OK, sorry, not trying to P/O people. If you assume Dayton RS225-8 Qt 0.33; I get about 21 L closed but the F3 is about 69Hz (I'm looking for mid to low 50's) so this doesn't work. But, if you use Qe of 0.44, I get 47 L and F3 about 51 Hz, also doesn't work. I no longer use Qt but now use what I call Qt' (Qt prime) based on the added DCR of all series inductors in the X/O. OK, lets try an average of Qt and Qe (say 0.39). I get 32 L and 58 Hz (still doesn't work for me). These Seas Prestige 4 layer CD22 and L22 seem to both calculate as to what I'm after here
Hi,
Your full of self justifying BS. You just want to be agreed with.
Your spouting meaningless numbers that are simply wrong.
Grow up. Probably a bit late for that. The Seas are good drivers
and so are the Daytons, but VFM, its not much of a contest.
rgds, sreten.
I've also built at least 50 speaker systems over the years as a hobbyist. I recently built a 3 way "bookshelf" speaker system that I'm quite pleased with.
The cabinet is a virtual copy of the old AR4X (closed box). Since I'm bi-amping it 4th order at 500HZ, just about any good 8 incher would be fine (resonances/breakups in the upper midrange are well attenuated). I also have active EQ to make it acoustically relatively flat down to 30HZ, with a 2 pole rolloff below that. Low bass is a huge feature IMO. I used an 8 inch kevlar cone Focal woofer that I already had (I'd use a Nomex cone Peerless 8 inch if I didn't already have the Focal - but with my active EQ and 4th order active crossover I don't think the woofer resonant F and such is nearly as critical - go for a good Xmax though).
At 500HZ I crossed over to a Vifa TC9 3 inch driver which is reasonably flat to about 14kHZ, and a tweeter takes care of the rest.
The big feature besides the low bass is the fact that I have no crossover between 800HZ and 6kHZ, where the ear is most sensitive, and most of the stereo effect is generated. Inter-aural crosstalk confuses imaging below 800HZ, so I really didn't want to have beam variations in the upper mids due to crossover phase shifting. I used a passive 1 pole for between the mid and tweeter. These speakers make me want to keep listening. Low bass notes may be rare, but percussion sounds better with the low bass, and every sound at any frequency comes in an envelop that often has spectra down to near DC, so the realism is really great.
The cabinet is a virtual copy of the old AR4X (closed box). Since I'm bi-amping it 4th order at 500HZ, just about any good 8 incher would be fine (resonances/breakups in the upper midrange are well attenuated). I also have active EQ to make it acoustically relatively flat down to 30HZ, with a 2 pole rolloff below that. Low bass is a huge feature IMO. I used an 8 inch kevlar cone Focal woofer that I already had (I'd use a Nomex cone Peerless 8 inch if I didn't already have the Focal - but with my active EQ and 4th order active crossover I don't think the woofer resonant F and such is nearly as critical - go for a good Xmax though).
At 500HZ I crossed over to a Vifa TC9 3 inch driver which is reasonably flat to about 14kHZ, and a tweeter takes care of the rest.
The big feature besides the low bass is the fact that I have no crossover between 800HZ and 6kHZ, where the ear is most sensitive, and most of the stereo effect is generated. Inter-aural crosstalk confuses imaging below 800HZ, so I really didn't want to have beam variations in the upper mids due to crossover phase shifting. I used a passive 1 pole for between the mid and tweeter. These speakers make me want to keep listening. Low bass notes may be rare, but percussion sounds better with the low bass, and every sound at any frequency comes in an envelop that often has spectra down to near DC, so the realism is really great.
I agree with your midrange ideas. To me, crossovers I avoid in 1KHz to 3KHz range. Many of my past 3-ways crossed at 300Hz to 5KHz for the mids. To my ears, the female singing voice and piano sound more natural when avoiding X/O's within this all important band. My ideas for this particular build is to get the F3 fairly low but keep the size low also. I would just be using an "office size" receiver, probably 20 to 25 Watts max. I have no plans for any active electronics, equalization, etc. Just a plain Jane small 3-way. My current office system goes pretty low because of "room gain" but I want to improve on this! Thanks everyone, NOT trying to start WWIII here!
In the designs I modeled for both the RS225 and SB23NAC, I included an added series resistance of 0.5 ohms to represent the d.c. resistance of a series crossover inductor; therefore, I modeled with Qts'. I don't have the model set up for the RS225 but I do for the SB23NAC and its Qts' is 0.458, and I think that for the RS225 was a bit lower but still above 0.4. Both woofers were modeled in the same sealed box having 26 liters (net) using actual measured T/S values, resulting in an F3 in the mid-50s for the RS woofer and just below 50 Hz for the SB woofer. I have no "horse in this race" and I don't give a diddly what you build, but both of these designs and woofers sure seem to fit exactly what you describe you're after, yet you continue to ignore facts and claim they won't. That's why you are P/O-ing people here after requesting their help and suggestions. BTW, you said you use what you call Qt' as if you invented it, which by itself says a lot about you.
Paul
Paul
Many drivers from many manufacturers don't adhere to their published specifications. Some are reasonably close while others are way, way off. If you want really accurate designs instead of just guessing like you seem to do, you need to measure your actual drivers and get some accurate modeling software.
Paul
Paul
Can we check the math?
Hi, I think Sreten's math for the RS225 is closer to being right (not the flames tho ), even if you are using highish DCR in your coils.... so let's check.
The voice coil Re is 6.5. Supposing you add a whopping 1.5 ohm DCR from the coils, Re' is now 8, this will change Qe from 0.46 to Qe' at about 0.6
IIRC adding series resistance does not change Qm, only Qe
So Qt' being the parallel value of Qm=1.46 and Qe'=0.6, the new Qt' will go from 0.38 to about 0.425
For sealed boxes Fs is dominated by the ratio of Vas to box volume, which in this case is exactly 2:1
For any given box volume, IIRC the multiplier (Alpha) is calculated by taking the square root of (1+Vas/BoxVolume). So in this case the Alpha would then be Sqrt (1+2) = Sqrt 3 = 1.73
So... with Alpha at 1.73...
- To find Fs' multiply driver fs by Alpha = 28.3 x 1.73 = 49Hz
- To find Qts' multiply Qt' by Alpha = 0.425 x 1.73 =0.73
Qts' is a little over the 0.7 mark..... so now add some stuffing.... the cabinet volume will look a little bigger (some say up to 1.4x, but in practice I have never seen it go beyond about 1.2) ...so... Vb=1 cub ft, and Vb' will be about 28.3x1.2 =close to 34 liters, subtract 3 liters for the driver, and Vb' is about 31 liters, this means the Alpha is more like Sqrt (1+56.8/31)=1.68
So, with that the Fs' becomes 47Hz - that's lower than 50, right?
And the Qt' is about 0.71
So you use coils with a slightly lower DCR, and you're in under the Qt'<0.7 barrier.
those numbers are pretty close to the stated goal aren't they?
.... assuming my rusty memory didn't completely bungle the formulae....
So, please don't take my word for it, even if I too have a BSEE (wow, so many of us have one! Gee, we must be impooortant!). I invite you to check the math once again.
If I'm right, so is Sreten.... and then you can apologize to him for being rude, and he can apologize to you for repeatedly pulling out his flamethrower and turning you crispy around the edges (ha! I doubt it).
Either way, the RS225 is a kick-donkey driver and you would do yourself a big favor to use it - and then say THANKS to Sreten for pointing it out to you.
peace - i hope
Hi, I think Sreten's math for the RS225 is closer to being right (not the flames tho ), even if you are using highish DCR in your coils.... so let's check.
The voice coil Re is 6.5. Supposing you add a whopping 1.5 ohm DCR from the coils, Re' is now 8, this will change Qe from 0.46 to Qe' at about 0.6
IIRC adding series resistance does not change Qm, only Qe
So Qt' being the parallel value of Qm=1.46 and Qe'=0.6, the new Qt' will go from 0.38 to about 0.425
For sealed boxes Fs is dominated by the ratio of Vas to box volume, which in this case is exactly 2:1
For any given box volume, IIRC the multiplier (Alpha) is calculated by taking the square root of (1+Vas/BoxVolume). So in this case the Alpha would then be Sqrt (1+2) = Sqrt 3 = 1.73
So... with Alpha at 1.73...
- To find Fs' multiply driver fs by Alpha = 28.3 x 1.73 = 49Hz
- To find Qts' multiply Qt' by Alpha = 0.425 x 1.73 =0.73
Qts' is a little over the 0.7 mark..... so now add some stuffing.... the cabinet volume will look a little bigger (some say up to 1.4x, but in practice I have never seen it go beyond about 1.2) ...so... Vb=1 cub ft, and Vb' will be about 28.3x1.2 =close to 34 liters, subtract 3 liters for the driver, and Vb' is about 31 liters, this means the Alpha is more like Sqrt (1+56.8/31)=1.68
So, with that the Fs' becomes 47Hz - that's lower than 50, right?
And the Qt' is about 0.71
So you use coils with a slightly lower DCR, and you're in under the Qt'<0.7 barrier.
those numbers are pretty close to the stated goal aren't they?
.... assuming my rusty memory didn't completely bungle the formulae....
So, please don't take my word for it, even if I too have a BSEE (wow, so many of us have one! Gee, we must be impooortant!). I invite you to check the math once again.
If I'm right, so is Sreten.... and then you can apologize to him for being rude, and he can apologize to you for repeatedly pulling out his flamethrower and turning you crispy around the edges (ha! I doubt it).
Either way, the RS225 is a kick-donkey driver and you would do yourself a big favor to use it - and then say THANKS to Sreten for pointing it out to you.
peace - i hope
Hey Paul, good comments... I'm curious about the mid-50's on the RS225, seems a little high - did you check the effect of stuffing?
What measured values did you get for the RS225? The ones I tried were very close to spec after about 50 Hours of 5Hz torture...
With sealed boxes the sensitivity to changes in Fs vs Qts is much less than in vented boxes...so OP is probably still in the OK zone (not the optimal one) to use factory specs...
Jack, your detailed response and mathematical derivation is clear and appropriate, but sometimes people ask questions for which they believe they already know the only correct answer and refuse to listen if anyone else has a different answer, regardless of the correctness of those answering or visible proof.
Paul
Paul
Jack,
I made no attempt to optimize the box by tweaking stuffing density or anything else as that was not my intent. The box was modeled with it entirely stuffed at a density of 0.5 lb/ft3. The actual measured T/S values I used for the modeling were: Fs=29.4 Hz; Re=6.6 ohms; Le=0.75 mH; Bl=8.13; Sd=216.4 cm2; Vas=67.6 liters; Qes=0.526; and Qms=1.357. These are reasonably close to the published values. Martin's modeling software automatically calculates Qts based on the values entered for Qes and Qms plus any resistance added in series with the driver. So, with a box having internal dimensions of 10" x 10" x 16", the predicted F3 was in the mid-50s.
Paul
I made no attempt to optimize the box by tweaking stuffing density or anything else as that was not my intent. The box was modeled with it entirely stuffed at a density of 0.5 lb/ft3. The actual measured T/S values I used for the modeling were: Fs=29.4 Hz; Re=6.6 ohms; Le=0.75 mH; Bl=8.13; Sd=216.4 cm2; Vas=67.6 liters; Qes=0.526; and Qms=1.357. These are reasonably close to the published values. Martin's modeling software automatically calculates Qts based on the values entered for Qes and Qms plus any resistance added in series with the driver. So, with a box having internal dimensions of 10" x 10" x 16", the predicted F3 was in the mid-50s.
Paul
Hi again acstcwrfrmn! Now there's a name that rolls off the tongue! I'd rather be on first name terms! 
Interesting thread this. I've never tried adding resistance to a woofer to get the reflex type Qts up to nearer 0.5, but it's a known technique.
But has anybody noticed how ghastly the frequency response of most of the reflex woofers suggested here is? I've looked up most of them, and they either have terrible 10dB peaks somewhere around 3kHz, or awful breakup higher up!
SEAS H1208-08 L22RN4X/P
SEAS H1192-08 CD22RN4X
SB Acoustics SB23NACS45-8
There are some that I quite fancy implementing without great filtering effort, even if they are fundamentally reflex drivers, because a peak is a peak, even if low down in level:
SEAS H1659-08 U22REX/P-SL
SEAS H1471-08 CA22RNY
I always like drivers with high Qms. 30L reflex is easy. Think eminence or scanspeak. But closed box, nobody seems to make them these days! Weird, eh! You can still get the famous old Vifa M21WG-09-08. Great revealing driver, but slightly problematic Qts of 0.7. But TBH, no-one ever complains about it.
Just to show off my latest 8" + 1" reflex, it has that lovely old efficient and cheap Sony corrugated surround woofer I enthuse about with the transparent cloth dustcap. Mylar tweeter. BW3. It's almost a PA woofer, and sounds fantastic. I'll talk about it elsewhere.
acstcwrfrmn, I think you should give up on the closed box idea for now. Unless someone has a brainwave. Just turn the bass down when you listen to Toccata and Fugue on your 30L reflex.
Interesting thread this. I've never tried adding resistance to a woofer to get the reflex type Qts up to nearer 0.5, but it's a known technique.
But has anybody noticed how ghastly the frequency response of most of the reflex woofers suggested here is? I've looked up most of them, and they either have terrible 10dB peaks somewhere around 3kHz, or awful breakup higher up!
SEAS H1208-08 L22RN4X/P
SEAS H1192-08 CD22RN4X
SB Acoustics SB23NACS45-8
There are some that I quite fancy implementing without great filtering effort, even if they are fundamentally reflex drivers, because a peak is a peak, even if low down in level:
SEAS H1659-08 U22REX/P-SL
SEAS H1471-08 CA22RNY
I always like drivers with high Qms. 30L reflex is easy. Think eminence or scanspeak. But closed box, nobody seems to make them these days! Weird, eh! You can still get the famous old Vifa M21WG-09-08. Great revealing driver, but slightly problematic Qts of 0.7. But TBH, no-one ever complains about it.
Just to show off my latest 8" + 1" reflex, it has that lovely old efficient and cheap Sony corrugated surround woofer I enthuse about with the transparent cloth dustcap. Mylar tweeter. BW3. It's almost a PA woofer, and sounds fantastic. I'll talk about it elsewhere.
acstcwrfrmn, I think you should give up on the closed box idea for now. Unless someone has a brainwave. Just turn the bass down when you listen to Toccata and Fugue on your 30L reflex.
Attachments
There's a simple reason for that, because it's not really an issue for a 3-way with a <<1kHz XO. Taming the breakup node either requires exotic construction which drives up the price and/or compromises the flatness of the response below the breakup node.But has anybody noticed how ghastly the frequency response of most of the reflex woofers suggested here is? I've looked up most of them, and they either have terrible 10dB peaks somewhere around 3kHz, or awful breakup higher up!
A 3kHz breakup node will be a fair way down with a low XO. Often with a 4th order acoustic slope a notch filter isn't even required. Worst case 3kHz breakup induced harmonic distortion will be 5th order at 600Hz.
Probably the best (worst?) example is the Peerless 830667 with breakup around 2.3kHz and a resonance around 900Hz. Still great value for a system crossing at 300-400Hz.
Practical experience of metal drivers tells me I couldn't agree less. Even with a low x/o and LR4 the resonance peak can have an audible effect in the treble; normally perceived as either an extra bit of sharpness or a bit of extra sibilance. The peak may be well down, but it's still there.
A notch filter is only two elements - is it really worth skimping?
A notch filter is only two elements - is it really worth skimping?
The reason the upper mid "breakup" resonance can be a problem, even with a lower xover freq. is because it amounts to ringing, which can elongate the time it exists when stimulated, thereby making itself more perceptable to the ear-brain mechanism. Harder cone drivers will have more resolution, but need lower xover ferqs and higher order rolloff rates, or the final result can really suck. Could sound good at first, but be tedious over time.
I also want to say that the ear-brain mechanism is much less sensitive to distortion at bass frequencies. Unless you are going active 4th order 500HZ on the crossover, I'd hesitate to recommend a hard cone woofer.
Ported box woofers can be pretty good, but getting them right, and keeping them tuned right over time can be tricky. The damping of the cone by the air in the box is said to be good only at the tuned frequency, and weak on both sides of that frequency, so active EQ is not recommended. With a closed box, the box is smaller, and you can build a little active EQ circuit that could run on a wall-wort, which would be inserted into the tape monitor loop on an old receiver, or between the preamp and power amp when that's an option, that would make it acoustically flat to 30HZ. To me, that's a huge feature. I've been doing variations of that for decades, and to me there's no going back.
I also want to say that the ear-brain mechanism is much less sensitive to distortion at bass frequencies. Unless you are going active 4th order 500HZ on the crossover, I'd hesitate to recommend a hard cone woofer.
Ported box woofers can be pretty good, but getting them right, and keeping them tuned right over time can be tricky. The damping of the cone by the air in the box is said to be good only at the tuned frequency, and weak on both sides of that frequency, so active EQ is not recommended. With a closed box, the box is smaller, and you can build a little active EQ circuit that could run on a wall-wort, which would be inserted into the tape monitor loop on an old receiver, or between the preamp and power amp when that's an option, that would make it acoustically flat to 30HZ. To me, that's a huge feature. I've been doing variations of that for decades, and to me there's no going back.
That more often than not comes from people assuming T/S are scalars when they are actually curves. It is very possible that the people who say they are way off are just measuring on a different part of the curve.
There are also sample-to-sample variations, from some manufactures these are huge, even good manfactures usually have significant ranges.
dave
FWIW, Jack, I took the T/S measurements I had for the RS225, adjusted Qes and Qts to account for the 0.5 ohms I added in series, giving a Qts' of 0.399, then ran through the manual calculations you detailed for the benefit of the OP, resulting in a predicted F3 of 52 Hz and a Qtc of 0.76 in the 1600 in3 box volume I modeled with. I then set up the model again and changed the resolution on the y-axis for the system bass SPL vs frequency so I could better eyeball the resulting F3, and I see it as 52-53 Hz. So, "doing the math" and creating the model resulted in the same F3. I also removed all stuffing from the model which didn't change F3 a bit (not surprising), but surely affected Qtc. Last, I replaced my measured T/S values with PE's published values in the model, resulting in exactly the same F3 but a slightly reduced sensitivity.
Paul
Paul
Last edited: