CBM-170 and CMT-340 101

[Nicholas Mosher]

Okee-Doke, if people will be generous enough to forgive my ignorance, I have some questions regarding these two speakers, and speaker design in general. I'm also hoping that it will help others who have similar question but are lurking and too afraid to ask.

Rergarding the spectral decays, are they graphs of how long the speaker takes to stop producing output at the various frequencies indicated by the x-axis? This would be the overshoot (or ringing) on material that abruptly stops just before z=0.00 right? This is a graph depicting how well the response is damped, correct? Would an ideal speaker have Y-axis values of say -80dB (if you were listening at 80dB) for all points? I guess where my confusion rests is the difference in where the Y-axis begins on the graphs. For example, many of the little peaks between 1kHz and 4kHz from 1.92 to 3.83 msec for the CBM-170 would not be shown (or be reduced by 6dB pictorally) if you negate graphing "noise" below -42dB as the CMT-340 graph does and use the same window... or am I reading this wrong? (help!)

BTW:
I read the CMT-340 FR graph wrong regarding the slope of the response curve above 10kHz, forgot the jump to the line after 10kHz jumps another 10kHz.

I also have a question regarding distortion. Its definetly something you can hear such as when you crank a guitar amp up and jam, but how is it actually measured and defined? I often see it listed as a percentage at a frequency and SPL.

[curtis]

So you are asking, if the dB scale were equal, wouldn't the 170 loose some of those peaks? If that is the case, then it would also show that the initial output (at 0) of the 170 were lower too...right?

[davef]

Hi Nicholas,

I have answered these questions many times already.... I appreciate and respect your desire for better understanding, but I have already explained this as best as I am able... There are some great books out there on loudspeaker measurements etc. that are written by people who might be able to explain these concepts in a more user friendly manner.

Rergarding the spectral decays, are they graphs of how long the speaker takes to stop producing output at the various frequencies indicated by the x-axis?

Ringing, resonance, transient response; are all revealed in a cumulative spectral decay (as well as other factors).

This would be the overshoot (or ringing) on material that abruptly stops just before z=0.00 right? This is a graph depicting how well the response is damped, correct?

No.... When Z=0ms this is the fundamental response. An impulse is sent to the speaker, the speaker instantly reproduces this impulse, the microphone picks up this response, the response is stored and referenced as the first slice, 0ms. The signal (the impulse sent to the speaker) is instantly stopped, and each additional measurement AFTER 0 ms (not before) is stored and plotted at its appropriate point in time (z-axis).

Would an ideal speaker have Y-axis values of say -80dB (if you were listening at 80dB) for all points?

No... Let’s pretend the speaker is a perfect digital device with no ringing (impossible).... Using your example, if you sent an instantaneous 1ms long signal to this digital device, you would see many slices all at the same amplitude from 0 to 1 msec, and then you would see no additional slices from 1 msec and later. The dB scale on the left, if you were to use absolute spl, would read 74dB down to 50dB in 6dB increments. Don’t ask, this is just how it works... The idea of the graph is a visual representation of sound vs. time for all frequencies… It is designed to give the best visual representation of the data.

I guess where my confusion rests is the difference in where the Y-axis begins on the graphs. For example, many of the little peaks between 1kHz and 4kHz from 1.92 to 3.83 msec for the CBM-170 would not be shown (or be reduced by 6dB pictorally) if you negate graphing "noise" below -42dB as the CMT-340 graph does and use the same window... or am I reading this wrong? (help!)

I know where your confusion lies and you are indeed reading the graph wrong. The scale between the two different graphs is exactly the same. The only difference is that the Y-axis range is different, and this is because the CMT-340 has higher sensitivity. The dB scale is absolute SPL, normalized to a reference level. The highest point of the dB range on the Y-axis is -6dB, (for the 340 graph) and this is approximately what the normalized absolute SPL level of the fundamental slice measured. The highest point of the dB range (y-axis) on the 170 measurement is -12dB and this is approximately what the normalized absolute SPL measured. Notice too on the CBM-170 graph that the 0ms slice appears slightly higher on the graph than the CMT-340.. This is because the absolute measurement should really be a dB or 2 higher than -12dB.. It is difficult to visualize dB levels, and this is because the dB level is not important. From that normalized absolute SPL level, the range is then generated in 6dB increments. CMT-340 = -6dB to -42dB. CBM-170 = -12dB to -48dB..

Now, examining the CBM-170 graph, look at the .5ms time slice at 900Hz. The bump you see is about 12dB down from the fundamental slice. Measure the distance from the first slice to the top of the bump at 900hz (which is at about .5 ms). Compare this distance to the Y-axis scale, about 12dB down right?

Do the exact same thing for the CMT-340 graph... Surprise! also 12dB down from the fundamental.

Now look at those "little peaks" you mentioned on the CBM-170 between 1.5kHz and 3 kHz starting at 2ms and continuing to about 3.5 ms. I will be specific here, on the CMT-340 graph, 3kHz at 2.88ms. The bump is about 34dB down from the fundamental. It looks to be about 2dB higher than -42dB (the bottom) = -40dB. Since the fundamental response is at about -6dB already, -40dB is 34dB less than -6dB. On the CBM-170, the same exact bump appears to be at least 6dB higher than the bottom which is at -48dB, so the bump is at -42dB. Since the fundamental response of the CBM-170 is at about -12dB, -42dB is 30dB down from -12dB. The same bump on the 340 is 34dB down compared to 30dB down on the CBM-170. Let's approximate this to a difference of 3dB... which magically coincides to HALF the acoustic level, or the advantage I explained a long time ago of two drivers yielding half the distortion at the same output level.. VOILA

I hope you can finally interpret the graphs properly now.... I can not spend any more time on this....

Regarding distortion, there are many, many different types of distortion. Each are measured differently, each sound different as well. Some are more audible than others. Distortion is measured and defined depending on what type of distortion you are researching.. Sorry for being so vague... There are some great books out there regarding loudspeaker measurements.

[davef]

So you are asking, if the dB scale were equal, wouldn't the 170 loose some of those peaks? If that is the case, then it would also show that the initial output (at 0) of the 170 were lower too...right?

Yup.. Curtis has it right, well sort of. the dB scale between the two is the same (6dB per division). The dB range is different. if the Range were the exact same, everything would just be shifted downward, including the bottom (what Nicholas incorrectly termed as the noise floor). If the range were the same between the two graphs, there would be no visual changes at all. The dB scale on the graphs we are discussing are exactly like a distance scale on a map, used as a visual guide to help calculate the distance between one place to another.

Here, look at the dB scale of this popular and expensive speaker...



I use a bit higher resolution (more slices) and a longer time window on my measurements, but looking at this CSD it is easy to spot some resonances and at what frequency they occur. Exactly what the measurement is designed to reveal. While the y-axis range is completely different than mine, the scale is the exact same. The engineer who took these measurements normalized absolute SPL to a different level than I... doesn't matter, the range is meaningless.....

[bikeman]

I'd like to see Dave F's technical explanations put in a single folder. In responding to new member's questions, it would be great if we didn't have to dig through old threads to find an appropriate response. I mostly can't remember in which thread Dave has given an explanation. So I havta dig. Digging for information is part of my job so I appreciate when it's made alot easier.

David

[curtis]

Wow....I was close to getting something right this week.

DavidB.....I could put together a read only technical thread if you help gather the info and format it. Ofcourse that is assuming it is OK with DavidF.

[Nicholas Mosher]

Very good explanation Dave, I appreciate the time you've taken to help me out with understanding the spectral decay graphs.

this is because the CMT-340 has higher sensitivity. The dB scale is absolute SPL, normalized to a reference level. The highest point of the dB range on the Y-axis is -6dB, (for the 340 graph) and this is approximately what the normalized absolute SPL level of the fundamental slice measured. The highest point of the dB range (y-axis) on the 170 measurement is -12dB and this is approximately what the normalized absolute SPL measured. Notice too on the CBM-170 graph that the 0ms slice appears slightly higher on the graph than the CMT-340.. This is because the absolute measurement should really be a dB or 2 higher than -12dB.. It is difficult to visualize dB levels, and this is because the dB level is not important. From that normalized absolute SPL level, the range is then generated in 6dB increments. CMT-340 = -6dB to -42dB. CBM-170 = -12dB to -48dB..

I see where I was going wrong! I now see that the fundamental response average Y-axis values for the 170 are about -12dB while the 340s are around -6dB. Now that I simply compare the difference in Y values of the fundamental response with the peaks after its stopped I see how even though the Y scale is shifted 6dB you have less noise in the (1.5kHz, y, 2ms) to (3kHz, y, 3.5ms) area. Very cool.

Now I'm even more curious how they sound when compared to the 170s...

Thanks again Dave, and sorry for bugging you!

[bikeman]

Now I'm even more curious how they sound when compared to the 170s...

OK. Nicholas pool #2. On what date will NM take delivery on his new 340's?
May 30th sounds about right to me.

Nicholas. Jump in the car and make a day trip to Syracuse. I'll take ya hiking after ya have a listen to the 340's. It'll be a short hike cause the sun will have gone down hours ago.

David

[curtis]

OK. Nicholas pool #2. On what date will NM take delivery on his new 340's?
May 30th sounds about right to me.

What was pool #1? I think it will be a little farther out. He has a second sub and HDTV in his plans.

[bikeman]

What was pool #1?

Nicholas bought a guitar while he was auditioning subwoofers and I speculated that it might not be long before he purchased a five string bass guitar that goes down to 30hz. The pool was how long it would take.

David