My New Try With Series Filter
Remember the first order (parallel) filter in my speaker I talked about? I rearranged the same components for series filter today (Am listening to it from different room).
Planet10 mentioned his effort with 1st order parallel that fell into place when the components were rearranged for series filter. I just did the same.
Somebody else had also mentioned about a working series filter that became worse when rearranged for parallel filter...
Of course, a working series filter will most probably get worse when rearranged into parallel filter. How about the opposite? Well of course the situation is better than doing the opposite (But it is still comparing 2 different designs!).
In my case, the series filter work just fine. I cannot "hear" the woofer. The resultant sound move but not far away from both drivers.
The first time I turned on the amplifier and heard the sound, I quickly recognized that sound character of series crossovers. Less sonic, less defined instrument sounds, a strange soundstage. The back EMF is probably the one responsible for that typical sound.
Yes, of course, the parallel version sounded better to my ears. But that's because I have successfully hit the "sweet spot" (My other tries of parallel filters are actually worse than the series one). So may be if similar "sweet spot" is hit by the series filter after tweaking, the series will be better? May be, but that less defined sound character is I think intrinsic to the topology.
So if I can get it right with 1st or 2nd order parallel, I will not consider a series filter, unless for economical and/or practical reasons. But when I have to consider 3rd or 4th order parallel, then that's the time I have to consider series filter first. But still I will have to build both versions and compare the result.
Remember the first order (parallel) filter in my speaker I talked about? I rearranged the same components for series filter today (Am listening to it from different room).
Planet10 mentioned his effort with 1st order parallel that fell into place when the components were rearranged for series filter. I just did the same.
Somebody else had also mentioned about a working series filter that became worse when rearranged for parallel filter...
Of course, a working series filter will most probably get worse when rearranged into parallel filter. How about the opposite? Well of course the situation is better than doing the opposite (But it is still comparing 2 different designs!).
In my case, the series filter work just fine. I cannot "hear" the woofer. The resultant sound move but not far away from both drivers.
The first time I turned on the amplifier and heard the sound, I quickly recognized that sound character of series crossovers. Less sonic, less defined instrument sounds, a strange soundstage. The back EMF is probably the one responsible for that typical sound.
Yes, of course, the parallel version sounded better to my ears. But that's because I have successfully hit the "sweet spot" (My other tries of parallel filters are actually worse than the series one). So may be if similar "sweet spot" is hit by the series filter after tweaking, the series will be better? May be, but that less defined sound character is I think intrinsic to the topology.
So if I can get it right with 1st or 2nd order parallel, I will not consider a series filter, unless for economical and/or practical reasons. But when I have to consider 3rd or 4th order parallel, then that's the time I have to consider series filter first. But still I will have to build both versions and compare the result.
Last edited:
Well I think some of the off topic stuff here is alright, the problem is the thread title and then what's going on in the thread, what is going on in this thread?😕
You seem to have posted so much irrelevant waffle into this thread though that it's become difficult to actually work out what it is you're trying to say, or what it is you're trying to ask.
The main point of the forum is to have a lively discussion where we talk about ideas and so forth, but this, with respect to this thread, should be done entirely on scientific proof and justification. This isn't a thread where you can be 'subjective' about things. You can have a subjective preference but you need to be able to back it up with scientific proof as to why. A thread that simply tries bumbling along mainly on subjectivism is no different from one of the endless cable threads.
You mention that no high order crossover has ever sounded natural to you, well hurrah. But no one gives damn about that. What we care about is what systems you actually liked and then trying to figure out why. I do not like bringing any commercial products into this discussion either because they are an unknown quantity (unless you designed them yourself). If one bases their opinion mainly off of listening to only commercial products then this in and of itself is bogus in my opinion. We should be discussing our own designs, where we had absolute control and are now therefore free to discuss without being compromised by any kind of legal rubbish.
John K posts from his own experience and from a position of scientific background. He will have his own preferences, but he backs up what he says with proof, either simulated or measured and from a lot of his own designs and design work.
He has posted, shared his opinion, made his points in a clear and easy to understand manner and has effectively communicated what it is he was trying to say.
We have had people posting simulations and giving descriptions of what the strengths and weaknesses are about certain type of filter. We've then had some discussion on their limitations or what could be limitations.
What is it that you specifically want to discuss in this thread speakerman19422? Because along the way it would seem that the meaning has been somewhat lost. So perhaps you could restate it in a clear and concise way.
Last edited:
I did not start this thread. Planet10 choose to split it off. I made post to a question about crossovers pertaining to a Kef driver.That thread was started in 2002 it was about using a Kef B200 in a t-line.
I suggest you read the thread.
http://www.diyaudio.com/forums/multi-way/3960-b200-kef-drivers.html
I was told I was wrong when I had seen the design implemented pertaining to my post in a 6db per octave crossover in parallel. It works even better in series.
.
Read the papers Formulas were given in the papers that are scientific proof.
One example is the time constant formula 159,155/Fo,
The time constant of a given frequency is 159,155 times the frequency of the inverse.
Since inductance is the inverse of capacitamce it explained the Fried designs.
Not to difficult to understand.
Richard Small 1971 was one of the first to take into account voice coil inductance in determining a drivers crossover frequency. And I am told I am wrong ha.
I suggest you read the thread.
http://www.diyaudio.com/forums/multi-way/3960-b200-kef-drivers.html
I was told I was wrong when I had seen the design implemented pertaining to my post in a 6db per octave crossover in parallel. It works even better in series.
.
Read the papers Formulas were given in the papers that are scientific proof.
One example is the time constant formula 159,155/Fo,
The time constant of a given frequency is 159,155 times the frequency of the inverse.
Since inductance is the inverse of capacitamce it explained the Fried designs.
Not to difficult to understand.
Richard Small 1971 was one of the first to take into account voice coil inductance in determining a drivers crossover frequency. And I am told I am wrong ha.
Just for the purpose of clarification,re #226;as I recall it was Norman Crowhurst who suggested including the voice coil inductance as part of the total 'choke' value in a crossover where the inductor was in series with the speaker.However,since the speaker inductance is neither pure or constant, it is best to follow current practice and use a zobel network to ensure a speaker imitates a resistor at the crossover frequency.
Last edited:
+ 1
The thing with this is that the loudspeakers inductance is always included when you're designing. The inductance is defined as a part of the impedance measurement, so regardless of what filter type you choose to use, you are working with the inductance as part of the design process. Of course this is only if you're using proper CAD software, but in this day and age who'd work without proper measurements and computers anyway?
'the companies he's associated with'? He was at Tymphany last I knew. No- drivers with wider bandwidth have a lower inductance. You also mean frequency response is driver-inductance related, not the other way around.
The Thiele/Small came partially from Richard Small- yes, so that's a given.
Series xovers, or 'series circuits'? Not same thing, please be specific. Le does not determine the xover frequency between 2 drivers. It tells you the rolloff of the driver in question, but many xover points can be made to work.
Le can be used to make a true zobel circuit yes, but actually using the 10kHz Le measurement is more beneficial if you are trying to use a CR-zobel in the true sense, or your impedance will actually be lower than flat as frequency increases. This is all moot, since zobels are not absolutely necessary all the time. Using a 'tweaked-zobel' CR shunt filter to damp the rolloff knee of a woofer or tilt the treble's rising response is a more typical issue.
Tweeter typical inductance is less than 0.1mH. I use that value a lot in tweeter networks of both topologies. Not sure how that relates to xover frequency, as this would place it a lot of times above 10kHz.
If you want to discuss theory- then let's do it! All I see is speculation, opinion, and the need to state you can't reveal any info to prove anything due to copyrights or patents.
I'm only discussing- I'm not calling you out. It also sounds like you think I am a fat invalid, and that is far from the truth. I like this hobby a lot, and I can say this- 'If replying to you infuriates you, you can ignore me and pick up a more challenging hobby.' You don't have to be so confrontational. It's a hobby, and if we want answers, why not answer them instead of implying you know what you're doing without proof?
I just don't understand your convictions in this, or the need to tout your speaker-beliefs as law to the masses.
I will say this though- Fried helped Oskar create some sweet tweeters!
Later,
Wolf
It's pretty hard to discuss theory when other forum users don't have the articles. I stated theory before posting formulas.
I know what a Zoebel does. It equalizes the impedance rise of the voice coil inductance. It is also shunting the tweeter in a series crossover.
Every Fried series crossover plan I have a copy of for the umptieth time uses the 6.5" Woofer mh and tweeter mh as the basis for the crossover frequency.. The value of the L1= woofer mh. The value of C1= tweeter mh.
Small referred to using woofer mh as the design basis in series crossover circuits.
Not all 6.5' woofers or dome tweeters have the same mh. values.
Wolf-teeth you have not carefully read the posts. How many times do I have to repeat the same thing.
What I interpreted from Small's AES paper was whole point was using tweeter mh as the crossover frequency for the tweeter.
Using the woofer mh as as the crossover frequency of the woofer.
Look at how all three tweeters have a different mh. C1 = tweeter mh in first order series crossovers The same can be said of all the woofers mh. L1 = woofer mh in first order series crossovers.
http://www.scan-speak.dk/datasheet/pdf/d2905-990000.pdf
ScanSpeak Classic D2008/8512 20 mm Dome Tweeter: Madisound Speaker Store
ScanSpeak Discovery D2608/9130 1" Textile Dome HDS Tweeter: Madisound Speaker Store
ScanSpeak Classic 18W/8545 7" Mid Woofer Paper Cone: Madisound Speaker Store
Eton 7-300/A8 Symphony, 7" Mid/Bass with Phase Plug: Madisound Speaker Store
I know what a Zoebel does. It equalizes the impedance rise of the voice coil inductance. It is also shunting the tweeter in a series crossover.
Every Fried series crossover plan I have a copy of for the umptieth time uses the 6.5" Woofer mh and tweeter mh as the basis for the crossover frequency.. The value of the L1= woofer mh. The value of C1= tweeter mh.
Small referred to using woofer mh as the design basis in series crossover circuits.
Not all 6.5' woofers or dome tweeters have the same mh. values.
Wolf-teeth you have not carefully read the posts. How many times do I have to repeat the same thing.
What I interpreted from Small's AES paper was whole point was using tweeter mh as the crossover frequency for the tweeter.
Using the woofer mh as as the crossover frequency of the woofer.
Look at how all three tweeters have a different mh. C1 = tweeter mh in first order series crossovers The same can be said of all the woofers mh. L1 = woofer mh in first order series crossovers.
http://www.scan-speak.dk/datasheet/pdf/d2905-990000.pdf
ScanSpeak Classic D2008/8512 20 mm Dome Tweeter: Madisound Speaker Store
ScanSpeak Discovery D2608/9130 1" Textile Dome HDS Tweeter: Madisound Speaker Store
ScanSpeak Classic 18W/8545 7" Mid Woofer Paper Cone: Madisound Speaker Store
Eton 7-300/A8 Symphony, 7" Mid/Bass with Phase Plug: Madisound Speaker Store
Back when Small and others wrote their papers I doubt there were very few tweeters if any this would work with using the woofers and tweeters that were available.
There would have been a large dip in the frequency respnse between the drivers. The first tweeter that this would have worked with was the Dynaudio D-28H that I am aware of.
There would have been a large dip in the frequency respnse between the drivers. The first tweeter that this would have worked with was the Dynaudio D-28H that I am aware of.
I've removed a bunch of recent posts with personal attacks. Some technical content was collateral damage. Cease and desist name-calling- it's OK and encouraged to criticize ideas.
1- It's "Zobel", please spell it correctly.
2- I never said that they did. it was a 'tolerance bracket' stating tweeters have typically <0.1mH inductance.
3- I don't have the AES papers, nor does everyone else. You have to make the assumption that not all posters have all the documentation if your proof is in them.
So- you are using the inverse of the tweeter inductance to find the C1 value?
Do you add a second coil for Baffle Step comp?
I may have to try and simulate this to see....
Later,
Wolf
C1 = tweeter mh
L1 = woofer mh
R1
R2
R and C in Zobel
I can't type as fast as I would like No spell check function.
Allen B did the simulation on post 112.
Read his posts 158,186
There is no second coil for baffle step up. Driver voice coil centers distance realtive to crossover frequency determine this to get the flattest response.
The formulas I use to figure this out are in many papers to go through.
I can't remember every formula off the top of my head. I am not being sarcastic.
There are bits and pieces of this concept that Fried pieced together from the AES papers. At least that is where I pieced it together from. He implemented their use around 83.
When he switched from parallel crossover to series crossover in his Model C2 satellite and O2 subwoofer all the values were the same. The layout was different.
I would be curious what your simulation R1 and R2 values come up with.
Old Colony Sound use to have all the papers in book. I think itis no longer available.
L1 = woofer mh
R1
R2
R and C in Zobel
I can't type as fast as I would like No spell check function.
Allen B did the simulation on post 112.
Read his posts 158,186
There is no second coil for baffle step up. Driver voice coil centers distance realtive to crossover frequency determine this to get the flattest response.
The formulas I use to figure this out are in many papers to go through.
I can't remember every formula off the top of my head. I am not being sarcastic.
There are bits and pieces of this concept that Fried pieced together from the AES papers. At least that is where I pieced it together from. He implemented their use around 83.
When he switched from parallel crossover to series crossover in his Model C2 satellite and O2 subwoofer all the values were the same. The layout was different.
I would be curious what your simulation R1 and R2 values come up with.
Old Colony Sound use to have all the papers in book. I think itis no longer available.
In a standard 2 way design the C2C spacing, regardless of the crossover used, between the tweeter and the mid/bass will have virtually nothing to do with, or to do with compensating for baffle step.
The enclosures are narrow. They are in transmission lines not a closed box.
They are phase coherent.
They are phase coherent.
Last edited:
None of post 235 has anything to do with correcting for baffle-step losses. The only relevant part is that the enclosures are narrow, effectively pushing the baffle-step frequency higher. But unless you've got an enclosure that's like 5cm wide then you can forget correcting for baffle step losses with a normal crossover to a tweeter.
They have a fairly flat frequency response according to simulation graphs which is the end result.
And I take it that your initial measurements taken for simulating include the effects of baffle-step?
I have all ready stated in prior posts that I have measured use warble tones with spl meter with my pyramid shaped satellites and subwoofer.
I calculated octave of roll off with subs also being disconnected from satellites givng 12db per octave after second octave .
The first octave cut off to sub was about 8- 10b. I am going by memory I have 2 sets of the satellite and sub models.
One crosses over at 100hz another at 200hz.
The meter accuracy is plus or minus 2 db from 32 -10,000hz
I calculated octave of roll off with subs also being disconnected from satellites givng 12db per octave after second octave .
The first octave cut off to sub was about 8- 10b. I am going by memory I have 2 sets of the satellite and sub models.
One crosses over at 100hz another at 200hz.
The meter accuracy is plus or minus 2 db from 32 -10,000hz
That isn't what I asked though. I asked you if your initial measurements, the ones you took of the filterless drivers in the enclosures, were done so as to factor in baffle-step losses.
You mentioned that the 'simulations' show a flat response, so you must have measured the drivers initially for those simulations right?
You mentioned that the 'simulations' show a flat response, so you must have measured the drivers initially for those simulations right?
- Status
- Not open for further replies.