Amplitude testing question

[Robert Lofgren]

Re: Impedance testing question

Have you actually done that?

Driving the speaker with 20V may get louder than is comfortable, and require a power amp, not just a signal generator.

I actually built this impedance feature into a piece of value test equipment I designed and sold back in the '80s.. I used a modest stimulus voltage (0dBu) and only a 1/4W sense resistor. I did have a precision voltmeter (actually a dB meter). It was test equipment after all...

JR

Maybe not obvious in my post but I assume your 50ohm resistor inline with the speaker and the bridge over the resistor. This will drive the 8ohm speaker with rougly 1.5Watt, somewhat loud but still within reason. Voltage should be able to be decreased down to 5v without any issues.

it is very long time since I was doing analog circuit design so I blame age if I say something wrong

 

[Art Welter]

Re: Impedance testing question

Hey Art,

If you look at the specs from Radian Audio, the response chart for the 5312 coax shows the ~10db dip at just under 1Khz that I talked about. This is caused by the natural rolloff of each driver. I tested them both full range and they are quite a few db down at 1khz. I find it a bit surprising that Radian's passive crossover can't bring up the response in the x-over region some more, but maybe it's really a driver limitation as opposed to a summing error.

Simon,

Your previous statement that an an "8db boost gives me a 2db result" would suggest a time alignment problem, the driver does look quite deep compared to most co-ax.
If the drivers were summing properly, an eq boost should result in an equal amplitude gain, assuming the boost is accurate.
Does an 8 dB boost at say 200 or 2000 Hz result in an 8 dB amplitude change?
The Radian spec sheet shows only the summed response using a passive crossover which may or may not use time correction which is difficult and expensive to implement.
I was surprised to see as big a hole as you described.
A picture is worth 1000 words, I'm interested in what you mean by "they are quite a few db down at 1khz".
With time correction and some EQ in both bands, if they are each only around 6 dB down at 900 Hz, flat phase and amplitude should be possible.
If the LF and HF are both 15 dB down at 900 Hz, could be a different story- raw response dictates what is possible.

Art

 

[Per Søvik]

Re: Impedance testing question

Here's what I do for measuring resonance frequency etc:
I take a 10 m speakon lead, connect a thin lead to the 1+ terminal on both ends, and feed those two connections to pin 2 and 3 of an xlr. Plugging that xlr into an interface or mixer, I get about -15 dBu at 100W into an 8 ohm speaker. The signal will be a measurement of the current going through the standard speakon lead, so will be a true representation in actual working conditions.
(Caution: make a mistake when making up the lead, and the potential for frying an input is fairly real)

Since few amplifiers will deliver a 100% stable voltage irrespective of load, for accurate measurement, one also need to log the voltage across the speaker terminals for accurate measurements of impedance.

 

[Simon Coffin]

Re: Impedance testing question

Hey Folks,
I finally captured some response graphs and starting working on what I see as the trouble areas. Unfortunately the demo time period for Fuzz Measure ended and I lost the last few graphs I made. Once I purchase the full version I'll rerun them.

The attached freq graphs show the full range response of each driver nearfield. The dip and spiking around 200-400hz was caused by reflections between the grill supports that I have since damped. The spike around 2.4khz appears to be a horn issue. The original Microwedge dsp settings have a -6db EQ here, so I imagine this can be dealt with. The port measurement was taken with the mic (a DBX measurement mic) flush to the port opening.

Hopefully later this week or this weekend I can rerun the tests and start to work on the crossover settings.


Ciao
Simon

 

[Art Welter]

Re: Impedance testing question

The HF driver has good response in the 900 Hz range, but the woofer appears to be dropping at around 24 dB per octave from 700 Hz.

Where was the mic position relative to the horn and woofer in the test?
Did you short out the unused driver in the tests?

 

[Simon Coffin]

Re: Impedance testing question

Hey Art,

The mic was positioned about 3 inches off the dustcap, pointing straight down the horn. No, the unused driver was not shorted, just muted in the dsp.

Given this response, I'm thinking a x-over in the 700hz range would be a good place to start. If the drivers sum well, I may be able to push it a little higher, but the woofer really falls of the cliff above 900hz. Given the choice, what would you prefer, to push the woofer higher, or drive the horn lower?

Ciao
Simon

 

[David Morison]

Re: Impedance testing question

No, the unused driver was not shorted, just muted in the dsp.

Presumably still connected to an amp channel that was powered up though, as you mention the muting being done in the DSP?
If so, the low output impedance of the amplifier will provide most of the damping that a dead short would.
Cheers,
David.

 

[Art Welter]

Re: Impedance testing question

Hey Art,

The mic was positioned about 3 inches off the dustcap, pointing straight down the horn. No, the unused driver was not shorted, just muted in the dsp.

Given this response, I'm thinking a x-over in the 700hz range would be a good place to start. If the drivers sum well, I may be able to push it a little higher, but the woofer really falls of the cliff above 900hz. Given the choice, what would you prefer, to push the woofer higher, or drive the horn lower?

Ciao
Simon

Given the measurement distance/position I doubt the woofer upper range measurement is accurate, as the mic may be in a null point.

With a 3" HF diaphragm, going as low as 700 Hz at 24 dB per octave probably won't be a problem.
The LF may not need a steep crossover to match, since it already rolls off steeply.
If possible, I'd try to make the acoustic cross over where the off-axis dispersion between the LF and HF matches.

 

[Simon Coffin]

Re: Impedance testing question

Hey Art,

I was puzzling over where to place the mic in order to get an accurate near field measurement. Closer seemed better with less room interference, but its hard to tell what is direct and what is reflected in my less than acoustically perfect basement. It's been so cold outside I haven't been able to get outside, making accurate tests difficult. If the weather clears up for a day or two I hope to take everything into the garden and rerun some of the basic tests and see what I get. The woofer may very well be able to go higher, we'll see.

I was curious about the dispersion angles of the two drivers as well. Getting the x-over point where the angles match makes a lot of sense as that gives an even soundfield. How would you go about measuring the angles? Just choose a likely freq and do a number of tests starting at say 60 degrees?

Incidently, my measurement setup uses a Macbook Pro running FuzzMeasure, connected to a Presonus StudioLive via firewire and an old XP laptop controlling the SpeakerPower dsp. Getting everything setup and working was a project in itself, but now that I know what I'm doing, it's pretty easy to run tests and make changes to the dsp.

Cheers,
Simon

 

[Simon Coffin]

Latest results

Hey Gents,

Had some time on saturday afternoon for more tests and got some interesting results.

After confirming the raw response of both drivers I started working on the eq and xover settings. This turned out to need a lot of trial and error to get the Q and xover points where I wanted them. The app to define the eq and xover filters is not that intuitive and take some time to figure out. After a while I started to get them behaving more or less how I wanted.

With the eqs and xovers applied and with both drivers on, but no delay, I got a huge 20db dip at around 657hz. This told me the drivers were summing destructively at this point. With a bit of math I figured out what a half wavelength of delay would be at 650hz and applied that to the lows. The results are pretty reasonable. After some more tweaking to the low response I ended up with the final graph.

Right now the wedge sounds a tad bright to me, vocals seem to be a bit out front, but this may be a good thing in a monitor. The lows are full and detailed and do a surprisingly good job with kick drum or bass. Once I get some more time, I'm probably going to bring the highs down a db or two and see what that does. Since the Speakerpower module can save two dsp setting, I'll probably set one as vocal specific, and another for more general listening.

More results to come whenever I get a chance.

Cheers
Simon

 

[Art Welter]

Re: Latest results

I was curious about the dispersion angles of the two drivers as well. Getting the x-over point where the angles match makes a lot of sense as that gives an even soundfield. How would you go about measuring the angles? Just choose a likely freq and do a number of tests starting at say 60 degrees?

With the eqs and xovers applied and with both drivers on, but no delay, I got a huge 20db dip at around 657hz. This told me the drivers were summing destructively at this point. With a bit of math I figured out what a half wavelength of delay would be at 650hz and applied that to the lows. The results are pretty reasonable. After some more tweaking to the low response I ended up with the final graph.

Right now the wedge sounds a tad bright to me, vocals seem to be a bit out front, but this may be a good thing in a monitor.

Did you try reversing HF polarity before delaying the lows?
Generally the HF needs delay, not the LF.
You may have delayed the LF by a wavelength (or two) which can make the response look smooth, even though the LF is lagging behind.

The Haas effect says what we hear first sounds louder, if the HF precedes the LF, it will sound brighter.
Can your system measure phase?

As far as the dispersion tests, use the same method as you are using now, though you should measure outdoors with the wedge on the ground and start on axis with the mic around head height.
Use a tape measure to the center of the HF horn, and move the mic off axis in 5 degree increments out to 50 or 60 degrees off axis, maintaining the same distance and angle to the face of the wedge.
Measure the raw response of both drivers (the level only needs to be above the ambient noise, a watt is probably plenty for the HF driver) then compare the charts, looking for the angle that both show -6 dB less response than their greatest response.

The LF will get more narrow in dispersion at higher frequencies, the HF will get wider at lower frequencies. The HF may have a slight on axis dip at some frequencies.

 

[drew gandy]

Re: Latest results

Art, I need to do some more experiments to confirm some things but I believe that Fuzzmeasure calculates the phase from the magnitude so it does get confused with fast phase flips. Several years ago I found this out when measuring a speaker with some dramatic phase at the crossover points. Spectrafoo showed the phase that I expected. Fuzzmeasure showed a surprisingly flat phase that I knew was wrong. Regardless, Fuzzmeasure does not have a way (that I've found) to show phase and magnitude together at the same time so it takes some cut and paste work (and other hassles) to put them together.

btw, I agree that you would likely need to delay the highs (with 4th order slopes especially) rather than the lows. I'm not sure about Art's Haas theory concerning brightness but I've certainly had the same thoughts from similar experiences. We need some research grant money to find out.

 

[Simon Coffin]

Re: Latest results

Hey Art,

On this coax the lows need to be delayed because the high driver is mounted to the back of the low driver and fires through the pole piece. Radian Audio's website doesn't have them anymore, but the old Microwedge dsp settings showed a delay on the lows as well.

No, I didn't try flipping the polarity. The amp module is held in with a dozen screws so it was easier to add the delay.

As drew mentions below, Fuzzmeasure doesn't seem to be able to measure phase very well. The big dip at 657hz was actually a nice tidbit of information because it told me what the phase was doing at that point, namely 180 out or very close to that. I want to play with delaying the highs a bit to see if I can increase the size of that dip, just to be sure I know where the 180 out point is, then delay the lows half a wavelength.

One day if I can borrow a Smaart rig I'll confirm where the phase is.

If it ever warms up outside I'll also get the dispersion measurements. Given the weather report, that will be in April.......

Cheers,
Simon

 

[Art Welter]

Re: Latest results

Hey Art,

On this coax the lows need to be delayed because the high driver is mounted to the back of the low driver and fires through the pole piece. Radian Audio's website doesn't have them anymore, but the old Microwedge dsp settings showed a delay on the lows as well.

No, I didn't try flipping the polarity. The amp module is held in with a dozen screws so it was easier to add the delay.

If you can't switch polarity in the DSP, I'd strongly suggest pulling the amp and covering the hole while you do your testing so you can check polarity.
The crossover hole without delay looked worse than Radian's passive crossover results, indicating that you probably need a HF polarity reversal, then a minor delay.
Given the depth of the driver, I could see delaying the LF slightly, in the .1-.3 ms range, but I'm guessing you delayed a wavelength +.

What delay time, crossover filters & PEQ did the Microwedge use, and what are yours?

 

[Ivan Beaver]

Re: Latest results

Hey Art,

On this coax the lows need to be delayed because the high driver is mounted to the back of the low driver and fires through the pole piece.

That is not always the case. In many cases with a coax that has the HF on the rear of the woofer-the HF (yes the highs) need to be delayed.

I know it sounds odd-and goes against "common sense", but the low pass filter and high inductance of the woofer often cause the woofer to have more delay than the physical placement would dictate.

Therefore the highs need to be delayed.

Now this is not always the case-but has been in a number of coax drivers that I have measured-in order to get the best phase and amplitude response.

PROPER measurement/alignmnet is the only way to find out for sure.

 

[Mark DeArman]

Re: Latest results

That is not always the case. In many cases with a coax that has the HF on the rear of the woofer-the HF (yes the highs) need to be delayed.

I know it sounds odd-and goes against "common sense", but the low pass filter and high inductance of the woofer often cause the woofer to have more delay than the physical placement would dictate.

Therefore the highs need to be delayed.

Now this is not always the case-but has been in a number of coax drivers that I have measured-in order to get the best phase and amplitude response.

Yep totally agree! Seen it many times. Normally with a DSP, especially with a HP filter, there is a huge amount of excess delay. Everyone should realize that when you place a 35Hz-60Hz HP on a woofer it adds delay.

PROPER measurement/alignmnet is the only way to find out for sure.

Double totally agree.

 

[Simon Coffin]

Re: Latest results

Hey Ivan,

You are of course correct, accurate measurement always trumps theory. Unfortunately, since I don't have an accurate phase trace, I have to infer what it's doing from the measurements I do have. Which actually makes this little exercise all the more fun because I have to start thinking beyond the graphs and try to understand what is going on with the drivers and how they are summing through the crossover region. Delaying the lows makes sense given the drivers physical setup and other dsp settings for this driver also show the lows delayed, so there is at least a presedence. It may be wrong, but at least I'm not just guessing.

My results may end up being way off, but I'm certainly learning a lot.

If someone with a Smaart setup is ever in the central PA area and willing to spend an afternoon making noise, I'll fill the fridge with beer and fire up the bbq.

Cheers,
Simon

 

[Ivan Beaver]

Re: Latest results

Hey Ivan,

You are of course correct, accurate measurement always trumps theory. Unfortunately, since I don't have an accurate phase trace, I have to infer what it's doing from the measurements I do have.

One of the things I always look for when doing an electronic crossover (passive can be different) is how well the different passbands sum together.

I save a trace of each passband by itself and then the summed response. I look an octave above and below the crossover point and see if I am getting summation.

Very often you will summation at crossover (lets say 100hz) but then cancellation notches at higher and lower freq. This is a good indicator that the delay is not set properly.

And you may get summation-but is the the maximum you can get? Sometimes you get so much summation you need to change the crossover points to "get rid of the excess level" buy raising or lowering one. I usually raise the highpass.

Of course once you change the crossover freq-the delay time will also change-so the summation will be different.

It is simply "a process" that you go through.

Sometimes it is pretty quick and easy-other times not so easy.

Of course when setting delay times-the only type product that will end up with a "proper" delay is a coax or a truly single point speaker.

If you have a separate woofer and horn-the physical distance will change as you move around-and therefore the delay time will need to be different.

So you have to choose a point and call it good. Generally this is somewhere on axis-on axis of what is often the question------------