I see what's up, it's eqed - it's really low efficiency in the bass that's eq'd to match the low midrange. So it's an "almost" horn loaded eq'd speaker . Slick setup for the small cabinet size. Still curious on the low frequency distortion and what levels it is referenced to.
Exactly. If you look at the un-equalized bass horn response, you'll see it's anything but flat - typical for a horn. The peak of the response is ~90Hz, falling away to either side of that. To get that flat, you have to push down the center of the response by at least 5dB, maybe more. Much like on open baffle, you have to trade away part of the response to get it to match the bass.
If the bass driver has a sensitivity of 96dB, and you have to push down 5 or 6dB to get flat, you'd be left with near 90dB/W. The horn loading seems to bring that back up quite a bit, if it really is 101dB/W
That raises the question: "How do you measure speaker sensitivity on an active EQ system?". If the signal going into the speaker is not linear, how do you determine power levels or SPL per Watt?
If the bass driver has a sensitivity of 96dB, and you have to push down 5 or 6dB to get flat, you'd be left with near 90dB/W. The horn loading seems to bring that back up quite a bit, if it really is 101dB/W
That raises the question: "How do you measure speaker sensitivity on an active EQ system?". If the signal going into the speaker is not linear, how do you determine power levels or SPL per Watt?
If it is an undersized horn than it is likely way down below the 96 db rating of the driver at the claimed 28 hz. My guess would be 6 db so it's likely 90 with a gain of 6-9 db at it peak over the 96 db rating so it 12-15 db down at the 28 Hz claim. Room gain will help it a bit. Just run a sine wave with 2.83 volts at 28 hz and measure it at a meter. There you have it.
The bass driver must have a lot of Xmax to not run out of gas at any appreciable power level, especially at that EQ. Do you have a high-pass on the bass driver in DSP?
Martagon LF discussions 1
Dear All,
....miracles do not come easily, but sound mathematics/physics helps to optimize.
In other words, a 400 liter 6th order bandpass pumps out equal SPL energy as my optimized horn,
but, there is more smearing over time('booming') and less articulation and 'peak-pressure'
- Distortion was measured at an at estimated 1 Watt (see later)
- Please find attached the specs end Small-Thiele paramter of the 18Sound 15MidBass1000.
Note that lineair excursion is +/- 6 mm (only); for 'damage' you may triple this.
A 'Mid-Bass' is choosen instead of subwoofer a.o. because of
* hight sensitivity (98 dB) and
* to design the resonance frequency in between impedance peaks of enclosure (see later)
- High X-max, high-pass?
* In horn cabinet, excursions are DRASTICALLY damped from some
200 Hz down to 28 Hz
the latter thanks thanks to the acoustical impedance / load of the 3
meter horn and 'pipe'
* with maximal excursion of 0.3 mm at 1 Watt for 60 Hz
* Below I will explain that for
the SuperSize Martagon there is
! NO EQ BOOST AT ALL but rather
! a MINUS - 4 dB low Q ('2') band-'stop'/attenuation at 65 Hz
! 2nd order Butterworth Low-pass at 124 Hz,
! resulting in a 3 dB points at 28 and 150 Hz
the FullSize Martagon there is
! a minor '+ 3 dB EQ boost at 35 Hz
! a MINUS - 4 dB low Q ('2') band-'stop' at 85 Hz
! 2nd order Butterworth Low-pass at 124 Hz,
! resulting in a 3 dB points at 35 and 150 Hz (thanks to upgoing
sensitivity slope)
* So are no spurious excursions over target bandwidth, so no large
Xmax needed
* Nevertheless (see below):
below 28 Hz is another story and, YES, a 2nd order Butterworth high-
pass at 25 Hz is required and applied
- Sensitivity (of LF)
* Room gain: target use of Martagon is on the floor and (almost)
against a wall, so indeed, theoretically this would boost at least 3 dB
compared to 'half-space'
* Calibrated sensitivity of LF is difficult to measure
! Truncated horn cabinets show tremenduous impedance peaks
! 2.83 Volt, given the substantial Impedance peaks and valeys is
only unambiguous if you use real 'Voltage-driven' amp with very
low internal impedance
! My LF amp is a low-cost 2 x 80 Watt Solid State;
so below 40 Hz there is certainly influence of the amp.
* The Freq-Transfer functions of the sole LF units are measured at the
mouth, so at '0' meter to avoid room acoustics (even in outside air).
Absolute SPL is relative, agreed.
Distorsion figures were at an estimated 1 Watt.
* ABSOLUTE SPL and Amplitude-Frequency characteristics
of total systems were measured in outside air at 2 meter (to avoid
'near-field effect of mid horn).
* Given the sensitivity of Fostex at (corrected) 106 dB,
estimating the MID at 106 - 2 dB and the LF at 106 - 4 dB
-->> an estimated LF sensitivity of > 100 dB is conservative even.
* ADD to this that in the target application of stereo, two LF units
couple acoustically in the verly low end, boosting another 3 dB
- Rationale for > 100 dB sensitivity of LF hybrid parabolic horn <-> 1/4 wavelength 'organ-pipe'
* I fully agree that for both the Martagon LF units
! the 'horn' amplification of up to 10 dB (and minimized excursions)
GRADUALLY takes place from some 60 Hz on due to the
'truncation' of the mouth-size (about 1 m**2) and length of the
horns (3 and 2 meter)
! but going down from this 60 Hz, again gradually, the sensitivity
goes up again,
to about 10 dB thanks to the load of the horn-pipe that acts as
a 1/4 wave-lenght resonator
* please inspect the simulation via 'AJHorn' of the (387 ltr) SuperSize
horn, displayed for floor-wall situation:
! Black line is unstuffed result, showing the hybrid nature
! Red line after light-stuffing (starting moderate at throat and
getting ligther going towards mouth)
! At impedance peaks, acoustical load is maximal and excursions
minimal
! LF pumps out 125 dB max within lineair excursion range at 200
Watt
! High-pass at 25 Hz very much required.
! Recall the Ratio of throat/mouth is 9.8 for the SuperSize,
so there is a LOT of chamber-compression to achieve this 10 dB
horn
amplification.
To get identiacl result for a 'plain' horn, the length / mouth increase that much
that the volume would go over 1000 ltr.
* second picture is 'FullSize'
Dear All,
....miracles do not come easily, but sound mathematics/physics helps to optimize.
In other words, a 400 liter 6th order bandpass pumps out equal SPL energy as my optimized horn,
but, there is more smearing over time('booming') and less articulation and 'peak-pressure'
- Distortion was measured at an at estimated 1 Watt (see later)
- Please find attached the specs end Small-Thiele paramter of the 18Sound 15MidBass1000.
Note that lineair excursion is +/- 6 mm (only); for 'damage' you may triple this.
A 'Mid-Bass' is choosen instead of subwoofer a.o. because of
* hight sensitivity (98 dB) and
* to design the resonance frequency in between impedance peaks of enclosure (see later)
- High X-max, high-pass?
* In horn cabinet, excursions are DRASTICALLY damped from some
200 Hz down to 28 Hz
the latter thanks thanks to the acoustical impedance / load of the 3
meter horn and 'pipe'
* with maximal excursion of 0.3 mm at 1 Watt for 60 Hz
* Below I will explain that for
the SuperSize Martagon there is
! NO EQ BOOST AT ALL but rather
! a MINUS - 4 dB low Q ('2') band-'stop'/attenuation at 65 Hz
! 2nd order Butterworth Low-pass at 124 Hz,
! resulting in a 3 dB points at 28 and 150 Hz
the FullSize Martagon there is
! a minor '+ 3 dB EQ boost at 35 Hz
! a MINUS - 4 dB low Q ('2') band-'stop' at 85 Hz
! 2nd order Butterworth Low-pass at 124 Hz,
! resulting in a 3 dB points at 35 and 150 Hz (thanks to upgoing
sensitivity slope)
* So are no spurious excursions over target bandwidth, so no large
Xmax needed
* Nevertheless (see below):
below 28 Hz is another story and, YES, a 2nd order Butterworth high-
pass at 25 Hz is required and applied
- Sensitivity (of LF)
* Room gain: target use of Martagon is on the floor and (almost)
against a wall, so indeed, theoretically this would boost at least 3 dB
compared to 'half-space'
* Calibrated sensitivity of LF is difficult to measure
! Truncated horn cabinets show tremenduous impedance peaks
! 2.83 Volt, given the substantial Impedance peaks and valeys is
only unambiguous if you use real 'Voltage-driven' amp with very
low internal impedance
! My LF amp is a low-cost 2 x 80 Watt Solid State;
so below 40 Hz there is certainly influence of the amp.
* The Freq-Transfer functions of the sole LF units are measured at the
mouth, so at '0' meter to avoid room acoustics (even in outside air).
Absolute SPL is relative, agreed.
Distorsion figures were at an estimated 1 Watt.
* ABSOLUTE SPL and Amplitude-Frequency characteristics
of total systems were measured in outside air at 2 meter (to avoid
'near-field effect of mid horn).
* Given the sensitivity of Fostex at (corrected) 106 dB,
estimating the MID at 106 - 2 dB and the LF at 106 - 4 dB
-->> an estimated LF sensitivity of > 100 dB is conservative even.
* ADD to this that in the target application of stereo, two LF units
couple acoustically in the verly low end, boosting another 3 dB
- Rationale for > 100 dB sensitivity of LF hybrid parabolic horn <-> 1/4 wavelength 'organ-pipe'
* I fully agree that for both the Martagon LF units
! the 'horn' amplification of up to 10 dB (and minimized excursions)
GRADUALLY takes place from some 60 Hz on due to the
'truncation' of the mouth-size (about 1 m**2) and length of the
horns (3 and 2 meter)
! but going down from this 60 Hz, again gradually, the sensitivity
goes up again,
to about 10 dB thanks to the load of the horn-pipe that acts as
a 1/4 wave-lenght resonator
* please inspect the simulation via 'AJHorn' of the (387 ltr) SuperSize
horn, displayed for floor-wall situation:
! Black line is unstuffed result, showing the hybrid nature
! Red line after light-stuffing (starting moderate at throat and
getting ligther going towards mouth)
! At impedance peaks, acoustical load is maximal and excursions
minimal
! LF pumps out 125 dB max within lineair excursion range at 200
Watt
! High-pass at 25 Hz very much required.
! Recall the Ratio of throat/mouth is 9.8 for the SuperSize,
so there is a LOT of chamber-compression to achieve this 10 dB
horn
amplification.
To get identiacl result for a 'plain' horn, the length / mouth increase that much
that the volume would go over 1000 ltr.
* second picture is 'FullSize'
Martagon LF sensitivity pictures
'AJHorn' of the (387 ltr) SuperSize horn
'AJHorn' of the (260 ltr) FullSize horn
'AJHorn' of the (387 ltr) SuperSize horn
An externally hosted image should be here but it was not working when we last tested it.
'AJHorn' of the (260 ltr) FullSize horn
An externally hosted image should be here but it was not working when we last tested it.
Last edited by a moderator:
Trouble seeing LF AJHorn designs of Martagon LF?
Please let me know if the 2 pics are visible?
Thanks
Please let me know if the 2 pics are visible?
Thanks
I had a look at the difference between the non-filtered and the crossover + EQ curves of the bass horn. From what we see there, I derived the filter function needed to get the flat (in band) response needed, using attenuation only. See below.
If that attenuation EQ/crossover is applied to the response of the woofer in horn, there is a resulting bandwidth attenuation of ~3.5dB on pink noise. I.E., if pink noise were measured and averaged with no crossover or EQ, it would be about 3.5dB higher than with crossovers and EQ. That's not as much attenuation as I figured.
Don't know absolute levels, tho.
If that attenuation EQ/crossover is applied to the response of the woofer in horn, there is a resulting bandwidth attenuation of ~3.5dB on pink noise. I.E., if pink noise were measured and averaged with no crossover or EQ, it would be about 3.5dB higher than with crossovers and EQ. That's not as much attenuation as I figured.
Don't know absolute levels, tho.
Martagon equalizing and sensitivity issues
Dear Pano and All,
thanks for estimating the equalizing filter(s) needed to cause the LF horns to give flat response.
You can start from the AJHorn simulation, or from the 'raw' measurements that incorporate panel-absorption and damping/stuffing materials.
To my opinion, the AJHorn simulation and 'raw' measurement are very much alike.
Also, Pano's equalizing and my equalizin by ONLY attenuation only differ 0.5 dB (in size).
So indeed, a well designed truncated horn requires only minimal attenuation of some 3 - 4 dB and low Q (to avoid extra filter-ringing.
!!
I forgot to mention (= advanced class, lol) that via AJHorn I build acoustic notch-filters in the cabinet to damp the residual oscilations in the 'raw' response.
* The SuperSize' has 2 notch-chambers (at 105 Hz and 150 Hz),
* The FullSize has 1 (at 133 with Hz, 1.25 ltr, tube surface 2.27 cm**2, length 2.3 cm).
All 3 acoustic notch filters have mild stuffing to reduce the Q-factor related to 'notch'.
Below is the effect of the FullSize:
* Black: stuffed but without acoustic notch; peak at 135 Hz is in pass-band.
* Red: acoustic notch at 133 Hz with no stuffing
* Green: with stuffing
ALSO, ...
there is are complex relationships between especially Fs, Qes, Qms, Vas and BxL.
Please do realize that I run over more than 500 (sub-woofers to come up with this result...
My general feeling....:
these truncated/hybrid horns work well ( for 15 Inch) when aiming at flat response ('HiFi'):
* high compression (250 cm**2 for 15 inch)
* ...and so large ratio throat/mouth to get decent horn amplification
* horn is 'parabolic' and has large length (2 - 3 mtr)
For (15 inch) speakers with Small Thiele:
* relatively high Fs (45 - 50 Hz) and
* Vas not too large (125 - 150 ltr), so high sensitivity
* moderately low Qts (0.3)
* not too strong BxL (20)
The interplay between Small Thiele and the enclosure parameters must be balanced as indicated above.
Note that in the AJHorn simulation the impedance peak of the speaker at 48 Hz has vanished, but it excersizes its influence, together with moderate Qts and Vas, to keep horn oscillations controlled and small, and large bandwidth extending to over 300 Hz.
Dear Pano and All,
thanks for estimating the equalizing filter(s) needed to cause the LF horns to give flat response.
You can start from the AJHorn simulation, or from the 'raw' measurements that incorporate panel-absorption and damping/stuffing materials.
To my opinion, the AJHorn simulation and 'raw' measurement are very much alike.
Also, Pano's equalizing and my equalizin by ONLY attenuation only differ 0.5 dB (in size).
So indeed, a well designed truncated horn requires only minimal attenuation of some 3 - 4 dB and low Q (to avoid extra filter-ringing.
!!
I forgot to mention (= advanced class, lol) that via AJHorn I build acoustic notch-filters in the cabinet to damp the residual oscilations in the 'raw' response.
* The SuperSize' has 2 notch-chambers (at 105 Hz and 150 Hz),
* The FullSize has 1 (at 133 with Hz, 1.25 ltr, tube surface 2.27 cm**2, length 2.3 cm).
All 3 acoustic notch filters have mild stuffing to reduce the Q-factor related to 'notch'.
Below is the effect of the FullSize:
* Black: stuffed but without acoustic notch; peak at 135 Hz is in pass-band.
* Red: acoustic notch at 133 Hz with no stuffing
* Green: with stuffing
An externally hosted image should be here but it was not working when we last tested it.
ALSO, ...
there is are complex relationships between especially Fs, Qes, Qms, Vas and BxL.
Please do realize that I run over more than 500 (sub-woofers to come up with this result...
My general feeling....:
these truncated/hybrid horns work well ( for 15 Inch) when aiming at flat response ('HiFi'):
* high compression (250 cm**2 for 15 inch)
* ...and so large ratio throat/mouth to get decent horn amplification
* horn is 'parabolic' and has large length (2 - 3 mtr)
For (15 inch) speakers with Small Thiele:
* relatively high Fs (45 - 50 Hz) and
* Vas not too large (125 - 150 ltr), so high sensitivity
* moderately low Qts (0.3)
* not too strong BxL (20)
The interplay between Small Thiele and the enclosure parameters must be balanced as indicated above.
Note that in the AJHorn simulation the impedance peak of the speaker at 48 Hz has vanished, but it excersizes its influence, together with moderate Qts and Vas, to keep horn oscillations controlled and small, and large bandwidth extending to over 300 Hz.
Thanks for the replies Wim. Most impressive.
I seem to recall from my (minimal) research that a Qes of between 0.2 and 0.3 is also desirable. Your driver uses 0.31. Was that part of your selection criteria as well?
I seem to recall from my (minimal) research that a Qes of between 0.2 and 0.3 is also desirable. Your driver uses 0.31. Was that part of your selection criteria as well?
Yes: but AJHorn takes them all into consideration.
Unfortunately, AJHorn is only available in German version (lol: not a problem for a Dutchman of course...)
Unfortunately, AJHorn is only available in German version (lol: not a problem for a Dutchman of course...)
How measure speaker sensitivity on an active EQ sys
Easy...(?).
A DSP EQ emulates physical filters (only). My amps are external, so gain up/down to 2.83 Volt at the speaker's connecors and measure SPL at 1 mtr..?
Of course you have to define the bandwidth limitation and tune over this range or use filtered noise.
Is this too obvious or do I miss something here?
With digital filters there is no loss due to physical filter components so in general you will meaure higher SPL
Easy...(?).
A DSP EQ emulates physical filters (only). My amps are external, so gain up/down to 2.83 Volt at the speaker's connecors and measure SPL at 1 mtr..?
Of course you have to define the bandwidth limitation and tune over this range or use filtered noise.
Is this too obvious or do I miss something here?
With digital filters there is no loss due to physical filter components so in general you will meaure higher SPL
Sensitivity EQ-ed speakers...
Silly me, you were absolutely right Pano to pose this question,
it is not so obvious....!
However, if you would
* measure the mean SPL after EQ in a given bandwidth, e.g. 104 dB
at 2.83 mean Volt connector excitation
* and then determine the mean (frequency-weighted) gain of the EQ filer(s) over this bandwidth (e.g. -2.7 dB)
-->> then,
the EQ corrected SPL/Sensitivity for the 'raw' would be 101.3 in this example.
Silly me, you were absolutely right Pano to pose this question,
it is not so obvious....!
However, if you would
* measure the mean SPL after EQ in a given bandwidth, e.g. 104 dB
at 2.83 mean Volt connector excitation
* and then determine the mean (frequency-weighted) gain of the EQ filer(s) over this bandwidth (e.g. -2.7 dB)
-->> then,
the EQ corrected SPL/Sensitivity for the 'raw' would be 101.3 in this example.
I looked up AJhorn this morning and it seemed to be available in english. Price was a little steep but well worth it. I downloaded the manual (in english) and will be looking at it in the next few days.
Thanks again!
Thanks again!
AJ-Systems - Loudspeaker simulation with AJHorn
In the upper right hand corner it looks like you can choose your language.
In the upper right hand corner it looks like you can choose your language.
I had never seen HJHorn, so had a good look yesterday. I like the look of it, and the easy to make chambers that Wim has used to tune the horn. The purpose the chambers wasn't clear at first, nice to read about what they are used for.
I'm not sure I have the tools to do this. But I also don't quite understand what we are looking for.
From the data published all I could do was determine the response of the woofer in the horn, and the response of the woofer after crossovers and EQ.
That gave me two filters.
- The woofer in the horn
- The woofer in the horn with EQ+Crossover.
- Filter 1 was applied to pink noise, to simulate the woofer horn response
- The filtered pink noise was analyzed for its RMS value. That's the baseline
- The baseline pink was filtered with filter 2, the EQ/crossover function
- Pink noise analyzed again and difference noted as -3.5dB. Brownian noise gave a difference of -3dB
What I don't know is the SPL of the woofer+horn with no external EQ. If the peak at 90dB is a known SPL, then gain could be calculated. The Eighteen Sound 15" woofer is nominally 96dB/Watt on an IEC baffle. Obviously, the horn gain is not flat across the woofer bandwidth, tho not all that bad.
- Status
- Not open for further replies.