Base recommendations regards smoothing

[charlie2]

What is acceptable smoothing in relation to frequency response when measuring ie 1/48 down to 1/3 etc

Been doing quite a bit of measuring lately with positive and dire results in the 1/24 smoothing range (compression drivers and assortment of horns)

So what is classed as good at what frequency range and dB +/- scaling

Flat response ideally before any crossovers or eq applied

I am only altering start frequency so far regards measuring

Just looked at one speaker on line and was showing +/- 2.5db (1/3 smoothinig) 1k-20k @0.75 meters

 

[EarlK]

I use 1/6 octave smoothing when designing networks ( if that's of any help to you ).

 

[ropf]

For speaker development no smoothing at all, instead windowing out the room reflections.

 

[eriksquires]

Hi Charlie:

In English, you mean "bass."

So you'll never get in-room flat response except in an anechoic testing chamber.

So, let's back up a bit, and ask WHY. Are you trying to design a crossover? Active or passive? EQ?

This all affects your choices in measurements.

Best,

E

 

[charlie2]

Hi Charlie:

In English, you mean "bass."

So you'll never get in-room flat response except in an anechoic testing chamber.

So, let's back up a bit, and ask WHY. Are you trying to design a crossover? Active or passive? EQ?

This all affects your choices in measurements.

Best,

E

Eric no not bass as in thud thud but a base starting point reference minimum datum data to work from.

i.e Graphs vs 50db 40dB 10db 5db 2db increment scaling
i.e graph smoothing 1/48 vs 1/24 down to 1/1

They can polish a turd

I am building my own horns nothing fancy just cardboard n glue and measuring assessing them and learning the effects hope this makes sense

 

[eriksquires]

Hi Charlie,

If you are trying to measure for speaker design, 1/6 is detailed enough for a woofer but, the key is in the environment. You need to remove the room in order to get repeatable, useful data. There's a number of ways. One is to use mixed far/near field (quasi-anechoic). Another is outdoor measurements, a third is ground plane measurements.

Measuring in-room will produce wildly variable results depending on exactly where your microphone and where your speakers are. For this reason this is avoided until you are ready to place, and perhaps deal with EQ and bass traps.

Measuring bass response with more than 1/6 or 1/12 octave is pretty much useless. If you measure in-room you are just seeing noise and room artifacts which will drive you nuts.

On the other hand, with proper near field or ground-plane measurements, the speakers look much smoother, and there's not much more data to be gained. Your 1/24th graph looks a lot like your 1/6th graph.

So, I would say, pick whatever resolution you like, but your measurement technique is going to matter a lot more.

In the mid-treble with gated measurements, high resolution becomes more useful and essential.

 

[gedlee]

Eric no not bass as in thud thud but a base starting point reference minimum datum data to work from.

i.e Graphs vs 50db 40dB 10db 5db 2db increment scaling
i.e graph smoothing 1/48 vs 1/24 down to 1/1

They can polish a turd

I am building my own horns nothing fancy just cardboard n glue and measuring assessing them and learning the effects hope this makes sense

I use a sliding scale for the bandwidth averaging that goes from about 1/3 octave at LFs to about 1/12 octave at HFs. If you can't do this (many measurement platform can) then stick to the 1/12. To me its the dB scale that matters, I want to see 2 dB resolution and no more than a 40 dB total scale. But to really nail things down you need to look at the total polar response to +- 90 degrees horizontal in high resolution of about 2 degrees, 5 at the most.

Attached is an example of a very high resolution measurement which shows pretty much all you need to know - 2 dB contours near zero dB and 2 degree polar resolution.

 

[charlie2]

Thanks Earl

this is the point I was getting at, and you have provided a very good and detailed response and your personal reference points to my enquiry cheers