I am still (and will most probably always be) a newbie in loudspeaker measurements. Since I am limited by the climate conditions to be able to measure outside to get rid of the reflections. Therefore this is what I do during the cold period to design crossovers for my (mostly) MEH projects. These are mostly symmetrical in horizontal and vertical plane, the round ones even in more planes at certain frequency bands. I tend to think that the symmetry allows getting better results with the method below than for e.g. a standard two way bookshelf speaker. Maybe it can be also considered a point source as a simplification in certain frequency areas?
I measure the bands and the whole speaker on axis as far as possible without seeing a strong reflection in the measurements. In general, this is always too close for serious measurements - I end up usually somewhere between 30 - 50 cm. I use the traces or live measurement (slightly smoothed for better view) to set the crossovers. The measured FRs even this close are usually in a good agreement with manufacturer datasheets and e.g. Hornresp models. Then I move the mic to the listening position, make a few measurements across the area to get an idea what is the speaker response and what is the influence of the room. Some further EQ may be added carefully to balance the averaged listening spot response.
This gives me usually quite listenable results. I would consider this method to be "better than nothing". Would it make sense to measure "polar response" in nearfield at the place of installation or would such a measurement be totally useless and way off from the proper anechoic measurement? From my experiments, especially if a model (like from ATH/ABEC) is available, I think I can see general trends in the measurements that follow the simulation results. All of that with heavy "brain" filtering and using various gating/smoothing settings on the measured data. For example, the midrange narrowing on my large 9 sided conical horn showed up at the frequencies, where it appears in the model.
My target is to identify areas where EQ is applicable and useful and areas, where EQ is contra productive and set up a DSP crossover. I wonder what are the limitations of such approach and where are the risks of being mislead by wrong assumptions. Any input on this topic is highly welcome.
I measure the bands and the whole speaker on axis as far as possible without seeing a strong reflection in the measurements. In general, this is always too close for serious measurements - I end up usually somewhere between 30 - 50 cm. I use the traces or live measurement (slightly smoothed for better view) to set the crossovers. The measured FRs even this close are usually in a good agreement with manufacturer datasheets and e.g. Hornresp models. Then I move the mic to the listening position, make a few measurements across the area to get an idea what is the speaker response and what is the influence of the room. Some further EQ may be added carefully to balance the averaged listening spot response.
This gives me usually quite listenable results. I would consider this method to be "better than nothing". Would it make sense to measure "polar response" in nearfield at the place of installation or would such a measurement be totally useless and way off from the proper anechoic measurement? From my experiments, especially if a model (like from ATH/ABEC) is available, I think I can see general trends in the measurements that follow the simulation results. All of that with heavy "brain" filtering and using various gating/smoothing settings on the measured data. For example, the midrange narrowing on my large 9 sided conical horn showed up at the frequencies, where it appears in the model.
My target is to identify areas where EQ is applicable and useful and areas, where EQ is contra productive and set up a DSP crossover. I wonder what are the limitations of such approach and where are the risks of being mislead by wrong assumptions. Any input on this topic is highly welcome.
Ok - that is clear. What if I wanted to measure within the coverage of the horn - does a close (30 - 50 cm) measurement provide some sort of valid information within that area? Let us assume that I am not interested for the pattern outside. For a dual flare MEH, this area would be defined as the angles of the primary flare - would the near measurement correlate with a proper anechoic measurement at least down to a certain frequency?
The minimum distance depends on the size of the driver or speaker. I think the rule of thumb is something like at least 2 to 3 times the diameter of the baffle is the minimum distance for measurements. For example, I measure a nude 6.5" driver at about 20" (0.5m). So measuring a MEH at only 30cm (12") is likely too close, assuming the mouth is at least that wide.
You can eliminate one boundary reflection by doing a groundplane measurement. Try to find some indoor space with a hard, flat surface (garage floor?) if you cannot do it outdoors.
For another take on the minimum measurement distance with some supporting info, see the section on measurement distance in this article, and especially Figure 2:
https://audioxpress.com/article/measurements-for-loudspeaker-modeling-files
In the article they state:
Last edited: