When I simulated what was needed to push my driver (PHL 3411) in WinISD to 43 Hz f3 in 30 a liters box, it never matched my measurement results. I had to use considerably more boost, which was no problem, because it is an efficient driver and I have 250 Watts powering it.
In any case, I was wondering if you could actually match the Hornresp simulation with your measurements?
In any case, I was wondering if you could actually match the Hornresp simulation with your measurements?
Actually, the BR tuning is off compared to the HR model (due to port shape). I used -16 dB , Q=3 low shelf, I moved the frequency until I got the peak aligned with the BR tuning.
I do not think I can do any distortion plot at home, I would need to take it out, but not before the winter ends. Or maybe I just do not know how to do that in room...
I cannot see any movement of the 15" cone while listening, so I do not expect amy distortion that could be noticeable.
I do not think I can do any distortion plot at home, I would need to take it out, but not before the winter ends. Or maybe I just do not know how to do that in room...
I cannot see any movement of the 15" cone while listening, so I do not expect amy distortion that could be noticeable.
Hi pelanj,
hey this might take some of the fun and adventure out of filter design, but after trying every combo of filters on low end response,
i think all roads lead to the same solution.
What i mean is, it doesn't matter if we use low shelving, PEQs, Linkwitz transforms, or whatever..... to boost the low end rolloff..
After we add high-pass filters, or attenuating PEQs or shelves, or again whatever filter(s) to limit over excursion.....
the final net filter of whatever we put together, for each and every method we try, will be the same for a given acoustic output.
So, I just take the filter route looks easiest to implement, knowing in the end there are no free lunches to be had.
hey this might take some of the fun and adventure out of filter design, but after trying every combo of filters on low end response,
i think all roads lead to the same solution.
What i mean is, it doesn't matter if we use low shelving, PEQs, Linkwitz transforms, or whatever..... to boost the low end rolloff..
After we add high-pass filters, or attenuating PEQs or shelves, or again whatever filter(s) to limit over excursion.....
the final net filter of whatever we put together, for each and every method we try, will be the same for a given acoustic output.
So, I just take the filter route looks easiest to implement, knowing in the end there are no free lunches to be had.
Same with the 10-inch. At levels beyond acceptable in the city, no cone movement at all (only af shifts/refocus):
Have no x-ray eyes, cannot see the voice coil working though.
@perrymarshall
You used speakerbuilder website in the article but on my computer it doesn't download driver data from server. You guys also experience this problem?
You used speakerbuilder website in the article but on my computer it doesn't download driver data from server. You guys also experience this problem?
SW270 in 30 litres, Fb 30Hz:
Now Hornresp filter:
and resulting max output (xmax limit):
group delay: (@perrymarshall - too high right?)
Now Hornresp filter:
and resulting max output (xmax limit):
group delay: (@perrymarshall - too high right?)
Last edited:
The shelving filter does not lower the group delay, but it does allow for a higher max SPL because it limits power below fb where cone excursion is high. Actually, what struck me, but I did not have the time to re-read the Keele paper and compare it with the approach from Perry Marshall, was that the latter suggests to boost at fb. This might be related with the boost-dip function of the shelving filter. Keele suggested otherwise: boost above fb. This results in a quasi-lower Q of the port function, expanding its range. And the resulting group delay is less, as the boost is not doubling up at the tuning frequency of the port itself, but is rather a bit distributed.
Could be that my reasoning here is erroneous, because in my simulations, I have neglected the additional high pass suggested by Keele, because I found it to be unnecessary for my listening situation.
Could be that my reasoning here is erroneous, because in my simulations, I have neglected the additional high pass suggested by Keele, because I found it to be unnecessary for my listening situation.
The shelving filter doesn't reduce group delay absolutely. It just creates a lot less group delay than a 24dB filter (half) and compared to a 30, 36, 48dB/octave standard high pass filter, it reduces it by 2/3 or 3/4ths.
48dB filters have a very audible amount of group delay. You can literally hear the "thump" of the bass drum arrive at a different date than the slap. Good song for testing this is Fear of a Blank Planet by Porcupine Tree.
When I did the math for the AX article, I found that immediately below Fb, you need to drop at least 14dB in the space of 1/2 octave, in order to ensure you're not wasting cone excursion on sub-band signals.
That's 28dB/octave so most people would just apply a 30dB/octave or steeper high pass filter.
But if you use a -16dB shelf filter with a Q of about 2, you get something like 20dB drop within that 1/2 octave... exactly what's necessary to get the job done and not a bit more. AND you get your +6dB boost at Fb as a free bonus. The filter below is perfect for a speaker with Fb = 68Hz.
Philosophically, in speaker design, I believe brute force backfires. Brute force is clumsy, not elegant. Shelf filters are elegant. Lots of DSPs have this exact shelf filter available, so it's super easy to execute.
@mark100 it's true that all identical curves have identical phase, but shelf filters have less phase shift because they level off at very low frequencies. So they're a different animal than standard High Pass.
@sheeple The #1 goal of my article is to squeeze as much Xmax & Vd as humanly possible out of a given woofer, and if that's your goal, then the point of max DSP boost should be exactly at Fb, not above or below. Fb is the reflex point of maximum leverage - you can dump power into the speaker, the port generates 85% of the output, and excursion is at a minimum. Of course you can get away with boosting it above Fb if you wish, but not much below.
You have to make sure your port can handle the volume of air movement. Most ports chuff loudly when you drive the system hard at Fb. Most ports are way too small. When your port is big enough you'll be surprised at how much a small woofer and box can put out, and how authoritative it sounds.
@jzagaja the speaker building web tool just displays a result, it doesn't save woofer data or export data. You can do more sophisticated things with other tools.
Also @pelanj you could easily get away with tuning your system to 25 or 30Hz if you wanted to - just add mass to the passive radiator or length to your port, and adjust the EQ accordingly.
48dB filters have a very audible amount of group delay. You can literally hear the "thump" of the bass drum arrive at a different date than the slap. Good song for testing this is Fear of a Blank Planet by Porcupine Tree.
When I did the math for the AX article, I found that immediately below Fb, you need to drop at least 14dB in the space of 1/2 octave, in order to ensure you're not wasting cone excursion on sub-band signals.
That's 28dB/octave so most people would just apply a 30dB/octave or steeper high pass filter.
But if you use a -16dB shelf filter with a Q of about 2, you get something like 20dB drop within that 1/2 octave... exactly what's necessary to get the job done and not a bit more. AND you get your +6dB boost at Fb as a free bonus. The filter below is perfect for a speaker with Fb = 68Hz.
Philosophically, in speaker design, I believe brute force backfires. Brute force is clumsy, not elegant. Shelf filters are elegant. Lots of DSPs have this exact shelf filter available, so it's super easy to execute.
@mark100 it's true that all identical curves have identical phase, but shelf filters have less phase shift because they level off at very low frequencies. So they're a different animal than standard High Pass.
@sheeple The #1 goal of my article is to squeeze as much Xmax & Vd as humanly possible out of a given woofer, and if that's your goal, then the point of max DSP boost should be exactly at Fb, not above or below. Fb is the reflex point of maximum leverage - you can dump power into the speaker, the port generates 85% of the output, and excursion is at a minimum. Of course you can get away with boosting it above Fb if you wish, but not much below.
You have to make sure your port can handle the volume of air movement. Most ports chuff loudly when you drive the system hard at Fb. Most ports are way too small. When your port is big enough you'll be surprised at how much a small woofer and box can put out, and how authoritative it sounds.
@jzagaja the speaker building web tool just displays a result, it doesn't save woofer data or export data. You can do more sophisticated things with other tools.
Also @pelanj you could easily get away with tuning your system to 25 or 30Hz if you wanted to - just add mass to the passive radiator or length to your port, and adjust the EQ accordingly.
Last edited:
There is a highly informative thread on ports thanks to the experimental nature of @stv, that shows 'big' ports create specific issues too, midrange leakage. Thread is long, this as a starter:
https://www.diyaudio.com/community/...-port-resonance-absorbers.388264/post-7509903
https://www.diyaudio.com/community/...-port-resonance-absorbers.388264/post-7509903
I would still suggest to check SPL requirements before implementing any sort of over excursion protection. It gets anecdotical now, due to the late hour, but just as an example: I was once airplaying a youtube video to my speakers listening to some sort of lecture, and dialogue was extremely low level, so I increased level a lot. What I did not know and neither expected that the stupid video ended on an electronic music jingle with considerable, I mean extremely higher level and a lot of bass. The speakers instantly killed me, the SPL was enormous, and after the statistical reaction time of 0.2 sec that we learned about in driving school, I managed to grasp the stupid iPhone and kill the horror that I brought upon me an my neighbors: the driver DID NOT COMPLAIN, but for the first time, I saw the cone move, considerably so. I would never, ever, reach such a level when listening, and the speaker was still... okayish with it. Maybe try without the over excursion protection if your woofer (mine: PHL 3411) has considerable xmax and you are listen to music at non deafening levels.
Last edited:
In fact, the tuning is more like 36 Hz. Lowering the tuning more would be a bit more difficult to fit the box and excursion would get out of hand.
The truth is I always try to aim for maximum possible output ending up using only a fraction of the capability. If I feel adventurous one day, I will try just the boost. For low levels there will be not much difference most probably. I just feel good knowing I can reach PA levels at home
The truth is I always try to aim for maximum possible output ending up using only a fraction of the capability. If I feel adventurous one day, I will try just the boost. For low levels there will be not much difference most probably. I just feel good knowing I can reach PA levels at home
Yeah, get yourself an SPL meter or whatever you find sufficient. Be honest. You do not need a Porsche if you cannot push the throttle.
I did not trust my voltage-adjusted measuring mic REW results first, but as it is said that SPL measurements on iPhones are kind of reliable (due to the standardized mics used in each model), I could confirm these values and I listen at levels so low I would have never expected, as I always thought I was terrorizing my neighbors already, and probably do so. 55 to 65 dB average, 80 max/peaks for the latter, at 2.7 meters distance, which is too loud already/drunk listening/holiday fun.
I will buy me a cheap Ali Express A/C weighted SPL meter some time, but do not expect the previously tested listening levels to be wrong. You don't do yourself a favor if you build a system to specs that you would never use. What matters is how the system performs in relation to you listening habits.
I did not trust my voltage-adjusted measuring mic REW results first, but as it is said that SPL measurements on iPhones are kind of reliable (due to the standardized mics used in each model), I could confirm these values and I listen at levels so low I would have never expected, as I always thought I was terrorizing my neighbors already, and probably do so. 55 to 65 dB average, 80 max/peaks for the latter, at 2.7 meters distance, which is too loud already/drunk listening/holiday fun.
I will buy me a cheap Ali Express A/C weighted SPL meter some time, but do not expect the previously tested listening levels to be wrong. You don't do yourself a favor if you build a system to specs that you would never use. What matters is how the system performs in relation to you listening habits.
For this reason, I like what I have done, which is the alternative to using a high pass, that is; use so much Sd, a steep High Pass isn't needed. Not always possible given the situation but in a lot of domestic situations, I think, most could. This is just 2 sealed 18"s at my ~54" listening position. With the intended 6-8 18"s, I don't thd will be an issue, in the absence of an HP, at higher levels.