Frequency Response Graph

[Tim Snyder]

Help me learn how to read a Frequency Response Graph - pretty please. I get the whole rise and fall in decibels along the frequency range.

However, what is confusing me is some manufacturers don't send the graph all the way to the 20 Hz line. And even some manufacturers don't do it consistently within their own company.

For example Shure:

SM-81 is listed as a 20 HZ to 20 KHZ microphone. When you look at the graph the line stops plotting around 40 hz.

KSM-137 is listed as a 20 HZ to 20 KHZ microphone. When you look at the graph the line plots all the way to 20 HZ.

So which is it? Does the KM-81 actually go to 20 HZ as listed or 40 HZ as the plot suggests?

Or are we to use the graph and draw the rest of the line ourselves following the previous information?

Just trying to learn more about how to read these things. Thanks much.

Tim

http://www.shure.com/stellent/groups...sm81_large.gif

http://www.shure.com/stellent/groups...m137_large.gif

[Audiop]

All microphones go to 20Hz.. or 10Hz.. but that means nothing unles you say at what level.

Lot's of gear is speced 20Hz-20kHz but it is totally useless info. Wee need something like +0/-3dB 20Hz-20kHz. That could mean that at 1kHz we have 0dB (a reference level, it does not mean that 0dB is dead silent as it could be 110dB SPL) and along the 20-20k scale we never se a level lower than -3dB.

The same device could be speced +/-1.5 dB 20Hz-20kHz.

The info above does not tell anything about the level at 20Hz and 20kHz though.

Electronics is typically flat within the passband and have a roll off at high and low frequncies. Such devices may be speced as being down let's say -1dB at 5Hz and 30kHz.


/Peter

[Audiop]

Originally Posted by Tim Snyder

So which is it? Does the KM-81 actually go to 20 HZ as listed or 40 HZ as the plot suggests?

Yes with a gradual roll off.

Or are we to use the graph and draw the rest of the line ourselves following the previous information?

Yes. An electromechanical device never drop like a brickwall such a digital filter can do.

A mic or speaker that is truly flat to 40Hz will have significant pickup/output at 20Hz.


/Peter

[opegas]

the basic problem you are pointing out is that there is no standard or mandatory anything. People can print whatever specs they want, accurate or not.

Sometimes mics have different roll offs on the low end so graphs only go so low. Some manufacturers don't have good testing facilities and may or may not be able to accurately test 20 hz. It is almost normal for everyone to just put 20-20k regardless of any meaning to the spec.
"frequency response: 20-20kHz" has no meaning

more accurate would be "frequency range 20-20khz" or

"frequency response 20-20kHz +/-2db" would tell you that somewhere in that range the frequency response is up or down 2 dB.

Lots of mics are not flat, sometimes by default sometimes by design. A LDC with a boost at 8k or so of 5-6 db might look bad as a spec if they wrote
"frequency response 20-20kHz +/- 6dB"

Sometimes its not the engineering department of a company, but the marketing people who take specs and change them or smooth out and "artistically" draw frequency response graphs for literature or websites.

Other things to consider for the low end are proximity effect on measurements. If mics are measured at 1 meter, but typically used at an inch, say like an sm58, the measurement at 1 meter might appear to roll off early/ higher in frequency.The manufacturer intended it to be used close where the bass will boost due to proximity.

Paul

[Audiop]

Originally Posted by opegas

Other things to consider for the low end are proximity effect on measurements. If mics are measured at 1 meter, but typically used at an inch, say like an sm58, the measurement at 1 meter might appear to roll off early/ higher in frequency.The manufacturer intended it to be used close where the bass will boost due to proximity.

Good points. It's worth to mention also that the proximity effect is something that happens with pressure gradient microphones ie. directional mics such as cardioids and fig. 8's.

Pressure microphones/omnis do not "suffer" from this.


/Peter

[John Willett]

Often when a manufacturer says 20Hz to 20kHz it means the limit of where he *measures* the response. The graph shows you the actual response of the mic.

To be honest I hate this way of describing a response - I would far prefer it for the manufacturer to state the frequency limits at the -3dB points both ends.

[John Willett]

Originally Posted by Audiop

Good points. It's worth to mention also that the proximity effect is something that happens with pressure gradient microphones ie. directional mics such as cardioids and fig. 8's.

Pressure microphones/omnis do not "suffer" from this.

Yes - and some manufacturers do mic. measurements closer than the standard 1m to disguise the fact that their mics are slightly lacking in the bass. tutt

[Tim Snyder]

Hey, thanks for the all the input. Just trying to clear my head of this stuff. I really don't ever look at this stuff, I tend to just listen to a piece of gear and make a decision based on that.

However, in this case these graphs just jumped out as odd when looking at them.

So anyway, thanks much.

tim

[chris319]

I find it helpful when a manufacturer normalizes a frequency response curve so that 1 kHz = 0 dB. Using Voxengo's Curve EQ I sometimes make an EQ curve which is the inverse of the published/measured frequency response curve.

[mathieujm]

Originally Posted by chris319

I find it helpful when a manufacturer normalizes a frequency response curve so that 1 kHz = 0 dB. Using Voxengo's Curve EQ I sometimes make an EQ curve which is the inverse of the published/measured frequency response curve.

I never found this method give me good results (even if I have the REAL response curves of my mics). There are so much parameters than only the response curve that make the final sound !

JMM

[Audiop]

It can be useful IMO but obviously you need to consider the FR off axis ie. polar pattern. At lest it is a good starting point if the published graphs have some connection with reality.


/Peter