|31st March 2013, 10:00 AM||#12|
Join Date: Jul 2004
je m' excuse
I'm terribly sorry for the 3-month wait. I haven't dropped the project, but a serious illment in the family really eats up my weekends. Now recently it looks like I can put some more time into this again.
I haven't sat still either: I have purchased my Onkyo TX-NR818 surround reciever and temporarly put a leftover pair of Akai speakers from the previous house owners on it. I stuffed them with about 80grams of polyfill to tame the boomyness, and the Audissey Room Correction from the amplifier makes it tolerable. (Bass driver is full ranged ??? and there is one cap protecting the 2 inch "tweeter".)
Now on to the passive QB5 crossover:
The passive crossover for a QB5 alignment
With the information above, one modifies the AkAbak script to the correct box volume and port length/area for the chosen QB5 alignment.
In our case this will be 9,7L and a port of 3*3*1,5cm. When modeling the SPL plot, you'll clearly notice the bump around 100Hz.
Next step is to find the impendance at 80Hz *in that box*. Inspect > Network Impendance and clicking on the graph at 80Hz reveals a value of 9
Ohm. This is the R value we must use in the formulae above.
C = 1 / ( 2 * p * Z * d * f ) = 1/ (2 * 3,14159 * 9 Ohm * 1,59 * 80 HZ) = 139 PicoFarad
L = ( Z * d ) / ( 2 * p * f ) = (9 Ohm * 1,59) / (2 * 3,14159 * 80Hz) = 14,31 / 502,65 = 28,5 MilliHenry
When we put this into an AkAbak script, we see that we don't get the expected flat response (like with an active filter.) This is due to the effect of the impendance on the passive components. (There are some good articles about that, and the Elliot Sound Page also has some about countermeasures.)
So what we need is a personalized crossover, dependant on the impendance- and spl-plot. Enter Speaker Workshop, a free program that accepts .frd (=spl plot) and .zma (=impendance plot) files, and has a brute force/Monte Carlo analysis tool to do the tedious work.
To get these files, one could use the export function of Akabak (and modify the data in a spreadsheet) and a tool to generate the phase data column. (Mentioned in "Crossover Simulation in Speaker Workshop - Tutorial",
Speaker Workshop Tutorial )
Alternatively, use HornResp and export the data. Remove the first lines from the file, and you have the needed .FRD and .ZMA files!
Follow the above mentioned tutorial to import the files, choose a standard/stock 2nd order crossover and optimize the 80-2000Hz for maximum flat response. If you get a bump around a certain freq, insert a "Impendance compensation - resonant peak" into the crossover and re-optimize.
This got me a ruler flat response from 80Hz till up. I put the resulting crossover in AkAbak and it looks good, but there are some differences with Speaker Workshop. Maybe someone can shed a light on this. I think AkAbak is more accurate, but with SW one can get about 95% of the way. (Or assume Speaker Workshop is spot-on, and be done with it!)
Anyway, with this I'll conclude the basic QB5 alignment optimization. The whole SW optimization process will have to be re-done once the diffraction, baffle step and internal resonances have been simultated, so no point in taking this further right now.
I'll just stop for a moment on the reduction in excursion by the electrical filter, and other benefits.
In the original unfiltered box, we see a 6dB bump. Once we filter this away, we see the responce is flat, and the excursion is reduced (at 107 Hz) by 25%. Obviously, we can now put in about 4 times more power before we reach xmax, resulting in our original 6dB increase.
Now how about the 25% excursion reduction?
After digging trough the 3-part "Klips Modulation Distortion" article, we can conclude that the reduction in Modulation distortion is lineair with excursion, resulting in a 25% decrease! This is beyond my wildest imagination. Surely a fullranger has its own issues, but once my whole analysis is done, I am truly confident I'll get the best possible response from it.
Sorry again for the very long wait, but I'll do my best.
Next up: internal reflections!
Here is the AkAbak script with the SpeakerWorkshop' calculated crossover. Included screenshot is from SW and then from AkAbak.
Node=2=0 L=11.38mH Rs=1ohm
Node=3=4 L=10.75mH Rs=5.36ohm
Duct 'Rear chamber'
Radiator 'Front Driver'
Radiator 'Bassport Exit'
-- If in doubt: overdesign!
|5th April 2013, 12:25 PM||#13|
Join Date: Nov 2006
To sum things up, so that I understand your goal:
You want to take a not so pricey, simple driver like the Monacor and get a QB5 Allignement with a passive Filtering (highpass with a C) inside a ported enclosure. Furthermore then deal with Bafflestep and reduce box-insider Reflexions as much as possible.
As a "bonus" you are looking for a "variable Q" (passive version)..
Correct so far?
|6th April 2013, 06:20 PM||#14|
Join Date: Jul 2004
The fullranger (and the emphasis on internal reflections) is for the merit of imaging for hometheatre. Max SPL will be limited, but the THX reference level is 85dB with 20dB of headroom, which is too loud anyway for a house with neighbours and my girlfriend... Also the sub crossover freq must be 80Hz, so the QB5 is to take the monacor's "extra" below 80Hz and push it up to trade it for about 6dB of extra output, less IM distortion, power compression, etc...
The "variable Q" is to be able to implement the desired Q from the alignment tables into a passive filter, but I think it is not feasable: in the HornResp thread I read that with an active filter there is no back-influence from the driver - easy, but obviously in a passive filter the impendance curve wil effect the Q of the crossover. I guess one should try to separate the desired Q (from the tables) into a sum of the crossover Q and the impendance curve'Q.
So no passive formula for everyone's benefit to make passive QB5's I guess...
Also, baffle step, beaming, back wall reflection... will also effect the unfiltered responce, so I realize now that the crossover will be the last thing to do and that it will be a fit-to-curve reiteration process, rather than a formula-based-nailed-it.
Trying something and realizing it will fail that way (for me) is what I learned with the passive attempt.
-- If in doubt: overdesign!
|6th April 2013, 07:28 PM||#15|
Join Date: Nov 2006
Ok, lets clear some stuff up, you have a lot of thoughts here.
This driver has very low xmax to do bass. Datasheet says 1mm, which is a conservative value, since monacor states the real geometrical xmax with these drivers. so 1,5mm might be realistic, if the driver is symmetrical in its movement (lokking at the age of the driver and the way its constructed - donīt count on it).
So SPLmax will be limited a lot. It doesnīt matter which alignement you chose, as long as its a ported enclosure (with or without a highpass C), max SPL at low frequnzies is always the same with the same driver.
Doing 80Hz with this driver will be very limited and will only work in a very small room if you are stisfied with low levels of volume.
QB5 Alignments are good with drivers having a very high Qts, since these drivers tend not to work in a "normal" ported enclosure, they have very boomy bass. QB5 with a proper C moves everything in place. But again - SPL is still limited, no matter what.
Since your port and the C have a close relationship in how they work together and since the port is always tuned to the same frequency (unless you buy a variable port and change it mechanicaly everytime), chaning C doesnt help a lot without changing the port. The variance of parameters where the values of the tuning and of C "make sense" are in a small range - so the myth of getting whatever response you want by simply changing the Q by changing port an C - has its limits in reality.
A variable C in a passive crossover is exepnsive, intertechnik sells some for speaker development, but doing this with normal capacitors requires buying a lot of them and buliding some kind of matrix switch to change the values...
QB5 Alignements have a high order slope on the lower end and therefore high phase changes. This could get in your way of adding a sub.
The big capacitor for the QB5 Alignemnt does some change to the impedance where baffle-step correction is suited - this has to be taken into account, you are right on this one. Its effect isnīt that great, Iīd guess you wonīt be needing impedance linearisation here, but to be sure, youīd have to simulate in Akabak to see the results. Of course, bafflestep depends on the size of the enclosure, so you have to think about this one, too.
with back wall reflection I guess you donīt mean whats inside the box but the wall behind the enclosure? How close do you want to get to the wall?
I have some of these drivers here - the wanted quality of a fullrange doing a good job in the mids and with voice is limited with those in my oppinion. Of course, they donīt cost much so its ok for the money you spend. But the lower end will be limited quite alot and midrange will suffer from distortion induced by too much excursion, doppler and intermodulation effects.
I get the impression you want the QB5 Alignement to do some magical trick which gets a lot more out of the speaker then another alignment - this isn't the case.
To calculate a QB5 Alignement, I have a formula somewhere in the drawer which is pretty simple to handle. I could look it up. But itīs designed for very high Qts (0,43 isnīt that high, I am talking about 0,7 and much higher). You will be needing a lot more volume, too than in a normal reflex enclosure.
You should be fine with a normal ported enclosure with this driver.
Last edited by Sabbelbacke; 6th April 2013 at 07:30 PM.
|8th April 2013, 06:31 PM||#16|
Join Date: Jul 2004
I already did some comparisons on small commercial sattelites. Some of them claim to go as low as 40Hz, but the sensitivity is then in the low 80's. I feel they swap efficiency for bass extention. Quite the opposite of what I try to do.
Sabbelbacke, Making everything (port and C) variable and adjustable to suit is a very practical approach, which I cannot do, but I guess you have already done that in the past. Cool that you have this facilities.
The living room here is 4x3m (more around 3x3m for practical purposes, hence space is at a premium), so the speakers will be hung on-wall. This will give some boost at the lower frequencies -- "reflector" element in AkAbak to test this --, and having 4 (well 5) speakers is +6dB from the quadruple cone area.
You did make me wonder about two things: How will the room help the speakers, and how much SPL is actually used?
Since I calibrated everything with Audissey, it advised a crossover freq of 40Hz??? I first tought this was an error, but according to their website this can be because of room augmentation. I run it at 80Hz anyway, but I measured the dimentions of my temporarly (AKAI M45-SW) speakers: 10L box and flared port tuned to about 50Hz. A bit on the optimistic side, but it seems in a practical enviroment the calibration microphone does think it is useable to 40Hz. I also think a rapid rolloff or mismatch between sub and sattelite slope will give lobing or diffraction effects, which I think shouldn't be such a problem at those low freq, and if they will, the Adyssey should iron them out. (I'm hoping the amplifier has a "brick wall" crossover so there should be no overlap, but I couldn't get that from the manual.)
SPL-wise I have a simple decibelmeter - my only measuring equipment - and I remembered some volume settings during normal radio listening and some "spirited" action movie scenes. I took my meter at hand and it seems ... 70dB A-weighted and 90dB C-weighted during peak moments are about the loudest I went (so far ;-) Verry reassuring I think.
Tomorrow I don't have to work and I'll try to do some scripting work again.
-- If in doubt: overdesign!
|Thread Tools||Search this Thread|
|Thread||Thread Starter||Forum||Replies||Last Post|
|QB5 crossover||Cordraconis||Multi-Way||13||4th August 2012 10:44 AM|
|2 or 3 way, bassreflex + horn||Pengvim||Multi-Way||4||4th October 2009 03:54 PM|
|Step-by-step ambitious first project! Suggestions?||ledhed2222||Multi-Way||7||13th August 2009 03:51 AM|
|QB5 Alignment||bigboy||Multi-Way||1||19th September 2008 09:16 AM|
|Modeling a Baffle Step Correction Circuit||Seadweller||Full Range||4||28th December 2007 09:51 PM|
|New To Site?||Need Help?|