Measuring my DIY Speakers for Crossover Design

[Nordo]

After a 30 year hiatus, I decided to get back into speaker building - as a hobby really.
I'm at the stage now where I need to measure the frequency response at various angles, as well as the impedance of each driver, using REW (which I've used many times in the past for room testing, etc.).
I understand how to do the impedance testing, but I'm not sure about the freq response testing.

I have been told that to get accurate results (I'll be testing outside to minimise reflections), I need to take measurements using a semi-dual or full dual measurement setup.
My problem is that I have a UMIK-1 usb microphone, and I'm not sure if I can do semi-dual or full dual measurements,

Can anyone explain how I could get accurate test results with my UMIK-1, and my laptop, which has a headphone out, and a HDMI socket.
I have Windows 10 and use ASIO.

Thanks

 

[John Mulcahy]

You can send signals to the speaker using the headphone output. When measuring with a USB mic REW generates a high frequency timing signal to act as a reference. The difficulty with polar measurements using that is the off axis measurements may not pick up the timing signal well or a side wall reflection may be larger than the direct path. If you are measuring outside the side wall problem goes away, but in any case I suggest you use a separate speaker to reproduce the timing signal so that stays constant while the speaker under test rotates. For the design work VituixCAD is well thought of, there is a very large thread on it at DiyAudio.

 

[Nordo]

You can send signals to the speaker using the headphone output. When measuring with a USB mic REW generates a high frequency timing signal to act as a reference. The difficulty with polar measurements using that is the off axis measurements may not pick up the timing signal well or a side wall reflection may be larger than the direct path. If you are measuring outside the side wall problem goes away, but in any case I suggest you use a separate speaker to reproduce the timing signal so that stays constant while the speaker under test rotates. For the design work VituixCAD is well thought of, there is a very large thread on it at DiyAudio.

Thanks John (a great help as always).
I'm not too sure what you mean by "use a separate speaker to reproduce the timing signal".
Where I'm planning to test the drivers (in their enclosure), is a location where the nearest object is over 25m away, and the ground also drops away in front of the test location.
My only problem is I can't rotate much more than 80-90deg. as I will definitely start to get some relatively close reflectable objects.

I'm following the threads regarding VituixCAD v2 on DiyAudio, as well as HTGuide.
Like you, Kimmo is always willing to support their software by answering what probably appears to be stupid questions from newbies like me.

 

[John Mulcahy]

The idea of the separate speaker is to ensure a stable timing reference that is unaffected by the orientation of the speaker under test. The timing reference spans 5 kHz to 20 kHz, so if it is coming from the test speaker it will be heavily attenuated at the extremes of the off-axis measurements.

 

[dcibel]

I recommend investing in a standard XLR mic and USB audio interface. Set up for measurement will be simple and with high reliability and repeatability. As well, you’ll be able to follow the guide for VituixCAD verbatim. Full dual channel measurement is available in REW 5.20.10 preview using the “timing with cal” measurement option.

 

[Nordo]

Thanks guys.
I had a XLR individually calibrated mic combined with a small mixer that gave me my phantom power.
This was connected to a Sound Blaster external sound card.
But the sound card was always playing up, so I sold the mic and mixer and bought my USB mic., which was still OK at the time as I was concentrating mainly on room measurments or on-axis testing.

Is this basically what you mean when "using a separate speaker"? (from the minidsp website).
Click on the "Measure" button (top left of the main window). Check the frequency sweep range and level. Set Timing (towards the top right of the window that pops up) to "Use acoustic timing reference."

Set both Output and Ref. Output to "L":

Now run a levels check and then the measurement sweep. You will hear a short "peep" from the left speaker, shortly followed by a full frequency range sweep. Rename the measurement to "Front Left".
Click on the "Measure" button again. This time, set Output to "R" but leave Ref Output at "L." You will use this setting for all remaining speakers. It looks like this:

Now run the measurement sweep and label it "Front Right."

Is there anywhere else I can find details on using a reference speaker, and where to locate it, etc.?

 

[John Mulcahy]

Yes, pretty much, but with a separate speaker as the reference it would be on a different channel to the measurement, e.g. reference on L, measurement on R. It doesn't matter too much where it is, easiest is probably to have it a similar distance from the mic as the speaker being measured and pointing at the mic. It can be placed closer if that is more convenient, but you want to avoid picking up reflections from it.

 

[sm52]

Nordo
Selling a good measuring microphone and buying a USB microphone is something you didn't have to do for measurements. And it was necessary to buy an external sound card with two outputs and two inputs. Not cheap. The Sound Blaster is not suitable for measurement.
As for the individual speaker. Your speakers can be used for acoustically synced measurements as long as they have a working tweeter. Or take a separate driver - a tweeter, put it somewhere not very far from the microphone, so that there are no reflective surfaces nearby, connect it to the sound card output and specify this sound card output channel as a reference in REW. And connect the second output channel of the sound card to the measured driver (speaker) and indicate in REW that this is an output (not a reference) channel.

 

[Nordo]

Thanks again guys.
I have some two way speakers left over from my Home Theatre days (recently moved house), so one of those would be perfect.
You have also confirmed that location of the timing reference speaker isn't critical (which was going to be my next question), so long as it doesn't create reflections,.

Regarding selling my previous mic setup, I was only using it to fine tune sound treatment and speaker location in my home theatre, and was getting sick of battling the external sound card I was using.
The USB mic was so much easier to use.
I'm trying not to spend any more money on this project as it is a one-off speaker project, hence the reluctance to purchase new test equipment.

 

[Nordo]

Finally finished the cabinet build and installed the drivers, and now actually ready to take my measurements.
I'm using my UMIK-1 with my Dell laptop.
The headphone outlet is no good for the output to the amp/speaker, so I have bought a U-Control UCA222 USB audio interface with digital output.
i.e. I am using one USB socket for the mic and another USB socket for the output to the amp and speakers.

My problem is I am still none the wiser regarding using the Timing Reference.
The REW instructions talks about 3 different methods, but after reading the instructions several times, I'm still in the dark and have no idea which method to use and how to actually use it.

I will be taking typical measurements on the horizontal plane (0 deg., 10 deg., etc.) and some in the vertical plane, for both the woofer and the tweeter, so I definitely need to use some sort of timing reference.

Any help would be greatly appreciated.

 

[dcibel]

With a USB mic the only option truly available to you is an acoustic timing reference.

 

[Nordo]

With a USB mic the only option truly available to you is an acoustic timing reference.

Yeah, I realised I couldn't do any sort of loopback.
BTW, if it makes any difference, my Dell laptop has two USB 3.0 (which I'm using for Mic and for output to the speakers) and a single USB 2.0.
I have the test speaker on the right channel and my reference speaker (with tweeter) on the left.
I have the reference speaker set for "acoustic timing reference".

My main confusion is the rest of the measurement window settings when incorporating a reference speaker.
Also are there any adjustments I need to do manually after the first test, or is the timing adjusted automatically for each subsequent test?

I've used REW over many years, but until now I've never used it to measure individual drivers for a crossover design.

 

[dcibel]

Acoustic reference sets t=0 for the measurement. In order to retain any consistent timing between multiple measurements, acoustic reference distance in relation to the mic must not change. That is, the mic and reference speaker remain stationary as you rotate the speaker through multiple measurements. Apart from the difference in timing loopback, you can follow the measurement guide for REW at the VituixCAD site for the rest of the process.

 

[Nordo]

Acoustic reference sets t=0 for the measurement. In order to retain any consistent timing between multiple measurements, acoustic reference distance in relation to the mic must not change. That is, the mic and reference speaker remain stationary as you rotate the speaker through multiple measurements. Apart from the difference in timing loopback, you can follow the measurement guide for REW at the VituixCAD site for the rest of the process.

You are a legend.
Everything else I've read covered the 3 timing reference options and just made it confusing.
I definitely intended keeping the mic and ref speaker fixed.
I've printed out Kimmo's instructions for testing using REW, and will be following them.
Of course he assumes you have a "proper" mic and can run dual channel, but I can still follow the main part of the procedure.

I've bought some bits to make up a jig for the impedance testing.
But I need to get through the acoustic testing first.
Thanks again.

 

[dcibel]

Proper mic with electrical timing reference is the preferred method for reliability and repeatability, and for simplicity. Acoustic timing reference requires more setup and is more prone to user error. When moving the mic between drivers you’ll have to be careful to move the acoustic reference as well, if possible it would be better to raise and lower the speaker cabinet instead so that mic and reference can remain constant.

 

[Nordo]

Proper mic with electrical timing reference is the preferred method for reliability and repeatability, and for simplicity. Acoustic timing reference requires more setup and is more prone to user error. When moving the mic between drivers you’ll have to be careful to move the acoustic reference as well, if possible it would be better to raise and lower the speaker cabinet instead so that mic and reference can remain constant.

Apart from what you've told me, I have found very little information on correct settings when using a Timing Reference.
e.g. not sure what to do when REW says that the mic is too low (sensitivity), and I'm not sure how to fix this.
But I am stuck with Acoustic Timing Reference (USB mic),
I have been intending (at least when doing horizontal plane measurements) to keep the mic and reference speaker fixed for all measurements including both tweeter and woofer measurements, even though the mic will be sitting on the tweeter's axis.

I am now realising that my intention to do some vertical plane off-zero measurments may be difficult to achieve, so I'll probably just have to settle for horizontal plane measurements. (0 to 90 deg in 10deg steps).

 

[dcibel]

Spatial data needs to be generated at each driver’s axis, and correct power & DI chart requires 360 degrees of measurement.

 

[Nordo]

Proper mic with electrical timing reference is the preferred method for reliability and repeatability, and for simplicity. Acoustic timing reference requires more setup and is more prone to user error. When moving the mic between drivers you’ll have to be careful to move the acoustic reference as well, if possible it would be better to raise and lower the speaker cabinet instead so that mic and reference can remain constant.

Apart from what you've told me, I have found very little information on correct settings when using a Timing Reference.
e.g. not sure what to do when REW says that the mic is too low (sensitivity), and I'm not sure how to fix this.
I am stuck with Acoustic Timing Reference (USB mic),
I have been intending (at least when doing horizontal plane measurements) to keep the mic and reference speaker fixed for all measurements including both tweeter and woofer measurements, even though the mic will be sitting on the tweeter's axis.

I am now realising that my intention to do some vertical plane off-zero measurments may be difficult to achieve, so I'll probably just habe to settle for horizontal plane measurements. (0 to 90 deg in 10deg steps).

Spatial data needs to be generated at each driver’s axis, and correct power & DI chart requires 360 degrees of measurement.

How important is it to achieve full Power and DI charts when using VituixCAD, i.e to measure passed 90deg. horizontally?
I intend taking all my measurements outside in front of my garage. The driveway slopes away and the nearest reflective object is around 30m away.
But once I measure passed 90deg. I'll start to get some near field reflections.

 

[dcibel]

How important is it to achieve full Power and DI charts when using VituixCAD, i.e to measure passed 90deg. horizontally?

That's really up to you to decide. Data may be perfectly "usable" for design, but won't be representative of the CTA-2034 standard that this chart is meant to represent. Suggest reading through the CTA-2034-A standard for more information on how this information is derived. You may also load up the example files available in VituixCAD help, compare using full 180 degrees of driver data vs only 90 deg.

I intend taking all my measurements outside in front of my garage. The driveway slopes away and the nearest reflective object is around 30m away.
But once I measure passed 90deg. I'll start to get some near field reflections.

I don't understand this comment. If the mic is stationary, distance of reflections remains the same regardless of which direction the speaker is pointing. Unless you've hoisted your speaker 30m in the air, the nearest reflective object is your driveway. Unless your speaker is enormous, such distances aren't strictly necessary anyway, many people including myself measure indoors where measurement window is limited to about 4ms, which is fine until you get to large drivers 10"+.

 

[dcibel]

For a normal small 2-way monopole speaker, you can see there is not much difference in overall shape of power & DI, however the slope is quite affected so it cannot be directly compared to other "spinorama / CTA-2034-A" data. My thought has always been that if you have the mic set up and went to 90 degrees, just take the extra couple minutes and measure full 180 degrees and do it right.

 

[Nordo]

I don't understand this comment. If the mic is stationary, distance of reflections remains the same regardless of which direction the speaker is pointing. Unless you've hoisted your speaker 30m in the air, the nearest reflective object is your driveway. Unless your speaker is enormous, such distances aren't strictly necessary anyway, many people including myself measure indoors where measurement window is limited to about 4ms, which is fine until you get to large drivers 10"+.

Sorry, I'm using my layman's logic.
Considering the horizontal plane from 0 to +/-90 degrees, there will be nothing in front of the speaker, apart from the driveway, and the driveway slopes down away from the speaker's test location (i.e. any close reflective objects are behind the speaker).
But once you move the speaker passed the 90 deg. orientation, there will be objects (like the front of the garage) that will receive the direct radiation from the driver being tested.
I just didn't want to include measurements where I know there will be very strong reflections from nearby objects.

 

[dcibel]

If the speaker is 1m off the ground, and 1m from the garage door, the ground is still the nearest reflection surface. At 1m mic distance it'll be about 3.6ms for the ground reflection, and more like 6ms for the door reflection to hit the mic.

You may find that the reflected sound from the garage door is much louder than the direct radiation from the speaker pointed away from the mic, but that's why you have that important timing reference.

 

[Nordo]

If the speaker is 1m off the ground, and 1m from the garage door, the ground is still the nearest reflection surface. At 1m mic distance it'll be about 3.6ms for the ground reflection, and more like 6ms for the door reflection to hit the mic.

You may find that the reflected sound from the garage door is much louder than the direct radiation from the speaker pointed away from the mic, but that's why you have that important timing reference.

Are you saying that with the speaker facing away from the garage (I'll probably also leave the roller door open), there will still be significant reflection from the garage reaching the mic?

I've added a photo of my proposed measurement location. The speaker will be sitting at the front of where the car is at the moment.
I've also added a photo of one of my speakers (with the reference speaker on top). When measuring, the reference speaker will be on a separate stand next to the speaker being measured.

 

[dcibel]

I've provided you the CTA-2034-A standard so you can read about the impact and weight of each measurement angle. As well I've pointed you to the example files for VituixCAD so you can see what real world measurements look like at each angle, frequencies and amplitudes of rear facing sound vs forward facing.

Measuring outdoors the bigger problem will be wind and annoying the neighbors. You've plenty of room to get a long measurement window, You've got the mic, you've got a speaker, and plenty of room to get a long measurement window at all angles. It's time to take some real measurements and get some hands on experience. I think you have enough information to decide on the impact of cutting corners in the measurement process.

 

[Nordo]

I've provided you the CTA-2034-A standard so you can read about the impact and weight of each measurement angle. As well I've pointed you to the example files for VituixCAD so you can see what real world measurements look like at each angle, frequencies and amplitudes of rear facing sound vs forward facing.

Measuring outdoors the bigger problem will be wind and annoying the neighbors. You've plenty of room to get a long measurement window, You've got the mic, you've got a speaker, and plenty of room to get a long measurement window at all angles. It's time to take some real measurements and get some hands on experience. I think you have enough information to decide on the impact of cutting corners in the measurement process.

Thanks so much for all the advice and the patience.

I've downloaded the CTA standard and will try to digest it fully.
I have been worrying about the wind (and birds, dogs, etc., etc.), and at the moment I'm waiting for the rain to stop.
But you're right. I've got to bite the bullet and do some testing.

This is just a one-off project, hence my tardiness to try to fully comprehend all of the theory behind speaker design.
I'm also hopeless at cabinetwork, and find that part of DIY speakers fairly painful.
But if I was going to take on speaker design and building as a ongoing hobby, I would buy better equipment - for the cabinetowrk as well as the audio design.
Thanks again for all your help.