Hi everybody!
Have made some homework and research to start to build decent speaker pair.
First of all, I found "original" Thor at Seas webpage. It seemed pretty good at the beginning.
Now I´v had some discussions and different toughts, result is Fat Thor ML.
Design is a bit changed - I´ll use 18mm first grade baltic birch plywood (one of the best in the world). At the top and sides I´ll use some extra parts, mostly for visual look. If it´s necessary to note, I have some woodwork skills as well and all pieces will be cnc cutted.
*Actually I´m not sure about crossover - do I use jimangie1973 last design or try to amp them active or something else.
*Should I fill top corners with some sand or something?
*Maybe could use port with flared ends?
*How much damping/stuffing?
*If I go passive, then crossover will be out of the enclosure.
*Should I made pedestal/stand or just adjustable spikes?
Any advice/tips/toughts are welcome!
Few rendered pictures, sorry for "not so good textures" (~90% edges/corners are 3mm rounded).
Back there is design with a bit different side panel.
Have made some homework and research to start to build decent speaker pair.
First of all, I found "original" Thor at Seas webpage. It seemed pretty good at the beginning.
Now I´v had some discussions and different toughts, result is Fat Thor ML.
Design is a bit changed - I´ll use 18mm first grade baltic birch plywood (one of the best in the world). At the top and sides I´ll use some extra parts, mostly for visual look. If it´s necessary to note, I have some woodwork skills as well and all pieces will be cnc cutted.
*Actually I´m not sure about crossover - do I use jimangie1973 last design or try to amp them active or something else.
*Should I fill top corners with some sand or something?
*Maybe could use port with flared ends?
*How much damping/stuffing?
*If I go passive, then crossover will be out of the enclosure.
*Should I made pedestal/stand or just adjustable spikes?
Any advice/tips/toughts are welcome!
Few rendered pictures, sorry for "not so good textures" (~90% edges/corners are 3mm rounded).
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Back there is design with a bit different side panel.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I just read this entire thread. Very interesting and thanks to those (you know who you are) who contribute so much to the community!
This seems very close to an ideal speaker project for me*, as I am very interested in TL, and 2-way MTMs but with a few caveats:
1. The design is over 10 years old now. That doesn't mean it's no longer a great speaker, of course, but it does make me wonder how much longer these drivers will be available, and whether there are improved drivers we should be looking at for a v2? Or maybe there are simply newer designs which might be worth a look.
2. Going through this entire thread I am still not clear at all about the sonic differences between the fat, short and small variants. It would be helpful to many people, I think, to compare the variants in terms of size, weight, cost and sound differences.
3. This might be heresy to some but I have a thing for ribbon tweeters, and would seriously consider replacing the tweeter with a ribbon or even AMT, which seems quite doable as long as the tweeter can be crossed over at 1.5kHz. Has anyone done this? One other member in this thread expressed an interest in doing the same thing. (Long term plan would be to bi-amp these speakers with digital crossovers and EQ).
* (years ago I had a pair of DIY bass reflex towers with Scanspeak 18W-8545 carbon fibre/paper bass mids in an MTM configuration with a TiO2 Focal tweeter. Great speakers they were too! They are missed.)
This seems very close to an ideal speaker project for me*, as I am very interested in TL, and 2-way MTMs but with a few caveats:
1. The design is over 10 years old now. That doesn't mean it's no longer a great speaker, of course, but it does make me wonder how much longer these drivers will be available, and whether there are improved drivers we should be looking at for a v2? Or maybe there are simply newer designs which might be worth a look.
2. Going through this entire thread I am still not clear at all about the sonic differences between the fat, short and small variants. It would be helpful to many people, I think, to compare the variants in terms of size, weight, cost and sound differences.
3. This might be heresy to some but I have a thing for ribbon tweeters, and would seriously consider replacing the tweeter with a ribbon or even AMT, which seems quite doable as long as the tweeter can be crossed over at 1.5kHz. Has anyone done this? One other member in this thread expressed an interest in doing the same thing. (Long term plan would be to bi-amp these speakers with digital crossovers and EQ).
* (years ago I had a pair of DIY bass reflex towers with Scanspeak 18W-8545 carbon fibre/paper bass mids in an MTM configuration with a TiO2 Focal tweeter. Great speakers they were too! They are missed.)
Last edited:
Mr. S.
Good looking design, Thor's have an unbelievable voiceing. keep us posted on your build. pictures are good.
10 years old or not, MTM TM are a nice speaker design. New drivers and designs are always being developed. You decide.
Size and weight go hand in hand with cost, material dependent. Sound differences? Bigger is bigger Bass for bigger listening spaces.
changing any driver in the design is a new design, mostly your on your own
All said and done, I've been very satisfied with the Thors. Only recently have I found a new speaker to build which will better the sound, IMO. It costs roughly 3 times more to build. I certainly don't expect 3 times the sound quality. The carpentry is much more difficult and that's a large part of the challenge for me. I'll still keep the Thor's though.
Good looking design, Thor's have an unbelievable voiceing. keep us posted on your build. pictures are good.
10 years old or not, MTM TM are a nice speaker design. New drivers and designs are always being developed. You decide.
Size and weight go hand in hand with cost, material dependent. Sound differences? Bigger is bigger Bass for bigger listening spaces.
changing any driver in the design is a new design, mostly your on your own
All said and done, I've been very satisfied with the Thors. Only recently have I found a new speaker to build which will better the sound, IMO. It costs roughly 3 times more to build. I certainly don't expect 3 times the sound quality. The carpentry is much more difficult and that's a large part of the challenge for me. I'll still keep the Thor's though.
Hi Folks!
I've just started a build thread on my 'curved side' Small Thor project - taking after Renron's excellent build thread. Started with the crossovers using jimangie1973's latest design, which I don't believe has been posted yet - or at least I couldn't find it - I contacted him directly via PM. If you are interested in seeing it go to:
http://www.diyaudio.com/forums/multi-way/269850-curved-small-thor-redux.html
I'll be putting up more pics of the construction as it progresses.
Thanks to all the hard work and sharing that has gone with the various Thor versions by Renron, Scottmoose, jimangie1973, Planet10 +++ - I hope to share a few humble experiences with my build thread.
Cheers
Tim
I've just started a build thread on my 'curved side' Small Thor project - taking after Renron's excellent build thread. Started with the crossovers using jimangie1973's latest design, which I don't believe has been posted yet - or at least I couldn't find it - I contacted him directly via PM. If you are interested in seeing it go to:
http://www.diyaudio.com/forums/multi-way/269850-curved-small-thor-redux.html
I'll be putting up more pics of the construction as it progresses.
Thanks to all the hard work and sharing that has gone with the various Thor versions by Renron, Scottmoose, jimangie1973, Planet10 +++ - I hope to share a few humble experiences with my build thread.
Cheers
Tim
Fat Thor CAD / CNC Drawings
Are there any members willing to share their Fat Thor AutoCAD drawings files with me please? I want to have the flats cut, but don't have the software or knowledge to make my own drawings on CAD and have given up on doing it on Word and Publisher.
I have completed a fantastic pair of Small Thor speakers and revel in their sound daily. The next pair will be for my brother, but I have decided to go the Fat Thor route, as he prefers more bass and he does have the space for large spreakers.
If willing, please send me a PM - Many thanks to all the senior members without whom I would never have been able to build my speakers.
Are there any members willing to share their Fat Thor AutoCAD drawings files with me please? I want to have the flats cut, but don't have the software or knowledge to make my own drawings on CAD and have given up on doing it on Word and Publisher.
I have completed a fantastic pair of Small Thor speakers and revel in their sound daily. The next pair will be for my brother, but I have decided to go the Fat Thor route, as he prefers more bass and he does have the space for large spreakers.
If willing, please send me a PM - Many thanks to all the senior members without whom I would never have been able to build my speakers.
Zobel Network for Thor/Odin Crossover
This is a bit of long post, but I think you might find it worth it. Please bear with me.
I built the Thor speakers pretty much the year the design was published in AudioExpress years ago. For the last decade or so, I have been experimenting with crossover modifications, just as pretty much everybody else who built them, it seems. I have never before contributed to the forum, but I have read many of the entries. I know that many of you are satisfied with your Thors, but like many others, too, I thought that the speakers never sounded right. I suppose this discussionforum and others like it all over the web testify to the fact that many feel the same: for what they are, these speakers are underperforming. Moreover, people rightly noticed that this must have something to do with the poor original crossover design, and some of you have spent many an hour and a great deal of money to improve it.
So didd I. First I proceeded to modify the original Xover design by following somebody's online advice. Then I experimented with different parts and tuned the speakers by ear. I even bought an expensive Vidsonic VCB-100 crossover design box. Time and again, I had the illusion of improving the sound, but usually nothing much changed: The speakers were shouting, sounded wooden, and were only good for short listening periods because they fatigued the ear. For a while I thought I was on the right track because in the last crossover incarnation I had ended up, among other things, with a 4 Ohm resistor in series with the tweeter, just like the latest design proposal by Jim, who has labored heroically over the years to improve the crossover.
Well, my search is finally over --- I have found what I was looking for. After nearly dozen years of tinkering, soldering, and listening, my speakers finally sound the way I had always hoped they would. If you have a resistor in series with the tweeter, I am almost certain that you, too, are not entirely satisfied with the sound of your Thors or Odins. In short, there should be no such resistor. It is the wrong cure for an entirely different problem, one that has to do with the woofers. Let me explain:
The resistor is there because, listening to the speakers, one thinks that the highs have to come down because the speaker is unbalanced with too much treble, and so one instinct is to bring the sound pressure level of the tweeter down. But a simple reflection shows that it is very odd to have to wire any series resistor at this point: For according to the manufacturer, the tweeter has a sensitivity of 88dB, and the woofer's sensitivity is 86dB. Now, since we have two woofers in parallel, their actual sensitivity is raised by ~6dB to 92dB. So the tweeter should have trouble keeping up with the woofers, and shouldn't need any attenuation! And indeed, once the cross over is designed properly, this is precisely what happens when the SPL is measured with a microphone (I have been using a Behringer ECM 2000). The crossover I propose here needs no such resistor; in fact the tweeter puts out slightly less sound pressure than the woofers, and the entire setup sounds absolutely glorious and totally different from what it was before.
So what is the problem?
If you look at the impedance curve of the woofers, you notice a steady rise of their impedance from about 6 Ohms at 200 Hz to nearly 20 Ohms at 10kHz. Plus, there is this notorious and pronounced peak between 4 and 5 kHz. If you let the impedance of the woofers rise in the way they do, the crossover network for them has a very difficult time taming the woofers. The rise of the impedance with increasing frequency means that there is a greater voltage across the speakers than there should be at higher frequencies, and with a fairly high crossover point anywhere in the region of 1500-2500 Hz, this means that the woofers are seriously interfering with the treble, what should be the sole domain of the tweeter. I believe that what we have all been listening to are the highs produced by the woofers, and we have been trying to fix those by dialing down the voltage across the tweeter. If so, it is not at all surprising that the whole setup sounded wanting, and in any case much worse than these fine speakers are capable of.
So what is the cure?
I decided start from scratch, using Rod Elliott's (from ESP) very good and detailed advice on how to design passive crossovers. The link is here:
Passive Crossover Network Design
As Rod Elliott makes quite clear, the most important thing to do first in designing a crossover network is to keep the impedance of the speakers more or less constant at frequencies above the crossover point. Only in that way can the networks serving the speakers work properly and function as designed. The way to do that is to put a so-called Zobel network in parallel with the woofers (and theoretically also across the tweeter, but that turned out to be unnecessary). A Zoibel network is a simple resistor (NOT a choke!) in series with a capacitor wired directly across the speaker (electrically speaking, not physically) which kicks in as frequency rises, thus counteracting the rising impedance of the speakers. Very simple. What is not so simple is to determine the precise values needed for this to happen, and this has to be done experimentally. Rod tells one exactly how to do that, and I am mot going to repeat his instructions here. In my case, it turned out that I needed a 10uF capacitor in series with a 4.3 Ohm resistor.
Now, you have to be quite exact with the measurement of these Zobel network components, esp. the resistor; the value is quite critical. Your numbers might differ from mine, as you may have a different run of production of woofers, or your cables (also part of the woofer impedance now) are slightly different. Also, make sure to note the number your Ohmmeter shows when the test leads are shortened; you need to subtract that from the measured value of the resistor in your test setup.
Now, with the Zobel network in place, you measure the combined impedance of the two speakers plus Zobel network in the region of your desired crossover frequency, an impedance that should remain more or less constant as frequency rises. You do the same for the tweeter, and these two measured impedances are now the basis for the calculation of your crossover network. You can use Rod Elliott's terrific crossover calculation spreadsheet to do that, or do the calculation by hand.
The true measured impedances at my crossover frequency (2400Hz) were: combined Zobeled woofers: 3.8 Ohms; tweeter 5.5 Ohms (without Zobel, see below).
Now, I wanted the network to be a 12dB/octave design, with a crossover frequency of 2400Hz. I chose that frequency because it is close to the frequency recommend by SEAS for the Millennium tweeter (2500Hz), and because it allowed my to use the expensive .5mH choke that I had bought some time ago. The values are then as follows: Woofer coil: 0.5mH; woofer capacitor: 8.7uF; tweeter capacitor: 6.03uF; tweeter coil: 0.729 mH. I ended up using 0.5mH; 8.7uF; 6.2uF and 0.7uH. At first I left room on the board for a series resistor in front of the tweeter, but measurements show that no such resistor is needed as the tweeter puts out about 2-3 dB less than the combined woofers. The tweeter also does not benefit from its own Zobel network (I wired one at first, consisting of a 8 Ohm resistor in series with a 1uf capacitor). The Zobel attenuated the treble SPL ever so slightly, and that is not what the tweeter needs.
The schematic will be posted separately once I have a good drawing of it; it could not be more straightforward, and I am sure that all of you who have been experimenting with these speakers over the years, or just bought the kit, can easily source the parts. Mine are, for the woofers, a 0.5mH copper foil inductor from Madisound, the original 8.2uF that came with the kit in parallel with a .47uF film cap; then, for the tweeter, a Solen 6.2uF terminated by a 0.7mH DIY air coil. The parts for the Zobel network don't have to be of any audiophile quality because they don't pass anything that will end up as sound waves. But it is important to use parts that remain stable over time; no electrolytics. My Zobel cap is a 10uF Wima MKP that I happened to have in the bin; you could use the 8uf cap that came with the original Madison kit and parallel it with the 2.2uf cap that also came with it. The Zobel resistor should be of a high wattage; I paralleled two 5 Watt resistors, 6.8 Ohm and 11.2 Ohm. Also, it is VERY important that, as with all 12dB crossover networks, the polarity of the tweeter be reversed.
The improvement of the sound is obvious and dramatic. I have never heard the Thors like this! This is the way they were supposed to sound. Subjectively, there is considerable more bass, but no more shouting in the midrange and highs, no harshness, no fatigue, and all because the woofers finally give the tweeters the necessary space to work their magic. Unlike ever before, I now find myself seriously listening to music, and enjoying every minute of it. N.B. I have not modified the TM enclosure in any way, except that I am using a bit less damping material than originally specified by Joe D'Apollito.
I have not read each and every post on the Thors in all discussion forums, though many of them, and I am pretty sure that this Zobel network has never been discussed (if I am mistaken, I apologize). I thought that given the number of people interested in these SEAS kits, this was too important an improvement not to share it on DIYAudio, in a forum that still seems to be active.
If you don't have access to the necessary measuring equipment but you have built the speakers and are not entirely convinced of their performance in your system, do yourself the favor and build the crossover network proposed here, simply taking my values as a basis (though it would be better to do your own measurements). In either case, you will be amazed, I promise.
This is a bit of long post, but I think you might find it worth it. Please bear with me.
I built the Thor speakers pretty much the year the design was published in AudioExpress years ago. For the last decade or so, I have been experimenting with crossover modifications, just as pretty much everybody else who built them, it seems. I have never before contributed to the forum, but I have read many of the entries. I know that many of you are satisfied with your Thors, but like many others, too, I thought that the speakers never sounded right. I suppose this discussionforum and others like it all over the web testify to the fact that many feel the same: for what they are, these speakers are underperforming. Moreover, people rightly noticed that this must have something to do with the poor original crossover design, and some of you have spent many an hour and a great deal of money to improve it.
So didd I. First I proceeded to modify the original Xover design by following somebody's online advice. Then I experimented with different parts and tuned the speakers by ear. I even bought an expensive Vidsonic VCB-100 crossover design box. Time and again, I had the illusion of improving the sound, but usually nothing much changed: The speakers were shouting, sounded wooden, and were only good for short listening periods because they fatigued the ear. For a while I thought I was on the right track because in the last crossover incarnation I had ended up, among other things, with a 4 Ohm resistor in series with the tweeter, just like the latest design proposal by Jim, who has labored heroically over the years to improve the crossover.
Well, my search is finally over --- I have found what I was looking for. After nearly dozen years of tinkering, soldering, and listening, my speakers finally sound the way I had always hoped they would. If you have a resistor in series with the tweeter, I am almost certain that you, too, are not entirely satisfied with the sound of your Thors or Odins. In short, there should be no such resistor. It is the wrong cure for an entirely different problem, one that has to do with the woofers. Let me explain:
The resistor is there because, listening to the speakers, one thinks that the highs have to come down because the speaker is unbalanced with too much treble, and so one instinct is to bring the sound pressure level of the tweeter down. But a simple reflection shows that it is very odd to have to wire any series resistor at this point: For according to the manufacturer, the tweeter has a sensitivity of 88dB, and the woofer's sensitivity is 86dB. Now, since we have two woofers in parallel, their actual sensitivity is raised by ~6dB to 92dB. So the tweeter should have trouble keeping up with the woofers, and shouldn't need any attenuation! And indeed, once the cross over is designed properly, this is precisely what happens when the SPL is measured with a microphone (I have been using a Behringer ECM 2000). The crossover I propose here needs no such resistor; in fact the tweeter puts out slightly less sound pressure than the woofers, and the entire setup sounds absolutely glorious and totally different from what it was before.
So what is the problem?
If you look at the impedance curve of the woofers, you notice a steady rise of their impedance from about 6 Ohms at 200 Hz to nearly 20 Ohms at 10kHz. Plus, there is this notorious and pronounced peak between 4 and 5 kHz. If you let the impedance of the woofers rise in the way they do, the crossover network for them has a very difficult time taming the woofers. The rise of the impedance with increasing frequency means that there is a greater voltage across the speakers than there should be at higher frequencies, and with a fairly high crossover point anywhere in the region of 1500-2500 Hz, this means that the woofers are seriously interfering with the treble, what should be the sole domain of the tweeter. I believe that what we have all been listening to are the highs produced by the woofers, and we have been trying to fix those by dialing down the voltage across the tweeter. If so, it is not at all surprising that the whole setup sounded wanting, and in any case much worse than these fine speakers are capable of.
So what is the cure?
I decided start from scratch, using Rod Elliott's (from ESP) very good and detailed advice on how to design passive crossovers. The link is here:
Passive Crossover Network Design
As Rod Elliott makes quite clear, the most important thing to do first in designing a crossover network is to keep the impedance of the speakers more or less constant at frequencies above the crossover point. Only in that way can the networks serving the speakers work properly and function as designed. The way to do that is to put a so-called Zobel network in parallel with the woofers (and theoretically also across the tweeter, but that turned out to be unnecessary). A Zoibel network is a simple resistor (NOT a choke!) in series with a capacitor wired directly across the speaker (electrically speaking, not physically) which kicks in as frequency rises, thus counteracting the rising impedance of the speakers. Very simple. What is not so simple is to determine the precise values needed for this to happen, and this has to be done experimentally. Rod tells one exactly how to do that, and I am mot going to repeat his instructions here. In my case, it turned out that I needed a 10uF capacitor in series with a 4.3 Ohm resistor.
Now, you have to be quite exact with the measurement of these Zobel network components, esp. the resistor; the value is quite critical. Your numbers might differ from mine, as you may have a different run of production of woofers, or your cables (also part of the woofer impedance now) are slightly different. Also, make sure to note the number your Ohmmeter shows when the test leads are shortened; you need to subtract that from the measured value of the resistor in your test setup.
Now, with the Zobel network in place, you measure the combined impedance of the two speakers plus Zobel network in the region of your desired crossover frequency, an impedance that should remain more or less constant as frequency rises. You do the same for the tweeter, and these two measured impedances are now the basis for the calculation of your crossover network. You can use Rod Elliott's terrific crossover calculation spreadsheet to do that, or do the calculation by hand.
The true measured impedances at my crossover frequency (2400Hz) were: combined Zobeled woofers: 3.8 Ohms; tweeter 5.5 Ohms (without Zobel, see below).
Now, I wanted the network to be a 12dB/octave design, with a crossover frequency of 2400Hz. I chose that frequency because it is close to the frequency recommend by SEAS for the Millennium tweeter (2500Hz), and because it allowed my to use the expensive .5mH choke that I had bought some time ago. The values are then as follows: Woofer coil: 0.5mH; woofer capacitor: 8.7uF; tweeter capacitor: 6.03uF; tweeter coil: 0.729 mH. I ended up using 0.5mH; 8.7uF; 6.2uF and 0.7uH. At first I left room on the board for a series resistor in front of the tweeter, but measurements show that no such resistor is needed as the tweeter puts out about 2-3 dB less than the combined woofers. The tweeter also does not benefit from its own Zobel network (I wired one at first, consisting of a 8 Ohm resistor in series with a 1uf capacitor). The Zobel attenuated the treble SPL ever so slightly, and that is not what the tweeter needs.
The schematic will be posted separately once I have a good drawing of it; it could not be more straightforward, and I am sure that all of you who have been experimenting with these speakers over the years, or just bought the kit, can easily source the parts. Mine are, for the woofers, a 0.5mH copper foil inductor from Madisound, the original 8.2uF that came with the kit in parallel with a .47uF film cap; then, for the tweeter, a Solen 6.2uF terminated by a 0.7mH DIY air coil. The parts for the Zobel network don't have to be of any audiophile quality because they don't pass anything that will end up as sound waves. But it is important to use parts that remain stable over time; no electrolytics. My Zobel cap is a 10uF Wima MKP that I happened to have in the bin; you could use the 8uf cap that came with the original Madison kit and parallel it with the 2.2uf cap that also came with it. The Zobel resistor should be of a high wattage; I paralleled two 5 Watt resistors, 6.8 Ohm and 11.2 Ohm. Also, it is VERY important that, as with all 12dB crossover networks, the polarity of the tweeter be reversed.
The improvement of the sound is obvious and dramatic. I have never heard the Thors like this! This is the way they were supposed to sound. Subjectively, there is considerable more bass, but no more shouting in the midrange and highs, no harshness, no fatigue, and all because the woofers finally give the tweeters the necessary space to work their magic. Unlike ever before, I now find myself seriously listening to music, and enjoying every minute of it. N.B. I have not modified the TM enclosure in any way, except that I am using a bit less damping material than originally specified by Joe D'Apollito.
I have not read each and every post on the Thors in all discussion forums, though many of them, and I am pretty sure that this Zobel network has never been discussed (if I am mistaken, I apologize). I thought that given the number of people interested in these SEAS kits, this was too important an improvement not to share it on DIYAudio, in a forum that still seems to be active.
If you don't have access to the necessary measuring equipment but you have built the speakers and are not entirely convinced of their performance in your system, do yourself the favor and build the crossover network proposed here, simply taking my values as a basis (though it would be better to do your own measurements). In either case, you will be amazed, I promise.
Garinel Crossover
Here is the schematic of the crossover for the Thor transmission line speaker.
Since the Thor and Odin kits share the same crossover design, I am assuming that it will work just as well for the Odins, but I have not verified that. The values of the components might come out slightly different, but the principle should be the same.
Regards to all who have contributed so much to this thread,
-garinel
Here is the schematic of the crossover for the Thor transmission line speaker.
Since the Thor and Odin kits share the same crossover design, I am assuming that it will work just as well for the Odins, but I have not verified that. The values of the components might come out slightly different, but the principle should be the same.
Regards to all who have contributed so much to this thread,
-garinel
I will give this a try. Will be using parts on hand that are close to your values. With the exception that the mid-woof circuit will have a 10mH
coil and 4u5F cap. as that is what I have on hand. So, the 5mH was doubled and the cap value was halved. My hope is this will give me a hint of what your cross over design is capable of.
Thanks for your effort.
Hi Ichiban,
the value of coil L1 is 0.5mH, not 5mH. So if you are using 10mH, this will be way out of line. I think you meant to write 1.0mH, but I just wanted to make sure.
Regards,
-garinel
I ran a quick sim on your x/o and this is what I got:
Not very promising really; these drivers deserve so much more.
An externally hosted image should be here but it was not working when we last tested it.
Not very promising really; these drivers deserve so much more.
This very helpful, Zuhl. You are quite right, there is this peak just below 5kHz, and my network does not address that issue. I did a couple of quick and rough measurements (unfortunately I don't have access to simulation software or an anechoic chamber): What I found was that although the voltage across the woofers was well down at 5kHz by more than -12dB, as it should be, the SPL was not. I did not measure hump at 4.5kHz that your simulation suggests, but it is true that the SPL of the woofers did not slope down the way it should have. which meant that woofers and tweeter overlapped for about an octave, and I think one can see that in your graph. So we need a notch filter, you are right.
By way of a quick test, I wired a 0.15mH - 7uF (f=4912Hz) across the woofers, which brought the SPL of the signal above the xover frequency more in line to what I had intended. The further improvement of the sound of the speakers was noticeable, but it was not as dramatic as the improvement I initially talked about.
So, I guess what I am proposing now is that the woofers get paralleled by both a Zobel network and a notch filter.
I am not sure how much the addition of a notch filter is going to affect the components of the other filters, esp. the xover filter. I am quite certain that I have not hit upon the ideal values. Maybe somebody with access to good simulation software can help here?
By way of a quick test, I wired a 0.15mH - 7uF (f=4912Hz) across the woofers, which brought the SPL of the signal above the xover frequency more in line to what I had intended. The further improvement of the sound of the speakers was noticeable, but it was not as dramatic as the improvement I initially talked about.
So, I guess what I am proposing now is that the woofers get paralleled by both a Zobel network and a notch filter.
I am not sure how much the addition of a notch filter is going to affect the components of the other filters, esp. the xover filter. I am quite certain that I have not hit upon the ideal values. Maybe somebody with access to good simulation software can help here?
Here's a quick flirt in XSim - with more time I could do better.
Crossover:
Frequency response:
Frequency response - reverse nul:
I won't post the impedence graph. It's a tough load below 4Ω most of the way, but it is fairly even.
Crossover:
An externally hosted image should be here but it was not working when we last tested it.
Frequency response:
An externally hosted image should be here but it was not working when we last tested it.
Frequency response - reverse nul:
An externally hosted image should be here but it was not working when we last tested it.
I won't post the impedence graph. It's a tough load below 4Ω most of the way, but it is fairly even.
Hi
I have used these drivers on various designs. I really like the tweeter - one of my favourites. The bass driver requires a notch filter to remove the resonance peak. To do this use the SEAS crossover as I don't think there is anything wrong with the design. The problem is that component tolerances can make the notch filter slightly wrong. (ie. Inductor, capacitor and driver tolerances)
I fix this problem by using an inductor slightly bigger than specified. 0.15mH.
Use 0.2mH - measure the driver output with mic close to the drivers. Slowly
unwind the 0.2mH until the peak is a minimum. You need to do this for each pair of W18E001 drivers so you get the optimum value for each speaker.
I have used these drivers on various designs. I really like the tweeter - one of my favourites. The bass driver requires a notch filter to remove the resonance peak. To do this use the SEAS crossover as I don't think there is anything wrong with the design. The problem is that component tolerances can make the notch filter slightly wrong. (ie. Inductor, capacitor and driver tolerances)
I fix this problem by using an inductor slightly bigger than specified. 0.15mH.
Use 0.2mH - measure the driver output with mic close to the drivers. Slowly
unwind the 0.2mH until the peak is a minimum. You need to do this for each pair of W18E001 drivers so you get the optimum value for each speaker.
Curved Thor's nearing completion!
Here's a couple of pics of my finished cabinets.
Am currently testing crossovers - jump over to my thread for progress on tuning next week once I get my parts in, if interested...initial impressions tweeters way too hot for both the Jimangie crossover and the SEAS version, but the SEAS version is less bright. Confirmed by Audiotools RTA and Impulse FFT measurements using my iPad.
http://www.diyaudio.com/forums/multi-way/269850-curved-small-thor-redux.html
Will be swapping the series resistor in the tweeter network with higher values next week, per advice from Jimangie. Will try a 2.7, 3.0, and maybe a 3.3 ohm if still too hot. Jim says he is using a 2.7 in his current setup. Jim exceptionally helpful, BTW - many thanks!
Woofers probably need to break in some more as well...plus all the wires coming out of the port are probably breaking up the port contribution - will try to tidy that up.
Here's a couple of pics of my finished cabinets.
Am currently testing crossovers - jump over to my thread for progress on tuning next week once I get my parts in, if interested...initial impressions tweeters way too hot for both the Jimangie crossover and the SEAS version, but the SEAS version is less bright. Confirmed by Audiotools RTA and Impulse FFT measurements using my iPad.
http://www.diyaudio.com/forums/multi-way/269850-curved-small-thor-redux.html
Will be swapping the series resistor in the tweeter network with higher values next week, per advice from Jimangie. Will try a 2.7, 3.0, and maybe a 3.3 ohm if still too hot. Jim says he is using a 2.7 in his current setup. Jim exceptionally helpful, BTW - many thanks!
Woofers probably need to break in some more as well...plus all the wires coming out of the port are probably breaking up the port contribution - will try to tidy that up.