gate time of 3.5ms just made the resolution woeful.

This is a measurement of both of the MW144's mounted running in parallel full range..... hmmm writing that statement, I just wondered whether or not the phase data is all over the shop because there are two drivers involved..... the measurement is taken midway (height wise) between the two drivers. maybe I should measure just one :smash:

Tony.
 

Attachments

  • mw144_gated.gif
    mw144_gated.gif
    46.6 KB · Views: 600

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
tcpip said:
I'm reading your thread very carefully, waiting for answers from others. This is one bridge I'll have to cross very soon now... I'll be doing SPL measurements for the first time in my life sometime in early September. :D And I too have Speaker Workshop and a Panasonic capsule, unmodified. :confused:

Hopefully we will both get something out of it :)..... I'm getting semi usefull results, but I'm not confident I can use them for crossover design, I have a glitch or two somewhere I'm sure!

I've successfully used SW for getting T/S params in the past, and I've played with accoustic measurements too and come to some usefull conclusions from what I have measured, but this is my first time designing a crossover, so I want to make sure I'm not working with completely bogus data ;)

I found with the mic capsule that the modification actually made things worse. Unless you are planning to test at high SPL's I wouldn't recommend doing the mod. It will reduce the gain of the mic considerably. for measurements at 1M you will require excessively high SPL's to overload an unmodified mic (If you don't have hearing protection on I think you will have damaged hearing ;) I can't see any reason for doing the mic mod personally unless you want to do high level nearfield testing.

Tony.
 
Probably not important right now, but I will in future be crossing (actively) over to stereo subs (and was planning to do it at about 300Hz).

What concerned me with these plots was the phase doing a reversal at exactly that point (ie 300Hz) when the driver is wired up with correct polarity.... I haven't built new cabinets for the subs yet so haven't done any measurements on them, they are currently the bass drivers in my existing 3 Ways , but because of the low freq's involved I figured that I could only use near field (or possibly ground plane) measurements for this part of the project. Sorry I didn't include that detail I moved this from the other thread I had going because I figured it wasn't on the topic of that thread, and I guess I forgot to put in some details

So in that context would a phase reversal at 300Hz be an issue?? (my feeling is that it is a big one) and if it is should I be running the midbass and subs in reverse polarity?? I guess this question is a little pointless without measurements from the subs.....

I also have a concern about the phase on my gated measurements... it just doesn't look right, its inverting all the time and is very jagged, like a saw tooth......

Just so you know it's is not actually "inverting". When the phase goes above 180 degrees it will look like it inverted, but it is simply going to 181 degrees. When you look at it try and mentally unwrap it. Hope that helps.
Joe
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
Thanks Joe :) that makes much more sense! I guess I've seen lots of stuff saying that the phase shouldn't be changing rapidly at the crossover point and thought it was these (seemingly odd) inversions.... thinking of phase as rotating through 360deg and eventually working its way back round to zero makes much more sense!

I still don't actually understand why the phase changes (non-linear reproduction of the sound by the driver??), but that certainly clears up one incorrect interpretation that I had :)

Tony.
 
You are right that it would be difficult to make an ON AXIS measurement at 300HZ. On most 3 way speakers with a 2nd order xover around 300Hz, the midrange is connected in reverse polarity. If you were to try to determine this by making nearfield measurements you wouldn't be succesful, as the mic is in a different place for each driver...therefore the phase data is not usefull.

Most of the time people will connect their sub polarity based on listening preference...set it for whichever sounds best.

As for the "zig zag" phase plot, this is due to the long path the signal must travel from driver to mic. Some call this the phase rotation from the Time of Flight...or EXCESS DELAY. You can remove this delay by using your pulse graph. In the example picture below, I show the original phase plot, then my pulse which starts at 8msec. Next you would hilight the original chart, select TRANSFORM/DELAY, then enter the estimated time of flight. In my case I enterred 7.4msec, which produced the bottom chart. NOTE: Be sure to UNCHECK the REMOVE EXCESS DELAY box when you do this.

[IMGDEAD]http://members.verizon.net/~pecker/midonaxis.gif[/IMGDEAD]
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
Thanks jbateman :D that little peice of information was invaluable and makes me feel a lot better about my measurements :) I now have a normal looking phase plot for the midbass measurement!

I guess with the subs I'll suck it and see, but I think that's a little way off yet :)

I'd looked at page 122 of the unoficial speaker workshop manual titled excess delay, and had stopped reading when it said it is not recommended to do this..... I hadn't read the next bit that said to untick the excess delay box, and clean up the phase of the individual driver!!! might dig up the sw manual two thread and add this in :)

I've noticed one thing though (questions questions)... in my pulse measurement the pulse starts at 3.1ms I tried putting that in and got rubbish, then I tried 3.0ms and still got rubbish, 2.9ms seems to give an ok result, if I change it to 2.8 it's also ok but different to 2.9..... top graph with delay set to 2.9ms bottom one set to 3.0ms.... I'm guessing going with the result closest (but less than the pulse start time) that gives a decent result is the way to go.

Tony.
 

Attachments

  • mw144_phase_fixed.gif
    mw144_phase_fixed.gif
    40.9 KB · Views: 522
I don't think it's too important the exact amount of delay you remove. The goal is to remove the same amount from your midwoofer and tweeter measurements, so that you can more easily compare the phase in the crossover region (when you develop the crossover)

In my example above, the little glitch in the impulse graph at about a half msec is some leakage between the 2 channels on my soundcard when the pulse is generated. So I subtract that from 8msec when the pulse arrives to get my time of flight.
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
That makes sense :) compare apples with apples, and it doesn't matter wether the phase is the actual phase or not provided both are the same relative to each other. It's just the relationship between the two drivers phases at a particular freq that we are interested in.

I'm glad I started this thread :)

Tony.
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
another glitch

OK here is another thing that has been bugging me.....

I got some really nice flat nearfield measurements of the drivers a while back. This was after re-soldering the connection to the mic capsule, they seemed to correspond nicely to the modeled response.

Then all of a sudden the curves weren't as flat any more, nothing changed in the test setup (The only thing I can think of is that there was some strain on the mic cable which possibly compromised the soldered joints), just the results which now have A sort of a ragged hump between 100 and 300Hz... no amount of resoldering or changing mic capsules seems to bring back the nice flat response (the green curve in the attached pic).

I'd had a similar though more exagerated repsonse curve problem when I joined two peices of my mic cable (which is really only shielded audio cable). At the time I put it down to stray capacitance because of the join.... I tried two mic capsules and got the same very ragged response (I'm talking about 6db dips and humps) all because of a join in the cable (red curve on the attached) and it wasn't due to dud soldering either..... I basically joined so I could test different capsules without desoldering and resoldering to the capsule itself... needless to say I scrapped that idea!!!!

So I'm thinking maybe the problem is cable related (it's possibly been stretched as the wand came off its mount on the tripod and the weight of the wand was pulling on the cable (1M length of 6mm brass tube).... any clues??? I'm thinking I'll go and buy some 4.5mm diameter mic cable to use instead of my 2+ M lenght of 2.5mm screened audio cable.....

in the picture green is good but no longer achievable measurement. blue is what I get now, red was with join in the cable!

Tony.
 

Attachments

  • mw144_nearfield_comparison.gif
    mw144_nearfield_comparison.gif
    31.4 KB · Views: 293

jomor

Member
2004-02-20 1:36 pm
Athens
About phase, you can put the first marker just before the main pulse instead of 0ms, this will clear the phase plot in a degree.

When it comes (and it will come soon) to crossover modeling, Speaker Workshop doesnt care about how the phase looks, but it cares MUCH about the following:

1) All pulse measurements of all drivers should have the same latency ( = the peak at Measurement.In.L file under the System Folder should happen at EXACTLY THE SAME TIME). This will ensure that
the phase data of each driver measurement is usable for crossover modeling.

2) The microphone and the speaker positions should be exactly the same for all drivers measurements.

3) The markers should be at exactly the same positions for all FFTs you make at all drivers (or for all Frequency measurements you make for all drivers - use the tweeter's pulse to set the markers, which is the closest driver to the mic) :)


About the near field problem I m not sure what causes the ripple, but it would be interesting to find out, sometimes a silly problem takes all the time :mad:
 
About phase, you can even not subtract the excess delay to have a better view of the drivers phase comparison: just zoom in the chart in the possible crossover region.

The main problem, for xover simulation, is too be sure that the impulse always starts at the same point, when repeating the measurement: in my HP there is a folder dedicated to this problem.

Regards.

Claudio
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
Thanks jomor :) I'll try resetting my minimum gate time before my next measurement and see how it looks... I'm not 100% sure what you mean with regards to the system measurement.in.l I thought that it was a calibration file... mine has a latency of around 0.17ms or I guess 170 micro seconds...

had two under control, but three is very helpfull :) I probably would have done separate pulse measurements for the tweeter and midbass, do I just copy and rename the tweeter pulse measurement to the name that matches the midbass?

SY yes there is another driver in reasonably close proximity (I'm running the midbass's in parallel so thought I should measure with both running)... I suspected this and tried the measurement without the second one running, but it wasn't the culprit. I think you may be onto something with the brass tube vibrating though! I just realised that I was originally holding the mic on the end of the wand with heat shrink, but after changing a few capsules I got sick of cutting it of and started using electrical tape.... could be the heat shrink was damping the tube a bit... I'll try a bit of polyester fill or some foam and maybe put the heat shrink back on :)

Thanks Claudio I'll go check out your site again, it's been a while since I looked there (basically had forgotten :rolleyes: ), and I hadn't been back since starting these types of measurements

but right now I need to get to bed ;) too many late nights are catching up!!!

Tony.

attached is a pic of my measurement.l.in
 

Attachments

  • measurement_in.l.gif
    measurement_in.l.gif
    10.4 KB · Views: 230

jomor

Member
2004-02-20 1:36 pm
Athens
wintermute said:
Thanks jomor :) I'll try resetting my minimum gate time before my next measurement and see how it looks... I'm not 100% sure what you mean with regards to the systme measurement.in.l I thought that it was a calibration file... mine has a latency of around 0.17ms or I guess 170 micro seconds...

attached is a pic of my measurement.l.in

This is important to understand, Latency setting at Properties is one thing, checking the Measurement.In.L is another.

When you are about to measure a woofer and a tweeter and build a crossover for them, you should make sure you have accurate phase data. This means (among the other things mentioned before) that the time each pulse that will be sent to the driver occurs, is exactly the same for all drivers. This is checked via the peak's time at Measurement.In.L file. You should check the peak's time each time you do an acoustic measurement to be the same.