That is agreeable. To low would show signs of non linear distortion. I think we have already dissected that issue. 200hz XO may have been pushing the limits as far as Linear Transient performance was concerned. Was never an issue for average spl. Raising the XO only increases headroom. At 220hz summing is still fine to my ears. I do not like the height of the tweeter but I will work on getting it lower plus I grow use to it over time. The major change has been the room. I went from a small living room with a loudspeaker placed away from walls to this corner loaded situation and it is Not the same. The difference in off axis character may be really showing its self in this situation. I am still tweaking things but the bass seems heavy regardless looking to be voiced flat....I keep giving my microphone the side eye, as the flat measurements do not seem to line up with perception, but it we've talked about how decay affects perception so I cannot say I should be shocked.... Whats really confusing is that if I measure levels right at the drivers, its obvious the woofers are at a much higher level than the tweeter, if I match levels this way, it sounds more balanced, but when taking measurements at the listening spot, it looks bass shy.... and then if I flatten things out at the listening spot, no it seems like its bass heavy. a work in progress....
Heres now
Heres the previous testing upstairs. (ignore the HF roll off justy a quick voicing to see cross band)
I had the system voiced flat for most of the testing, upstairs, but this is what I have.... Below is shot of the just horn when it was upstairs.
Either way, things are much dryer down here.... until the rom modes kick in.
Last edited:
Camplo is still considering crossing the high frequency driver at 220Hz.
Two years ago, a few weeks after we were discussing SPL (your own hand clap reaching 125dB peak, etc.) you posted this graph showing at 220Hz, the high frequency horn/driver's second harmonic distortion is -22dB from the 110dB fundamental, ~8% distortion.
As I recall, you still decided that was enough for your revised design goals, though one decent front loaded 15" would easily have +20dB more output with less distortion at that SPL level.
I still think the IMD of a HF compression driver forced to cover two decades of bandwidth at anything approaching live acoustic instrument peak SPL sounds terrible compared to what it sounds like if the cone drivers below "take the weight off".
Art
Last edited:
@weltersys Thank yo for debating me, I do appreciate it.
We also had the discussion surrounding incorrect thd measurements of horns due to low level like that in the cutoff. That measurement was also 3db above target. It does have the actual 220hz response
I wanted to reach 115db per channel, it made sense. As the challenges unfolded I started back off. Maybe 115db accumulated would be good enough, I said. That only requires 112db from the each channel approximately. I also starting investing gating thd taken without the horn but with appropriate voltage. I think the pic below is near 3volts... Putting max linear performance closer to 110db. per channel... to which then I said... well 95db is max safe average....plus 15db 110db walla. I only care about 1meter listening as it is.
Moving forward I still wanted to try and get to the goal anyway. I mean maybe I would like it or it would make a real world difference, consider that I concluded no real world deficit with the original arrangement.
Lately Ive been working here. Effectively crossing somewhere around 280hz or so. I'm still learning the proper terminology. Even though my filter is much lower, the effective roll off and knee is much higher.
Here there is likely no issues with headroom, given the size of the horn and lf extension of the driver.
Right now I am baffled by what my measurements show versus what I hear... possibly my kids goofed my measurement mic, then I looked at prices of the umik-2 and thought, maybe I better make sure lol...
You guys know a sure fire way of testing a measurement mic?
We also had the discussion surrounding incorrect thd measurements of horns due to low level like that in the cutoff. That measurement was also 3db above target. It does have the actual 220hz response
I wanted to reach 115db per channel, it made sense. As the challenges unfolded I started back off. Maybe 115db accumulated would be good enough, I said. That only requires 112db from the each channel approximately. I also starting investing gating thd taken without the horn but with appropriate voltage. I think the pic below is near 3volts... Putting max linear performance closer to 110db. per channel... to which then I said... well 95db is max safe average....plus 15db 110db walla. I only care about 1meter listening as it is.
Moving forward I still wanted to try and get to the goal anyway. I mean maybe I would like it or it would make a real world difference, consider that I concluded no real world deficit with the original arrangement.
Lately Ive been working here. Effectively crossing somewhere around 280hz or so. I'm still learning the proper terminology. Even though my filter is much lower, the effective roll off and knee is much higher.
Here there is likely no issues with headroom, given the size of the horn and lf extension of the driver.
Right now I am baffled by what my measurements show versus what I hear... possibly my kids goofed my measurement mic, then I looked at prices of the umik-2 and thought, maybe I better make sure lol...
You guys know a sure fire way of testing a measurement mic?
I agree that in sound reinforcement, there are better choices than using a compression driver very low at peak SPL, although there are drivers that can handle a low crossover point. And I am familiar with your extensive experience in this field and certainly respect that. For modest SPL like audio at home it is no problem to cross the CD low if that brings benefits.
Could you repost the above graph with higher resolution spl scale, please? It will be more revealing to see what happens. And smoothing no more than 1/6th, preferably 1/12th?
@Horneydude Surely you don't mean every measurement, any one in particular? These are taken at my usual ~1m away listening position
I think I have fixed some of my issues concerning why things sounded off... I didn't exactly voicer the system as I new how.... I started with at listening position measurements. That, and I don't think LF shelfs being used to counteract the Lf roll off of the drivers, without a high pass was the wisest. So this time I just use a peak filter around 30hz and another one to squash the typical HF rising of a low qes driver. THis is a measurement taken right at the front of the slot. After adjusting levels to match the horn at the listening position things sounds a lot better than whatever I was doing before.
This is a voicing I am playing with... subject to change every 15 minutes...
All Measurements raw accept for the usual smoothing on distortion graphs.
I think I have fixed some of my issues concerning why things sounded off... I didn't exactly voicer the system as I new how.... I started with at listening position measurements. That, and I don't think LF shelfs being used to counteract the Lf roll off of the drivers, without a high pass was the wisest. So this time I just use a peak filter around 30hz and another one to squash the typical HF rising of a low qes driver. THis is a measurement taken right at the front of the slot. After adjusting levels to match the horn at the listening position things sounds a lot better than whatever I was doing before.
This is a voicing I am playing with... subject to change every 15 minutes...
All Measurements raw accept for the usual smoothing on distortion graphs.
Thanks for the additional graphs, I did not mean every measurement, just the latest one. But it looks a lot better now, don't give up and enjoy the ride! I am glad you find it sounding a lot better too, congrats. Looking forward to more reporting👍
Something is wrong
THis is a measurement of my small two way... If I were to make this flat, It would, on screen, match the voicing of my large speaker. I know that this speaker is voiced relatively flat ~100hz and tp 19khz.... Because this aligns more with my experience. I can conclude that my microphone isn't working properly anymore. I ruled out the operating system by starting up in an older one and running REW. The spl meter on my phone also did not support the measurements. I wanted a new microphone but but not by force.... lol
THis is a measurement of my small two way... If I were to make this flat, It would, on screen, match the voicing of my large speaker. I know that this speaker is voiced relatively flat ~100hz and tp 19khz.... Because this aligns more with my experience. I can conclude that my microphone isn't working properly anymore. I ruled out the operating system by starting up in an older one and running REW. The spl meter on my phone also did not support the measurements. I wanted a new microphone but but not by force.... lol
Its just a normal sine sweep measurement that REW does.... The small two-way measured, is voiced flat. Not by me but by the factory. I've measured it before and this isn't what it should look like. The HF is hyped in this measurement and the low end is weak. SO when I use this to voice my large speaker I end with attenuated highs and hyped lows... thats exactly what I've been hearing. That also means I have even more headroom than seen so far,
Z scale? I thought you were using a logarithmic which would show a rise as the frequency increases, power density increases per octave.
Rob 🙂
Too flat of a rising response over the entire frequency spectrum to be a microphone defect.
To rule out the "operating system", you need to measure line in/out directly (no microphone) it should appear as a straight line with an upward, flat or downward slope depending on the signal and measurement screen used.
Your flat two-way speaker has a rising response with whatever filter you have placed in series with it.
The response looks similar to what might be used to flatten the upper response of your AXI2050 horn/driver.
White noise on a linear bandwidth screen would appear to be flat, as it has equal energy per frequency.
Pink noise (filtered white noise for equal energy per octave) would appear to drop at -3dB per octave.
White noise on a logarithmic scale RTA screen (as you are, or should be using) rises at +3dB per octave, pink noise or swept sine would be flat.
Your "flat" speaker is rising at about 4.25dB per octave, ~30dB per decade.
Good luck sorting out what your measurement/voicing errors are due to.
Art
its just a normal rew measurement. In my experience log vs linear usually applies to frequency where as amplitude can be scaled identically. Log or linear display of frequency can be used with 1/1octave or 1/3octave scaling. Non the less... REW doesn't offer a change for its normal measurements screen... it is 1/1 octave scaling with logarithmic frequency scaling to which white noise appears flat.... thats how its always been.
There is no "normal" presentation, REW offers a choice between "RTA" (Real Time Analyser) or "Spectrum" mode.
Using "Spectrum", rather than "RTA" will make the measured response look like it rises 10dB per decade when it is flat.
Looks like you have selected "Spectrum", rather than "RTA" for your presentation of the measurement of your desk monitors in post 14,348.
You should be using RTA and pink noise to try to match your speakers response to them.
The REW manual explains why:
The different presentations mean the same signals will look different on the plot.
In spectrum mode the frequency content of the signal is split up into bins that are all the same width in Hz. White noise has the same energy at each frequency.
On a spectrum plot, which shows the energy at each frequency, the white noise plots as a horizontal line.
On a RTA plot white noise appears as a line that rises with increasing frequency, as each RTA bin gets wider it covers more frequencies and so has more energy.
Pink noise has energy that falls 3 dB with each doubling of frequency.
On a spectrum plot the pink noise line falls at that 3 dB per octave rate, on an RTA plot it is a horizontal line as the energy in the signal is falling at the same rate as the bins are widening.
In RTA mode the bin widths are an octave fraction, so their width in Hz varies with the frequency. The plot shows the combined energy at each frequency within each bin. This is closer to how our ears perceive sound.
On an RTA plot white noise appears as a line that rises with increasing frequency, as each RTA bin gets wider it covers more frequencies and so has more energy.
White noise sounds "hissy", as it has more energy at higher frequencies, 10,000 "units of noise" in the octave between 10kHz and 20kHz, while there are only 100 "units of noise" between 100 and 200Hz .
Anyway, your mic is not responsible for your proper mode selection.
Or your choice of of a proper crossover frequency...
Art
Oh really, and where do I go to adjust the setting?
I am not talking about the RTA but the measurements display.
Everything else you are explaining about the rta is old news.... being able to adjust scaling on the measurements tab is new news for me. To be specific
I press the measurements button to start the process
Then I go over to the spl/phase to view results, which is the norm, btw...
And you suggesting that the settings for RTA are going to change my results here? Do tell... I will check and see when I get home. I hope you are right.
Last edited:
Since your swept sine measurement display looks like white noise, which it shouldn't with a flat response speaker, you need to determine why.
Since that response does not look like a microphone defect, you need to measure line in/out directly (no microphone) to see what the interface frequency response looks like.
There are many computer settings that could affect interface response.
For all we know, an RIAA equalization curve (or similar filter) may have been inserted.
Anyway, until you have tested the interface, it is premature to blame the mic.
Good luck sorting out what is causing your measurement anomaly.
Art