I'd like to begin construction of the Seas Classic 3 Way - flat baffle version, not the current edition - with drivers from the Seas Excel line - 8 in woofer with Nexel coating [W22NY-001] and Excel T25CF-002 tweeter; still searching for a midrange so this might retain the original to the design. I realize that Troels states any change invalidates the crossover configuration, but am looking for opinions on just how out of bed these substitutions may take the sound "quality" of the speaker system. Slightly, so as to be a decent tuning start point or horrific, don't even bother?
I've searched Madisound archives for a configuration with these driver updates, but no luck. Someone will doubtless point to several software options for crossovers to start from scratch. As a long term aerospace engineer, I'm familiar with the issue of folks believing they are experts because they can simply run a piece of software; I have no illusions that I'll quickly learn the empirical guidelines readily to design a crossover well, just because I can input parameters - so I'm looking for a decent system starting point.
I've searched Madisound archives for a configuration with these driver updates, but no luck. Someone will doubtless point to several software options for crossovers to start from scratch. As a long term aerospace engineer, I'm familiar with the issue of folks believing they are experts because they can simply run a piece of software; I have no illusions that I'll quickly learn the empirical guidelines readily to design a crossover well, just because I can input parameters - so I'm looking for a decent system starting point.
The crossover topology should be transferable to the Nextel Excel drivers. Troels has what, 8 different driver versions using the basic design, but component values would certainly need tweaking, and maybe also the box volume and tuning frequency. P.S. the W15 Nextel midwoofer is on sale at Madisound atm; that should work for the mid. There's also the new edge coated 4", but that's even more expensive. I remember Troels once said the M15 Nextel dedicated mid was a bit tricky to work with (can't remember if it was the Cyclop design or one of the 4 ways).
AllenB's suggestion is a good next step. At least compare the curves of the original drivers to your intended replacements, and sim the general response using the MK1 crossover with the Nextel drivers. Cross reference it also to the Scan Discovery 3WC MK1; I think that XO is now public, too, although it uses a smaller mid and different XO points. You should be able to sim something reasonable to determine feasibility. Look at the various XO frequencies Troels settled on for different designs where he uses the Crescendo tweeter to get a view on the range.
These are dang expensive drivers, though. I don't see great value for performance in the tweeter and the 8". It would also be a bit of a risk to contemplate a fully realized design using different drivers without measurement gear and XO design experience. Not to say it can't work.
AllenB's suggestion is a good next step. At least compare the curves of the original drivers to your intended replacements, and sim the general response using the MK1 crossover with the Nextel drivers. Cross reference it also to the Scan Discovery 3WC MK1; I think that XO is now public, too, although it uses a smaller mid and different XO points. You should be able to sim something reasonable to determine feasibility. Look at the various XO frequencies Troels settled on for different designs where he uses the Crescendo tweeter to get a view on the range.
These are dang expensive drivers, though. I don't see great value for performance in the tweeter and the 8". It would also be a bit of a risk to contemplate a fully realized design using different drivers without measurement gear and XO design experience. Not to say it can't work.
Last edited:
And you have AllenB helping. Personally, I'm quite interested to learn how to replicate a given transfer function with different drivers without measurement gear. I've also heard the Nextel drivers in the Diapason Dynamis at a show, and well, that sounds bloody amazing.
@AllenB here's the existing design: http://www.troelsgravesen.dk/3WClassic.htm
I modeled it here with FRD and ZMA files for the exact MK1 drivers from dibirama - not in baffle measurements but overall it seems usable as a starting point.
Attached the vituixcad project file as a zip.
Dibirama doesn't have FRD and ZMA files for the Nextel drivers, but I can work on those from the spec sheets.
@AllenB here's the existing design: http://www.troelsgravesen.dk/3WClassic.htm
I modeled it here with FRD and ZMA files for the exact MK1 drivers from dibirama - not in baffle measurements but overall it seems usable as a starting point.
Attached the vituixcad project file as a zip.
Dibirama doesn't have FRD and ZMA files for the Nextel drivers, but I can work on those from the spec sheets.
Attachments
What you are proposing doesn't make much sense as I think you realise. You either follow someone else's design with minor cosmetic changes or you pickup the task of sorting the design yourself which requires a reasonable grasp of the relevant engineering. Another thing making life difficult is picking expensive drivers first and then trying to design a speaker around them (assuming you opt to design yourself) rather than picking a speaker configuration/design and then looking for appropriate drivers that fit the design. A well sorted design tends to bring more to the sound quality than expensive drivers and this is more so with a 3 way where the drivers are less required to work outside a reasonable operating range than a 2 way.
What is the spec for your speaker? Is it an 8" 3 way monitor like the Seas Classic? A speaker using Excel drivers? A better monitor than Seas Classic?
Thanks, I think I follow. Troels published the measured responses. I guess I need a way to import that response per curve, then tweak the Nextel drivers crossover values to match it. Not sure how to do that import step yet. Will investigate.
Here are the FRD and ZMA files I created from the Seas spec sheets (Nextel W22, W15 and T29 Crescendo).
And how they look directly swapped into the original crossover as is - it looks a lot more workable than I expected; maybe cause it's from the same company, even though the tweeter is bigger.
Here are the FRD and ZMA files I created from the Seas spec sheets (Nextel W22, W15 and T29 Crescendo).
And how they look directly swapped into the original crossover as is - it looks a lot more workable than I expected; maybe cause it's from the same company, even though the tweeter is bigger.
Attachments
Okey doke. Will create files from the original drivers from the spec sheet shortly.
P.S. also read the new desired tweeter wrong (got the T29 Crescendo mixed up with the T25 Milennium - Troels usually uses the Crescendo with the Nextel drivers; have to do that one, too)
To confirm, should I be trying to make the new drivers match the final measured responses, per driver?
Or should I be tweaking the existing XO to make the new drivers match the XO's response in my sim (i.e. don't need to use the measured responses at all)?
Thinking about I guess it would be the latter - I suppose we are saying the baffle step and diffraction will be similar/constant.
Original drivers' response with published XO:
Nextel drivers' response with published XO (T29 tweeter for illustration. The T25 Millenium is quite a bit less sensitive which might be an issue):
P.S. also read the new desired tweeter wrong (got the T29 Crescendo mixed up with the T25 Milennium - Troels usually uses the Crescendo with the Nextel drivers; have to do that one, too)
To confirm, should I be trying to make the new drivers match the final measured responses, per driver?
Or should I be tweaking the existing XO to make the new drivers match the XO's response in my sim (i.e. don't need to use the measured responses at all)?
Thinking about I guess it would be the latter - I suppose we are saying the baffle step and diffraction will be similar/constant.
Original drivers' response with published XO:
Nextel drivers' response with published XO (T29 tweeter for illustration. The T25 Millenium is quite a bit less sensitive which might be an issue):
Last edited:
Motoko & AllenB - I'm extremely appreciative of the of your inputs, modeling and suggestions! Good to hear that the alternate driver show promise. I'll look up the referenced software files to attempt becoming at least familiar enough to follow along and, hopefully, interact better with your experienced comments. It's also a boost that you've heard the Seas Nextel products as I'm going strictly from other comments and the idea that improvements over the years in their line could yield a desirable assembly. I picked the Seas Classic as a basis from positive reviews I've read and the fact that [I believe so far] they have not outsourced their products. From stepping out of aerospace a while to run a commercial design engineering department who dealt with offshore supply, I know the quality control in that path can vary greatly. I began searching for a current replication of the KLH Model 5 [dating myself here] and evolved into the 8 in woofer version of a 3 way from reading about the Classic.
I've read to same [perhaps] comments you cited on the Nextel 5 in midrange - giving me a little pause to say it should be the one, but will look for more inputs - to be clear I think it's what you used in the simulation above, correct - Seas Excel M15CH-002 or did I misinterpret that? Anyway, that's how I came to consider keeping the Classic's Seas MCA12RC midrange. Looking forward to further inputs.
As a bit of background, I'm entirely comfortable on the cabinet side being an experienced woodworker and cabinet maker for my own renovation projects. My son has a commercial shop with CNC router, vacuum bagging and access to BB ply and veneers. I worked on commercial video equipment repairs in college and built a couple of the early Hafler amp & preamp kits - but there's big difference in following a manual / installing pre-defined mods in the Halfers and design from scratch - mechanical part, I got that - electronics, reaching out to the community here for assistance.
I've read to same [perhaps] comments you cited on the Nextel 5 in midrange - giving me a little pause to say it should be the one, but will look for more inputs - to be clear I think it's what you used in the simulation above, correct - Seas Excel M15CH-002 or did I misinterpret that? Anyway, that's how I came to consider keeping the Classic's Seas MCA12RC midrange. Looking forward to further inputs.
As a bit of background, I'm entirely comfortable on the cabinet side being an experienced woodworker and cabinet maker for my own renovation projects. My son has a commercial shop with CNC router, vacuum bagging and access to BB ply and veneers. I worked on commercial video equipment repairs in college and built a couple of the early Hafler amp & preamp kits - but there's big difference in following a manual / installing pre-defined mods in the Halfers and design from scratch - mechanical part, I got that - electronics, reaching out to the community here for assistance.
Sorry, I don't understand what this thread is heading to.
Simulation is one way to go, done right, but not from a dead start.
Any drivers response is dependent of the location it is mounted in. So the only way to start such a design is to choose a baffle and install the drivers. Then measure and take these measurements for a simulation. Measuring gear, that is perfectly accurate for such a task, costs less than a fraction of one of the projected drivers. This consists of a calibrated microphone and (if the mike is not USB) a sound card, even as most better laptops have usable inputs. Inputs can be tested for linearity in two minutes.
If you were going to use a few cheap drivers, such coarse simulation, as even the most complicated ones will be just that, may be OK. Even as I would oppose this opinion in any case. If you are going to spend more than a thousand $ on chassis, why not get measuring equipment for $150? Free software has come a long way and is 100% of what you need, easy to learn, too.
No one constructing serious loudspeakers does not measure from the beginning to the end of a project.
My personal opinion, in your case, go for an active setup with a DSP. If you have some talent, this will get you a perfect result or at least near to it.
What may surprise you, this option is much cheaper than the passive x-over, as you do not need to buy a few dozens of try and error capacitors, coils and resistors. These once cheap parts have ten fold in price (for us end users) without any technical advance, since I started to build speakers, while active gear has reached a ridiculous low price point with a quality never seen before. Some good Class D amp modules and one or two DSP will beat any amp you own, hands down. Not considering your "simulated", expensive x-over.
Please think about it: DSP's are for beginners while passive cross over design is something only few people on this planet really can do. Like Troels. Not the other way round. The great advantage of such an "in home made" active speaker, you can match it to your listening room. Think how funny this is, other people build the best passive speaker kits and then buy an expensive DSP system to fit it into their listening environment.
Please do not fall for the voodoo, passive parts would add magic touches to the speaker chassis. Any capacitor, coil or resistor degrades sound. Some of the most linear passive speakers, using complicated multi part x-overs, sound dead in real life, while a single broad band driver may be fantastic fun.
If you decide to go active, you will find all the information and help you need at this forum.
Trust me, installing some D-amp modules in a case is much simpler than designing a passive x-over and less frustrating.
Maybe time to think your way to design a speaker over?
Simulation is one way to go, done right, but not from a dead start.
Any drivers response is dependent of the location it is mounted in. So the only way to start such a design is to choose a baffle and install the drivers. Then measure and take these measurements for a simulation. Measuring gear, that is perfectly accurate for such a task, costs less than a fraction of one of the projected drivers. This consists of a calibrated microphone and (if the mike is not USB) a sound card, even as most better laptops have usable inputs. Inputs can be tested for linearity in two minutes.
If you were going to use a few cheap drivers, such coarse simulation, as even the most complicated ones will be just that, may be OK. Even as I would oppose this opinion in any case. If you are going to spend more than a thousand $ on chassis, why not get measuring equipment for $150? Free software has come a long way and is 100% of what you need, easy to learn, too.
No one constructing serious loudspeakers does not measure from the beginning to the end of a project.
My personal opinion, in your case, go for an active setup with a DSP. If you have some talent, this will get you a perfect result or at least near to it.
What may surprise you, this option is much cheaper than the passive x-over, as you do not need to buy a few dozens of try and error capacitors, coils and resistors. These once cheap parts have ten fold in price (for us end users) without any technical advance, since I started to build speakers, while active gear has reached a ridiculous low price point with a quality never seen before. Some good Class D amp modules and one or two DSP will beat any amp you own, hands down. Not considering your "simulated", expensive x-over.
Please think about it: DSP's are for beginners while passive cross over design is something only few people on this planet really can do. Like Troels. Not the other way round. The great advantage of such an "in home made" active speaker, you can match it to your listening room. Think how funny this is, other people build the best passive speaker kits and then buy an expensive DSP system to fit it into their listening environment.
Please do not fall for the voodoo, passive parts would add magic touches to the speaker chassis. Any capacitor, coil or resistor degrades sound. Some of the most linear passive speakers, using complicated multi part x-overs, sound dead in real life, while a single broad band driver may be fantastic fun.
If you decide to go active, you will find all the information and help you need at this forum.
Trust me, installing some D-amp modules in a case is much simpler than designing a passive x-over and less frustrating.
Maybe time to think your way to design a speaker over?
@TroutBear the mid I'm thinkng of here is the W15LY-001. The MCH15 I remember Troels said it needed a lot more work for the chamber and the crossover in a particular design. I'll try to find that source. If you're going to use the W22 Nextel, I think it makes sense to use the matching cone material from the 5" rather than the MCA12 or MCA15. Relatedly, the Millenium tweeter is an interesting choice. I've also heard this in the Joseph speakers. It is 3db less sensitive than the 27TFFC, though, and also less sensitive than the flagship and nutso expensive T29 Crescendo, which Troels always uses with the Nextel designs, and which is also used in the Diapason and Kudos speakers. Another alternative for you is the T25CF001 which is a lot cheaper, more sensitive, and easy to work with (well, I've simmed it in the past). Actually, I think these are used in my ATC SCM19 v1s, and they sound good to me!
A point of interest, I'm not entirely sure how much design vs manufacturing Seas does natively anymore, rather than outsource things to Kurt Muller. Who knows.
@Turbowatch2 - well sure. That was my first comment response. Can this swap be done? Yes. Is it done optimally without measuring gear? Maybe not, but as per @AllenB's reply, I infer that the replication of transfer functions from simulation, using apples to apples measurements from Seas themselves, should work fine. I find this an intriguing notion as it would allow me to explore the same principle in a design I had in mind.
DSP needs a lot work too with multiples amps, software fiddling, and the vagaries of destructive ADC/DAC from cheapo ADC/DAC, on top of low-end class D amps, like you'd find in low-end studio monitors. That's not my cup of tea.
Your other point that passive speakers are for Albert Einstein experts only, and DSP is therefore the way to go for us plebs. Hmm, I don't agree with this at all. The whole sticky design without measurements thread would be meaningless otherwise, and it clearly begins not by denying you don't want measurements, but that it's still possible to design and then tweak a good sounding speaker yourself using maths and acoustics knowledge. Also, if you read say that Hypex Fusion amps thread, it seems to me a lot of users basically all suffer from the filter design software being absolutely shite and buggy.
I'm still waiting for the reply on my question to explain better why this endeavor here is still credible, and how to take the sims forward.
The other thing I'm still dodgy about is the W22 Nextel models quite differently than the CA22 for box volume; quite a lot less bass with 37L.
A point of interest, I'm not entirely sure how much design vs manufacturing Seas does natively anymore, rather than outsource things to Kurt Muller. Who knows.
@Turbowatch2 - well sure. That was my first comment response. Can this swap be done? Yes. Is it done optimally without measuring gear? Maybe not, but as per @AllenB's reply, I infer that the replication of transfer functions from simulation, using apples to apples measurements from Seas themselves, should work fine. I find this an intriguing notion as it would allow me to explore the same principle in a design I had in mind.
DSP needs a lot work too with multiples amps, software fiddling, and the vagaries of destructive ADC/DAC from cheapo ADC/DAC, on top of low-end class D amps, like you'd find in low-end studio monitors. That's not my cup of tea.
Your other point that passive speakers are for Albert Einstein experts only, and DSP is therefore the way to go for us plebs. Hmm, I don't agree with this at all. The whole sticky design without measurements thread would be meaningless otherwise, and it clearly begins not by denying you don't want measurements, but that it's still possible to design and then tweak a good sounding speaker yourself using maths and acoustics knowledge. Also, if you read say that Hypex Fusion amps thread, it seems to me a lot of users basically all suffer from the filter design software being absolutely shite and buggy.
I'm still waiting for the reply on my question to explain better why this endeavor here is still credible, and how to take the sims forward.
The other thing I'm still dodgy about is the W22 Nextel models quite differently than the CA22 for box volume; quite a lot less bass with 37L.
Last edited:
Found the observation from Troels against the M15. It's from the redacted Cyclop design he never fully finished.
"Why a section on the midcab alone? The M15CH002 is a tricky fellow with a potential "problem" in the 1-2 kHz range, that is when the baffle isn't right. We have a dip here and it adds a significant fingerprint to vocal presentation if you have flat upper midrange up to 1 kHz followed by a ~3 dB dip in the 1-2 kHz area. Using the M15CH on a wide baffle didn't work - I've tried the PMS cabs and had to abandon the PMS-NEXT project - it didn't sound right. Too much presence - in the wrong way. Read below."
"Why a section on the midcab alone? The M15CH002 is a tricky fellow with a potential "problem" in the 1-2 kHz range, that is when the baffle isn't right. We have a dip here and it adds a significant fingerprint to vocal presentation if you have flat upper midrange up to 1 kHz followed by a ~3 dB dip in the 1-2 kHz area. Using the M15CH on a wide baffle didn't work - I've tried the PMS cabs and had to abandon the PMS-NEXT project - it didn't sound right. Too much presence - in the wrong way. Read below."
The FR published by Seas are measurements of drivers on a specific box. The Dibirama ones are on a IEC baffle, and the Troels one are on the intended box (in this case the Seas 3W Classic). Hard to compare them together.
The Seas 3W Classic has the mid operating between 500 and 3500 Hz, and great part of the musical content lies there. If you plan to retain the mid I really don't see the point in changing woofer and tweeter, with all the trouble in measuring and designing a new and appropriate crossover. IMHO the original prestige tweeter is better in many ways than the excel.
Ralf
Suspending the (valid) concerns for the moment, @AllenB can you confirm I'm taking the right approach whereby:
- I load the Seas datasheet specs for the drivers
- I model the published XO as-is
- I tweak the new drivers to match the curve of the original XO and drivers
OK here's my first pass at the woofer:
Some observations:
The W22 reaches 89db at 120hz vs 87.6db for the CA22 - is attenuation needed or do you go with it for the new system?
The W22 doesn't have the breakup peak of the CA22 at 2.8khz.
The W22 doesn't need the parallel resistor to track the curve on the downward slope
Some observations:
The W22 reaches 89db at 120hz vs 87.6db for the CA22 - is attenuation needed or do you go with it for the new system?
The W22 doesn't have the breakup peak of the CA22 at 2.8khz.
The W22 doesn't need the parallel resistor to track the curve on the downward slope
My continued thanks Motoko for the models and your inputs on the tweeter & mid. I had misinterpreted on the mid and agree with your comments there on the W15 - and it's nice that Madison has it on sale. Having used the suggested tweeter in your work, I'm very open to that substitution; my selection on the T25 Millenium was base on must be a reason it evoked. If fact, I believe that I've seen the [older] T25CF-001 in Madison crossover archives. Looks encouraging on the W22 woofer above.
Turbowatch2 - thanks for your perspective. I am, however, continuing on the passive path here. I personally consider high quality components in passives to be a very valid route still in use today - of course I believe the turbocharged 3.0L 5 cylinder Mercedes Diesel engine in the old W123 chassis series a crowning achievement in automotive history and the FJZ80 series solid front axle Toyota Landcruisers to be the last great one in the series, just to offer a calibration... I'll be driving these with a Hafler DH200 until I can locate a McIntosh MC2100 or perhaps slightly higher power model in that same vintage range.
Turbowatch2 - thanks for your perspective. I am, however, continuing on the passive path here. I personally consider high quality components in passives to be a very valid route still in use today - of course I believe the turbocharged 3.0L 5 cylinder Mercedes Diesel engine in the old W123 chassis series a crowning achievement in automotive history and the FJZ80 series solid front axle Toyota Landcruisers to be the last great one in the series, just to offer a calibration... I'll be driving these with a Hafler DH200 until I can locate a McIntosh MC2100 or perhaps slightly higher power model in that same vintage range.