Room EQ Wizard Doubles as Crossover Design/Measurement Tool

23629 Views 33 Replies 6 Participants Last post by Anthony, Aug 17, 2007

aktiondan

Discussion Starter · Jun 17, 2007 (Edited)

Is anyone else using REW to design crossovers? I've been working on a new set of MTM speakers for a friend's home theater and discovered that REW works great for measuring crossover networks. To some degree it will even do impedance plots and measure Zobel networks. See the example below. Though I don't believe the absolute impedance value is correct (i.e., the graph shows the resonant frequency of the woofer [red] to be at 39Hz and a dB level that might correspond with an impedance of only 8.7 ohms, when in fact it's more like twice or three times that (you'd need to convert the log function to a linear function), however the profile of the curve is accurate). Notice the classic impedance rise which is easily combated with a simple Zobel network. Using just 15uF cap and 8 ohm resistor I was able to measure the transfer function [purple] of the Zobel (into an 8 ohm resistive load) and then see what the resultant impedance plot of the woofer [green] looks like - much better! (The black line is my reference, which is an 8 ohm resistive load). From here I can easily adjust the capacitor and resistor to tweak the profile and achieve something that looks even better.

Anyway, this also works for crossover networks, (I've got a bunch of plots of those, too) as well as measuring the near-field response of woofers and tweeters. So far I haven't seen any inherent limitation in REW that would indicate it isn't capable of doing these kinds of measurements (other than the proper scaling for impedance). But there is one thing I would love to see - the ability to export the graph to a .csv file. That way impedance plots and measured FR plots of drivers could be imported into Passive Crossover Designer 5.00 and then used for actually designing crossover networks (instead of relying on mfg's data) and then once the crossovers are designed, REW can be used to measure them and tweak them. And of course once the speaker is built, you can measure the resultant 20-20kHz response of whole speaker.

Just thought I'd share, REW has proved to be invaluable to me, and I am glad I ran across this site. Sorry this is my first post and kinda long, but this forum is great.

See less See more

1 - 20 of 34 Posts

brucek

· #2 · Jun 17, 2007

Welcome Dan..... yeah, we use REW for all kinds of in/out measurements. I don't remember anyone using it for crossovers, so that's a new one for sure. I love the tuning of the Zobel - very cool.

We definitely measure receiver/ processor, line driver and BFD equalizer responses. One of the things you can do when you use an ECM microphone and preamp for audio measurements is to put the preamp in the cable loop when you are creating the soundcard.cal file and it will increase the accuracy of your measurements by compensating for any response HPF effects from the preamp electronics when the file is created....

I used REW to find the correct spot on the poorly marked dial of my X-30 crossover box that corresponded to the crossover frequency I needed.

brucek

Chrisbee

· #3 · Jun 22, 2007

Dan

How exactly did you do the impedance plot?

No great accuracy required just a bump in roughly the right place to check Fs.

Thanks

aktiondan

Discussion Starter · #4 · Jun 23, 2007

Put a 100 ohm, 10W resistor in series with your driver. Then connect the input of your sound card across terminals of the driver (in place of your mic/SPL meter). Make sure you turn off any mic/SPL calibrations curves, but be sure to use the sound card cal. Then you set up REW just like you would for a regular measurement and hit go. The graph that results has just measured impedance instead of SPL. Hope that was descriptive enough, but it's not too tough and works great for determining fs of a driver.

Chrisbee

· #5 · Jun 23, 2007

Perfect! Thanks. :T

terry j

· #6 · Jun 23, 2007

not sure I followed it exactly, will probably make sense when I do it for real.

Would that work once the driver is in the box??? ie would it tell you the tuning of the box be it vented or sealed or whatever?

aktiondan

Discussion Starter · #7 · Jun 23, 2007

It should make sense once you do it, but I'll try and take a picture or do a quick drawing or something. Yes, it works for a complete speaker. Here's a plot I just ran of a pair of two-way monitor speakers I built a year ago. Fb of the box lies between the first two dips, in this case it would be about 35Hz. I believe I was shooting for a tuning of 36Hz or so. If this were a sealed box, there would be a single peak at the low end, and Fb would be at the top of that peak.

See less See more

Chrisbee

· #8 · Jun 23, 2007

Dan

Just for clarification: The driver is put in series with the resistor between the input and output of the soundcard?

I'm getting a tiny peak at 50Hz which doesn't make any sense at all.

I'm also getting a constant buzz on the driver.

aktiondan

Discussion Starter · #9 · Jun 23, 2007

The driver is connected to your amp/stereo with the 100 ohm resistor in series with the (+) terminal on the driver. The line out from the sound card goes into the CD/AUX in on the stereo. This is the conventional method for using REW I believe. Then you connect a pair of wires to the terminals on the speaker and run that to the line in on your sound card. Most sound cards are only line-level out (about 1Vrms) which isn't enough to drive the speaker straight - you need the external amp to do that. Let me know if that doesn't get you going.

Chrisbee

· #10 · Jun 23, 2007

aktiondan said:
The driver is connected to your amp/stereo with the 100 ohm resistor in series with the (+) terminal on the driver. The line out from the sound card goes into the CD/AUX in on the stereo. This is the conventional method for using REW I believe. Then you connect a pair of wires to the terminals on the speaker and run that to the line in on your sound card. Most sound cards are only line-level out (about 1Vrms) which isn't enough to drive the speaker straight - you need the external amp to do that. Let me know if that doesn't get you going.

I wish you'd mentioned the amplifier in your first post.
I have just spent 5 hours trying to get a peak straight off the soundcard.

aktiondan

Discussion Starter · #11 · Jun 23, 2007

Sorry about that, I thought it was understood that an external amp was used when using REW. Have you tried it again with the amp?

aktiondan

Discussion Starter · #12 · Jun 23, 2007 (Edited)

Chrisbee, well I felt bad for steering you astray and causing you to waste your time on this test, so I thought I'd try it out your way, and connect my speaker straight to the sound card (with the series resistor still), and it actually still works. So maybe there's something else not right with how you're hooking it up? Without using the amp, I've got the volume control full blast on the WAV as well as the main output, and the line in is set at about 1/3 the way up. When I go into to check levels, it comes up around -14dB. Then I just hit measure. I've got it set to measure a 1M log sweep to 20kHz. It runs a sweep, I can hear the speaker go through the sweep, and the resultant plot comes looking like an impedance plot.

Most of the other settings in REW are at their default value, so I'm not sure if there's something not set up right there? Anything else I can clarify to get this working for you?

Here's a quick hook-up diagram (w/o the amp):

Line Out (+) -------------------/\/\/\/\------------------Speaker (+)-------------------Line In (+)
Line Out (-)--------------------------------------------Speaker (-)--------------------Line In (-)

terry j

· #13 · Jun 24, 2007

whilst this procedure can't give absolute values to the measurements (unless s/one with the suitable mathematical skills can work out a correct formula-is that possible?)I'm also guessing that it may help us determine the correct amount of stuffing in a sealed box for example?

IIRC the resonance (and therefore the impedance) of the system drops with stuffing and will cease falling and start rising when the correct amount has been added. I would have to hunt around to get the exact data, but I think that's right??

If that is kinda correct then it might take the guesswork out of how much stuffing to put in. Maybe it would help with the vented boxes and LLT's too, one for Steve Callas perhaps, tho he's probably way ahead of such crude techniques.

Am I on the right track here?? cause I'm soon to start building my subs and so it would be a timely discovery.

aktiondan

Discussion Starter · #14 · Jun 24, 2007

2

It was years ago that I built this little impedance analyzer, and the method for measuring impedance was much more time consuming. It involved using a signal generator and a DMM and you had to go through and manually test and measure each frequency. REW just replaces the signal generator and DMM, and it does it quite nicely too. Since all it's really doing is comparing an output voltage to an input voltage.

I've got a little 1/8" mono plug that runs to an RCA jack with a couple of wires soldered to it, and that hooks up right across the woofer/speaker. The 8 ohm resistor can be used to calibrate the measurement, so you have a reference to 8 ohms on the graph in REW. Though it shows up as dB, you can convert back to ohms with the formula Z=CALz*10^(A/10) where CALz is your reference load (8 ohms in this case) and A is the delta dB from the 8 ohm reference. For example, if you set it up so that 80dB is your 8 ohm reference, then if the plot rises 3 dB to 83 dB, that would mean the impedance actually doubles from 8 ohms to 16 ohms. A 6dB rise would be 32 ohms. The trick would getting the plot into Excel so you can manipulate the data.

I hadn't thought about using it to evaluate the ideal amount of stuffing, but it certainly can help you "see" what you're doing by adding/removing stuffing. For measuring the tuning frequency of a newly built sub this method is ideal. A more accurate way to measure the Fb of a ported box is to plot the impedance and measure the frequency at which each of the two lower peaks occur and label them Fl and Fh. Then seal the ports, and measure the frequency at which the single peak occurs and label it Fc. Run the numbers through the formula Fb = SQRT(FL^2 + FH^2 - FC^2) and you get Fb right on the nose.

See less See more

Chrisbee

· #15 · Jun 24, 2007

2

I have copied your soundcard-only circuit exactly but I'm still getting only the slightest peak at around 31.5 Hz.

I've double checked and my resistor still measures 105 Ohms.

Only if I mess about with the graph limits can I make it look as if that peak actually means something.

These AE IB15 drivers are supposed to have a 16Hz Fs!

See less See more

Chrisbee

· #16 · Jun 24, 2007

Okay, at last. Impedance curve using an old 30WPC stereo amp I dug out of storage.

I didn't want to risk my 450WPC IB amp on this test with all those bare wires dangling all over the place..

I think this is absolute confirmation that my AE IB15 drivers don't remotely match the claimed 16Hz spec. The green line is my soundcard-only plot.

I suppose I ought to test all four just to double check because I get a very different response between the two pairs in the same manifold.
They don't even look the same in some minor details.

See less See more

Chrisbee

· #17 · Jun 24, 2007

Confirmation all 4 drivers have much the same impedance curve all peaking at ~32Hz.

I switched off the amplifier and readjusted REW SPL levels between each sweep.
This accounts for the slight variation in level between each driver.

See less See more

brucek

· #18 · Jun 24, 2007

The trick would getting the plot into Excel so you can manipulate the data.

Dan, REW has an Export Measurement as Text feature under the File pulldown. Copy and paste the results of this feature's text file into Excel and use the Text Data to Columns feature in Excel to separate it using the space as the delimiter. This yields a very resolute (frequency and SPL level) set of columns................ Here's a quick export and paste into excel showing how resolute the results of a measurement are between 10Hz and 15Hz. You get the idea anyway.

Text Line Parallel Font Number

I have copied your soundcard-only circuit exactly but I'm still getting only the slightest peak

Chrisbee, the line driver circuitry in the soundcard line-out is meant to operate as a voltage bridge sourcing high impedance loads. A speaker and 100 ohm resistor don't really qualify. Especially (as it is in your case I believe), if you're driving and measuring with long lines, which would add their own reactance to the mix. Best stick with using a power amp to do your test ...

brucek

See less See more

aktiondan

Discussion Starter · #19 · Jun 25, 2007

brucek, thanks for posting how to export the data, not sure why I didn't notice that option before, guess I just wasn't clicking around enough. So I took the plot I had measured, pulled the data into Excel, converted to ohms, and plotted the results and compared them against the manufacturer's data. With the exception of the absolute impedance values seeming a bit exaggerated, the resonant frequency is right on - considering the driver is brand new out of the box, it reads a bit higher than one that's been broken in.

Chrisbee, that's weird about your drivers reading 32Hz when they're suppose to be 16Hz. Are the drivers mounted in an enclosure? I'm guessing they are in some sort of infinite baffle configuration, just based on the name of the driver, but if you're measuring the speaker in their enclosure, even if it's an IB enclosure, you end up measuring the Fb of the box more so than the Fs of just the driver. If the box is sufficiently large enough, the Fb of the enclosure should approach the Fs of the driver, so it should effectively be the same thing as measuring the driver in free air, but maybe not. And that's how the driver should be measured, no baffle, suspended in free air at least 3 feet away from other objects. Otherwise the enclosure comes into play and drives the impedance peak (Fs) upward.

See less See more

Chrisbee

· #20 · Jun 25, 2007

The drivers were tested in their open-sided 4' high IB manifold with the door right next to the box wide open. There should be no real effect on Fs with the front and backs of the drivers exposed to the same volume of air. Worst-case path length between the front and back of the cones was only about 4 feet. The impedance peaks look far too close to each other in frequency for there to have been much variation in air loading despite different path lengths.

1 - 20 of 34 Posts

This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.