Hey everyone,
I wanted to post a think-piece about learning to do speaker analysis. In particular for a beginner starting out with a 2-way system and using in-room measurements may be the very best way to get started.
Of course there will be room interactions not fully accounted for, but maybe a measurement which gates the tweeter/mid-woofer crossover areas and then blends that with the mid-woofer's low end response is the best way to get started?
With 1m measurements there are a few complications we avoid (compared to quasi-anechoic), such as:
I wanted to post a think-piece about learning to do speaker analysis. In particular for a beginner starting out with a 2-way system and using in-room measurements may be the very best way to get started.
Of course there will be room interactions not fully accounted for, but maybe a measurement which gates the tweeter/mid-woofer crossover areas and then blends that with the mid-woofer's low end response is the best way to get started?
With 1m measurements there are a few complications we avoid (compared to quasi-anechoic), such as:
- Integrating the port (if any) with the woofer response
- Adding baffle step calculations
- Understanding boundary reinforcement and how that will shape the final outcome.
- Measuring the acoustic offsets
- Measuring off-axis to pick ideal crossover location.
There's a soundfield created by two sources, in stereo.
There are no room interactions, it's the sound that propagates and collides on hard surfaces and refracts, reflects, diffracts at angles and holes.
What you catch at 1 m on axys is only the amplitude ( magnitude- scalar) of a vector.
There are no room interactions, it's the sound that propagates and collides on hard surfaces and refracts, reflects, diffracts at angles and holes.
What you catch at 1 m on axys is only the amplitude ( magnitude- scalar) of a vector.
I would suggest reading a book as the efficient tried and tested way to get from knowing nothing to having an idea of where to look for guidance on what is not understood.
Of course. My point was meant to ask, for absolute beginners, whether or not attempting to do a fully quasi-anechoic analysis of a speaker is the best approach or not.
I'm very much afraid the spirit and background of my original post is getting lost. When I talk to a lot of DIY'ers who are just starting to mod speakers it seems to me their enthusiasm is somewhat blunted by the difficulty of complete measurements. Given that and my desire to see more people enjoy the hobby I posited that encouraging messy but good approaches is useful.
My opinion...
Yes, given today's tools it doesn't make sense to not take good on- and off-axis measurements with a dual channel setup. The goal is to make something DIY as good, or better, than commercial offerings. But if you aren't going to do it right, you can get perfectly acceptable performance from new, or especially used, speakers.
I wouldn't bother buying a book. There are 5 to 10 page Internet write-ups that explain what needs to be done.
Jeff Bagby's straightforward suggestions (which is exactly what they are) remain about the best 'simple' guide that I'm aware of, in terms of producing data that's actually useful:
http://audio.claub.net/software/FRD...curate In-Room Frequency Response to 10Hz.pdf
http://audio.claub.net/software/FRD...curate In-Room Frequency Response to 10Hz.pdf
I think probably the most important thing is technique and discipline. You need to be able to reliably make repeatable measurements. If you can't do that your are toast. So I would put emphasis on practicing as a beginner and experimenting with distance, measurement spots in the room and so on so you can find the "sweet spot" where you have the least amount of room influence.
Once you find a spot you can just tape off reference points so you can set -up quickly and get reliable and repeatable results.
Rob 🙂
Once you find a spot you can just tape off reference points so you can set -up quickly and get reliable and repeatable results.
Rob 🙂
I think it is important to recognize and acknowledge that different peoples have different styles of learning. Some people want or need to grasp the theory of how something works before taking action. Other people learn best by taking action first, and after gaining a sense of how something works, then diving into the theory.
@eriksquires - your idea makes sense for some people.
@eriksquires - your idea makes sense for some people.
Definitely agree with this. I am in the camp of normally reading a lot to learn, but when it comes to entirely new topics like speaker measurements, everything I know about theory goes out the window and I just need to play with fast and easy techniques to better understand the concepts, before I can spend time reading again about the right way to do things.
2 cents: Just get a free Android APP called Frequency Sound Generator and listen to finger-swept tones, of as many speakers/setups as possible. Their most apparent sound-quality issues become exposed and correlated to their frequency response. Then one learns quickly how to make such measurement-by-ear reliable and repeatable, more broadly useful and actionable for diy.
Some people prefer a monkey see / monkey do type approach. You don't need to understand why you do something... just do it. Of course as soon as we tray off this path and need problem solving, we're stuffed... but if for newbies, we can make it easy on ourselves and choose for example easily available, cost effective and well behaved drivers in a knockdown or standard pre-cut panel size cabinet with limited tools required.
With learning styles, there's the obvious audio (text), vs visual (video) and kinesthetic (doing). Assuming kinesthetic requires hands on with a tutor, we lean towards likely video (backed with a numbered step guide).
We then need to break it down into sections, a bit like the stickies do here.
So I suppose you are wanting to jump in at (say) chapter 4 or so with measurement approaches and get a more specific guide on that?
If we assume a single channel setup (USB), then it's simply
1. Measure at a distance
2. Learn about gating, why we do it and how we apply it
3. Repeat off axis
4. Repeat above for each driver (not moving mic nor changing power levels)
5. Dual driver measurement (for Z)
6. Z adjustment to align individual drivers to combined #5 response.
We then need to pick the software we use to demo this (measurement and XO modeling), drivers and potentially microphone (and stand).
With learning styles, there's the obvious audio (text), vs visual (video) and kinesthetic (doing). Assuming kinesthetic requires hands on with a tutor, we lean towards likely video (backed with a numbered step guide).
We then need to break it down into sections, a bit like the stickies do here.
So I suppose you are wanting to jump in at (say) chapter 4 or so with measurement approaches and get a more specific guide on that?
If we assume a single channel setup (USB), then it's simply
1. Measure at a distance
2. Learn about gating, why we do it and how we apply it
3. Repeat off axis
4. Repeat above for each driver (not moving mic nor changing power levels)
5. Dual driver measurement (for Z)
6. Z adjustment to align individual drivers to combined #5 response.
We then need to pick the software we use to demo this (measurement and XO modeling), drivers and potentially microphone (and stand).
Yeah I know what you mean as a lab partner in college I was the guy setting up the experiments in Physics or lighting the bunsen burner in Chemistry. Same in electronics hands on all the way!
Rob 🙂
I'm a complete beginner! My first question is: if you say 1 m does it mean that the baffle is the zero / starting point?
Does it have to be really millimetrical and perpendicular? ( Pun intended: if I listen between the two I'm off axis, so it would be more important the -30°)
There are many questions. Generally I see that it leads to accepted misconceptions, which are not serious as they might be intended as colloquial or shortcut for something more complex. When you read from the start that two sources 'sum' you know that two waves superimpose.
I read often standing waves and internal reflections inside a box, I always say that maybe the room, not the box.
I presume you mean the title "Testing Loudspeakers." It generally sells for $40+ but I see two copies for under $30 through bookfinder.com.
What if the system is not LSI ?
And the stimulus? As music contains decaying signals..err, notes. Not the steady signal.
But let's talk about the room, and the anechoic room where the perfect measurement set up operates. The semi-anechoic means what?!
If sound in an anechoic room is spoiled, does it have the same attributes that are accredited to the sound we perceive normally? Is eco localization an important attribute?
So I return to post #2: what does represent a single point in a sound field?
Can a scalar magnitude represent a vector that needs 3 numbers to be defined?
If you're a complete beginner, then I suggest you forget everything about 'scalar magnitude represent[ing]... vector' and simply read the link I posted to Jeff's short writeup on taking quasi anechoic measurements, which offers a very simple step-by step guide through taking measurements themselves, with highly practical (pragmatic) examples and suggestions.
When you've done that, if you are using a USB setup that doesn't take flight-time into consideration (no time lock), this gives a walk-through, in the same straightforward manner, on how to establish the relative acoustic offsets (aka phase relationships): http://techtalk.parts-express.com/filedata/fetch?id=1149302 for use in filter design. Jeff was a superb designer & educator, and was always looking for simple solutions to practical problems. He's a great loss to the community. The methods he outlines in those two short writeups also works with other software like XSim, WinPCD, VituixCAD etc.
As a start, a couple of pointers.
When you've done that, if you are using a USB setup that doesn't take flight-time into consideration (no time lock), this gives a walk-through, in the same straightforward manner, on how to establish the relative acoustic offsets (aka phase relationships): http://techtalk.parts-express.com/filedata/fetch?id=1149302 for use in filter design. Jeff was a superb designer & educator, and was always looking for simple solutions to practical problems. He's a great loss to the community. The methods he outlines in those two short writeups also works with other software like XSim, WinPCD, VituixCAD etc.
As a start, a couple of pointers.
- Generally you will take all measurements from a given point in space, usually on the intended listening axis, and you will not move the microphone unless you have a fixed test setup / chamber etc. Usually, it's on the tweeter, with the occasional exception of large 4 - 5 way designs (and sometimes not even them). For off-axis measures, it's generally easier for DIYers (and many professionals for that matter) to move the speaker, be it on a turntable, adjustable feet or whatever. The most important point there though is to do what works for you, and you can do repeatably / reliably.
- Getting the microphone aligned properly is critical for best results. A laser tape-measure can be useful here, but the old-fashioned way still works, obviously. 😉
- 1m is a bit of a red-herring, as Jeff says. For smaller speakers, it can end up being excessive & you'll probably end up with excess reflections. As an ROT, farfield starts from 3x - 5x the effective radiating diameter of the (LF) driver away, and at least 2x the baffle width to properly account for step-loss. So for e.g., if you've got a driver with a 5in piston diameter on a 9in wide baffle, 20in would be ample in most cases (since it's more than 3x the piston diameter, and more than 2x the baffle width). With larger speakers, with drivers distributed over a greater distance, you may need to be further out for proper summing -unfortunately that does tend to mean more room modes, but that's just nature of the beast & not much you can do about it. In those cases, the above guidelines hold, with the addition of it also being at least as far out as the distance from the listening axis to the centreline of the furthest driver away from that on the baffle.
Last edited:
The modern era seems filled with online material where we can learn off each other. And one of the best ways, if you are a visual/auditory learner, is via video
This is great "introductory course" because it doesn't get into the weeds with complete polar measurements and diffraction or distortion, which are also important, but can lose people who are just starting out.
It's like a video tutorial of practical things to do- how to measure at 1m (or less), how gate to how to remove the influence of room reflections, how to drive your software simulator and experiment with caps, resistors and inductors, how to choose the right box for your woofer etc. Like what the late Jeff Bagby or Sigfried Linkwitz would have on YouTube, if someone taped their live tutorials and uploaded them into chapters...
That's enough for most people to sink their teeth in, without losing them.
If you've already built/modded a few speakers in Speakerworkshop, lspCAD and familiar with terms like baffle step loss compensation and vertical polar response and you get excited about diffraction mitigation , or only ever measured on-axis and couple of other angles as a spot check?
15 pages of VituixCAD2 quick start guide is what you want...
Last edited: