I've built and listened to many speakers over the years, but have recently come up with a design that is really very good. It is based on a Jordan JX92S transmission line by Carolina Audio ( http://www.carolinaaudio.com/ ; Ronnie is a very nice guy to deal with) and an ESg2 ribbon distributed by E-speakers.com ( http://www.e-speakers.com/ ; Andre is also very helpful). The speaker fulfills many criteria that I had: use minimal number of drivers of high quality, shielded drivers for potential home theater, no MTM or dipole configerations, time/phase aligned, 1st order x/o, play loud, have great detail and imaging, have flat frequency and impedance plots, and not be too large.
The speakers fit all of these criteria and sound wonderful with all types of music. (Although be aware that the Jordans definitely require some serious break-in time.) The JX92S + ESg2 are flat from 50 Hz to 20+ kHz (limit of my measuring system) within 2 dB. (The ribbons are flat to 30 or 40 kHz.) With a pair of subs (12" BagEnd x/o at 100 Hz, 1st order passive going into my main amps for the JX92S), the system is flat to 25 Hz, being -3 dB in my room at 20 Hz. I've played them at 100 dB continuous, measured from the listening position 2 meters from the speakers and they sounded very smooth and dynamic (powered with 350 W/8 ohm monoblocks). In fact, they are so good, I'm selling my Quad 988 that I compared them to when I built them.
The crossover is very simple, but only arrived at after many weeks of measurements and listening tests using impulse, freq resp, and in room warble testing. While the JX92S appears flat on the manufacturer's graphs, I believe the upper frequencies are only flat due to cone breakup distortions adding to the signal. Removing them using the x/o described below, helped improve them. The freq response is also flat to 45 degrees off axis. Since this is not a dipole room interaction is minimized. Imaging is superb. The speakers can be aimed to cross in front of you for a wide sweet spot extending from speaker to speaker, or aimed to cross outside your head for better depth and a wider soundstage while making the sweet spot narrower.
The x/o uses a 1.5 mH inductor in series with the JX92S (I tried series x/o but I could not get it to measure or sound proper). Also, a Zobel consisting of a 4.7 microfarad capacitor in series with a 6 ohm resistor is used on the Jordan. The ribbon is crossed over with a Auricap ( http://www.audience-av.com/passive.htm ): 3 microfarad/200V with an L-pad of 5 ohms in series and 3 ohms in parallel with the 8 ohm ribbon. This creates 8.5 dB of attenuation with a tweeter impedance of 7.2 ohms. The overall speaker impedance is about 5.5 ohms and is ruler flat from 100 Hz to 20 kHz.
The ESg2 ribbon is recessed on top of the cabinet by about 18 mm to time align it with the Jordan. The whole package is about 42 inches tall, 7 inches wide and 9 inches deep, so the speakers are not intrusive (not counting of course the subs). The Jordans are about 36 inches from the floor, so head level falls between the Jordan and ribbons, depending upon your sofa/chair height.
I think this about covers everything. Good luck if you give them a try.
Regards, Robert
The speakers fit all of these criteria and sound wonderful with all types of music. (Although be aware that the Jordans definitely require some serious break-in time.) The JX92S + ESg2 are flat from 50 Hz to 20+ kHz (limit of my measuring system) within 2 dB. (The ribbons are flat to 30 or 40 kHz.) With a pair of subs (12" BagEnd x/o at 100 Hz, 1st order passive going into my main amps for the JX92S), the system is flat to 25 Hz, being -3 dB in my room at 20 Hz. I've played them at 100 dB continuous, measured from the listening position 2 meters from the speakers and they sounded very smooth and dynamic (powered with 350 W/8 ohm monoblocks). In fact, they are so good, I'm selling my Quad 988 that I compared them to when I built them.
The crossover is very simple, but only arrived at after many weeks of measurements and listening tests using impulse, freq resp, and in room warble testing. While the JX92S appears flat on the manufacturer's graphs, I believe the upper frequencies are only flat due to cone breakup distortions adding to the signal. Removing them using the x/o described below, helped improve them. The freq response is also flat to 45 degrees off axis. Since this is not a dipole room interaction is minimized. Imaging is superb. The speakers can be aimed to cross in front of you for a wide sweet spot extending from speaker to speaker, or aimed to cross outside your head for better depth and a wider soundstage while making the sweet spot narrower.
The x/o uses a 1.5 mH inductor in series with the JX92S (I tried series x/o but I could not get it to measure or sound proper). Also, a Zobel consisting of a 4.7 microfarad capacitor in series with a 6 ohm resistor is used on the Jordan. The ribbon is crossed over with a Auricap ( http://www.audience-av.com/passive.htm ): 3 microfarad/200V with an L-pad of 5 ohms in series and 3 ohms in parallel with the 8 ohm ribbon. This creates 8.5 dB of attenuation with a tweeter impedance of 7.2 ohms. The overall speaker impedance is about 5.5 ohms and is ruler flat from 100 Hz to 20 kHz.
The ESg2 ribbon is recessed on top of the cabinet by about 18 mm to time align it with the Jordan. The whole package is about 42 inches tall, 7 inches wide and 9 inches deep, so the speakers are not intrusive (not counting of course the subs). The Jordans are about 36 inches from the floor, so head level falls between the Jordan and ribbons, depending upon your sofa/chair height.
I think this about covers everything. Good luck if you give them a try.
Regards, Robert
robert wrote....
use minimal number of drivers of high quality, shielded drivers for potential home theater, no MTM or dipole configerations, time/phase aligned, 1st order x/o, play loud, have great detail and imaging, have flat frequency and impedance plots, and not be too large.
this seems to be my goal too!
2 questions....
1. why not dipole/bipole?
2. what if one were to use 2 92S s then one would need less attenuation. if one were to use 2 92s would you recomend a 2.5 way of a 2 way with both drivers in parallel? I assume that a large part of the 8.5db attenuation is due to baffle difraction loss. If I would venture to a 2.5 way would you recomend the 125 as the second woofer.
I am looking for just such a design for 2 reasons that are not usually applicable to the rest of you.
1. Indian customs limit the number of drivers you import as they consider anything more than 6 as commercial qualtities. Using 4 92S and 2 ribbons would make a nice pair. I could then use a single 92S + ribbon for surround and center and that would be 6 drviers too.
Lastly this question is to Dave....
start with this design
add second woofer push pull in the rear of the cabinet mate both woofers with a torsion brace.
You get the difraction compensation as well as a little bit of dipole effect from the rear facing woofer.
what do you think?
I have considered a design using the JX125 or bigger woofers but lack of how to control the cone break up (and the complicated crossover required to do so) limited me.
Thanks Robert!
use minimal number of drivers of high quality, shielded drivers for potential home theater, no MTM or dipole configerations, time/phase aligned, 1st order x/o, play loud, have great detail and imaging, have flat frequency and impedance plots, and not be too large.
this seems to be my goal too!
2 questions....
1. why not dipole/bipole?
2. what if one were to use 2 92S s then one would need less attenuation. if one were to use 2 92s would you recomend a 2.5 way of a 2 way with both drivers in parallel? I assume that a large part of the 8.5db attenuation is due to baffle difraction loss. If I would venture to a 2.5 way would you recomend the 125 as the second woofer.
I am looking for just such a design for 2 reasons that are not usually applicable to the rest of you.
1. Indian customs limit the number of drivers you import as they consider anything more than 6 as commercial qualtities. Using 4 92S and 2 ribbons would make a nice pair. I could then use a single 92S + ribbon for surround and center and that would be 6 drviers too.
Lastly this question is to Dave....
start with this design
add second woofer push pull in the rear of the cabinet mate both woofers with a torsion brace.
You get the difraction compensation as well as a little bit of dipole effect from the rear facing woofer.
what do you think?
I have considered a design using the JX125 or bigger woofers but lack of how to control the cone break up (and the complicated crossover required to do so) limited me.
Thanks Robert!
navin said:
push-push not push-pull. Bipole not dipole (you would have zero bass if wired as dipole). I have many of these floating around in my head so yes i think it is a great idea. You can tune the bi-pole effect with an optional inductor on the back driver (with the inductor tuned to the baffle-step frequency you have a 2.5 way -- but with no phase anomolies above turnover).
dave
JLM-ESg2
The Jordan cabinets are the JLM model on Carolina Audio's web site. These are narrow and deeper than the original version. The front vertical edges have a radius of 1/2". The same radius is on the top front edge.
I have plans for enclosing the ribbon and will be building the enclosures over the next week or so. I'm sure Carolina Audio could whip up something if there is enough interest, else you're on your own for the ribbon cabinet (it's only for appearance as the rear of the ESg2 ribbon is sealed and requires no chamber).
A couple of things I forgot to mention is the inductor is an air-core from Parts-Express ( http://www.partsexpress.com/pe/show...61&St3=41906512&DS_ID=3&Product_ID=5816&DID=7 ), part 260-735. I tried foil, which cost more, but the air core sounded better and their slightly greater resistance is probably important. The resistors for the L-pad are also from Parts Express: Mills 12W (005-3 and 005-5).
Additionally, I lined the upper chamber of the TL cabinet with 40 lb. felt and with a strip of felt down the rear chamber wall extending to the speaker terminals. I moved the stuffing that came with the speaker, placing it in the exit port near the base of the speaker.
Finally, one other note, while the polarity of the ribbon may be correct, I needed to reverse it when running the impulse testing to properly generate an impulse and a great looking step response.
The combination of a fast, small Jordan aluminum driver, that can easily make it to 100 Hz, mates perfectly with this superb, ultra-fast, ribbon. The sound is incredible and the package is nice and small. I'm now in the process of making another pair for a rear channel surround. BTW, I ordered one of the satellites (model JSM) from Carolina as a center channel. The exact same ribbon and x/o work here too. I'm laying this smaller cabinet on its side, and re-orienting the ribbon to maintain the same horizontal dispersion pattern as the fronts. This even smaller package will easily fit beneath a large display.
A friend came over this weekend to hear what I've been working on. He's heard my Quads, and rather liked the set-up. After hearing the JLM-ESg2, he's selling his Martin Logan Aerius speakers. The JLM-ESg2 are so much more dynamic than either of these electrostats, yet image better (less beamy) without any loss of microdynamic detail.
Regards, Robert
The Jordan cabinets are the JLM model on Carolina Audio's web site. These are narrow and deeper than the original version. The front vertical edges have a radius of 1/2". The same radius is on the top front edge.
I have plans for enclosing the ribbon and will be building the enclosures over the next week or so. I'm sure Carolina Audio could whip up something if there is enough interest, else you're on your own for the ribbon cabinet (it's only for appearance as the rear of the ESg2 ribbon is sealed and requires no chamber).
A couple of things I forgot to mention is the inductor is an air-core from Parts-Express ( http://www.partsexpress.com/pe/show...61&St3=41906512&DS_ID=3&Product_ID=5816&DID=7 ), part 260-735. I tried foil, which cost more, but the air core sounded better and their slightly greater resistance is probably important. The resistors for the L-pad are also from Parts Express: Mills 12W (005-3 and 005-5).
Additionally, I lined the upper chamber of the TL cabinet with 40 lb. felt and with a strip of felt down the rear chamber wall extending to the speaker terminals. I moved the stuffing that came with the speaker, placing it in the exit port near the base of the speaker.
Finally, one other note, while the polarity of the ribbon may be correct, I needed to reverse it when running the impulse testing to properly generate an impulse and a great looking step response.
The combination of a fast, small Jordan aluminum driver, that can easily make it to 100 Hz, mates perfectly with this superb, ultra-fast, ribbon. The sound is incredible and the package is nice and small. I'm now in the process of making another pair for a rear channel surround. BTW, I ordered one of the satellites (model JSM) from Carolina as a center channel. The exact same ribbon and x/o work here too. I'm laying this smaller cabinet on its side, and re-orienting the ribbon to maintain the same horizontal dispersion pattern as the fronts. This even smaller package will easily fit beneath a large display.
A friend came over this weekend to hear what I've been working on. He's heard my Quads, and rather liked the set-up. After hearing the JLM-ESg2, he's selling his Martin Logan Aerius speakers. The JLM-ESg2 are so much more dynamic than either of these electrostats, yet image better (less beamy) without any loss of microdynamic detail.
Regards, Robert
why not bipole/dipole? room interaction, diffuse imaging. I don't believe it necessary. I understand the idea of trying to design for a market (I too used to be in the industry), but now I'm designing for myself. I wouldn't have a speaker that has too many large drivers (you get lobing, which is why I detest MTM speakers--I even used to produce one, but I didn't keep any of them.
The JLM-ESg2 is very smooth and its tonality does not change with standing or moving abou†. You cannot say this about most non-time aligned speakers or dipole/bipole, and you can definitely not say this about speakers with huge arrays. Unless they are small, like the JX53, which will not destructively interfer at normal listening positions 2 meters away due to their small size.
The 8.5 dB attenuation is due to design: I made an L-pad to purposely reduce the output of the ribbon. The ribbon is 96 dB/W; the JX92S is 88 (probably a bit less in the TL cabinet). Attenuation = 20 log (3/3+5) = -8.5 dB. Imped = [1/(1/3 + 1/8)] + 5 = 7.2 ohms.
Robert
The JLM-ESg2 is very smooth and its tonality does not change with standing or moving abou†. You cannot say this about most non-time aligned speakers or dipole/bipole, and you can definitely not say this about speakers with huge arrays. Unless they are small, like the JX53, which will not destructively interfer at normal listening positions 2 meters away due to their small size.
The 8.5 dB attenuation is due to design: I made an L-pad to purposely reduce the output of the ribbon. The ribbon is 96 dB/W; the JX92S is 88 (probably a bit less in the TL cabinet). Attenuation = 20 log (3/3+5) = -8.5 dB. Imped = [1/(1/3 + 1/8)] + 5 = 7.2 ohms.
Robert
navin,
you've lost me. what is a 'Irage' driver? If you're asking about the JX92S, it is about 4 inches in diameter. As I look at your question, I think it's a typo and you meant how large is large. In that case, I was trying to use a midrange that was no more than 5". Anything else, like the pervasive 7" (actually 6.5") Scanspeaks that I used to use, break up horribly above 2 khz. This does not lend itself to a first order x/o. I wanted something that went higher. Break-up is a physical phenomenon directly related to diameter, although it can be somewhat exteneded by a phase plug (and the JX92S does have a modest plug).
I'm also not sure what you mean by baffle diffraction. The baffle for the JLM is only about 1" wider on each side than the driver itself. Considering that the cabinet is 3/4 inch MDF and radius, there is actually only about 1/2 inch on either side of the driver. This is about as close to the edge as you can make a cabinet. So diffraction is about as minimal as you can get. Now if you're asking why not an infinite baffle? Well that would be nice as you get very low extension. But I wouldn't want it in my house (nor would my wife).
regards, Robert
you've lost me. what is a 'Irage' driver? If you're asking about the JX92S, it is about 4 inches in diameter. As I look at your question, I think it's a typo and you meant how large is large. In that case, I was trying to use a midrange that was no more than 5". Anything else, like the pervasive 7" (actually 6.5") Scanspeaks that I used to use, break up horribly above 2 khz. This does not lend itself to a first order x/o. I wanted something that went higher. Break-up is a physical phenomenon directly related to diameter, although it can be somewhat exteneded by a phase plug (and the JX92S does have a modest plug).
I'm also not sure what you mean by baffle diffraction. The baffle for the JLM is only about 1" wider on each side than the driver itself. Considering that the cabinet is 3/4 inch MDF and radius, there is actually only about 1/2 inch on either side of the driver. This is about as close to the edge as you can make a cabinet. So diffraction is about as minimal as you can get. Now if you're asking why not an infinite baffle? Well that would be nice as you get very low extension. But I wouldn't want it in my house (nor would my wife).
regards, Robert
rljones said:
Navin isn't talking about edge diffraction, but Baffle Step Diffraction, the 6 dB loss you when the driver transitions from radiating into 2 pi stereradians to 4 pi stereradians which if one assumes your cabinet is 6" wide would have a -3dB pint of about 760 Hz. The Wiki has a little article on this too.
dave
rljones said:
Congratulations on your successful project and thanks for sharing your XO details and impressions on the speakers
Now this is an interesting crossover you ended up with: 1st order acoustical low pass with cutoff @ 900Hz; high pass cutoff @ 6500Hz but not acoustical first order so the system's total phase is not kept at 0 degrees.
I have played with Jordan 92S and Esg2 ribbon also some time but haven’t decided on the final topology. I have thought of MT, MMT, MTM or 2,5 way, bipole etc.
I think both these drivers have great potential to play well together if mated properly.
I also tend towards 1st order XO (and I mean acoustic 1st order) but to achieve one, seems to be rather complicated task.
Anyway I will certainly try out your XO also.
Argo
Argo,
You're correct in pointing out that the x/o is not intuitively obvious. That is, the high pass is around 6500 Hz (the ribbon is an 8 ohm load), but you are incorrect about the low pass. The JX92S driver is closer to a 4 ohm load, so 1.5 mH would be crossing at 450 Hz or so, rather than at 900 Hz. This makes the current x/o even more unlikely.
What I found was that the Jordan seems to have a built-in 6 dB rise (I believe to counter the baffle effect described above; there is some elaboration on this if you dig through the Jordan web site and its links). However, once I made a low pass x/o in the 5 to 6 kHz range, I was left with a broad midrange peak from 300 to 3 kHz. I did not like the sound, especially on orchestral works. I then began progressively increasing the inductor value from an initial 0.15 mH to the present value. This not only merged better with the ribbon, but removed the broad peak. (I also tried an incredible amount of other x/o components and variations like notch filters, higher order filters and the like; they were all horrible.) I would encourage any of you to play with this inductor value; maybe you'll like 2 mH better.
One of my criteria, besides what I listed in the initial post, was to have one driver cover the crucial midrange from 100 Hz to 5 kHz. I believe this is essentially what is done. As for phase, I have not run true phase tests, but as I said earlier, the step response and impulse response look great. More importantly (sinse computers and test equipment don't have to listen to the speakers), they sound very good. They are musical and can give the old 'goosebump' effect with the right source.
On other comment about the ESg2: they have user replaceable ribbons if you damage them. So far, they seem really robust, I have played them VERY loud without problems. (They do have the additional 1st order high pass at 100 Hz, as described above, to protect them from low frequencies.) Someone asked me about the polarity issue. I did check with a 9V battery the JX92S; they are correctly marked for +/- polarity. I did not similarly check the ESg2 for 2 reasons: you cannot see the ribbon move and DC voltages will destroy them. I based the polarity check of the ESg2 on the measurement system.
Regards, Robert
You're correct in pointing out that the x/o is not intuitively obvious. That is, the high pass is around 6500 Hz (the ribbon is an 8 ohm load), but you are incorrect about the low pass. The JX92S driver is closer to a 4 ohm load, so 1.5 mH would be crossing at 450 Hz or so, rather than at 900 Hz. This makes the current x/o even more unlikely.
What I found was that the Jordan seems to have a built-in 6 dB rise (I believe to counter the baffle effect described above; there is some elaboration on this if you dig through the Jordan web site and its links). However, once I made a low pass x/o in the 5 to 6 kHz range, I was left with a broad midrange peak from 300 to 3 kHz. I did not like the sound, especially on orchestral works. I then began progressively increasing the inductor value from an initial 0.15 mH to the present value. This not only merged better with the ribbon, but removed the broad peak. (I also tried an incredible amount of other x/o components and variations like notch filters, higher order filters and the like; they were all horrible.) I would encourage any of you to play with this inductor value; maybe you'll like 2 mH better.
One of my criteria, besides what I listed in the initial post, was to have one driver cover the crucial midrange from 100 Hz to 5 kHz. I believe this is essentially what is done. As for phase, I have not run true phase tests, but as I said earlier, the step response and impulse response look great. More importantly (sinse computers and test equipment don't have to listen to the speakers), they sound very good. They are musical and can give the old 'goosebump' effect with the right source.
On other comment about the ESg2: they have user replaceable ribbons if you damage them. So far, they seem really robust, I have played them VERY loud without problems. (They do have the additional 1st order high pass at 100 Hz, as described above, to protect them from low frequencies.) Someone asked me about the polarity issue. I did check with a 9V battery the JX92S; they are correctly marked for +/- polarity. I did not similarly check the ESg2 for 2 reasons: you cannot see the ribbon move and DC voltages will destroy them. I based the polarity check of the ESg2 on the measurement system.
Regards, Robert
rljones said:
Baffle Diffraction Step can be viewed as a 6 dB rise in response above the baffle step frequency. So with the low XO point you could just be adding baffle step compensation (BSC). Rune Aleksanderson used the same technique (althou not quite as dramatic) to deal with BS in his JX150/JX53 TL.
dave
many years ago (1976 or 78 or there abouts) I built a 2 way using a 8" and a 1". trying to adhere to teh KISS priciple (few drivers and few XO parts to cover 40-20k). The drivers were from Philips (W8065 and T0162 in case anyone cares) India.
what i found too was something similar. while i did not know about baffle step response at that time I found that as I increase the series inductor in the XO the sound became more listenable. I had no measurements to define why and what of this phenomenon.
The Tweeter if I remember correctly was XOed at about 6k (1st order) and the woofer if I remember sounded better and better as i XOed it lower I thinkI finally stopped at 1.2k or there abouts out of fear that the hole between the woofer and tweeter would get to big.
Today I feel that this was due to the following:
Baffle step (which causes a fall at 6 db per decade starting from F/W and ending at about F/10W) in this case starting at 1300Hz and ending at about 100Hz
The fact that "audiophile sound" is often defined as laid back or recessed midrange (a hole between 2.5k and 5k would create this effect).
room effects
As DIYers we have one advatage no speaker manufacturer has. We can build the speaker to compensate for some of the room anomalies.
Hope this helps. Does any of this make sense?
what i found too was something similar. while i did not know about baffle step response at that time I found that as I increase the series inductor in the XO the sound became more listenable. I had no measurements to define why and what of this phenomenon.
The Tweeter if I remember correctly was XOed at about 6k (1st order) and the woofer if I remember sounded better and better as i XOed it lower I thinkI finally stopped at 1.2k or there abouts out of fear that the hole between the woofer and tweeter would get to big.
Today I feel that this was due to the following:
Baffle step (which causes a fall at 6 db per decade starting from F/W and ending at about F/10W) in this case starting at 1300Hz and ending at about 100Hz
The fact that "audiophile sound" is often defined as laid back or recessed midrange (a hole between 2.5k and 5k would create this effect).
room effects
As DIYers we have one advatage no speaker manufacturer has. We can build the speaker to compensate for some of the room anomalies.
Hope this helps. Does any of this make sense?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.