Hi,
Somebody round here is in cuckooland. I haven't asked you any
direct questions, you haven't answered anything. Its rude in my
book to expect somebody to respond to meaningless ramblings.
rgds, sreten.
How anyone with an ounce of technical integrity could link
to that Magnastand page is beyond me. Far too much BS,
and "magic" claims such as his x/o adds 6dB to a Maggies
sensitivity, it cannot, and he even has the cheek to suggest
the model he doesn't like this mysteriously doesn't happen.
Last edited:
As already stated Thiel uses parallel. They aim for a 1st order 6dB/octave slope though and they obtain that by using an ungodly amount of passive components to get the final acoustic response to 1st order. Much better in my opinion would be using less components and aiming for 3rd or 4rth order.Yes they use more in trying to hit a 1st order slope.
Your thinking all metal cones sound "tinty" is probably you using series crossovers that don't allow any peak damping circuits and just ignoring the peaks in the midrange/woofer. I guess us parallel crossover people who can smooth out any peaks are living in the past 😛 Do you use ANY measurement equipment or ANY simulation software? Don't blame "tinty" sound on the driver, blame it on the designer of the finished product.
DavidL I never used metal cones. The Thiel metal cone speakers do sound tinty to my ears. The few metal cone speakers I have heard all sound tinty. By the way they are not my precious series crossovers . Fried designed them.
The series crossover values on the internet are from AES papers. I guess most forum users think Thielle , Small. Lipshitz, Ashley and Vanderkooy's work on series crossovers are wrong.
Using a 6db per octave crossover has been put down by many forum users. I know Jim Thiel uses 6 db per octave crossovers. It's Thiel and Partner that use 6db per octave series crossoveers. Poor choice of wording about Jim Thiel I did not mean to mislead any forum users.
http://www.humblehomemadehifi.com/download/Humble Homemade Hifi_Soup Sandwich_copy.pdf
It would seem Fried's designs using series crossovers and t-lines is not politically correct on this forum.
The series crossover values on the internet are from AES papers. I guess most forum users think Thielle , Small. Lipshitz, Ashley and Vanderkooy's work on series crossovers are wrong.
Using a 6db per octave crossover has been put down by many forum users. I know Jim Thiel uses 6 db per octave crossovers. It's Thiel and Partner that use 6db per octave series crossoveers. Poor choice of wording about Jim Thiel I did not mean to mislead any forum users.
http://www.humblehomemadehifi.com/download/Humble Homemade Hifi_Soup Sandwich_copy.pdf
It would seem Fried's designs using series crossovers and t-lines is not politically correct on this forum.
Last edited:
That's not what's being said. What's being said is that you are not explaining their methods with anything but random occurences, and haphazard guesses.
Noone is disputing their work, just your hunches.
Later,
Wolf
It is not hunches. Small was the first to state to subtract the voice coil inductance of the woofer being crossed over to the tweeter. Many have disputed this. From having many of the driver specs and crossover plans I know what the company did.
When Fried used parallel crossovers at 6db per octave he used a 12 db per octave layout on the B-2 and C models. He stated in his literature they were 6 db per octave to both tweeter and mid bass. Both drivers were oem and had the same excat voice coil inductnace as L1 .
A few years back Parts Express had a buy out of some Vifa drivers the company used. I also have those crossover plans using the ring radiator tweeter. The company did the same with L1 but in a 4 ohm design. I never purchased any since the former owner stated they were inferior to the composite cones I currently use.
Also using these designs in t-lines flatten the response. Have you seen any of the Fried series crossover plans and driver specs?
Vifa PL18WO09-04 7" Woofer 4 Ohm 299-225
When Fried used parallel crossovers at 6db per octave he used a 12 db per octave layout on the B-2 and C models. He stated in his literature they were 6 db per octave to both tweeter and mid bass. Both drivers were oem and had the same excat voice coil inductnace as L1 .
A few years back Parts Express had a buy out of some Vifa drivers the company used. I also have those crossover plans using the ring radiator tweeter. The company did the same with L1 but in a 4 ohm design. I never purchased any since the former owner stated they were inferior to the composite cones I currently use.
Also using these designs in t-lines flatten the response. Have you seen any of the Fried series crossover plans and driver specs?
Vifa PL18WO09-04 7" Woofer 4 Ohm 299-225
I think what most people are tired of is you making like series crossovers are WAY better than parallel then failing to show how they are and in fact showing just the opposite. The time to come up with a proper design is longer, you are stuck using certain drivers, you change one component and the whole response go to crap and you have to start over, well that's enough examples.
The GR Research Paradox 1 crossover was also designed by one of the companies engineers.
When a person eats humble pie they can get more of their questions answered as I did.
It's gotten to the point so what. If most forum users are satisfied with their high order parallel crossovers it's their loss.
Paradox 1
When a person eats humble pie they can get more of their questions answered as I did.
It's gotten to the point so what. If most forum users are satisfied with their high order parallel crossovers it's their loss.
Paradox 1
Exactly...so what? 🙄
"If most forum users are satisfied with their high order parallel crossovers it's their loss."
It's THAT sort of response that makes you look like an audio snob know it all...lighten up please. Please explain HOW using a parallel circuit with even greater roll off of the drivers is our loss? This should be good.
"If most forum users are satisfied with their high order parallel crossovers it's their loss."
It's THAT sort of response that makes you look like an audio snob know it all...lighten up please. Please explain HOW using a parallel circuit with even greater roll off of the drivers is our loss? This should be good.
DavidL you are not stuck using certain drivers. I have used many drivers over the years and tested to till I obtained the flattest response possible. Vifa, Audax, Carboneau, Dynaudio, Seas drivers using bextrene, polypropylene, paper cones. Aluminum, textile domes, paper cone tweeters from Parts express , Madisoun, Zalytron, Martin Sound and Speaker City.
Unfortunately most of the drivers I have used were discontinued except the Hiquphon tweeters .
DavidL read many of the prior posts and AES papers and other books where this topic has been discussed.
Buy the disc from Old Colony Sound of Speaker Builder back issues. There are numerous articles and debates over series compared to parallel crossovers.
Unfortunately most of the drivers I have used were discontinued except the Hiquphon tweeters .
DavidL read many of the prior posts and AES papers and other books where this topic has been discussed.
Buy the disc from Old Colony Sound of Speaker Builder back issues. There are numerous articles and debates over series compared to parallel crossovers.
Last edited:
I see that it was discontinued because Focal changed the tweeter to a cheaper product.
The new model they came out with the N1 uses a parallel network hmmmm I guess they found it easier to design and get good results going that route.
Well you mentioned using lower order series crossovers so if you use 1st order 6dB per octave slopes you are stuck using certain drivers if you want decent results
Just what was your "flattest response possible" by the way? Was it trial and error design or computer sims to help out?.
I already have all the Speaker Builder issues and have had them for years. That doesn't change my possition on parallel vs series.
The man that designed the crossover has other obligations.
He is involved with the new Fried company.
Would you want to compete with your own company.
He is involved with the new Fried company.
Would you want to compete with your own company.
Last edited:
I have already answered many of your questions in prior posts. Some have a 6and 12 db per octave. Some have a 6 and 18 db per octave.
What speakers with series crossovers have you listened to? There were many speaker reviews of the company in back issues of magazines you would have to purchase.
A few issues of Speaker Builder had some heated articles about the company and Gary Galo? There were some informative artilcles about series crossovers in the early 80's.
The Fried designs sound a lot like the Von Schweiikert and Meadowlark designs.
I am not an audio snob. I live close to where Fried assembled his speakers and became a kit customer when I was a teenager. I asked the man many questions and was given some very complicated answers.
Plus or minus 2 db from 30to 10,000hz. That is the range of the spl meter using warble tones.
What speakers with series crossovers have you listened to? There were many speaker reviews of the company in back issues of magazines you would have to purchase.
A few issues of Speaker Builder had some heated articles about the company and Gary Galo? There were some informative artilcles about series crossovers in the early 80's.
The Fried designs sound a lot like the Von Schweiikert and Meadowlark designs.
I am not an audio snob. I live close to where Fried assembled his speakers and became a kit customer when I was a teenager. I asked the man many questions and was given some very complicated answers.
Plus or minus 2 db from 30to 10,000hz. That is the range of the spl meter using warble tones.
Last edited:
Series or parallel?
I've used both to good effect, but only when the drivers are suitable. I'll explain.
A proper first order crossing pretty much requires drivers with wide bandwidth at least 3.5X (preferably 5x) on either side of the xo point, in which case the actual acoustic phase relationship between them allows the use of a simple 1st order, series or parallel, as long as the drivers acoustic positions are equidistant from the listening position. When these caveats are met, and if the driver impedances are very uniformly resistive, the responses between series or parallel would be identical.
However, if we are using real world drivers, most of which exhibit large resonant impedance peaks and increasing HF impedance due to vc inductance, then the interaction between the passive components and the driver impedances might favor the series arrangement. The parts count can be much lower. I believe this is what Fried did with the subs on his model H subs, and it worked very well, whereas a parallel implementation would not have worked without extremely complex and costly conjugate impedance networks.
1) For example, try using a single 10mH choke in a "parallel filter" configuration for the woofer ... then apply it to a 40Hz vented woofer and model the interaction... not pretty. Both the simulation and measurements will show gross deviations from flat response.
Trust me, I've done this. It was baaad. And you know what really hurt?
The speaker company I was working for forced me to release the product anyway!......because the marketing dept really liked that +9dB peak at 70Hz, even if there was nuthin' left below 55Hz. Then they made me use a HUGE mouth port which brought the peak up to 80Hz, at +12dB. Then they fired me for complaining about it. Yup. That was fun.
Adding insult to injury, that "passive" sub went on to become a best-seller and a consumer's digest best buy!! Go figure...
I wish they'd let me spend the extra pennies to implement the series version of it and kept the response flat to 40Hz... hmmm....probably would not have sold at all!
So on that front, score one for series networks, they will often win on lower parts counts for low bass xo's
2) Now, what if you want to cross over a 6 " driver with a nice clean built-in "natural" 4rd order 3.5KHz roll-off, and still want to use a simple 1st order electrical network?
You use a tweeter with a sub 1Khz roll-off (preferably something like the Scan-speak Aircirc which reaches to near 500Hz) and use a single cap to feed the tweeter, a single coil to feed the woofer and then "time-align" the tweeter by pushing it back in space by about 5cm (from actual vc location) to compensate for the approximately uniform group-delay caused by the woofer's natural top end rolloff. Then you compare that parallel implementation to the series version and.... in terms of performance, and without zobels or other impedance conjugates, due to the woofer's inductance, the series network will usually be better once again, good performance, less parts. Score another one for series.
3) Wanna save a part? Use the same woofer as above and drive it directly, with no inductor, and drive the tweeter with same single cap feeding the tweeter. Now, the tweeter only gets pushed back by about 3.75cm to get correct phase/time alignment on-axis. This configuration will only work for "parallel" connection, and there will be some fairly large impedance fluctuations near the crossover region. But you save a part, and very often that "direct"drive will bring the midrange alive in a very big way.
So on this front, the advantage probably goes to the "parallel" implementation.
So, then: when faced with wanting to reduce part count and cost, in SOME situations, and providing the drivers BW is conducive to it, series can be a real winner. But if your drivers have nasty peaks that need taming near the xo frequency, fuhggedaboutit.
Hope this helps reconcile the two sides. War is no fun, except for the warriors. Everyone else gets splattered if they dare come too close.
Best Regards!
I've used both to good effect, but only when the drivers are suitable. I'll explain.
A proper first order crossing pretty much requires drivers with wide bandwidth at least 3.5X (preferably 5x) on either side of the xo point, in which case the actual acoustic phase relationship between them allows the use of a simple 1st order, series or parallel, as long as the drivers acoustic positions are equidistant from the listening position. When these caveats are met, and if the driver impedances are very uniformly resistive, the responses between series or parallel would be identical.
However, if we are using real world drivers, most of which exhibit large resonant impedance peaks and increasing HF impedance due to vc inductance, then the interaction between the passive components and the driver impedances might favor the series arrangement. The parts count can be much lower. I believe this is what Fried did with the subs on his model H subs, and it worked very well, whereas a parallel implementation would not have worked without extremely complex and costly conjugate impedance networks.
1) For example, try using a single 10mH choke in a "parallel filter" configuration for the woofer ... then apply it to a 40Hz vented woofer and model the interaction... not pretty. Both the simulation and measurements will show gross deviations from flat response.
Trust me, I've done this. It was baaad. And you know what really hurt?
The speaker company I was working for forced me to release the product anyway!......because the marketing dept really liked that +9dB peak at 70Hz, even if there was nuthin' left below 55Hz. Then they made me use a HUGE mouth port which brought the peak up to 80Hz, at +12dB. Then they fired me for complaining about it. Yup. That was fun.
Adding insult to injury, that "passive" sub went on to become a best-seller and a consumer's digest best buy!! Go figure...
I wish they'd let me spend the extra pennies to implement the series version of it and kept the response flat to 40Hz... hmmm....probably would not have sold at all!
So on that front, score one for series networks, they will often win on lower parts counts for low bass xo's
2) Now, what if you want to cross over a 6 " driver with a nice clean built-in "natural" 4rd order 3.5KHz roll-off, and still want to use a simple 1st order electrical network?
You use a tweeter with a sub 1Khz roll-off (preferably something like the Scan-speak Aircirc which reaches to near 500Hz) and use a single cap to feed the tweeter, a single coil to feed the woofer and then "time-align" the tweeter by pushing it back in space by about 5cm (from actual vc location) to compensate for the approximately uniform group-delay caused by the woofer's natural top end rolloff. Then you compare that parallel implementation to the series version and.... in terms of performance, and without zobels or other impedance conjugates, due to the woofer's inductance, the series network will usually be better once again, good performance, less parts. Score another one for series.
3) Wanna save a part? Use the same woofer as above and drive it directly, with no inductor, and drive the tweeter with same single cap feeding the tweeter. Now, the tweeter only gets pushed back by about 3.75cm to get correct phase/time alignment on-axis. This configuration will only work for "parallel" connection, and there will be some fairly large impedance fluctuations near the crossover region. But you save a part, and very often that "direct"drive will bring the midrange alive in a very big way.
So on this front, the advantage probably goes to the "parallel" implementation.
So, then: when faced with wanting to reduce part count and cost, in SOME situations, and providing the drivers BW is conducive to it, series can be a real winner. But if your drivers have nasty peaks that need taming near the xo frequency, fuhggedaboutit.
Hope this helps reconcile the two sides. War is no fun, except for the warriors. Everyone else gets splattered if they dare come too close.
Best Regards!
Last edited:
speakerman 19422,your sojourn in the wilds seem to have imbued you with the hunted prey instinct of dodging and weaving to avoid capture.It does not work for you a hunter' as you keep shooting yourself in the foot.In referencing Richard Small and I would think in particular,his paper "Constant Voltage Crossover Design" the section dealing with Network Termination and Driver Equalization;where the combination of the woofer coil inductance as part of the total inductance is discussed,a reference to the source of this method is given to N.H. Crowhurst,via Audio Engineering,1970.The crucial point I wish to make is that this was suggested for a parallel network.From personal experience,it does not actually work and a Zobel network or better still a fully implemented impedance correcting cct is better and has become the universal approach.
VaNarn where did you start reading the thread? The designs use Zobels. The designs use the mid bass driver voice coil mh as L1. A Vifa 6.5 with voice coil mh of .5mh is the L1 value.
Hi,
This is precisely what I found with my 3-Way 1st order crossover Speaker (300Hz/3KHz crossover). Getting it to work with a simple parallel crossover required an LCR trap and Zobel on the midrange (the tweeter as ribbon is resistive) and the woofers crossover also was problematic.
Reconfiguring the crossover to series operation allowed me to drop all the conjugates and have a "pure" crossover (only some EQ and Level matching for the tweeter) and it produces both flat response and impedance (excluding bass loading peaks).
So my Bass crossover now consists of two Copper Foil Chokes (a pair of 1.5mH in series) and two film capacitors (68uF Auricap - I had them around).
For the "theoretically perfect crossover simulation" brigade, this crossover does NOT follow any standard math, but is the result of iterative optimisation.
I had the reverse problem (my Midrange is actually a fullrange driver with reasonable response to 1KHz, but the tweeter drops below 3KHz at a 2nd order rolloff.
And again, parallel 1st order networks showed poor integration, series worked better. This in my case uses a copper foil coil and Tinfoil/Polypropylene Capacitors (a lot in parallel).
Yes, precisely. You need drivers that are essentially flat by themselves and have wide bandwidth. In my speaker the "Woofer" is fat to above 1KHz with a claan rolloff, the "Midrange" is really a fullrange that build in covers 80Hz-15KHz. Only the tweeter is bandwidth challenged at the lower end...
It is all bout context. In most cases I find parallel types are the only ones that can be made work correctly...
Ciao T
This is precisely what I found with my 3-Way 1st order crossover Speaker (300Hz/3KHz crossover). Getting it to work with a simple parallel crossover required an LCR trap and Zobel on the midrange (the tweeter as ribbon is resistive) and the woofers crossover also was problematic.
Reconfiguring the crossover to series operation allowed me to drop all the conjugates and have a "pure" crossover (only some EQ and Level matching for the tweeter) and it produces both flat response and impedance (excluding bass loading peaks).
So my Bass crossover now consists of two Copper Foil Chokes (a pair of 1.5mH in series) and two film capacitors (68uF Auricap - I had them around).
For the "theoretically perfect crossover simulation" brigade, this crossover does NOT follow any standard math, but is the result of iterative optimisation.
I had the reverse problem (my Midrange is actually a fullrange driver with reasonable response to 1KHz, but the tweeter drops below 3KHz at a 2nd order rolloff.
And again, parallel 1st order networks showed poor integration, series worked better. This in my case uses a copper foil coil and Tinfoil/Polypropylene Capacitors (a lot in parallel).
Yes, precisely. You need drivers that are essentially flat by themselves and have wide bandwidth. In my speaker the "Woofer" is fat to above 1KHz with a claan rolloff, the "Midrange" is really a fullrange that build in covers 80Hz-15KHz. Only the tweeter is bandwidth challenged at the lower end...
It is all bout context. In most cases I find parallel types are the only ones that can be made work correctly...
Ciao T
I hope you meant to write: In most cases I find series types are the only ones that can be made work correctly....
If that's the case, your experience (similar to mine quoted earlier for a 2-way) will be largely ignored by the "theoretically perfect crossover simulation" brigade, like I was. 🙂
Hi,
No. Many modern drivers have behaviours that are not compatible with series X-Overs, from rising midrange response to severe resonances. The circuitry needed to fix these often does not cooperate well well with series crossovers.
One may argue these drivers are "defective by design" however the current trend to ever stiffer and even less well damped cones made out the latest version of BovineExcrementium (R) together with a disregard for flat in-band response seem to have become the current standard.
It will be ignored anyway, as much of what I have to say runs counter to what most people want to desperately believe in. So anything that does not fit this belief system is rejected anyway...
Ciao T
No. Many modern drivers have behaviours that are not compatible with series X-Overs, from rising midrange response to severe resonances. The circuitry needed to fix these often does not cooperate well well with series crossovers.
One may argue these drivers are "defective by design" however the current trend to ever stiffer and even less well damped cones made out the latest version of BovineExcrementium (R) together with a disregard for flat in-band response seem to have become the current standard.
It will be ignored anyway, as much of what I have to say runs counter to what most people want to desperately believe in. So anything that does not fit this belief system is rejected anyway...
Ciao T
Well, I guess I somehow misundertood your earlier post. You extolled the virtues of series xo's in your experience and then you write that last sentence that said the opposite. Confusing indeed 😕
I shared the details of my design and testing. How about you?
I shared the details of my design and testing. How about you?
- Status
- Not open for further replies.