Got a little time to try this out today with Baffle Step modified XO. with results below. I have to admit , this testing stuff gets to be fun in its own right. I'm sure that I'm only scratching the surface of this Clio box's capabilities , with MLS and Log Chirp looking like they might be worth learning how to use with averaging of multiple samples in this very noisy environment but the learning curve is steep and I have to use the time with it as best I can before returning to owner.
So anyway, here's what I did today.
First is both drivers with XO (white trace is THD).
Next is impedance of both with xo .
Next two are drivers alone at lower volume (but still at 1 Metre) to protect the AMT(The distortion for the AMT in blue was easy to see on the Clio's laptop but I can't see it on mine.)
Edit: Added expanded view of AMT distortion. Now I see reason why crossing it higher is recommended by some.
Thanks
So anyway, here's what I did today.
First is both drivers with XO (white trace is THD).
Next is impedance of both with xo .
Next two are drivers alone at lower volume (but still at 1 Metre) to protect the AMT(The distortion for the AMT in blue was easy to see on the Clio's laptop but I can't see it on mine.)
Edit: Added expanded view of AMT distortion. Now I see reason why crossing it higher is recommended by some.
Thanks
Attachments
Last edited:
I have a couple of questions about the results so far , not certain it's best to continue here but as I'm still working with L-Pad on series XO I'll go ahead and post.
First is the big null at 90Hz. shown in the first of the images directly above but also new one below. I tried some other tests with the mic farther away from the speaker and it's still there. Would that be caused by cancellation related to the distance of the woofer to the floor?
Also. I got a chance to explore the Clio a little , trying the MLS and LogChirp functions and now wonder what the best way to test a speaker is, Sine sweep, MLS or Log Chirp. I lean away from the sine as the other two offer the option of averaging multiple sweeps and as I'm testing in an apartment with uncontrollable noises of all types I believe averaging of a bunch of samples is most likely to minimize the influence of the noises on the result.
The manual is long and fairly complex and I would focus on one of the three if you have a recommendation for what is generally the most useful (for this or other projects in future. )
Thanks very much.
First is the big null at 90Hz. shown in the first of the images directly above but also new one below. I tried some other tests with the mic farther away from the speaker and it's still there. Would that be caused by cancellation related to the distance of the woofer to the floor?
Also. I got a chance to explore the Clio a little , trying the MLS and LogChirp functions and now wonder what the best way to test a speaker is, Sine sweep, MLS or Log Chirp. I lean away from the sine as the other two offer the option of averaging multiple sweeps and as I'm testing in an apartment with uncontrollable noises of all types I believe averaging of a bunch of samples is most likely to minimize the influence of the noises on the result.
The manual is long and fairly complex and I would focus on one of the three if you have a recommendation for what is generally the most useful (for this or other projects in future. )
Thanks very much.
Attachments
90Hz is most likely a floor bounce.
It's almost impossible to test speakers indoors and get some reliable results, especially at the low end. You need to place the mike very close to the cone to get any meaningful reading and still the great outdoors are your best option.
Don't know about CLIO, never used it. Played with ARTA, REW and HOLMImpulse. The last one is the easiest to use.
Nowadays only use Spectroid, a small Android app. It's enough for me.
It's almost impossible to test speakers indoors and get some reliable results, especially at the low end. You need to place the mike very close to the cone to get any meaningful reading and still the great outdoors are your best option.
Don't know about CLIO, never used it. Played with ARTA, REW and HOLMImpulse. The last one is the easiest to use.
Nowadays only use Spectroid, a small Android app. It's enough for me.
I wouldn't expect 90Hz to be any single surface per se, the room is quite involved with these lower frequencies.
MLS is a useful noise signal because unlike regular noise, it's possible to establish timing. The same can be done with a sine sweep, but with a greater S/N ratio.
MLS is a useful noise signal because unlike regular noise, it's possible to establish timing. The same can be done with a sine sweep, but with a greater S/N ratio.
After posting I found this chart and measuring the height it seems to correspond. The other nulls I will have to learn more to figure out.
Stanislav - I heard your comment about the two drivers not being a good fit for a 2-way and if I end up keeping the woofers here for longer will consider either the mid for a 3-way. One issue would be that any mid would have to be in its own enclosure and just stacked on top of the woofer's cab. This makes looking at a different HF driver for a 2-way more attractive to me.
When Pano helped me with a crossover some years ago I used REW , Holm and FuzzMeasure but he guided me through everything and on my own am not particularly good with Windows directory stuff, so the closer to a plug in the mic and push a button I can get , the better. This Clio's not bad that way , though I have no idea if I'm doing what's needed to set it up properly.
AllenB- Thanks for the direction. I'll learn what I can about using the MLS. One of the settings in the menu is size and window - default is 16K and Rectangular - Good to stay with that? (Options are size from 4K to 2M and for window Hanning, Hanning/2, Blackman, Blackman/2)
Stanislav - I heard your comment about the two drivers not being a good fit for a 2-way and if I end up keeping the woofers here for longer will consider either the mid for a 3-way. One issue would be that any mid would have to be in its own enclosure and just stacked on top of the woofer's cab. This makes looking at a different HF driver for a 2-way more attractive to me.
When Pano helped me with a crossover some years ago I used REW , Holm and FuzzMeasure but he guided me through everything and on my own am not particularly good with Windows directory stuff, so the closer to a plug in the mic and push a button I can get , the better. This Clio's not bad that way , though I have no idea if I'm doing what's needed to set it up properly.
AllenB- Thanks for the direction. I'll learn what I can about using the MLS. One of the settings in the menu is size and window - default is 16K and Rectangular - Good to stay with that? (Options are size from 4K to 2M and for window Hanning, Hanning/2, Blackman, Blackman/2)
IIRC longer windows are needed to manage lower frequencies. The windowing type is a compromise which tends to trade, for example, the precision in frequency against amplitude. If unsure there's usually one recommended.
When you say IIRC, I wonder if that means you don't use tests like these anymore.
I'm not looking to become more knowledgeable than I need to, just follow the recommendations of those who know what to do. It gets to be a bit much. Searching on window type, it looks like firm ground is small.
https://audioroundtable.com/misc/Mueller.pdf
I could think that at my level they're all good enough but the one reservation I have started with seeing the large variation I got when measuring impedance with different levels of resolution. The peaks were at the same frequencies but the level varied from 60 Ohm to 130 Ohm. That's really what got me looking for more technical advice.
Below are various sweeps at different resolutions and different windows. (Rectangular , Hanning, Hanning/2, Blackman, Blackman/2) First four are with mic close to woofer. Hardware set-up unchanged.
Last image is a compression driver on a small horn. Sine sweep 1/6 Octave resolution are the very top trace and THD Dark Blue right at bottom. Fuscia (FR) and Baby Blue (thd) are 1/48 Octave.
It looks like choosing the right settings are important , on this Clio box at least.
I'm not looking to become more knowledgeable than I need to, just follow the recommendations of those who know what to do. It gets to be a bit much. Searching on window type, it looks like firm ground is small.
https://audioroundtable.com/misc/Mueller.pdf
I could think that at my level they're all good enough but the one reservation I have started with seeing the large variation I got when measuring impedance with different levels of resolution. The peaks were at the same frequencies but the level varied from 60 Ohm to 130 Ohm. That's really what got me looking for more technical advice.
Below are various sweeps at different resolutions and different windows. (Rectangular , Hanning, Hanning/2, Blackman, Blackman/2) First four are with mic close to woofer. Hardware set-up unchanged.
Last image is a compression driver on a small horn. Sine sweep 1/6 Octave resolution are the very top trace and THD Dark Blue right at bottom. Fuscia (FR) and Baby Blue (thd) are 1/48 Octave.
It looks like choosing the right settings are important , on this Clio box at least.
Attachments
Just starting to understand the acoustic vs electrical idea I have a question about safety for the tweeter. First image is of the CD on 18 Sound horn with nothing else. (Green trace is THD) It looked to be a perfect replacement for the AMT in the above circuit so swapped it in with attenuating and impedance levelling resistors and got the curve in second image. The question is, does a driver's natural rolloff (with a horn in this case) actually protect it as well as an electrical filter would?
Attachments
In a series xo the high pass function is delivered by the inductor that is in parallel with the tweeter. What protects the tweeter is the low DCR of that coil. If the DCR is high it will present too much DC to the voice coil. You cannot really replace the xo inductor with resistors as it will either have a high value therefore high DC, or too low a value and it will look like a short for the amplifier.
So you need to have a proper high pass reactive element.
The attenuating resistors are just for that attenuation.
Your CD's response at the lower end will be elevated because of the horn so you'll need to attenuate it more and that will additionally protect the tweeter.
All this may sound convoluted but it's not.
Let me know if it's helpful.
So you need to have a proper high pass reactive element.
The attenuating resistors are just for that attenuation.
Your CD's response at the lower end will be elevated because of the horn so you'll need to attenuate it more and that will additionally protect the tweeter.
All this may sound convoluted but it's not.
Let me know if it's helpful.
I see I didn't explain myself well. I did use the full crossover just as AllenB gave it (attached again here). There was just something about the ease with I was able to exchange the drivers , simply padding down the CD a little more with series R and using a parallel resistor as described in the Intro to XO Design without Testing thread . It seemed too easy.
It does sound pretty good though with added woofer zobel circuit the whole circuit changes and there is much greater rolloff at high frequency. I can really appreciate AllenB's comment re: working with parallel crossover being more straightforward. (my wording )
And one question from wanting to know how your experienced eyes look at the trace in the second image, where my first thought was to take advantage of the existing knee at 1KHz would you set the FC higher , sharpening the slope?
It does sound pretty good though with added woofer zobel circuit the whole circuit changes and there is much greater rolloff at high frequency. I can really appreciate AllenB's comment re: working with parallel crossover being more straightforward. (my wording )
And one question from wanting to know how your experienced eyes look at the trace in the second image, where my first thought was to take advantage of the existing knee at 1KHz would you set the FC higher , sharpening the slope?
Attachments
They're TAD TD-2002 with 18 Sound XT-120 . I have searched for info on the TAD but not found anything other than some basic figures and the owner couldn't find the sheets that came with these ones.
Perhaps mostly useless info to you below but better give everything I can and let you decide.
The Clio gives .029mH for Le but in the same window it measures 23 Ohms for Re which must be impedance not DCR . It does measure a 1Ω resistor accurately though.
Extech LCR meter has +/- 930uH at 1KHz. 3.5mH at 120Hz.
So I have nothing for Le I'm sure of.
For Re they measure 5.76Ω and 5.62Ω with DVM and about the same at 1KHz with LCR meter. (20Ω at 120Hz which is close enough to Clio)
The horn was recommended by Pano for higher frequency fill use when the owner's whole system was being stored here and Pano very kindly helped me get them set up. The owner recently took the woofer cab and main Horn+Driver back, leaving the 2002 here for the time being. . . . .
. . . . .So, with the FR graph posted above I thought it looked worth taking out my AMT's and putting them in, even though the 18Sound horns are not supposed to be used below 1.5KHz. I just figured I might as well try them while they're here.
They seemed not bad with listening and my guesswork measurements until I started reading more and put a zobel on woofer and R across the 2002. The HF fell off and everything came up indicating "Start Over". : )
The horn seems like a good one. I did try a simple version of off axis measurements by moving the mic around and while the traces drop with SPL they stay mostly parallel to each other, which I took to be a good thing. I can post more test images if interested but will hold off making things busier than they are already for now.
Thanks
Perhaps mostly useless info to you below but better give everything I can and let you decide.
The Clio gives .029mH for Le but in the same window it measures 23 Ohms for Re which must be impedance not DCR . It does measure a 1Ω resistor accurately though.
Extech LCR meter has +/- 930uH at 1KHz. 3.5mH at 120Hz.
So I have nothing for Le I'm sure of.
For Re they measure 5.76Ω and 5.62Ω with DVM and about the same at 1KHz with LCR meter. (20Ω at 120Hz which is close enough to Clio)
The horn was recommended by Pano for higher frequency fill use when the owner's whole system was being stored here and Pano very kindly helped me get them set up. The owner recently took the woofer cab and main Horn+Driver back, leaving the 2002 here for the time being. . . . .
. . . . .So, with the FR graph posted above I thought it looked worth taking out my AMT's and putting them in, even though the 18Sound horns are not supposed to be used below 1.5KHz. I just figured I might as well try them while they're here.
They seemed not bad with listening and my guesswork measurements until I started reading more and put a zobel on woofer and R across the 2002. The HF fell off and everything came up indicating "Start Over". : )
The horn seems like a good one. I did try a simple version of off axis measurements by moving the mic around and while the traces drop with SPL they stay mostly parallel to each other, which I took to be a good thing. I can post more test images if interested but will hold off making things busier than they are already for now.
Thanks
I wouldn’t use a measurement at 120Hz for assessing Le, it’s too close where the impedance is dominated by Fs.
I’ll have a chance to look at the horn and driver to assess the low end and the place to cross, later today.
The TAD can comfortably play at around 1khz but with a different horn. Try the SB Acoustics horn, working down to 600hz.
I am not a fan of zobels in series xo. It looks good on paper but it doesn't sound right to me.
Parallel xo definitely benefits from zobels but still doesn't sound good.
I am not a fan of zobels in series xo. It looks good on paper but it doesn't sound right to me.
Parallel xo definitely benefits from zobels but still doesn't sound good.
This is something I wanted to ask about but was afraid to get too much going on at the same time. Something AllenB said in the "Introduction to designing crossovers without measurement" thread (IDCWM thread next time) indicated that impedance could be dealt with in the crossover itself and I thought his wording carried the hint that a better sound could be got that way. At least that's how I read it. So I'm interested in how the process would change if you were looking to do it that way.
Certainly what I've done so far doesn't sound as good with zobel in. I put it on the woofer because the measurement of THD peaks was insane - 200% @300Hz, 600% @30Hz. and 1100% at about 18Hz. The zobel dropped the percentage by a lot , though it added spikes between 80-100Hz. Still , the sound was better without , . . . even at the very high measured distortion.
AllenB,
Not quite sure how to read your comment on 120Hz & Fs. Here's impedance sweep of the 2002 on horn ( NOT in XO) and with 15 Ohm across it.
Certainly what I've done so far doesn't sound as good with zobel in. I put it on the woofer because the measurement of THD peaks was insane - 200% @300Hz, 600% @30Hz. and 1100% at about 18Hz. The zobel dropped the percentage by a lot , though it added spikes between 80-100Hz. Still , the sound was better without , . . . even at the very high measured distortion.
AllenB,
Not quite sure how to read your comment on 120Hz & Fs. Here's impedance sweep of the 2002 on horn ( NOT in XO) and with 15 Ohm across it.
Attachments
Last edited:
Zobels on the woofer in classic series crossovers can cause some odd behaviour with the tweeter where some can plateau or have a peak before roll off or a very shallow slope. OK in an AR series crossover as can help to achieve good phase over the crossover to achieve a deep reverse null.
The inductor parallel tweeter really needs a low DCR and I aim for less than 5% of the tweeter Re. Never had a tweeter problem in 20 years of series crossovers.
The inductor parallel tweeter really needs a low DCR and I aim for less than 5% of the tweeter Re. Never had a tweeter problem in 20 years of series crossovers.