I was considering a DIY project about 2 years ago but have been too busy until now; I posted about it at that time.
My ideas have firmed up and I'm posting them here for a sanity check.
Background:
Room size 18ft x 14ft x 7.5ft.
Amp is a 300B SET of about 7 watts (subjectively far more powerful than you'd expect). I really love this SET sound.
I don't want headbanging volume; equivalent to 3/4 of the way back in a concert hall is the most I want.
Musical tastes are classical, folk, jazz, some rock and popular; and with a special liking for good female vocals.
Speaker size is limited (for WAF reasons) to about 1000mm high, 210mm wide, about 265mm deep. Regard this as a 'hard' constraint.
Current speakers are IPL S3TL trans lines with Seas drivers; 6.5 inch bass and 1 inch tweeter (H400 or similar). I like the sound but I'm sure I can get more ......
General Plan:
The size constraints limit the bass driver to 6.5 inch. I want good sensitivity and reasonable bass; for this there is simply no substitute for driver area, which suggests an MTM design. I have checked my amp driving weird loads (speakers in parallel etc) and it seems OK; I can use a 4 ohm tap if necessary.
MTM for SET is often used to get sensitivity - note Eros, Living Voice Avatar, North Creek Catamount and others.
It is difficult to make a TL work well with twin drivers, and the pipe area would mean a box well outside my limits; so I give up on this and will look to a normal vented box, tuned for my kind of music.
Kits or Established Designs:
If the North Creek Catamount were still available, I'd be tempted. MTM with Scanspeak 5.25inch slit cone drivers, said to be ideal for SETs.
Others, such as Eros, I'm somehow not totally convinced; remember I'm looking for a beautiful midrange that does wonders for female vocals (as well as as much of the rest as I can get).
Solen's SSUR seems to have promise, but evaluations of the design here on this board (and a builder's comments) leave me underwhelmed.
The nearest good design I have found so far is Zaph's large MTM using Vifa XG18 and Seas 27TDFC.
My Proposal:
MTM using Vifa XT18WH09-08 and Seas 27TDFC. 1000mm x 210mm x 265mm, with curved sides giving some extra width. Allowing for bracing this should give 40 litres. Ported to 38Hz.
This is of course a unique design, but very much inspired by the floor standing version of Zaph's large MTM, and to some extent by Newfinish's project using similar drivers, and by Pjay's work with the same bass unit.
The volume of 40 litres is a lot less than that used by Zaph, and about the same as Vas for *one* driver. It may be an issue. However, modelling it suggests it may be OK. The bass roll off doesn't look bad. F3 is about 50Hz; at 40Hz it gives -6.2dB which is 2.2 lower than Zaph's large ported design. Now I reckon to place the speakers marginally closer to the rear wall than ideal, and I would expect this to compensate for the 2 to 3 dB loss given by less than ideal box size. I'm not a bass freak, (not into rock), but do want a reasonably tuneful bass. This box and tuning give a group delay of 10mS at 40Hz which I believe to be reasonable. Naturally I'm prepared to play with port size to get ythe sound I want, and if push comes to shove could convert the box to an IB of Zaph's volume though I think I'd lose some bass.
I've mainly considered bass here; but I understand and expect that these drivers can give an excellent midrange. If I get that, and compromise the bass slightly, that will be acceptable to me.
I propose to borrow Zaph's Xover point and alignment; 2nd order L-R at 1700 Hz. I can see why this fits these drivers very well so why change this. I expect to start with Zaph's exact design, but with two small changes. First, a low DCR choke on the bass, for a marginal increase in sensitivity and to partially compensate for the highish Zout of my amp. And a lower value series R on the tweeter, to compensate for the lower DCR choke and also the slight extra sensitivity of the XT18 over the XG18; about 2dB in total.
Some modelling of the Xover may be useful especially if I can factor in the Zout of my amp .....
I'm prepared to buy some test equipment to help with final tuning, and I'm OK at this kind of thing (built my own amp, I have a signal generator and a couple of scopes so you get the idea).
Anyway ... does this look like a rubbish project or does it have some promise?
Cost and VFM are not big issues.
Any advice welcomed ... but do please note the hard constraint on size, and the fact I am totally a SET fan!
My ideas have firmed up and I'm posting them here for a sanity check.
Background:
Room size 18ft x 14ft x 7.5ft.
Amp is a 300B SET of about 7 watts (subjectively far more powerful than you'd expect). I really love this SET sound.
I don't want headbanging volume; equivalent to 3/4 of the way back in a concert hall is the most I want.
Musical tastes are classical, folk, jazz, some rock and popular; and with a special liking for good female vocals.
Speaker size is limited (for WAF reasons) to about 1000mm high, 210mm wide, about 265mm deep. Regard this as a 'hard' constraint.
Current speakers are IPL S3TL trans lines with Seas drivers; 6.5 inch bass and 1 inch tweeter (H400 or similar). I like the sound but I'm sure I can get more ......
General Plan:
The size constraints limit the bass driver to 6.5 inch. I want good sensitivity and reasonable bass; for this there is simply no substitute for driver area, which suggests an MTM design. I have checked my amp driving weird loads (speakers in parallel etc) and it seems OK; I can use a 4 ohm tap if necessary.
MTM for SET is often used to get sensitivity - note Eros, Living Voice Avatar, North Creek Catamount and others.
It is difficult to make a TL work well with twin drivers, and the pipe area would mean a box well outside my limits; so I give up on this and will look to a normal vented box, tuned for my kind of music.
Kits or Established Designs:
If the North Creek Catamount were still available, I'd be tempted. MTM with Scanspeak 5.25inch slit cone drivers, said to be ideal for SETs.
Others, such as Eros, I'm somehow not totally convinced; remember I'm looking for a beautiful midrange that does wonders for female vocals (as well as as much of the rest as I can get).
Solen's SSUR seems to have promise, but evaluations of the design here on this board (and a builder's comments) leave me underwhelmed.
The nearest good design I have found so far is Zaph's large MTM using Vifa XG18 and Seas 27TDFC.
My Proposal:
MTM using Vifa XT18WH09-08 and Seas 27TDFC. 1000mm x 210mm x 265mm, with curved sides giving some extra width. Allowing for bracing this should give 40 litres. Ported to 38Hz.
This is of course a unique design, but very much inspired by the floor standing version of Zaph's large MTM, and to some extent by Newfinish's project using similar drivers, and by Pjay's work with the same bass unit.
The volume of 40 litres is a lot less than that used by Zaph, and about the same as Vas for *one* driver. It may be an issue. However, modelling it suggests it may be OK. The bass roll off doesn't look bad. F3 is about 50Hz; at 40Hz it gives -6.2dB which is 2.2 lower than Zaph's large ported design. Now I reckon to place the speakers marginally closer to the rear wall than ideal, and I would expect this to compensate for the 2 to 3 dB loss given by less than ideal box size. I'm not a bass freak, (not into rock), but do want a reasonably tuneful bass. This box and tuning give a group delay of 10mS at 40Hz which I believe to be reasonable. Naturally I'm prepared to play with port size to get ythe sound I want, and if push comes to shove could convert the box to an IB of Zaph's volume though I think I'd lose some bass.
I've mainly considered bass here; but I understand and expect that these drivers can give an excellent midrange. If I get that, and compromise the bass slightly, that will be acceptable to me.
I propose to borrow Zaph's Xover point and alignment; 2nd order L-R at 1700 Hz. I can see why this fits these drivers very well so why change this. I expect to start with Zaph's exact design, but with two small changes. First, a low DCR choke on the bass, for a marginal increase in sensitivity and to partially compensate for the highish Zout of my amp. And a lower value series R on the tweeter, to compensate for the lower DCR choke and also the slight extra sensitivity of the XT18 over the XG18; about 2dB in total.
Some modelling of the Xover may be useful especially if I can factor in the Zout of my amp .....
I'm prepared to buy some test equipment to help with final tuning, and I'm OK at this kind of thing (built my own amp, I have a signal generator and a couple of scopes so you get the idea).
Anyway ... does this look like a rubbish project or does it have some promise?
Cost and VFM are not big issues.
Any advice welcomed ... but do please note the hard constraint on size, and the fact I am totally a SET fan!
Looks like a feasible project. By experimenting with port length you'll find tuning that suits you even with a less than ideal cab volume.
You'll be okay with using Zaph's XO for his XG18/TDFC design. But I strongly suggest experimenting with the tweeter's padding and its lower rolloff. You can adjust these by changing the R2 and C4 values in Zaph's schematic. Although the XT18 looks a bit more sensitive than the XG18 on paper, you never know the right amount of padding that will suit your taste unless you actually try different values. Use two resistors in parallel for higher power handling and finer possible values. For example, you can try two 3 ohm resistors for 1.5 ohms, 3 ohm and 4 ohm resistors for 1.71 ohms, two 4 ohm resistors for 2 ohms, 4 ohm and 5 ohm resistors for 2.22 ohms, and so on. This is not a waste of money. Don't rule out values greater than 2 ohms for the reason I said above.
And according to my experience, adjusting the tweeter's lower rolloff is very important in the voicing process for 2-way speakers. Buy 10 uF cap and a few 1 uF caps. Then you can try 11 uF, 12 uF, 13 uF, and 14 uF. By increasing the value, you will have higher upper midrange SPL. I personally like a bit strong upper midrange for classical music. But you may have a different personal taste.
This fine tuning won't much affect designed phase alignment between drivers. So don't worry about it. If you have any doubt, try xo simulation with different component values I suggested above. But you can trust me. I've done this before. ^_^
Take a look at my web page: http://www.geocities.com/woove99/Spkrbldg/DesigningXO.htm
Scroll down to the section "Voicing your speakers." You'll find some more information about how to fine-tune your speakers to your ear.
Good luck!
You'll be okay with using Zaph's XO for his XG18/TDFC design. But I strongly suggest experimenting with the tweeter's padding and its lower rolloff. You can adjust these by changing the R2 and C4 values in Zaph's schematic. Although the XT18 looks a bit more sensitive than the XG18 on paper, you never know the right amount of padding that will suit your taste unless you actually try different values. Use two resistors in parallel for higher power handling and finer possible values. For example, you can try two 3 ohm resistors for 1.5 ohms, 3 ohm and 4 ohm resistors for 1.71 ohms, two 4 ohm resistors for 2 ohms, 4 ohm and 5 ohm resistors for 2.22 ohms, and so on. This is not a waste of money. Don't rule out values greater than 2 ohms for the reason I said above.
And according to my experience, adjusting the tweeter's lower rolloff is very important in the voicing process for 2-way speakers. Buy 10 uF cap and a few 1 uF caps. Then you can try 11 uF, 12 uF, 13 uF, and 14 uF. By increasing the value, you will have higher upper midrange SPL. I personally like a bit strong upper midrange for classical music. But you may have a different personal taste.
This fine tuning won't much affect designed phase alignment between drivers. So don't worry about it. If you have any doubt, try xo simulation with different component values I suggested above. But you can trust me. I've done this before. ^_^
Take a look at my web page: http://www.geocities.com/woove99/Spkrbldg/DesigningXO.htm
Scroll down to the section "Voicing your speakers." You'll find some more information about how to fine-tune your speakers to your ear.
Good luck!
Thanks!
Hi Jay,
Thanks for the encouraging and most helpful reply - info of this quality is why I love this board!
I've seen your site and in fact printed off your notes on tweaking the Xover, as it's logical and understandable; it gave me some confidence that I might, with some effort and a bit of luck, be able to do some of this. I'm an engineer so I'm used to the cycle of design/model/build/test/measure/correlate to the model/redesign.
A bit more background. Few ready made speakers fit my exacting needs; one is the Living Voice Avatar. So why not buy one? Well, I'm sure it's possible to do better; the bass drivers are decent enough but perhaps a bit run of the mill ... probably not of the standard of the Vifa XG or XT. I wonder if the Xover freq/tweeter choice has been done as effectively as Zaph did in his MTM. And of course I can build to a high standard, with laminated curved sides and a lot of bracing. So I think this XT18/27TDFC MTM has the potential to improve on the Avatar. Whether I have the ability to do so is another question! And I'm sure I know the answer (no ... or not yet).
I have to aim high, as the IPL isn't bad. But modern drivers seem to be somewhat better (the 27TDFC improves on the H400 for a start). However, I see no point in using 'average' drivers as I doubt if they will give much improvement over what I have now.
Hence the choice of drivers.
And above all, it's *fun* !!!!
Hi Jay,
Thanks for the encouraging and most helpful reply - info of this quality is why I love this board!
I've seen your site and in fact printed off your notes on tweaking the Xover, as it's logical and understandable; it gave me some confidence that I might, with some effort and a bit of luck, be able to do some of this. I'm an engineer so I'm used to the cycle of design/model/build/test/measure/correlate to the model/redesign.
A bit more background. Few ready made speakers fit my exacting needs; one is the Living Voice Avatar. So why not buy one? Well, I'm sure it's possible to do better; the bass drivers are decent enough but perhaps a bit run of the mill ... probably not of the standard of the Vifa XG or XT. I wonder if the Xover freq/tweeter choice has been done as effectively as Zaph did in his MTM. And of course I can build to a high standard, with laminated curved sides and a lot of bracing. So I think this XT18/27TDFC MTM has the potential to improve on the Avatar. Whether I have the ability to do so is another question! And I'm sure I know the answer (no ... or not yet).
I have to aim high, as the IPL isn't bad. But modern drivers seem to be somewhat better (the 27TDFC improves on the H400 for a start). However, I see no point in using 'average' drivers as I doubt if they will give much improvement over what I have now.
Hence the choice of drivers.
And above all, it's *fun* !!!!
Hi Mike,
The choice of 27TDFC is excellent. No alternatives to 27TDFC under 100 US dollars except the Peerless HDS 810921. But I don't think the more expensive Peerless is necessarily better than the TDFC.
The XT18 is also good. But as you may have noticed, the XT18's distortion performance in bass and lower midrange is only average according to Zaph's test. Also, Mark K's multi-tone test show that it---XG18 in this case, but I believe they share the same motor---is not a top performer even in the midrange.
Have you considered the new Seas ER18RNX? Its bass and lower midrange performance is among the best. Upper midrange distortions are a bit higher than the class leading SS Revelator or the Usher 8945P. But this won't be noticed much if you use a low crossover point (e.g., 1.5 kHz) with the TDFC. It is also suitable for a small vented design.
Just my 2 cents.
The choice of 27TDFC is excellent. No alternatives to 27TDFC under 100 US dollars except the Peerless HDS 810921. But I don't think the more expensive Peerless is necessarily better than the TDFC.
The XT18 is also good. But as you may have noticed, the XT18's distortion performance in bass and lower midrange is only average according to Zaph's test. Also, Mark K's multi-tone test show that it---XG18 in this case, but I believe they share the same motor---is not a top performer even in the midrange.
Have you considered the new Seas ER18RNX? Its bass and lower midrange performance is among the best. Upper midrange distortions are a bit higher than the class leading SS Revelator or the Usher 8945P. But this won't be noticed much if you use a low crossover point (e.g., 1.5 kHz) with the TDFC. It is also suitable for a small vented design.
Just my 2 cents.
Significant Rethink
Thanks for the suggestion of ER18RNX; it is now a strong contender for this reason.
I have factored in the amp Zout into the simulations. Some good technical info suggests 3.8 ohms for a 300B SET Zout - in my case less as I have a higher ratio OPT. So 3.0 ohms will be close for Zout and cable R (I have long cables) and I've used this.
This has a profound effect on Qts !
The XT18WH goes from 0.37 to 0.47
The XT18WO goes from 0.37 to 0.54
Modelling the XT18WH in a box of 42 litres and various ports gives what I think are somewhat poor results. Tuned to 30Hz gives a peak of 1dB at about 80Hz; F3 about 52Hz, F6 about 42Hz. That's about the best I can get .... I don't think I'd like any more of a peak than that.
The XT18WO gives worse results.
The ER18, with Qts corrected for Zout, gives what seems to me to be a better curve. Box is 40 litres, port 30Hz. There is a peak but it's less, about 0.2dB, and broader - roughly 100 to 150Hz.
F3 about 48Hz, F6 about 38Hz, and F9 about 30Hz.
The nature of the response curve for this design with ER18 seems a lot better to me than the XT designs (but what do I know!).
I have a long way to go to get a design out of this, but hey I think I'm learning.
I know the Zout has an effect on the Xover design, so I can factor this into some simulations ans see where I get to.
I'm finding it very interesting, 'seeing' the effect of Zout on the speaker design!
Thanks for the suggestion of ER18RNX; it is now a strong contender for this reason.
I have factored in the amp Zout into the simulations. Some good technical info suggests 3.8 ohms for a 300B SET Zout - in my case less as I have a higher ratio OPT. So 3.0 ohms will be close for Zout and cable R (I have long cables) and I've used this.
This has a profound effect on Qts !
The XT18WH goes from 0.37 to 0.47
The XT18WO goes from 0.37 to 0.54
Modelling the XT18WH in a box of 42 litres and various ports gives what I think are somewhat poor results. Tuned to 30Hz gives a peak of 1dB at about 80Hz; F3 about 52Hz, F6 about 42Hz. That's about the best I can get .... I don't think I'd like any more of a peak than that.
The XT18WO gives worse results.
The ER18, with Qts corrected for Zout, gives what seems to me to be a better curve. Box is 40 litres, port 30Hz. There is a peak but it's less, about 0.2dB, and broader - roughly 100 to 150Hz.
F3 about 48Hz, F6 about 38Hz, and F9 about 30Hz.
The nature of the response curve for this design with ER18 seems a lot better to me than the XT designs (but what do I know!).
I have a long way to go to get a design out of this, but hey I think I'm learning.
I know the Zout has an effect on the Xover design, so I can factor this into some simulations ans see where I get to.
I'm finding it very interesting, 'seeing' the effect of Zout on the speaker design!
If you look at Zaph's t/s parameter measurements, you'll find that the XT18/XG18's measured Qts is even higher than the ones published by the manufacturer. In contrast, the ER18's measured Qts is close to the one measured by the Seas.
I think the XT18/XG18 are more appropriate for a sealed design for two reasons: average bass/midbass performance and highish Qts (plus you have a high output Z of your amp).
ER18 will be an excellent choice for your project. It has high efficiency, top-performing bass / lower midrange, and low Qts like other Seas 6.5"s. Required cabinet volume can be even as small as 30 liters. 35 to 40 liters will be better in your case since you can get deeper bass and don't need high power handling.
If you're interested, I have a preliminary crossover design for ER18RNX/27TDFC 2-way at my website. This is for a TM vented 2-way. But I also have an MTM version of it (not on my website yet)---it turns out to be simpler than the TM version! Believe it or not, I'm pretty sure that this preliminary design will require only fine-tuning process to sound good. The reason is that I used Zaph's same measurements to design and voice my other 2-way projects. John uses exactly the same measurement settings in each of his driver test groups. So it is possible to obtain high consistency among different xo simulations if we use his data for the same kind of designs. I bet by tweaking the capacitor and padding resistor values of the tweeter network, you can make them sound and measure right to your ears and eyes. But I know that you want to do it yourself in every step. Anyway, this will be used at least as a reference or a starting point for you. If you want, I can send the xo design and its predicted responses (or my Speaker Workshop file itself). Let me know.
I think the XT18/XG18 are more appropriate for a sealed design for two reasons: average bass/midbass performance and highish Qts (plus you have a high output Z of your amp).
ER18 will be an excellent choice for your project. It has high efficiency, top-performing bass / lower midrange, and low Qts like other Seas 6.5"s. Required cabinet volume can be even as small as 30 liters. 35 to 40 liters will be better in your case since you can get deeper bass and don't need high power handling.
If you're interested, I have a preliminary crossover design for ER18RNX/27TDFC 2-way at my website. This is for a TM vented 2-way. But I also have an MTM version of it (not on my website yet)---it turns out to be simpler than the TM version! Believe it or not, I'm pretty sure that this preliminary design will require only fine-tuning process to sound good. The reason is that I used Zaph's same measurements to design and voice my other 2-way projects. John uses exactly the same measurement settings in each of his driver test groups. So it is possible to obtain high consistency among different xo simulations if we use his data for the same kind of designs. I bet by tweaking the capacitor and padding resistor values of the tweeter network, you can make them sound and measure right to your ears and eyes. But I know that you want to do it yourself in every step. Anyway, this will be used at least as a reference or a starting point for you. If you want, I can send the xo design and its predicted responses (or my Speaker Workshop file itself). Let me know.
Hi Jay,
If you're curious, here's how your crossover models using actual in-box measurements. (the design axis was the tweeter @ 1m) In this box, it was a 9" baffle, and the tweeter offset 1" which is probably pretty close. Note that this particular 27TDFC has a slightly different top end than the one I posted a year or two ago in the mishmash, but the sensitivity, impedance and low end response were the same. Your model looks pretty close. Differences are probably related to an increased resolution using near field summing techniques for a more accurate curve, some cavity effect from the woofer's cone on the tweeter's response, and potentially the couple of degrees difference in the off axis response of the woofer on the tweeter axis. There might be some mild level differences in there also. But overall, what you got with tracing and the FRD tools looks pretty close to reality.
Your crossover was almost exactly the same as one of my iterations. I've got several versions done myself and I'm trying to decide which one I like the best. I might end up crossing a bit higher than 1.5kHz as the tradeoff between some other design elements seems to be worth it.
My plot also reflects my bass tuning preference. (this is a .7 cu ft box) This "lean and low" tuning sounds pretty good in a medium sized room.
In regards to the XT and XG woofers, yes Qts has traditionally been a bit higher than spec making for a sealed box preference. I also recommend a single Scanspeak aperiodic vent with these in a smaller sealed box. Also, a bucking magnet really helps these speakers. I think I mentioned the Qts improvements on the PE board somewhere. Performance of the XT and XG woofers is much more level dependent than some other woofers. At lower levels the midrange performance is excellent, but it goes away quickly at higher levels. Generally, people who like it loud are better off with a Seas driver. I don't listen loud myself, and as such the XG18 was still one of the best midranges I've heard. (and tested)
If you're curious, here's how your crossover models using actual in-box measurements. (the design axis was the tweeter @ 1m) In this box, it was a 9" baffle, and the tweeter offset 1" which is probably pretty close. Note that this particular 27TDFC has a slightly different top end than the one I posted a year or two ago in the mishmash, but the sensitivity, impedance and low end response were the same. Your model looks pretty close. Differences are probably related to an increased resolution using near field summing techniques for a more accurate curve, some cavity effect from the woofer's cone on the tweeter's response, and potentially the couple of degrees difference in the off axis response of the woofer on the tweeter axis. There might be some mild level differences in there also. But overall, what you got with tracing and the FRD tools looks pretty close to reality.
Your crossover was almost exactly the same as one of my iterations. I've got several versions done myself and I'm trying to decide which one I like the best. I might end up crossing a bit higher than 1.5kHz as the tradeoff between some other design elements seems to be worth it.
My plot also reflects my bass tuning preference. (this is a .7 cu ft box) This "lean and low" tuning sounds pretty good in a medium sized room.
In regards to the XT and XG woofers, yes Qts has traditionally been a bit higher than spec making for a sealed box preference. I also recommend a single Scanspeak aperiodic vent with these in a smaller sealed box. Also, a bucking magnet really helps these speakers. I think I mentioned the Qts improvements on the PE board somewhere. Performance of the XT and XG woofers is much more level dependent than some other woofers. At lower levels the midrange performance is excellent, but it goes away quickly at higher levels. Generally, people who like it loud are better off with a Seas driver. I don't listen loud myself, and as such the XG18 was still one of the best midranges I've heard. (and tested)
Attachments
I've got HD plots of the XT18 from Hobby Hifi and Klang&Ton at different and multiple levels that nicely illustrate the level dependency of the distortion. Not sure if I could post them here without violating some sort of copyright. (mods what do you think?)
At 80dB/1m, there is almost no better midrange than what comes out of a XT or XG woofer. At 90 dB that lead disappears. My 6.5" HD plots have the level set at 92 dB @1/2m, which falls between the HH and K&T plots I've seen.
At 80dB/1m, there is almost no better midrange than what comes out of a XT or XG woofer. At 90 dB that lead disappears. My 6.5" HD plots have the level set at 92 dB @1/2m, which falls between the HH and K&T plots I've seen.
Thanks for the post, Zaph.
Yes, your simulation of my xo using your in-box measurements is close to my simulation result that used my modeled in-box responses. Indeed they're pretty darn close to the extent that even I am surprised!
In fact, I wanted to use a primary inductor value of 2.8 mH and a shunted cap value of 19 uF in the woofer network to more finely match my voicing preference I had found in my previous building experience. If I did, the measured response would be even flatter than the above! But I simply used the values of available stock components.
BTW, I really appreciate your contribution to the DIY speaker building community. Without your driver test data, my project would not have been possible. I look forward to seeing your ER18RNX 2-way design at your website.
Yes, your simulation of my xo using your in-box measurements is close to my simulation result that used my modeled in-box responses. Indeed they're pretty darn close to the extent that even I am surprised!
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.
In fact, I wanted to use a primary inductor value of 2.8 mH and a shunted cap value of 19 uF in the woofer network to more finely match my voicing preference I had found in my previous building experience. If I did, the measured response would be even flatter than the above! But I simply used the values of available stock components.
BTW, I really appreciate your contribution to the DIY speaker building community. Without your driver test data, my project would not have been possible. I look forward to seeing your ER18RNX 2-way design at your website.
Thanks - and Progress
Thanks for great info.
I'm starting to engage my brain and going more logically.
Instead of taking my box volume and fitting a particular driver into it, I need to analyze what parameters the driver needs and look for a good fit.
In fact, Jay has done this for me.
My amp's Zout will be just under 2 ohms, and another OPT I have has a 4 ohm tap which I will use, also giving just under 2 ohms. With cables, this gives 2.1 ohms and I've redone the models with this. Qts does not change quite as much.
The design with XT18WH still does not look good to me. There is a peak about 100Hz. I assume this peak indicates an over resonant bass (seems to agree with high Q, low damping), possibly a loose bass; anyway, not right.
But with the ER18RNX in a 40 litre box tuned to about 32 to 35 Hz, results look very decent to me. No peak; F3 just under 50Hz; F6 about 37Hz, F9 about 32Hz; a steady smooth curve with a nice rolloff. I hope this may work well.
I will still look at other drivers but this Seas seems to fit the bill very well all round.
More later - I'm away next week.
Thanks for great info.
I'm starting to engage my brain and going more logically.
Instead of taking my box volume and fitting a particular driver into it, I need to analyze what parameters the driver needs and look for a good fit.
In fact, Jay has done this for me.
My amp's Zout will be just under 2 ohms, and another OPT I have has a 4 ohm tap which I will use, also giving just under 2 ohms. With cables, this gives 2.1 ohms and I've redone the models with this. Qts does not change quite as much.
The design with XT18WH still does not look good to me. There is a peak about 100Hz. I assume this peak indicates an over resonant bass (seems to agree with high Q, low damping), possibly a loose bass; anyway, not right.
But with the ER18RNX in a 40 litre box tuned to about 32 to 35 Hz, results look very decent to me. No peak; F3 just under 50Hz; F6 about 37Hz, F9 about 32Hz; a steady smooth curve with a nice rolloff. I hope this may work well.
I will still look at other drivers but this Seas seems to fit the bill very well all round.
More later - I'm away next week.
hi mike
Imade some changes with the tweeter I used before the 27tdc and now I'm with the 9700 of scan speak.alot more open and realistic timbres,but all this depends on taste...
my xovers i recon to have 18db at 1800hz on the woofs and 12db at close to 4000 on the tweets.
the sound is really natural,for voices like patricia barber,norah jones,rebecca pidgeon...etc is just divine.
i used the abc type of box by using a total of 2 cubic feet,and split it in 2/3-1/3
the drivers in the biggest chamber with a 3"x7" port,and the inside port is the same size,and the bottom port for the small chamber is also the same size.
for a room like i had 14'x16'x8' was excellent.
now i have 13'x38'x7' which the speakers now are lacking bass....
but play the rest of the frequencies very well.
by using the scan speak tweet i don't need any pad.
a lot of fun.if you want i changing my setup if you would like the drivers i could make you a very good deal in the whole kit,with either the seas 277tdc or the 9700
they are very easy on small amps,my guess is they are in the 91 to 92db effecient
Imade some changes with the tweeter I used before the 27tdc and now I'm with the 9700 of scan speak.alot more open and realistic timbres,but all this depends on taste...
my xovers i recon to have 18db at 1800hz on the woofs and 12db at close to 4000 on the tweets.
the sound is really natural,for voices like patricia barber,norah jones,rebecca pidgeon...etc is just divine.
i used the abc type of box by using a total of 2 cubic feet,and split it in 2/3-1/3
the drivers in the biggest chamber with a 3"x7" port,and the inside port is the same size,and the bottom port for the small chamber is also the same size.
for a room like i had 14'x16'x8' was excellent.
now i have 13'x38'x7' which the speakers now are lacking bass....
but play the rest of the frequencies very well.
by using the scan speak tweet i don't need any pad.
a lot of fun.if you want i changing my setup if you would like the drivers i could make you a very good deal in the whole kit,with either the seas 277tdc or the 9700
they are very easy on small amps,my guess is they are in the 91 to 92db effecient
Where I'm At
I'm probably more confused than when I started - but I guess that's because I could not see some of the compexities. But I'm learning.
I seem to have found something about the interaction between SETs and speakers; a speaker tuned for a low Zout amp but used with an SET seems likely to have a hump in the bass. Typically I see a dB or two, fairly flat, 80 to 150Hz or so; could this explain the well known 'warmth' of valve amps? I guess it might ...
The XT18 seems a poor choice for a BR enclosure with SET amp. With my Qts, I get a broad hump, +0.5dB from 90Hz to 150Hz, max +0.65dB. That's about the best I can model, with box tuning 32Hz. I assume a warm slightly loose bass, perhaps a touch of boom.
Even so, I don't yet rule out the XT18 for this reason. I don't listen at high volume, and of course they are doubled up in the MTM, so they will be working at low volume where they sound very good. The excellent midrange might outweigh a *slight* bass problem. I don't listen to much rock, and a slight bit of bass warmth just might suit my tastes. And if it's bad (excess boom) then I can convert it to IB.
The ER18RNX is a very good 'fit' to the overall design, and measures very well. I guess I just need some confirmation that this means it sounds good, especially for vocals ..... I'll do some more looking around.
Whatever drivers I use, I have to do a lot of work on the Xover, as amp Zout has an effect. Jay is being fantastically helpful, but I know I have to do more analysis, experimentation and measurement.
'Newfinish': thanks for your input! I'm familiar with your XT18 MTMs. I doubt if my boxes will have enough volume for the coupled cavity bass; more reading upon the subject required.
I also find your experience with the 9700 tweeter interesting ... I'll look it up on the tymphany site and see how that might work with either of my choices of bass driver.
I have some tricky choices but I feel that any of the choices of driver could give me a good result.
I'm probably more confused than when I started - but I guess that's because I could not see some of the compexities. But I'm learning.
I seem to have found something about the interaction between SETs and speakers; a speaker tuned for a low Zout amp but used with an SET seems likely to have a hump in the bass. Typically I see a dB or two, fairly flat, 80 to 150Hz or so; could this explain the well known 'warmth' of valve amps? I guess it might ...
The XT18 seems a poor choice for a BR enclosure with SET amp. With my Qts, I get a broad hump, +0.5dB from 90Hz to 150Hz, max +0.65dB. That's about the best I can model, with box tuning 32Hz. I assume a warm slightly loose bass, perhaps a touch of boom.
Even so, I don't yet rule out the XT18 for this reason. I don't listen at high volume, and of course they are doubled up in the MTM, so they will be working at low volume where they sound very good. The excellent midrange might outweigh a *slight* bass problem. I don't listen to much rock, and a slight bit of bass warmth just might suit my tastes. And if it's bad (excess boom) then I can convert it to IB.
The ER18RNX is a very good 'fit' to the overall design, and measures very well. I guess I just need some confirmation that this means it sounds good, especially for vocals ..... I'll do some more looking around.
Whatever drivers I use, I have to do a lot of work on the Xover, as amp Zout has an effect. Jay is being fantastically helpful, but I know I have to do more analysis, experimentation and measurement.
'Newfinish': thanks for your input! I'm familiar with your XT18 MTMs. I doubt if my boxes will have enough volume for the coupled cavity bass; more reading upon the subject required.
I also find your experience with the 9700 tweeter interesting ... I'll look it up on the tymphany site and see how that might work with either of my choices of bass driver.
I have some tricky choices but I feel that any of the choices of driver could give me a good result.
If you look further for more 7" options around $100 or less, I'm pretty sure it should be one of the following---I know this because I did the same thing :
1. Usher 8945P - On par with (or better than) the ScanSpeak Revelator 7" in distortion performance according to Zaph and Mark K's test.
2. Usher 8945A - Very clean bass. Only a little worse than the P version in midrange performance.
3. Peerless HDS Nomex Exclusive 830883 - Best of the Peerless HDS 7".
4. Peerless XT18 or XG18 - Midrange among the best when played at a moderate volume level.
5. Seas ER18RNX - Lowest distortions of all Seas Prestige 7".
6. Dayton RS180S-8 - Distortion performance on par with the Peerless HDS Exclusive or the Seas ER18RNX at half the price, but not easy to work with due to its breakup nodes.
If your goal is lowest possible distortion, look no further than the Usher 8945P, which is what I ended up with (currently on sale at PE; well worth the price even considering shipping cost to UK). Otherwise, the ER18RNX looks like an excellent choice for your application---low Qts, highish Re (good for MTM use), and good efficiency. XT18 should still be a good choice, too, if its weakness doesn't bother you---every speaker design is a result of compromise, anyway.
:1. Usher 8945P - On par with (or better than) the ScanSpeak Revelator 7" in distortion performance according to Zaph and Mark K's test.
2. Usher 8945A - Very clean bass. Only a little worse than the P version in midrange performance.
3. Peerless HDS Nomex Exclusive 830883 - Best of the Peerless HDS 7".
4. Peerless XT18 or XG18 - Midrange among the best when played at a moderate volume level.
5. Seas ER18RNX - Lowest distortions of all Seas Prestige 7".
6. Dayton RS180S-8 - Distortion performance on par with the Peerless HDS Exclusive or the Seas ER18RNX at half the price, but not easy to work with due to its breakup nodes.
If your goal is lowest possible distortion, look no further than the Usher 8945P, which is what I ended up with (currently on sale at PE; well worth the price even considering shipping cost to UK). Otherwise, the ER18RNX looks like an excellent choice for your application---low Qts, highish Re (good for MTM use), and good efficiency. XT18 should still be a good choice, too, if its weakness doesn't bother you---every speaker design is a result of compromise, anyway.
Getting Clearer
I've already looked at most drivers of this size; some comments follow.
It's nice of course to keep cost down to say $100, but not essential. If I get a really good result for $2000 total cost, that's fine. Or if I get what I really want for $3000, I'll be very happy ... because I can't walk into a shop and buy what I really want.
Odd comments about driver choice, posted here in case they may be of use to other SET fans if they are correct ... and if not correct, please let me know!
Drivers with moderate Qms and moderate Qes seem to be less affected by the change in Qes due to high Zout of the amp. Intuitively, they are well damped mechanically, so the loss of eectrical damping has less effect.
Drivers with high Qms (like say 1.5 and above) end up with high Qts when amp Zout is factored into Qes.
In my case, a high Qms driver does not match my design well. It may do for other people - but still needs careful consideration of the true Qts when driven by an SET.
Drivers:
SS8531 - an excellent driver; but with Qms 5.1 and Vas 59 litres it does not look at all promising for me. To use SS drivers, I'd need to use the 150mm ones - and I'd lose efficiency.
Usher 8945P - an excellent driver; I've modelled it with my Zout but it needs a larger box than I can manage. In my preferred box, tuned to 32Hz it's not bad but has a peak max 1dB 70 to 150Hz (worse than the XT18).
Peerless 830883 - highish Qms but lowish Vas, seems to give a good result in 40 litre box tuned to 32Hz with very minor peak at 120Hz, f3 48Hz, f6 35Hz, f9 28Hz (approx figures). Very promising.
So it seems to come down to the Peerless 830883, ER18RNX or XT18. The ER18 still seems to model the best ....
I know all of these drivers are excellent so I doubt if I'll go far wrong.
Tweeter choice remains 27TDFC with SS 9700 in contention. I know I have to consider the Xover point and slope as the major factor in this decision.
I've already looked at most drivers of this size; some comments follow.
It's nice of course to keep cost down to say $100, but not essential. If I get a really good result for $2000 total cost, that's fine. Or if I get what I really want for $3000, I'll be very happy ... because I can't walk into a shop and buy what I really want.
Odd comments about driver choice, posted here in case they may be of use to other SET fans if they are correct ... and if not correct, please let me know!
Drivers with moderate Qms and moderate Qes seem to be less affected by the change in Qes due to high Zout of the amp. Intuitively, they are well damped mechanically, so the loss of eectrical damping has less effect.
Drivers with high Qms (like say 1.5 and above) end up with high Qts when amp Zout is factored into Qes.
In my case, a high Qms driver does not match my design well. It may do for other people - but still needs careful consideration of the true Qts when driven by an SET.
Drivers:
SS8531 - an excellent driver; but with Qms 5.1 and Vas 59 litres it does not look at all promising for me. To use SS drivers, I'd need to use the 150mm ones - and I'd lose efficiency.
Usher 8945P - an excellent driver; I've modelled it with my Zout but it needs a larger box than I can manage. In my preferred box, tuned to 32Hz it's not bad but has a peak max 1dB 70 to 150Hz (worse than the XT18).
Peerless 830883 - highish Qms but lowish Vas, seems to give a good result in 40 litre box tuned to 32Hz with very minor peak at 120Hz, f3 48Hz, f6 35Hz, f9 28Hz (approx figures). Very promising.
So it seems to come down to the Peerless 830883, ER18RNX or XT18. The ER18 still seems to model the best ....
I know all of these drivers are excellent so I doubt if I'll go far wrong.
Tweeter choice remains 27TDFC with SS 9700 in contention. I know I have to consider the Xover point and slope as the major factor in this decision.
Hi Mike,
Your research is careful and thorough! I like it.
As for the tweeters, I also did comprehensive research on them. If you want to cross below 2 kHz---this seems to be your case if you choose the ER18RNX or the Peerless HDS because good tweeters' distortion performance below 2 kHz is better than that of these midwoofers---, basically we have the following choices:
1. Seas standards (27TDFC or 27TBFC/G)
2. Dayton RS28A-4 (Zaph's tested drivers were from the problematic initial production batch; since then the problem was corrected.)
3. Peerless HDS 810921
If you want better distortion performance than these tweeters', you will need to use Scan Speak AirCirc 6600 or D2904/7100. But the improvement will be only marginal and not very much audible.
Mark Krawiec performed thorough tests of the SS 9300, 9500, 9700, Usher 9930, 9950, and all the tweeters I listed above. If you look at his distortion data carefully, you will find that the weakness of the SS's old designs and also the Usher's is their low end distortion performance. Not bad tweeters by any means, but not quite up to the above choices.
Your research is careful and thorough! I like it.
As for the tweeters, I also did comprehensive research on them. If you want to cross below 2 kHz---this seems to be your case if you choose the ER18RNX or the Peerless HDS because good tweeters' distortion performance below 2 kHz is better than that of these midwoofers---, basically we have the following choices:
1. Seas standards (27TDFC or 27TBFC/G)
2. Dayton RS28A-4 (Zaph's tested drivers were from the problematic initial production batch; since then the problem was corrected.)
3. Peerless HDS 810921
If you want better distortion performance than these tweeters', you will need to use Scan Speak AirCirc 6600 or D2904/7100. But the improvement will be only marginal and not very much audible.
Mark Krawiec performed thorough tests of the SS 9300, 9500, 9700, Usher 9930, 9950, and all the tweeters I listed above. If you look at his distortion data carefully, you will find that the weakness of the SS's old designs and also the Usher's is their low end distortion performance. Not bad tweeters by any means, but not quite up to the above choices.
Decision Made (ER18RNX)
I realised that I made an error in my analysis of the effect of Zout on Qes and Qts. My analysis was correct for one driver; but of course I will use two in parallel.
My way of correcting for the two drivers driven by 2 ohms is this. If I regard it as :
one driver driven by an amp with 4 ohms Zout
in parallel with exactly the same again
then I get the two drivers in parallel driven by a 2 ohm Zout.
So I have repeated the analysis, on one driver driven by 4 ohm Zout amp.
This seems to have made the 'marginal' cases into 'no go'.
Leaving the ER18RNX still looking acceptable, with a slight peak of just under 1 dB at 90Hz; 0.5dB up at 70Hz and 140Hz approx.
Bass rolloff is steady, no hard knee, F3 about 47Hz, F6 about 37Hz, F9 about 30Hz. Group delay is within Zaph's limit.
So technically this driver seems a good fit to my wants and constraints. Yes, I know it can do better in a better aligned box, but that's not an option.
I will be able to try port dimensions and possibly stuffing the port for tuning; and if I'm still not happy I can try a sealed box ... my proposed dimensions are not far off and I can of course reduce the volume. I'll check how this might work out before I build.
Now the other issue; does the ER18RNX sound good? I know it measures very well - but I don't always trust measurements alone. Remember I use an amp that measures about the *worst* that any amp could do .... but to me, with music I like, gives me the best sound I've ever had.
Anyway, google searches suggest that indeed this Seas does sound very good indeed, and what 'character' it has may suit me. I found a very interesting thread on diymobileaudio.com .
So .... measures well, sounds good, fits my tentative design; that's IT!
I intend to use the 27TDFC tweeter - though I'll look into the HDS for interest. But there's so much good info on this Seas and how to use it that I'm sure it's the best one for me to start with. And I doubt if I'll find any desire to change.
I guess this wraps up this thread for now as far as I'm concerned. I have ideas how to proceed.
I've already checked the BSC I need and have the curve.
I know of the modeller for room response and can try this.
Combining the results of these two should give me a target curve.
Then I need to plan a Xover to get close to this. Jay has provided a good one - but I'll try more modelling (while doing a lot of research into the subject).
When the speakers are built and run in, I plan to measure the responses in the box; then try to correlate what I measure with what I hear (and I have some golden eared friends to help here, you know who you are). With the help of Jay's great details on tuning, I hope to 'get there'. If I do .... I'm getting a warm feeling, this could be awesome.
Thanks guys!
I realised that I made an error in my analysis of the effect of Zout on Qes and Qts. My analysis was correct for one driver; but of course I will use two in parallel.
My way of correcting for the two drivers driven by 2 ohms is this. If I regard it as :
one driver driven by an amp with 4 ohms Zout
in parallel with exactly the same again
then I get the two drivers in parallel driven by a 2 ohm Zout.
So I have repeated the analysis, on one driver driven by 4 ohm Zout amp.
This seems to have made the 'marginal' cases into 'no go'.
Leaving the ER18RNX still looking acceptable, with a slight peak of just under 1 dB at 90Hz; 0.5dB up at 70Hz and 140Hz approx.
Bass rolloff is steady, no hard knee, F3 about 47Hz, F6 about 37Hz, F9 about 30Hz. Group delay is within Zaph's limit.
So technically this driver seems a good fit to my wants and constraints. Yes, I know it can do better in a better aligned box, but that's not an option.
I will be able to try port dimensions and possibly stuffing the port for tuning; and if I'm still not happy I can try a sealed box ... my proposed dimensions are not far off and I can of course reduce the volume. I'll check how this might work out before I build.
Now the other issue; does the ER18RNX sound good? I know it measures very well - but I don't always trust measurements alone. Remember I use an amp that measures about the *worst* that any amp could do .... but to me, with music I like, gives me the best sound I've ever had.
Anyway, google searches suggest that indeed this Seas does sound very good indeed, and what 'character' it has may suit me. I found a very interesting thread on diymobileaudio.com .
So .... measures well, sounds good, fits my tentative design; that's IT!
I intend to use the 27TDFC tweeter - though I'll look into the HDS for interest. But there's so much good info on this Seas and how to use it that I'm sure it's the best one for me to start with. And I doubt if I'll find any desire to change.
I guess this wraps up this thread for now as far as I'm concerned. I have ideas how to proceed.
I've already checked the BSC I need and have the curve.
I know of the modeller for room response and can try this.
Combining the results of these two should give me a target curve.
Then I need to plan a Xover to get close to this. Jay has provided a good one - but I'll try more modelling (while doing a lot of research into the subject).
When the speakers are built and run in, I plan to measure the responses in the box; then try to correlate what I measure with what I hear (and I have some golden eared friends to help here, you know who you are). With the help of Jay's great details on tuning, I hope to 'get there'. If I do .... I'm getting a warm feeling, this could be awesome.
Thanks guys!
Good choice, Mike.
Seas 7" woofers are many people's long-time favoirte. I believe the ER18RNX was designed on this success, simply with a different kind of cone material and an improved motor.
Good luck with your building. And if you have any questions on crossover design and tuning, let me know.
Seas 7" woofers are many people's long-time favoirte. I believe the ER18RNX was designed on this success, simply with a different kind of cone material and an improved motor.
Good luck with your building. And if you have any questions on crossover design and tuning, let me know.
Any MTM designs with LR type xover suffer from lobing errors (narrow horizontal listening window) when placed horizontally. To mitigate this problem, some people use inverted "V" shape driver positions, minimize woofers' center-to-center distance, or use a low crossover frequency. My ER18 MTM design uses a relatively low 1.6 kHz crossover point. If listeners sit within +/- 15 degrees from the speaker's center line, its frequency response will still remain acceptably flat.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.