VIFA tweeter response

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
hi all,

I have done some frequency response measurements on my modified Vifa Vivace speaker setup. Now i am trying to interpret the VIFA tweeter XT25SC30-04 measurement results, but I am unable to come to justified conclusions.
I have done a "in house" (freq.response)measurement and a outside measurent. The results quite differ. The tweeter is flush mounted on a 12" or 30 cm bafle.
The lowest red graph is the "in house" measurement, blue and green are the outside. Measurements where taken on 1 meter and on .25 meter distance.
Could somebody please comment on the results?
 

Attachments

  • Tweeter reponse.png
    Tweeter reponse.png
    44.7 KB · Views: 419
  • xt25sc-30-04.pdf
    717.1 KB · Views: 42
Is your in house measurement gated? It looks as its not. Outside one also appears to have been taken slightly offaxis, you need to make sure your mic is properly centered. The bump at what appears to be 1.4Khz looks like a diffraction product.

What exactly bothers you and can you provide more information on how you measured them, what measurement rig you use and is it calibrated?
 

ICG

Disabled Account
Joined 2007
The rise in the lower frequency range is normal and caused by the baffle step. On the upper end it looks like you've not been exactly on axis with the mic because it got a similar drop off in the datasheet at 30°. I do not suspect you to be that much off of the angle but the short measuring distance isn't very forgiving on errors there at a driver which beams that much.

For the impulse response: It reflects the different measurement distance. But it looks like it was heavily smoothed, so there's not that much information you can read out of it. And there's another important error in it: It shows the response on the measurement BEFORE the impulse (that would be the 0ms mark). That can, per definition, not be the actual impulse response since the response can't happen before the impulse.

How does your setup look like? Do you have a DSP somewhere in the signal chain?

edit: too slow.. :rolleyes:
 
First of all, i use Holmimpulse. The testrig is E-MU 0202 soundcard and a calibrated Samson MM01 microphone. The testrig is fully functional and calibrated. After a good thought i realized that the inhouse measurements were taken with a logarithmic sine sweep and the outside measurements were taken with a square noise (improved MLS). The woofer and mid in and outside measurement were consistent, there were only minor differences.
Because I took the inhouse measurements more than six month ago, there might be an small issue with a on axis error.

@Maria Pankov
The outside measurements were gated and the microphone was (fairly) good centered. I would agree, the 1.5kHz bump looks indeed a diffraction product. The flange is absolutely not perfectly countersunk.
What bothers me is the quick and steep roll-off at fairly low frequencies, I would expect the tweeter to extend to a far greeter range.
Which of the measurements would you consider the most valid?
@ICG
The baflestep frequency is expected to be in the 400Hz range, and definitely not in the 1kHz range.
The preset recording delay of 1500ms causes the impulse to show-up before zero.
There is no DSP in the signal chain.
1/N (N=3) frequency smoothing was applied to the graphs.

Thanks for your comments!
 

Attachments

  • Tweeter resp.png
    Tweeter resp.png
    12 KB · Views: 360

ICG

Disabled Account
Joined 2007
@Maria Pankov
The outside measurements were gated and the microphone was (fairly) good centered. I would agree, the 1.5kHz bump looks indeed a diffraction product. The flange is absolutely not perfectly countersunk.
What bothers me is the quick and steep roll-off at fairly low frequencies, I would expect the tweeter to extend to a far greeter range.
Which of the measurements would you consider the most valid?
@ICG
The baflestep frequency is expected to be in the 400Hz range, and definitely not in the 1kHz range.
The preset recording delay of 1500ms causes the impulse to show-up before zero.
There is no DSP in the signal chain.
1/N (N=3) frequency smoothing was applied to the graphs.

I misinterpreted your post, I was thinking about measurement with and without baffle. The driver without baffle would lead to an increase level in the mentioned range.

Generally you can only compare exactly measurements which are made under the same circumstances and settings. Even the air pressure and temperature can change a lot (any sun shining on the mic?), if not any other thing changed, especally in the range the mic got the greatest tolerances. Did you put a foam cap on the mic outside? (Guess no) You still haven't told how you measured it. Height from the ground? Distance to the next reflective surfaces? Any surroundings, table, tripod, clamps etc? What else changed?

The E-MU 0202 actually IS a dsp, on the digital side EVERYTHING runs through it. So even slightly changes on the settings can actually change a lot.
 

ICG

Disabled Account
Joined 2007
The highpass function is indeed on and set to 1kHz. You propose to setup the measurement without the highpass function?
But I guess you mean without smoothing .

The reason to measure is to get informations of the driver and to compare it. If you constantly change the measuring circumstances, you can't compare the measurements. Why don't you etablish a standard measuring method? A hp of 1 kHz is too high.

What measuring amp for the driver do you use? (Don't use any class d amp)
 
@Mario Pankov

Which of the measurements would you consider the most valid?

None, Lojzek and ICG have already explained why. You must measure under the same conditions with no alterations to your setup and using the same settings for all drivers. The in house measurement should be gated. You can achieve close to anechoic measurements outside if you can raise the speaker some 2-2.5m above ground and there nothing in close proximity although with today`s technology this is barely required.
 

ICG

Disabled Account
Joined 2007
Just another thing about the impulse response: On all measurements the first amplitude is negative, that means you've switched the polarity somewhere in the chain. That though doesn't change the frequency response measurement. I'd suggest you re-check the complete measurement chain.

@Lojzek: That should be at least fs/2.
 
The outside situation:
The speaker was placed on a stand, 75cm above the ground.
Distance from the tweeter to ground level : 1.65 meters.
There were no reflective surfaces in the vicinity within a distance of 2,5 meters and on the microphone was a foam dust cap in place.
 
A couple of thoughts:

- For sweeps turn off the highpass altogether and set the start frequency of the sweep around 100 Hz. I do this all the time in HolmImpulse and never have had a problem.

- I don't see any gate markers in your first post.

- For your outside situation you shouldn't see any ground reflections for 8 msec if the mic is 1 meter from the speaker. In your impulse response you should be able to verify that no reflections occur until after 8 msec.
 

ICG

Disabled Account
Joined 2007
The outside situation:
The speaker was placed on a stand, 75cm above the ground.
Distance from the tweeter to ground level : 1.65 meters.

Ah, thank you.

There were no reflective surfaces in the vicinity within a distance of 2,5 meters and on the microphone was a foam dust cap in place.

On stone, concrete etc. ground you'd have to calculate with the 1,65m, which would be fine too in that range. On grass or sand, you can use the 2,5m excapt for the bass.

You were right to use the foam cap, you have to use it to eliminate wind noise. The microphone foam cap does change the frequency response of the microphone though and that might be one of the main reasons for different response measurements, it filters the treble (how much depending on the foam material) and it increases the 'baffle' of the microphone. You could do all measurements with the cap on to have comparable measurements, but that means you don't have a flat mic response because it differs from the calibration file. That's not ideal but no that bad either. Or you could measure the same speaker twice, one time with and one time without the cap and edit the calibration file for the differences - but you have to select the correct calibration file for each measurement then.
 
In my setup the foam cap did introduce shifts in the recorded frequency response and I threw it in the trash :D Another thing you may want to consider is proper positioning of the mic stand. Best is to put it as far away as possible and the mic to be mounted as on-axis to the holder (so to look like an extension to it) as possible. Cable should be twisted around the mic holder and stand. You can also add absorbing material around these, including the microphone (for indoors use) to reduce reflections from the pencil that holds the capsule.

You can do a somewhat accurate measurement outdoors by putting the speaker to lay on a large concrete/tarmac plane, and putting the mic some 2-2.5m away lying on axis with the speaker. It is an old method that was widely used in the past.
 

ICG

Disabled Account
Joined 2007
Cable should be twisted around the mic holder and stand.

No, don't do that. It puts stress on the cable and twisting the cable itself happens automatically. Instead use gaffa or, even better, cable clips.

You can also add absorbing material around these, including the microphone (for indoors use) to reduce reflections from the pencil that holds the capsule.

Don't do that! There's a very good reason the mic got that shape, it's to avoid building up pressure, the absorbing material does not help you in any way but the 'baffle' of the mic increases. It affects the linearity and the distortion of the mic.

You can do a somewhat accurate measurement outdoors by putting the speaker to lay on a large concrete/tarmac plane, and putting the mic some 2-2.5m away lying on axis with the speaker. It is an old method that was widely used in the past.

That's called a groundplane measurement. But you cannot lay the mic on axis, it got to be placed flat on the ground and that is a MUST! You get a higher level (+6dB) and it's not ideal for tweeter measurement because of the distance of the tweeter to the ground (it becomes 'mirrored' from the ground). That is especially a problem with wider speakers. A GP measurement is great for subs or bass, where the reflection of the ground would otherwise be a big issue, for other uses 4pi (free field) measurement is a lot better.
 
I think that ring-radiators like the Vifa XT25 in single (30) or dual magnet (60) versions are a great idea. I say this without ever having heard one. :eek:

What we know about them, is they are non-ferrofluid and low impedance, hence have some impedance issues around 600Hz, which is fixable with crossover design.

Dome tweeters, especially soft-domes, go to pieces above a certain frequency. The big 3" ATC mids show this in spades at 5kHz on the waterfall.

There are certain theoretical mathematical grounds for thinking spherical domes are dreadful once they go into breakup. Exemplified by Leonard Susskind in his string theory lectures. A ring-radiator is a toroidal solution. IMO, it works better. In essence it pins the centre of a spherical dome, thus eliminating the troublesome breakup.

All is geometry in audio. :D
 

Attachments

  • Vifa_XT25TG30-04.PNG
    Vifa_XT25TG30-04.PNG
    40.4 KB · Views: 126
  • Vifa_XT25BG60-04_Picture.PNG
    Vifa_XT25BG60-04_Picture.PNG
    118.4 KB · Views: 125
  • Jeff Bagby ATC.JPG
    Jeff Bagby ATC.JPG
    113.2 KB · Views: 106
  • Waves on a sphere Prof. L Susskind.JPG
    Waves on a sphere Prof. L Susskind.JPG
    33 KB · Views: 99
  • Gaussian Curvature in Geometry.jpg
    Gaussian Curvature in Geometry.jpg
    101.1 KB · Views: 105
There are certain theoretical mathematical grounds for thinking spherical domes are dreadful once they go into breakup. Exemplified by Leonard Susskind in his string theory lectures. A ring-radiator is a toroidal solution.
Please stop with that nonsense which you are proclaiming over and over. I have to quote myself answering you from other, earlier thread:
Not true. Please bear in mind that some of us in this forum actually are educated in physics and mathematical topology. In fact, string theory and Prof. Susskind lecture has nothing in common with this topic. You have to understand what torus compactification and Ricci-flat actually means, before claiming that toroidal solution is more optimal. Ring from XT-25, or more precisely annulus, topologically is equivalent to an open cylinder, not a torus! Or, if "ring tweeter diaphragm" is not punctured, than it is topologically equivalent to a dome or cone, not a torus!
 

ICG

Disabled Account
Joined 2007
Dome tweeters, especially soft-domes, go to pieces above a certain frequency. The big 3" ATC mids show this in spades at 5kHz on the waterfall.

I don't see a problem there - that's already outside of its usable range, regarding f-response aswell as beaming or useful upper crossover frequency. They are mid drivers after all.

There are certain theoretical mathematical grounds for thinking spherical domes are dreadful once they go into breakup. Exemplified by Leonard Susskind in his string theory lectures. A ring-radiator is a toroidal solution. IMO, it works better. In essence it pins the centre of a spherical dome, thus eliminating the troublesome breakup.

In essence, it's a coil moving two huge cloth surrounds. That description might not appear to be appealing but the result speaks for itself. IMO an excellent concept.

All is geometry in audio. :D

I couldn't agree more! Thank you!
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.