|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
69.3.223.219
In Reply to: RE: programs for testing speaker response posted by floydandrews on May 17, 2016 at 11:50:52
F.
I have used REW and it works quite well, but there was a learning curve going from the old school way of making freq. resp. graphs using the warble oscillator test tracks on the Stereophile Test CD, test mic with a VU meter and graph paper.
Good news: You can generate a lot of data very quickly on your laptop with REW.
Bad news: You can generate a lot of data very quickly etc.
Interpreting all the data is still the difficult part. It took me awhile to understand why the measurements were quite different between REW and my old graph paper data. The reason is largely due to the tendency of the warble oscillator and VU meter to average out and smooth the response. You will need to select an averaging option with the graph REW creates, and 6 dB per octave is a good start as Atkinson usually uses this in his frequency response graphs in Stereophile (when this is specified). Using no smoothing will provide a graph that looks like a field of grass with too much micro information. Using a high order of smoothing, like 24 dB per octave, will make the results unrealistically smooth (read: too good to be true). What you want to look for is a signature, or the lay of the forest rather than overly focusing on particular trees in it. I have obtained similar results to Atkinson's measurements when measuring the same speakers he has tested in Stereophile
Make a paper copy of the REW instructions as you'll need it while your laptop is running REW. Also there's a good instructional video on Youtube about using REW. The Home Theater Shack website (where you can download REW for free) is mainly concerned with measuring room response which can be addressed with active EQ, and not so much with what the speakers themselves are doing without what the room is adding.
A bit of history: Richard C. Heyser invented Time Delay Spectrometry which he used to test speakers in Audio magazine back in the day. The first TDS unit cost $30,000 to cobble together out of various computer and pieces of scientific equipment. Nowadays a functional facsimile program is a free download, but we're still scratching our heads about the data! Heyser saw similarities among the discrepancy between what we hear and the measurements we obtain, and similar discrepancies between measurement data in Quantum mechanics.
Good Luck!
Paul
Follow Ups:
Hi Paul,
Fyi, smoothing usually works the opposite of what you've said. It's specified in fractions of an octave - 1/6 octave, 1/24 octave, etc. and 1/6 is more smoothing than 1/24. 1/6 octave smoothing is a good place to start. Less smoothing, like 1/24 octave can be useful if you're measuring outdoors far from reflections and actually want to see the finer detail of the system you're working on.
To the OP, I've used Speaker Workshop quite a bit over the years (it's quite old). It also can measure impedance and do crossover design. It's harder to pick up than current programs - it has a few bugs and you need to know a bit more to use it. But it can be very powerful. I'd also recommend that if you're trying to design a system from scratch, you go to the trouble of getting good outdoor measurements in order to design the system, then do fine tuning in your room. I find I get much better results that way.
The speakers are built. They are 2 way with an active crossover. I would like to make some basic measurements to refine the crossover. I live in an apartment. Outside measurements would take a little more planning than for someone with a backyard.
If your speakers are built, then you've fixed a couple of the variables - what the drivers are, the raw response they'll have, and their physical location relative to each other. The remaining work is to design filters that make the drivers work together in the best way possible. This is typically what determines a lot of how good a speaker will be. If you use the best components in the world but design bad filters for the drivers, it will still sound like crap.
I'm not saying you have to measure outside, but I've found it much easier to get good results if I design the filters for the drivers in an environment where I don't see the effect of the room and then tweak the response of the total system in the room without having to worry about whether I've got the crossover right or not.
Parking lots are usually my go-to places, like school parking lots on the weekend. I'll do a ground plane measurement where you place the mic on the ground to eliminate reflections up to a fairly high frequency. I typically use a laptop and a UPS to power my measurement system and amp. If you get a decent sized one it will go for a couple of hours. You definitely want to have your process down cold before you go do that though. You also want to be aware of what the noise floor is - take a measurement without any of the drivers playing and compare it to what you get when you measure the drivers.
This makes sense to me. I can access the roof of our building, and we live in a relatively quiet neighborhood. The speakers are in 2 boxes, woofer and fullrange, so shouldn't be too hard to haul them up and back. Thank you for weighing in. We have some very knowledgeable folks on this board (you being one of them) and I always appreciate their willingness to share and teach.
Happy to help.
The roof is a bit different situation, but potentially better. What I would try is positioning your speaker at the corner of the roof pointing away from the building and put the mic on a boom of some sort out into the air. If there's nothing flat behind the speaker, you should mostly eliminate reflections that way. You would get a reflection from the ground, but if you're high enough up, you could gate that out and get good LF resolution.
THIS is ONE of the reasons why audiophile-land hasn't changed in 45 years. I feel like I'm reading an article from 1972!!
I'm not exactly sure what you mean, but if you're saying that people not measuring the speakers they're building is like being back in the 70's, I suppose I'd agree although I'd say it's like any time before ~2000. Of course the difference now is that it's fairly straight forward to get a good measurement program, a decent mic and sound card, and start learning how to incorporate measurements into building speakers if one is inclined to put in the effort.
Hi John
With REW an initial un-smoothed measurement will provide a useless "too much grass" type of graph. A 6 dB smoothed graph will make the same measurement look similar to the frequency response graphs in Stereophile, though the scaling is different. At 24 dB smoothing with REW most of the detail is lost. Perhaps things are different with Speaker Workshop, which I've never used. Maybe it's a difference in nomenclature between the two systems.
As to doing measurements outdoors, that's "nice work if you can get it" as the song says. I live on a gentrifying block with at least two or three new or rehab house projects going on at any time, I'm a block away from a train line which is also next to the elevated train line, which is all under a glide path for O Hare airport. The presidential helicopter frequently flies over my house.
Don Keele observed that one of the weaknesses with these time delay sampling techniques was the one-size-fits-all sampling rate, and he suggested that the sampling rate should be longer at the lower frequency end, and higher at the high frequency end. REW uses the same same rate as far as I recall, perhaps some of the more sophisticated (non-free) programs use a sampling rate adjusted to the frequency. Keel's AES paper Time-Frequency Display of Electro-Acoustic Data Using Cycle-Octave Wavelet Transforms is a good read. He comes to a similar conclusion that Heyser did in that time-frequency displays suffer from quantum-like uncertainty because you can't get arbitrarily precise in one domain without getting less precise in the other.
I miss your JHS website!
Paul
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: