I've done it similarly using the free version of Praxis for splicing the NF+diffraction that allows you to add delay while doing so. Most of the time now I import a spliced response into SoundEasy ignoring phase of the NF part (since I still prefer LAUD for measurement), then match HBT phase of the measured quasi-anechoic part and use the final HBT with the previously set excess-delay. You could do that with CALSOD as well. Several ways to skin a cat.
Dave
It does require a different approach, because you have to consider both the front and the back side responses, but you don't have the to include a diffraction signature. John K described a method that works reasonably well, but it's slightly different depending on whether you're measuring a pure open baffle or some form of framed dipole, such as H-frame or U-frame.
For a pure open baffle I used the method at John's page. Since this is a dipole with a thin baffle, the combined, inverted response with the dipole moment added is a reasonable assumption as long as the rear opening of the driver is not restricted.
For the framed low frequency drivers, the mic is placed at a point centered on the opening (directly in front of one driver, mid-point of two) on the plane of the front opening, measure both front and rear, then combined, including the dipole moment. Since this is for very low frequency, you have a limitation due to cavity resonances, but that can be taken into account. John provides a good Excel spreadsheet for this design consideration. It was invaluable to me when doing my current design.
Dave
But you have to assume a known dipole moment right? This is a source of error - first it may not be obvious and second, it may change with frequency. But, sure, its better than assuming a monopole. I just don't like "assumptions".
Sure-fire way to screw things up is to assume things. I.E. listening tests - "I know what I hear". Thats a valid asumption right?
Sure-fire way to screw things up is to assume things. I.E. listening tests - "I know what I hear". Thats a valid asumption right?
Yes, you have to assume the dipole moment, but for a circular dipole source such as a dynamic driver, in practice taking the center of the driver works reasonably well, especially at low frequencies. At the higher frequencies where the wavelengths might be an issue, it's in the area above the dipole step. Taking John's recommendation of crossing below the dipole peak helps to limit this influence.
I'm not arguing that it's necessarily better than your method, but then I don't have access to any software that implements your method. That method might also run into similar issues when applying it to dipoles with dynamic drivers, since the motor structure will complicate the rear response.
I'm not one to make blanket statements such as "I know what I hear". I use measurements to the limits of my measurement capabilities as much as possible. At times that still requires some adjustment from what I hear, not that this is proof of anything. I don't expect to change that, I simply try to find the reason using measurements. I'm just limited by doing this DIY in my own room. If you find fault with that, so be it.
Dave
Dave
I was not criticizing you, it was a generic statment of what most audiophiles are willing to assume to get answers.
Linking subjective and objective is the key. If you can do that then things are fine. It's when "listening" is omnipotent and has no connection to measurements that things have failed. I have screwed up both sides of this equation, the listening and the measaurement, but never, to my knowledge, screwed up both at the same time in precisely the same way. Thats the key. Failure of either is likely, but there are an infinite number of ways to fail. For both the subjective and objective to fail and still agree will thus be extremely unlikely. But when both agree, then the confidence level is justifiably very high.
Sorry - just more of my rants about "listening tests" - back to the point.
In my procedure I do need to assume the radius of the hypothetical sphere. SInce it is hypothetical the assumed radius can't actually be wrong, and it is easy to find the center of rotaion of the table. But what can be wrong is the assumed center of this sphere. I need to work out what this error means to the data. For the far field stuff it is obviously benign due to the distances involved, but not so for the near field.
I was not criticizing you, it was a generic statment of what most audiophiles are willing to assume to get answers.
Linking subjective and objective is the key. If you can do that then things are fine. It's when "listening" is omnipotent and has no connection to measurements that things have failed. I have screwed up both sides of this equation, the listening and the measaurement, but never, to my knowledge, screwed up both at the same time in precisely the same way. Thats the key. Failure of either is likely, but there are an infinite number of ways to fail. For both the subjective and objective to fail and still agree will thus be extremely unlikely. But when both agree, then the confidence level is justifiably very high.
Sorry - just more of my rants about "listening tests" - back to the point.
In my procedure I do need to assume the radius of the hypothetical sphere. SInce it is hypothetical the assumed radius can't actually be wrong, and it is easy to find the center of rotaion of the table. But what can be wrong is the assumed center of this sphere. I need to work out what this error means to the data. For the far field stuff it is obviously benign due to the distances involved, but not so for the near field.
Hi Earl,
It is not necessary to know the dipole moment with my method. What is necessary is that the far field measurement extend into the region where SLP decreases by 6dB/octave. When you make the near field measurement you can pretty much assume any dipole moment as long as it places the dipole peak sufficiently above thelow frequency extent of the far filed data. When you over lay the dipole corrected near field data with the far field measurement and scale the amplitude of the near field data to match the far field it is pretty apparent how to combine them. There is usually a fairly wide over lay region where both responses are similar indicating dipole behavior.
It is not necessary to know the dipole moment with my method. What is necessary is that the far field measurement extend into the region where SLP decreases by 6dB/octave. When you make the near field measurement you can pretty much assume any dipole moment as long as it places the dipole peak sufficiently above thelow frequency extent of the far filed data. When you over lay the dipole corrected near field data with the far field measurement and scale the amplitude of the near field data to match the far field it is pretty apparent how to combine them. There is usually a fairly wide over lay region where both responses are similar indicating dipole behavior.
Oops, I was mixing the modelling with the measuring. Thanks for the correction. I've got to be more careful.
Dave
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.