DIY Audio Projects Forum
http://www.diyaudioprojects.com/Forum/

How To Interpret Speaker Frequency Response Graph
http://www.diyaudioprojects.com/Forum/viewtopic.php?f=11&t=6731
Page 1 of 1

Author:  Joegtech [ 24 Feb 2020, 18:40 ]
Post subject:  How To Interpret Speaker Frequency Response Graph

Reputable mic and speaker manufacturers provide a frequency response graph for their products. For those not familiar with them, the data is typically plotted in log graphs, not liner graphs. Refer to the attached graph of my EV ND357 mic below

However most manufacturers do not display the entire top octave--10khz to 20khz. They instead provide an abbreviated view of the top octave, also seen below.

In contrast Alesis provides a rare graph with the entire top octave present.
http://kb.inmusicbrands.com/media/images/FREQ3.jpg
https://www.alesis.com/kb/article/2227

Can anyone help me to interpret the abbreviated version of the top octave in the attached graph?

My best guess is that the vertical line between 10k and 20khz represents 15khz but I've not found any confirming information. Does anyone have a link to information explaining the nature of the abbreviated octave?

I'm trying to prepare calibration (Cal) files for two mics that I own. I want to use the REW app to evaluate a few of my blown or lacking speakers as I attempt to repair or rebuild them. I'll document my project in another discussion. Those who are new to speaker building will likely benefit.

Two of my mics have printed frequency graphs for those individual mics, not just generic graphs for those models.

Neither mic--EV ND357 shown below, and Equitek E-100--is ideal for speaker testing. They are supercardioids, not omnis. The E100 is very accurate from 10hz to 5khz. It does not have much of a "presence" peak above that but is weak above 10khz. The ND357 is good for a hand-held dynamic but far from ideal for my purpose.

I also own an uncalibrated old Radio Shack 33-2050 SPL meter. There are generic Cal files available online for that model, but they vary to some extent.

I have access to an uncalibrated Behringer ECM8000 reference mic. There are also generic Cal files available for that mic. That model is typically plenty accurate for my purposes from 50hz to 5khz, has a 5db boost at around 10khz and has a good amount of "air" above 10k.

I plan to use a pair of JBL PA speakers in good condition as well as my Alesis Monitor One home studio speakers for the tests. Hopefully a comparison of the tests with the four mics will provide reasonably similar results and leave me reasonably confident in the results.

Thanks in advance.

Joe

Author:  gofar99 [ 24 Feb 2020, 21:36 ]
Post subject:  Re: How To Interpret Speaker Frequency Response Graph

Hi, I would expect that line to be 15K. It is spaced about right for a similar one in the 1K to 2K set of lines. One thing to be aware of (you probably already are) is that the room will have a tremendous effect on the sound. Between peaks and nulls you can get all sorts of odd measurements. One diyer suggested that you take the speaker outside and just ignore the bass region response measurements. Good luck.


Good listening
BRuce

Author:  Joegtech [ 26 Feb 2020, 01:28 ]
Post subject:  Re: How To Interpret Speaker Frequency Response Graph

Bruce, thanks for the reply.

Thanks for noticing the similar layout of the graph in the 1k to 2k region.

I did some preliminary testing of my setup with a very modest setup in my bedroom. It is a pair of old Harman Kardan computer speakers with tone control and a Dell Sub. The sub's amp was gutted and I am powering it with an inexpensive sub amp board I got from Parts Expresss.

The setup is only intended to play soft, mellow background music--KLove FM radio.

The FM receiver's EQ settings must be boosting the octave below 100hz because the sound of the music has more deep bass than the graph suggests. However I was not surprised to see a bump in the central bass. I was surprised to see such a big bump.

The first pic below is without any smoothing of the curve, lots of "peaks and nulls" to use your terms. The differences in the highs are due to the tone control in the H&Ks.

The second pic has smoothing, probably 1/12th for the sub's curve and probably 1/6th for the H&K speakers.

The recordings were done at 1 foot and at around 70-75db according to my Radio Shack SPL meter.

The mic was the Behringer ECM8000 with a generic Cal file that I found on the net.

I made Cal files for the two calibrated mics that I own-- E-100 and ND 357--and will start testing them later in the week.

I also made Cal files for an AT 4033 and Shure SM 81 based on the manufacturer's advertised frequency response graph. I don't have much hope for these mics even though they both have wide frequency response and are fairly accurate old studio mics.

Finally I made the Cal file for my Radio Shack RS 33-2050 SPL meter. I chose the most credible of the Cal file contents I found on the net. One of the sites had data for an unbelievably accurate mic in the top two octaves. This SPL meter is using a C weighting so it can't be providing much in the top octave.

Author:  Joegtech [ 28 Feb 2020, 10:41 ]
Post subject:  Re: How To Interpret Speaker Frequency Response Graph

The yellow-brown line in the pic below is with the Harmon-Kardon PC speakers (H&K) and Dell sub in their typical locations. The mic was 8 feet away at the seat where I usually listen to the relaxing music.

Also the Sub has some foam in the rear port. That tightened up the bass. You can see it extended the bass response a quarter of an octave or so in comparison to the bass in the pic above.

The boom box amp that I use to play FM radio has EQ curves that are not represented by these curves. I'll need to come up with an FM transmitter to document the effect of the EQ.

Author:  Joegtech [ 28 Feb 2020, 11:54 ]
Post subject:  Re: How To Interpret Speaker Frequency Response Graph

Smoothing for the Graph's Curves:

REW app has a feature called "Measured Data Smoothing" It is in the "controls" as seen in the pic below.

In my first graph above you can see the curves without any "smoothing" of the curve.

Generally I've been using "1/6th Smoothing" I assume that is based on 1/6th octave. Please correct me if I am mistaken.

A combination of data smoothing and the zoom feature can really allow you to deceive someone about the frequency response of a speaker.

Check out the second pic below. It is the same data that is in the green curve in the "No Smoothing" pic above! What a difference! The curve below only has 1/6th smoothing; however if you look at the left side, the graph has 50db per horizontal line, not the more typical 10db per line.

You can see how specs can lie. You have to look at the specs carefully.

Does anyone know which data smoothing option most manufacturers use? I saw 1/6 smoothing listed recently when I was looking at speaker frequency response curves.

Page 1 of 1 All times are UTC - 6 hours [ DST ]
Powered by phpBB® Forum Software © phpBB Group
http://www.phpbb.com/