The higher the upper f3 the steeper the slope. IIRC the lower the group-delay distortion (at the upper end of the transmission band) the more "straight" the line appears. What is going on afterwards is determined by the lower cutoff frequency and the crossovers involved.
I.e. the whole frequency response is in fact included in the step- response but it it won't be easy to interpret it just by casual inspection.
Regards
Charles
I.e. the whole frequency response is in fact included in the step- response but it it won't be easy to interpret it just by casual inspection.
Regards
Charles
gedlee said:
In hindsight, I agree with this. The adaptive approach would only add value if this were needed to run in real time. With post processing, all the info is available in the two files to capture the impulse response directly, from the IFT.
Dr. Geddes,
How would "doppler" (phase modulation) distortion play into your dynamics testing? I expect it would not show up as a long term decay of response aberrations like the temperature of the motor, but only briefly as the driver's peak excursion is high or changing (decreasing) rapidly as you end the noise burst.
And the effect may, in reality, be small. But why not throw it into the mix since you're testing large signals and, presumably, large excursions?
Ref
Linkwitz: http://www.linkwitzlab.com/frontiers.htm#J
Elliott: http://sound.westhost.com/doppler.htm
Still learning,
Matt
How would "doppler" (phase modulation) distortion play into your dynamics testing? I expect it would not show up as a long term decay of response aberrations like the temperature of the motor, but only briefly as the driver's peak excursion is high or changing (decreasing) rapidly as you end the noise burst.
And the effect may, in reality, be small. But why not throw it into the mix since you're testing large signals and, presumably, large excursions?
Ref
Linkwitz: http://www.linkwitzlab.com/frontiers.htm#J
Elliott: http://sound.westhost.com/doppler.htm
Still learning,
Matt
DDF said:
Dave
Not to argue more, but I do think that the adaptive approach has more than just a real-time benefit. It can "update" the data, as you said before, without averaging and with a low number of DOF it would ignore the noise and focus on the system changes that we are inteested in. I really like this idea, its just that it is more work than the FFT approach that I used and I don't know if I have the time.
My FFT approach takes about 1-2 seconds of data. If I want to look at ms. changes I screwed.
Do you know MathCad? Doesn't it have built in algorithms for this? If so then I can impliment it without much trouble. But if I can't use MathCad then its a lot of work since everything that I do in signal processing is geared arround that program.
y8s said:
Matt - I would tend to put this into my category called "true, but irrelavent". Doppler distortion and phase modulation, along with amplitude modulation would all exist to some extent in any loudspeaker. But, only in a full range loudspeaker would they be significant and even there I would doubt that their audibility is of much concern.
Most people don't comprehend the fact that nonlinearity (as all of the above are) become bigger and bigger issues when as the loudspeakers bandwidth is increased. This is why I generally don't worry about modulation distortion in a subwoofer because its bandwidth is so limited. I have had some small concerns for my designs at 1 kHz because I use a 15" woofer over an unusually large bandwidth (about 80 Hz. - 1 kHz.) This means that the upper edge of the woofers passband could see some significant modulation by the woofer cone. But then this is exactly why we only make closed boxes. This design limits the cone excursions to relatively small excursions owing to the use of relatively small closed boxes. Trade-off is LF extention.
Far better than trying to extend the woofer response lower - and the resulting dramatic increase in excursion - is to just augment the system with subs - lots of distributed subs. Better bass, better mids, whats not to like?
Gentlemen - could you please move this non-thermaly related discussion to another post as it has nothing to do with the original topic - which I am still interested in.
What you say is that if a tweeter is crossed very low close to its limit, then thermal compression is likely to occur, no matter that measured non/linear distortion is low ?
How does a ribbon react to thermal compression ?
How does a ribbon react to thermal compression ?
Hi Earl
Thanks for the reply, I guess I will have to break out the mathcad book and see what else it can do or more correctly what can I figure out to do with it.
I mentioned the power compression thing I worked on in the 80’s, one test I did was to produce a band limited pink noise signal (keeping in mind this was for subwoofers) and used the TEF machine as a graphic level recorder by setting its window bandwidth to be larger than the range in question. With noise as the signal, the TEF would be set to do say a 15 Second sweep. As it “detected” all of the produced bandwidth, it recorded the average SPL over time. One simply set the noise drive level to “act like” some specific power level.
A test which might be useful is to use appropriately band limited pink noise which is increased in level. Against the increasing drive, the output spectrum is compared. This could be done rapidly, slowly and so on, to track either a pulse or sustain envelope..
At least for woofers it is easy to plug in a new value for Rdc and look at response and output, one can figure it has doubled its resistance by about 230C (well below rated power of modern drivers)
While not thermal as it relates to VC heating, I think there might be value in looking at the mechanical dynamic behavior. In particular, I would bet that drivers that depend highly on mechanical damping, also might show radiator behavior which is level dependent.
I know some of the goo’s that you can put on domes and cones in production are both temperature dependent and seem to have an envelope “history” effect where for some short time after a large envelope, its properties are different (less internal loss).
I would bet this is also a “thermal thing” but due to the absorption of energy within the damping.
Best,
Tom
There are a couple recordings of trains from the back yard at the new web site if interested. Use headphones first.
http://www.danleysoundlabs.com/technical downloads.html
Thanks for the reply, I guess I will have to break out the mathcad book and see what else it can do or more correctly what can I figure out to do with it.
I mentioned the power compression thing I worked on in the 80’s, one test I did was to produce a band limited pink noise signal (keeping in mind this was for subwoofers) and used the TEF machine as a graphic level recorder by setting its window bandwidth to be larger than the range in question. With noise as the signal, the TEF would be set to do say a 15 Second sweep. As it “detected” all of the produced bandwidth, it recorded the average SPL over time. One simply set the noise drive level to “act like” some specific power level.
A test which might be useful is to use appropriately band limited pink noise which is increased in level. Against the increasing drive, the output spectrum is compared. This could be done rapidly, slowly and so on, to track either a pulse or sustain envelope..
At least for woofers it is easy to plug in a new value for Rdc and look at response and output, one can figure it has doubled its resistance by about 230C (well below rated power of modern drivers)
While not thermal as it relates to VC heating, I think there might be value in looking at the mechanical dynamic behavior. In particular, I would bet that drivers that depend highly on mechanical damping, also might show radiator behavior which is level dependent.
I know some of the goo’s that you can put on domes and cones in production are both temperature dependent and seem to have an envelope “history” effect where for some short time after a large envelope, its properties are different (less internal loss).
I would bet this is also a “thermal thing” but due to the absorption of energy within the damping.
Best,
Tom
There are a couple recordings of trains from the back yard at the new web site if interested. Use headphones first.
http://www.danleysoundlabs.com/technical downloads.html
Tom
Look into Readwav and Writewav functions in MathCAD. Read in a wav file and then play arround at looking at the data values etc. Before long it will become obvious how to manipulate these files.
All of our psychoacoustical work is done with wav files manipulated with Mathcad. This allows for very precise changes to be made which can then be auditioned over headphones comparing the changed and unchaged wav files. I have a canned program that will track the users responses and calculate the stats.
SPEAK has always had the ability to look at changes in an enclosure tuning with temperature. It changes the magnet flux, VC Re, etc.
Once I have a test that I like, that is reasonable, and shows differences in systems, I will generate some wav files and do a test for audibility. This will then allow me to scale the actual audibility of the thermal changes that is correlated with an objective test. A very useful tool I am hoping.
Look into Readwav and Writewav functions in MathCAD. Read in a wav file and then play arround at looking at the data values etc. Before long it will become obvious how to manipulate these files.
All of our psychoacoustical work is done with wav files manipulated with Mathcad. This allows for very precise changes to be made which can then be auditioned over headphones comparing the changed and unchaged wav files. I have a canned program that will track the users responses and calculate the stats.
SPEAK has always had the ability to look at changes in an enclosure tuning with temperature. It changes the magnet flux, VC Re, etc.
Once I have a test that I like, that is reasonable, and shows differences in systems, I will generate some wav files and do a test for audibility. This will then allow me to scale the actual audibility of the thermal changes that is correlated with an objective test. A very useful tool I am hoping.
gedlee said:
First I would like to point out that I am in disagreement with the write up on Doppler distortion at Rod Elliot's site (assuming he hasn’t changed it since it was first posted). My response to his comments can be found here
http://www.geocities.com/kreskovs/Doppler1.html
where I show that the maximum frequency shift does indeed occur at the point of max cone velocity.
Second, re nonlinear distortion, Martin Colloms has a good discussion of this in Highperformance Loudspeakers, 5th Edition with numerous references. Test conducted by Moir showed that HD at low frequency isn’t very audible. Above 400 Hz it is suggested that levels should be maintained below 1%. The IM distortion threshold for untrained listeners was in the 4% to 5% while for experienced listeners it dropper to 2% to 4%. On the other hand, Doppler distortion was found to be quite noticeable with a pure tone modulated by a 20 Hz signal. using music (Sorry I don't know the details of the test speaker configuration) level of 0.1% to 1% were reported to be detectable by experience listeners. The obvious suggestion is to use big woofers to reduce excursion, the Doppler distortion. I think much of this re-enforced Earl’s position that for the most part a most nonlinear distortion is not really an issue in a well designed speaker.
As another point, more in line with this thread, with dynamics it would seem that for a give system the issues are play back level and duration. So why wouldn't it be possible to just use and MLS pulse. Say the sampling rate is 50k. For a give MSL signal the duration and level can be independently controlled. Level directly by signal level and duration by the length of the sample taken.
Long play back periods could be investigated by first playing music at the desired play back level to bring the system up to temperature before testing. I haven’t read all the posts and I am just thinking freely out loud.
Sorry that we did not know you just wanted to limit the subject to thermal induced effects on dynamics.gedlee said:
When thermal comppression testing is done using white or pink noise one needs to also measure the amp response and subtact it from the measured speaker response to make sure amp compression is taken out of the factor.
Hi John
“As another point, more in line with this thread, with dynamics it would seem that for a give system the issues are play back level and duration. So why wouldn't it be possible to just use and MLS pulse. Say the sampling rate is 50k. For a give MSL signal the duration and level can be independently controlled. Level directly by signal level and duration by the length of the sample taken.
Long play back periods could be investigated by first playing music at the desired play back level to bring the system up to temperature before testing. I haven’t read all the posts and I am just thinking freely out loud.”
I am not that familiar with MLS but could you send a long series of signals, which would be displayed like a waterfall, to show any changes in frequency response starting from time zero to the end.? In other words say a continuous frequency response waterfall of 2 – 5 seconds of signal.
For a woofer, there are several different time constants involved with the VC one being a “short” time constant (seconds) all the way to the magnet /frame heating after a many many minutes. I would think generally for proportionally smaller motor systems, the times would be all that much shorter.
What ever one looked at, it is probably useful to start looking in the time scale of musical crescendo’s or short envelopes.
Perhaps someone skilled in the art of computers could correlate and multiply a source signal amplitude envelope with the resultant signal amplitude envelope and via the product having the sum and difference component and in that difference component at least find a low frequency signal corresponding to the amplitude modulation element present in the result and not the source.
I suppose the source signal could be a tone burst or band limited pink noise in an amplitude envelope etc.
Anyway, interesting topic and comparing amplitude modulated envelopes may be useful..
Best,
Tom Danley
I am with you on “Big woofers”, although portable, this one was too big for my home haha. Check the construction video.
http://www.danleysoundlabs.com/matterhorn.htm
“As another point, more in line with this thread, with dynamics it would seem that for a give system the issues are play back level and duration. So why wouldn't it be possible to just use and MLS pulse. Say the sampling rate is 50k. For a give MSL signal the duration and level can be independently controlled. Level directly by signal level and duration by the length of the sample taken.
Long play back periods could be investigated by first playing music at the desired play back level to bring the system up to temperature before testing. I haven’t read all the posts and I am just thinking freely out loud.”
I am not that familiar with MLS but could you send a long series of signals, which would be displayed like a waterfall, to show any changes in frequency response starting from time zero to the end.? In other words say a continuous frequency response waterfall of 2 – 5 seconds of signal.
For a woofer, there are several different time constants involved with the VC one being a “short” time constant (seconds) all the way to the magnet /frame heating after a many many minutes. I would think generally for proportionally smaller motor systems, the times would be all that much shorter.
What ever one looked at, it is probably useful to start looking in the time scale of musical crescendo’s or short envelopes.
Perhaps someone skilled in the art of computers could correlate and multiply a source signal amplitude envelope with the resultant signal amplitude envelope and via the product having the sum and difference component and in that difference component at least find a low frequency signal corresponding to the amplitude modulation element present in the result and not the source.
I suppose the source signal could be a tone burst or band limited pink noise in an amplitude envelope etc.
Anyway, interesting topic and comparing amplitude modulated envelopes may be useful..
Best,
Tom Danley
I am with you on “Big woofers”, although portable, this one was too big for my home haha. Check the construction video.
http://www.danleysoundlabs.com/matterhorn.htm
- Status
- Not open for further replies.