Measuring frequency response in my garage

[spacelabstudio]

http://youtu.be/duMACTs8Ra8

[jjasilli]

Interesting.  My middle son & I are beginning to explore the process of measuring frequency response for hi-fi -- for building, re-building and fine-tuning amps and speakers.  It seems that sub-woofers are a special problem.  There's a bewildering array of software, for PC's but mostly for MAC's.  The software generates pink & white noise; analyzes frequency down to a 24th of an octave; etc.  It may contain DAC's and the reverse: analog to digital converters for digital eq and crossovers.  At this point just getting mentally prepared.  It seems that studio recording software can also be used to measure and design speakers & crossovers. 

Compared to just a few years ago, it seems that more such tools are becoming available to the diy market.

[spacelabstudio]

Interesting.  My middle son & I are beginning to explore the process of measuring frequency response for hi-fi -- for building, re-building and fine-tuning amps and speakers.  It seems that sub-woofers are a special problem.  There's a bewildering array of software, for PC's but mostly for MAC's.  The software generates pink & white noise; analyzes frequency down to a 24th of an octave; etc.  It may contain DAC's and the reverse: analog to digital converters for digital eq and crossovers.  At this point just getting mentally prepared.  It seems that studio recording software can also be used to measure and design speakers & crossovers. 

Compared to just a few years ago, it seems that more such tools are becoming available to the diy market.

Awesome.  This is pretty rough.  If I needed something more accurate I'd probably use a recording ADC and do some number crunching in the software.  If you like Python, numpy is a good package and includes an FFT implementation that you could probably use to do some distortion analysis.

Chris

[jjasilli]

Programming & FFT is above my paygrade.  I'd need a plug & play solution.  Even that would probably involve a learning curve.  Here's an example of something which looks suitable at least for measurement:  http://www.trueaudio.com/rta_abt1.htm

Presumably this would identify resonant frequency peaks that might need to be notch filtered out; and help determine the order of crossover needed.  There are dedicated, multi-channel eq boxes, or active crossover boxes that can substitute for good, old-fashion, hard-wired crossover networks.  Or can be used to design a hard-wired crossover.  Alternatively, studio recording software can do this through a computer.  My son's music collection is all CD so it starts off as digital signal anyway.

This requires bi-amping at a minimum.  One amp to power a subwoofer. Another to drive midrange and tweeter.  The reason is to avoid a 3-way passive crossover.  3-way interactions are too complex and unpredictable so as to be unsuitable for DIY.  (And, yes I'm still talking about electronics    )

[PRR]

4 billion transistors to measure *amplifier* freq-response?

I always did it with 5 transistors and two tubes.

Yeah, I had to make marks on paper if I wanted a graph.

But I see you already made lots of Python marks on screen, more characters than I've ever hand-plotted.

FWIW: my basic sweep on g-amp is to find clipping at 700Hz (and be sure it's near what I expect), turn to 70Hz and see what bass drop/rise is, turn to 7KHz and likewise, then 100 200 400 1K 2K 4K to look for anomalties. With tone-stack I expect a lot of slants but if it's reasonable then I assume it is working as designed and further measurements are a waste of time, on to ear-test.

Observation: nothing in a guitar amplifier will be jaggy like that. The steepest corner (not counting Fender tonestack 700Hz dip) will be -1dB at 0.5F, -3dB at F, -5dB at 2F (or vice versa); no more than 2dB change per octave at the corner and not-even 6dB/oct as it begins its fall to infinity. (Possibly 30dB/oct beyond 50KHz, but who cares?)

Observation: there's a freeware, RightMark Audio Analyzer, which will do a better curve and quite quickly. You sure will have to adapt your sound-card to/from guitar-amp levels, but you may have already done that.

(It may complain about the extreme un-flatness of a guitar amp.)
________________

frequency response for hi-fi -- ....and speakers.

Speakers used to be an impossible problem. Narrowband noise was slight help. Now with computers an such you can capture an impulse before the room slap-back, neat. That gets crammed with deep bass in small rooms.

More than one researcher has buried the box in the ground, and waited for trucks and airplanes to go by. Of course this is dubious for boxes which expect/need room boundary reinforcement, which is most home woofers. Mostly if you can find the cone and port resonant frequencies, this with driver data and box size confirms an Alignment which can be plotted (above my pay-grade). Assuming no big leaks, but these usually scream CHUF when ear-tested.

[LHPcope]

Home Theater Shack has a freeware program: Room Equalizer Wizard which works with PCs very well that will do what I think you are after, and a lot more.  Its a Java based program and relies on a quality sound card or outboard sound blaster to work well.

http://www.hometheatershack.com/

The attachment is an image of one of my stompboxes' outputs with a 100 Hz input from ROOM EQ Wizard.

[jjasilli]

Thanks for this info!