David,
Thanks, I chose the ribbon stripe sepele for that reason. The ribbon stripe is quarter sawn which makes it very dimensionally stable. I intend to take 1 inch thick boards and side glue them to make the individual panels. I have spoken with several wood working shops and have been told it will work well. I intend to seal the wood very well.
Thanks, I chose the ribbon stripe sepele for that reason. The ribbon stripe is quarter sawn which makes it very dimensionally stable. I intend to take 1 inch thick boards and side glue them to make the individual panels. I have spoken with several wood working shops and have been told it will work well. I intend to seal the wood very well.
I missed this one before.
That is NOT the driver in free air (..and therefor is NOT the driver itself). In fact the caption under it reveals what it is:
"Right: Impedance of drivers (JA8008 (blue) IN CABINET). Green (JA8008) and yellow (TW034) is phase."
(..emphasis added in all-caps.)
That is the driver behaving somewhat poorly due to the transmission line's loading. (..and even then I've seen a LOT worse.)
It's a matter of trade-offs for the design. Put up with some ripple to get higher spls from the line OR heavily damp the line with almost no supplemental output and have a "cleaner" looking response from the driver.
Frankly it's rather obvious, it's stated, and it make me wonder about the motivation of this post..
Also you have stated that Troels receives compensation for the kits, and presumably the drivers?
Is this correct? How much?
I don't think I've ever read anything to suggest that he does. Jantzen on the other hand most certainly are.
If you "cut it up and glue it back together" somewhat like "butcher-block" it should work well. In fact most of my furniture is manufactured like this.. with wood species I don't want to even think about..
Note that the width of board is usually varied - you need to specify your requirements. Use the lowest width - which is apparently 4 inches. Also consider getting it edge-planed by them - it reduces the amount of material you need to take care of, but still expect to use a planer or router for precise joining. Consider "dowel" joining instead of "biscuits" for this work.
A lot of work.. but it's pretty repetitive so it shouldn't be that difficult.
Staining to get an even finish will (by far) be the most work.. and you should have multiple "samples" for practice to get the finish you want (before you work on the actual finish).
Veneer should be less expensive and more uniform however.
Here is a 4X8 sheet of sapele ribbon for 70 US:
http://www.woodcraft.com/Family/2008679/2008679.aspx
You would still want the hardwood for all of the edges.. but that's a LOT less surface area.
ScottG,
Thanks, I expect it to take me a long time to complete. Did you see the picture of the guitar? I really like the variation in the grain. I think it looks a lot less common than most speaker cabinets. I guess not everyone will care for it.
I was actually looking at the sapele pomele veneer, which isn't cheap. To be honest, I enjoy the woodworking portion of the build. I haven't done this before, so I should learn something from it. Hopefully, it won't get more expensive by me messing up. I will layer the baffles with either mdf or plywood after sealing it.
I will need to use a filler and then sand prior to staining to get that super smooth finish. I intend to go for a high gloss luster with several coats of poly.
Thanks, I expect it to take me a long time to complete. Did you see the picture of the guitar? I really like the variation in the grain. I think it looks a lot less common than most speaker cabinets. I guess not everyone will care for it.
I was actually looking at the sapele pomele veneer, which isn't cheap. To be honest, I enjoy the woodworking portion of the build. I haven't done this before, so I should learn something from it. Hopefully, it won't get more expensive by me messing up. I will layer the baffles with either mdf or plywood after sealing it.
I will need to use a filler and then sand prior to staining to get that super smooth finish. I intend to go for a high gloss luster with several coats of poly.
ScottG,
Yes you're right. I also think that on the DTQWT wb page where Troels is talking about the front drivers going in to the project, that he shows a graph for the tweeter in cabinet alongside this one and these measurements appear to me to be driver response measurements IN CABINET, as you say, but ALSO without an appropriate crossover which can be used to smooth the result - hence the claim on the x-axis of the graph of "UNSMOOTHED".
I suspect that if we saw the response with crossover applied it would be sufficiently smooth.
Yes you're right. I also think that on the DTQWT wb page where Troels is talking about the front drivers going in to the project, that he shows a graph for the tweeter in cabinet alongside this one and these measurements appear to me to be driver response measurements IN CABINET, as you say, but ALSO without an appropriate crossover which can be used to smooth the result - hence the claim on the x-axis of the graph of "UNSMOOTHED".
I suspect that if we saw the response with crossover applied it would be sufficiently smooth.
DTQWT Thoughts/Questions
I've been following this thread with great interest for a while. I have also communicated with Troels via email and found him to be very responsive, knowlegeable, and willing to share ideas. I do have a couple of comments / questions regarding drivers and enclosure design for this speaker.
1. Someone pointed out earlier that there was some 'peakyness' of the impedance response. There is a 1kHz peak in the impedance of the JA8008 (driver, not driver + enclosure). It would seem that there are many high-end drivers that suffer from one or two of these impedance blips, (reflections from spider, spider resonance, suspension resonance, etc.). Are there any comparable drivers at this performance level and price point that are better in terms of low distortion, sensitivity, low compression, SQ and octave reach?
2. In the DTQWT version a pair of Emminance 'helper woofers' is crossed over at ~180Hz with a 6db low pass in a bipolar fashion on the rear. Someone made the comment earlier that they where unsure of the Emminacne drivers, I agree that the Emminence bass drivers seem to have a ragged impedance plot and FR. Implemented in this fashion, the bass drivers are ~ 27db down at ~2,880 Hz, is this low enough to not interfere with the midrange SQ? Fs is also not very low leading me to believe that the bass response may not extend very low. I've never build a TQWT or DTQWT before and am more familiar with BR and Sealed enclosures. I was told that the F3 of the design would have very little to do with Fs of the bass drivers and that line length of the enclosure and displacement, sensitivity, Mms and BL of the bass drivers were more important. Also, what's the phase impact of a rear mounted bipolar driver?
3. The spec sheet for the JA8008 says "Be aware that normal box-simulations do not apply to this kind of driver." What does this mean specifically?
4. I have not been able to find the Xmax spec for the JA8008. How loud can it get down to its cut-off before it runs out of steam?
5. Regarding the enclosure, since the skinny portion of the line is stuffed is this more of a ML-TQWT design? I notice that it differs from many TQWT designs I've seen - note: I am not an experienced TQWT builder, what's the impact of folding the quarterwave pipe and what about the open end versus a tuned port? Does this design act like a cross between an ML-TQWT and a horn because of the large open bottom? Can it be modeled in Mathcad?
Thanks again for any replies.
I've been following this thread with great interest for a while. I have also communicated with Troels via email and found him to be very responsive, knowlegeable, and willing to share ideas. I do have a couple of comments / questions regarding drivers and enclosure design for this speaker.
1. Someone pointed out earlier that there was some 'peakyness' of the impedance response. There is a 1kHz peak in the impedance of the JA8008 (driver, not driver + enclosure). It would seem that there are many high-end drivers that suffer from one or two of these impedance blips, (reflections from spider, spider resonance, suspension resonance, etc.). Are there any comparable drivers at this performance level and price point that are better in terms of low distortion, sensitivity, low compression, SQ and octave reach?
2. In the DTQWT version a pair of Emminance 'helper woofers' is crossed over at ~180Hz with a 6db low pass in a bipolar fashion on the rear. Someone made the comment earlier that they where unsure of the Emminacne drivers, I agree that the Emminence bass drivers seem to have a ragged impedance plot and FR. Implemented in this fashion, the bass drivers are ~ 27db down at ~2,880 Hz, is this low enough to not interfere with the midrange SQ? Fs is also not very low leading me to believe that the bass response may not extend very low. I've never build a TQWT or DTQWT before and am more familiar with BR and Sealed enclosures. I was told that the F3 of the design would have very little to do with Fs of the bass drivers and that line length of the enclosure and displacement, sensitivity, Mms and BL of the bass drivers were more important. Also, what's the phase impact of a rear mounted bipolar driver?
3. The spec sheet for the JA8008 says "Be aware that normal box-simulations do not apply to this kind of driver." What does this mean specifically?
4. I have not been able to find the Xmax spec for the JA8008. How loud can it get down to its cut-off before it runs out of steam?
5. Regarding the enclosure, since the skinny portion of the line is stuffed is this more of a ML-TQWT design? I notice that it differs from many TQWT designs I've seen - note: I am not an experienced TQWT builder, what's the impact of folding the quarterwave pipe and what about the open end versus a tuned port? Does this design act like a cross between an ML-TQWT and a horn because of the large open bottom? Can it be modeled in Mathcad?
Thanks again for any replies.
Last edited:
The peakyness Syncroniq mentioned were the small steps circa 150Hz-400Hz which he attributed to this. The 1Khz peak I would agree is as you say. I looked for a substitutable driver for this project (given the cost of the JA8008 is, I think, a bit steep) which I am saving for, but I couldn't find one. The measurements were IN CABINET. There is a blurb on Troels site about the JA8008 which shows measurement in free-air where this impedance peak is about 42Hz: JA8008
I think these are best asked to Troels himself. Certainly the plots suggest bass does get quite low with this design, if not quite full range. Also, I think there is a comment that the Eminence is used intentionally up to 2560Hz before rolling off and I think the reason is stated in the paragraph starting "In many ways life gets easier" on Troels DTQWT main build page. It sounds to me as if we are simply using more drivers to shift more air, thus need less driver excursion to achieve a given volume. From that, I infer we are going to divide power across more drivers operating with less excursion and, as a side effect, get less distortion than if using a single driver. However Troels does say this is not the ideal situation as the drivers are covering bass and midrange duties somewhat - at least that's how I read it. I am new to TQWTs and Voigt Pipes too, so your estimate is as good (or probably better) than mine on more important driver parameters, but from my reading these TQWTs and VPs are different beasts and much of their required design is quite different to typical thinking applied to bass reflex and sealed enclosures.
I'd like to know too. I can see that being a valid comment for the DTQWT or a TQWT etc, but not for typical BR or SE.
Good question. On Troels site he mentions about designing a driver is full of compromise and how certain things like a shorter voice coil will reduce xmax, but xmax is never given. I re-read the spec sheet at Jantzen and it's not there either. It's a fairly notable absence. For a driver that is much more costly than the SEAS it derives from, it should show what xmax is.
Like you, I'm not au fait with DTQWT or TQWT, but I can imagine what physically happens. So on that basis, I would say yes, it is more like a cross between a folded ML-TQWT and a horn. As for modeling it in Mathcad.... you just won the masochist of the year award!
Okay, not that bad really. Here's the basic math sheet for TQWTs:MathCad Computer Models : Upgraded Versions
More specifically:
http://www.quarter-wave.com/Back_Door/TL_ML_Corner_7_03_09.pdf
TQWT Enclosure calculator:
http://fullrangedriver.com/singledriver/TQWP.xls
Not math, but you may wish to read Thorsten's comments on the TQWT and horns:
Tweakers' Asylum: Some notes on High Sensitivity Speakers by Thorsten
If I can dig up anything more helpful, I'll let you know.
Okay, this is my first speaker project and I have quickly realized that even though I've done a lot of reading, most of you are on a completely different level. I did some reading last night on voight pipes / tqwt design and came across this bit on impedance issues with a tqwt. It describes how a dtqwt counters this issue. It also discusses the bass response. Hopefully, this is helpful.
The TQWT is really good at a few things. It manages very well to combine a horn and transmission line (and also some reflex loading). TQWT's TL behavior offers a nice acoustic load for the driver over most of it's operating range and in addition the bass reflex reinforcement can be tuned pretty low so that a pretty deep and tight bass can result.
The main problem of the TQWT is that it will produce a comb-filter effect due to line resonances at lower frequencies, certain bass notes would be near inaudible, other would boom way too much, a very uneven and lumpy bass response. The only way to reduce these line resonances is to stuff the line severely, which changes air speed and therefore acts as a prolongation line, but at the same time kills efficiency. The notches are caused by line harmonics canceling the output from the drivers front. They correspond to impedance curve peaks which can be easily seen on a plot.
There is fortunately another option. Instead of troublesome tuning, two or more pipes can be employed in one enclosure, possibly even with different drivers. One pipe should be tuned exactly 1/2 octave below the other, once the minimal stuffing in the throat (the narrowest point in the tapered tube) is taken into account. The trick is to have a major port resonance at the lowest frequency desired (essentially the driver's resonance). The pipe resonance will likely be somewhat higher than that. The port then needs a second "parasitic" (pipe type) resonance at the second harmonics of the line resonance. Normally this (2nd harmonic) resonance is out of phase with the driver's front radiation and hence creates a cancellation, the source of the deepest and lowest notch. This means the "parasitic" and normally unwanted resonance of a classic port is desired. What it does is to simply remove energy from the pipe at the the 2nd harmonic. In any case, now one pipe will have a "peak" wherever the other has a dip. If this is well calculated such enclosure will deal well with the resonances, having a much flatter impedance curve and a fairly flat frequency response. This procedure should provide a deep and tight bass with a pretty even response.
I am no expert on TQWT designs. This argument makes sense to me. Does anyone have an opinion on this?
I also realized, that TQWTs are often used with a single full range driver, ie lowther drivers. I know this may be common knowledge to many here but I'm learning as I go. Anyway, one of the main critiques that I read on these full range drivers is that they often need some reinforcement at the extremes of deep bass or high frequencies. I am assuming that this is why Troels chose this group of drivers for his DTQWT. Am I correct in my assumption?
As a note, I am completely unfamiliar with lowther designs besides what I've read. I understand they have quite a following.
The TQWT is really good at a few things. It manages very well to combine a horn and transmission line (and also some reflex loading). TQWT's TL behavior offers a nice acoustic load for the driver over most of it's operating range and in addition the bass reflex reinforcement can be tuned pretty low so that a pretty deep and tight bass can result.
The main problem of the TQWT is that it will produce a comb-filter effect due to line resonances at lower frequencies, certain bass notes would be near inaudible, other would boom way too much, a very uneven and lumpy bass response. The only way to reduce these line resonances is to stuff the line severely, which changes air speed and therefore acts as a prolongation line, but at the same time kills efficiency. The notches are caused by line harmonics canceling the output from the drivers front. They correspond to impedance curve peaks which can be easily seen on a plot.
There is fortunately another option. Instead of troublesome tuning, two or more pipes can be employed in one enclosure, possibly even with different drivers. One pipe should be tuned exactly 1/2 octave below the other, once the minimal stuffing in the throat (the narrowest point in the tapered tube) is taken into account. The trick is to have a major port resonance at the lowest frequency desired (essentially the driver's resonance). The pipe resonance will likely be somewhat higher than that. The port then needs a second "parasitic" (pipe type) resonance at the second harmonics of the line resonance. Normally this (2nd harmonic) resonance is out of phase with the driver's front radiation and hence creates a cancellation, the source of the deepest and lowest notch. This means the "parasitic" and normally unwanted resonance of a classic port is desired. What it does is to simply remove energy from the pipe at the the 2nd harmonic. In any case, now one pipe will have a "peak" wherever the other has a dip. If this is well calculated such enclosure will deal well with the resonances, having a much flatter impedance curve and a fairly flat frequency response. This procedure should provide a deep and tight bass with a pretty even response.
I am no expert on TQWT designs. This argument makes sense to me. Does anyone have an opinion on this?
I also realized, that TQWTs are often used with a single full range driver, ie lowther drivers. I know this may be common knowledge to many here but I'm learning as I go. Anyway, one of the main critiques that I read on these full range drivers is that they often need some reinforcement at the extremes of deep bass or high frequencies. I am assuming that this is why Troels chose this group of drivers for his DTQWT. Am I correct in my assumption?
As a note, I am completely unfamiliar with lowther designs besides what I've read. I understand they have quite a following.
Hi,
Yep that all looks right to me (including your final assumption). I did say this thing acts like a comb-filter and that's where those steps/peaks/dips come from and not from the driver itself as Syncroniq was trying to allude to.
It is also why it needs damping heavily compared to a more contemporary vented or sealed box.
I've tried brushing up too. It seems the best writings on this type of enclosure are written in this book:
Amazon.com: Great Sound Stereo Speaker Manual (TAB Electronics) (9780071348744): David B. Weems, G. R. Koonce: Books
Yep that all looks right to me (including your final assumption). I did say this thing acts like a comb-filter and that's where those steps/peaks/dips come from and not from the driver itself as Syncroniq was trying to allude to.
It is also why it needs damping heavily compared to a more contemporary vented or sealed box.
I've tried brushing up too. It seems the best writings on this type of enclosure are written in this book:
Amazon.com: Great Sound Stereo Speaker Manual (TAB Electronics) (9780071348744): David B. Weems, G. R. Koonce: Books
I'd add... people read about TQWT shortcomings and/or not very good bass depth and don't persist with overcoming this.
The book above focuses more on full-range drivers, which one reviewer on Amazon was clearly unhappy about together with using cheap drivers. However, it's not how costly the drivers are, it's how they perform in the cabinet at completion time that matters. In most cases, a properly engineerd cabinet and crossover with very cheap drivers will outperform a poorly designed cabinet and crossover with expensive drivers.
Given room acoustics is the major obstacle to decent sound in most people systems, I find it incredulous that people dont realise the cabinet is effectively a small-room the drivers are integrated in to and thus, getting the cabinet (the other 'room') engineered right is as critical as room correction. Expensive drivers are only going to be fitting for the very best cabinets, or tried and tested models that will yield additional benefit in simulation.
The book above focuses more on full-range drivers, which one reviewer on Amazon was clearly unhappy about together with using cheap drivers. However, it's not how costly the drivers are, it's how they perform in the cabinet at completion time that matters. In most cases, a properly engineerd cabinet and crossover with very cheap drivers will outperform a poorly designed cabinet and crossover with expensive drivers.
Given room acoustics is the major obstacle to decent sound in most people systems, I find it incredulous that people dont realise the cabinet is effectively a small-room the drivers are integrated in to and thus, getting the cabinet (the other 'room') engineered right is as critical as room correction. Expensive drivers are only going to be fitting for the very best cabinets, or tried and tested models that will yield additional benefit in simulation.
Oh.. I think it's actually the first edition of the above book that is more technical in covering TQWT and indeed the Weems Pipe.
The following mentions some sources and these are mainly all about transmission lines, which this design is, so worth reading perhaps:
Quarter_Wave_Resonators
Usmcjlp... I think you cited here:
https://mywebspace.wisc.edu/jstalnak/web/VoightPipeSpeakers/
The following mentions some sources and these are mainly all about transmission lines, which this design is, so worth reading perhaps:
Quarter_Wave_Resonators
Usmcjlp... I think you cited here:
https://mywebspace.wisc.edu/jstalnak/web/VoightPipeSpeakers/
Last edited:
Hey,
Well, my motives are questioned here. That was not the point of my post.
I simply couldnt imagine that a cabinet could do this damage to an driver. Thats why i automaticly thought that it was the driver. Anybody could have made that misstake.
I have asked about wether hes eaning money or not, and hes not any more. He was until 2010. Nobody told me, so i didnt know. There was alot of fuss about it here ind Denmark, thats why i got to know about it.
But, i can see that its not liked to be abit critic about Troels outside Denmark. So, ill stop.
Well, my motives are questioned here. That was not the point of my post.
I simply couldnt imagine that a cabinet could do this damage to an driver. Thats why i automaticly thought that it was the driver. Anybody could have made that misstake.
I have asked about wether hes eaning money or not, and hes not any more. He was until 2010. Nobody told me, so i didnt know. There was alot of fuss about it here ind Denmark, thats why i got to know about it.
But, i can see that its not liked to be abit critic about Troels outside Denmark. So, ill stop.
I think your wording is part of the issue, not any persons love of Troels. You laid claim to something without investigating it and said it was a poor design and you are still using words like 'damage' above when the overall impedance change in the driver is far greater than any small peaks/dips that are not going to be audible and certainly outdone by room response. This happens in lots of designs and was a graph from development without crossover to ameliorate it. These ripples are actually quite common unless a cabinet is perfectly acoustically dead, which very few are.
Martin King over at Quarter Wavelength Loudspeaker Design has done extensive research in to this form of transmission line furthering the science of this type of enclosure. I asked him about them and he said:
"The [cabinet] resonances are not very strong since the peaks are so much smaller then the two bigger peaks lower in frequency. All enclosures will exhibit some internal resonances, if more stuffing were added they would be better damped and not so obvious. They do not look significant to me. Any ripple in the SPL response caused by these small peaks should be very small, probably in the mud. Compared to the peaks and dips caused by the room they should be insignificant."
When you made a point of discussing driver parameters earlier you pointed us at a website for enclosure/cabinet design based on driver parameters. This is probably fine for nominal sealed/vented/reflex and bandpass designs, but I bet it doesn't cover the math for quarter wave transmission lines, Voigt tubes and folded QWT's and therefore plugging in the driver parameters probably gave measurements not in keeping with understanding this design properly.
I also looked at your website from your signature (Speakerbuilder) and the links to Henrik Hoegh's design on the Elysium Passage 2.0 speaker. Clearly there is a similar styled peak at 150Hz on the frequency and impedance graphs. Given these sorts of peaks happen frequently is why this is stated on your website:
"You can make an aditional filter called CRAN that removes the crossover resonans on the impedance profile. Below, you can see the effect of a proper CRAN filter on Elysium Passage 2.0. "
The FILTERED graph is still the one with the peak at 150Hz after it's been treated with CRAN.... guess your friend doesn't make speakers without resonances in then, but uses electronics to smooth them out (exactly what happens in most other designs and Troels). As it says at the bottom "* CRAN = LRC filter, stands for Crossover Resonance Absorption Network, and is used to remove the resonans that crossovers produces on the impedance. "
However, I would have said that a properly engineered crossover removes cabinet resonances, not that an additional LRC filter was required above and beyond the crossover (or to remove crossover resonances since crossovers should not introduce resonances) - indeed an LRC can be part of the overall crossover design.
It seems your wording is the issue here and you have plainly accused Troels design of being poor because of these resonances which your own website shows Henrik Hoegh using the so-called CRAN filter to remove. This is what makes us question your motives. If Troels designs are poor for exhibiting these resonances, equally are your friends for requiring a CRAN filter to deal with them.
Hey,
I think you are mixing things here....
The peak in the impedance at 150Hz is the enclosure, the peak higher is coursed by the crossover as most do. One is electrical, the other acoustical coursed. The CRAN removes the electrical one made by the crossover. The other one is best removed by using more dampening material in the bottom of the enclosure. Alot like i now know Troels could do.
In a normal box (like the one at speakerbuilder) there is one resonance. In troels contruction there are multible.
Its ofcourse a matter of taste, but i dont like resonances. They do damage the timedomain.
The design at speakerbuilder is made by me. Im Henrik. The requered CRAN is optional, and most crossoverbased speakers will have this peak, that a cran can remove. Is not duo to bad design, but how crossovers work.
I think you are mixing things here....
The peak in the impedance at 150Hz is the enclosure, the peak higher is coursed by the crossover as most do. One is electrical, the other acoustical coursed. The CRAN removes the electrical one made by the crossover. The other one is best removed by using more dampening material in the bottom of the enclosure. Alot like i now know Troels could do.
In a normal box (like the one at speakerbuilder) there is one resonance. In troels contruction there are multible.
Its ofcourse a matter of taste, but i dont like resonances. They do damage the timedomain.
The design at speakerbuilder is made by me. Im Henrik. The requered CRAN is optional, and most crossoverbased speakers will have this peak, that a cran can remove. Is not duo to bad design, but how crossovers work.
Actually criticism is a big part of diyAudio. But criticizing both *designers*, rather than the particular designs asked for here, and then suggesting faulty engineering and undisclosed "pay-off"s while providing another commercial design - is in all: bad form. Compounding this is the basis for your criticism - which was wrong (..and not just with respect to Troel's driver either). For instance Zaph's modeled and measured responses of the SB12.3 are very useful and you have a good idea of linearity deviations down to 1 db. The measurements he provides are even at 2.5 meters rather than 1 meter (..and 2.5m is about the distance most would consider a "critical" listening distance for this type of design). The 12.3's lower freq. response *modeled* response is only marginally useful - but that's standard.
As to *disclosure* on payments for the driver (and presumably the waveguide).. regrettable if correct. (Again though, you've not provided verification.) But if it's no longer the case then why bring it up?
Mistakes are to be expected.. but please look into things a bit more before you suggest incompetence on the part of Seas, Troels, or anyone.
Btw, there are several things I don't like about Troel's design.. but overall I think it's both a good design and *unique*, and I think it's better than the SB12.3 and numerous other designs (..including the one you posted).
Henrik,
Well I have no desire to slide this thread downwards in frenzied debate over the speaker design itself. I am trying to understand your position, but it does seem to myself and others to be somewhat Troels design bashing.
FWIW, I believe your speakers and Troel's DTQWT have totally different design goals in mind and that Troels succeeded in achieving a high efficiency design (95db/W) to the Elysium Passage 2.0 (87 db/W???). That was his design goal but not necessarily yours.
I think Troel's design goal, as stated, was to achieve a speaker that will have high enough efficiency to be used with fairly low powered single-ended triodes. In this he succeeded. As you probably know, the higher order a crossover is, the more efficiency we will lose, which is why I suspect Troels has the simple first-order low slope crossover he does. This does however let through those impedance aberrations you dislike so much. If he had used a better crossover, or 18db/octave filter, he would have sacrificed the primary design goal (of maintaining efficiency) and possibly had some other measured aberrations to deal with instead.
The JA8008 is, when measured in free-air, an improvement on the well respected SEAS model on which is is based. It happens to be a rare enough type that in costing this project, I couldn't find a substitute driver with similar efficiency that maintained that efficiency as widely as the JA8008. I looked because the JA8008 is too expensive IMO and I sought an alternative - without success.
However, due to the apparent efficiency considerations, I don't think it fair to compare your design to Troels. If I was looking for a speaker for solid-state I *would* choose yours. If I wanted a speaker usable with low-powered SETs, I would choose Troels.
On the crossover resonance element, I have never heard crossover resonance explained in the manner I infer from you ("thats how crossovers work"). Crossovers attenuate frequencies above/below a set frequency at a particular dB rate, or over a range. In the crossover range, the frequency outputs are 'summed' (constructive frequency interference between drivers reproducing the same frequency). There is always a non-flat result that needs fixing or ignoring (a.k.a. 'summing errors'). We can fix them through additional filters, higher order crossovers, or (my choice) a constant-voltage crossover which will ensure a flat in-phase response. In all cases, these destroy efficiency and move us away from supporting low powered valve-amps.
I suspect (but am not sure) that your CRAN filter is just a notch filter for impedance? Perhaps you could explain it a bit more. Your website has a "The business and me" link under "Speaker projects" so is it just a marketing term for what is nothing more than a notch filter on impedance rather than frequency?
Either way, I don't see it fair to compare your speaker with Troels. You do mention towards the end of the Elysium Passage 2 page that the impedance has been flattened and "your tube amp will love you for it", but I cannot see how if the efficiency is anything like 87db/W with nominal impedance at 3.5 ohms??
I suspect that your software from the link you posted doesn't apply for QWT transmission lines (http://graph.flexunits.com:8080/index.htm). To simulate/evaluate a QWT you would need Martin King's MathCAD models. If you design all your speakers using the software on that link provided, then I would think it would give you erroneous facts if putting in driver parameters from a QWT design in the "speaker parameter" and "vented box calculations" sections.
From my perspective, I understand the following:
1) Troels fundamental design goal is a transmission line with high efficiency for low-powered tube amps including low powered SETs
2) Efficiency is dictated by the drivers selected
3) Complicated or higher-order crossovers will lower efficiency
4) Efficiency can be maintained in a transmission line (TL)
5) If using a TL, line-length is an issue unless we go QWT but then bass depth may become a problem, but we can tune the opening for a particular frequency, plus, shift enough air to get good in-room response without concern over baffle step diffraction or efficiency losses overcoming it.
The trade-offs in this are:
1) A limited crossover will let through some impedance issues that would appear poor on a more conventional box
2) A DTQWT has resonances in the cabinet, which are alleviated with appropriate damping, much of which is predictable (from math), but experimentation and measurement is required to get this as good as possible
3) There may be slow phase variation over the bandwidth of the drivers.
With respect to (acoustic) phase, my understanding is that it is certainly audible, but phase has to shift markedly over a short period to be so. We don't hear absolute phase, we hear differences of phase between two sources and/or rapid phase shifts from a single source.
Would the DTQWT benefit from additional filtering from something like your CRAN? Possibly, but if it kills the overall speaker efficiency then we fail our primary design goal. What is important is the actual acoustic measurements and their acoustic phase response. In this respect, the DTQWTs measure well. The electrical impedance is always 7 or more ohms (up to what looks like 16 ohms). In the DTQWT this is not flat, but for an amp, it is not sinister enough to require additional work if the acoustic response is already up to par. The Ze does look like a notch that a notch filter could remove. Your Passage 2 has a Zmin of 3.5 ohms and at 87dB/W efficiency is a much harder load period.
Because of these different approaches, requiring different considerations and solutions, resulting in different speakers with different cabinet types and efficiencies, I don't think we can realistically compare them. They are built for different end users. Unless your speakers are similarly efficient to be used with low-powered SETs. Troels speakers are well-designed from a QWT / TL high-efficiency perspective. Yours are designed to appear a fairly constant load to an amp, but nowhere near as efficient to be available to such a wide range of tube amp users. You also use more typical sealed/vented enclosures which require different thought paradigms and math to Voigt tubes/ TLs /QWTs in their design. It is this, which frankly makes discussion over whether an impedance variation should or should not be there, a silly discussion as it is dependant on understanding the enclosure, the maths, the required trade-offs and the design goal.
As I said, for a low-powered tube design I would opt for Troels design. It's very hard to build a 95db/W speaker with good bass depth and to be an easy load for nearly every amp. However, for most solid-state amps, your design would certainly seem better, but I suspect your design would struggle compared to Troels if being driven by low to medium power tube amps (up to 25W) which like to see 4 ohms or more. That's not to say that at minimal volume with these amps yours would not sound as good, because they almost definitely will.
In conclusion, your speakers are good-to-excellent from a typical consumer standpoint (solid-state amps, high powered entry level tubes). Troels DTQWT are not aimed at this user, but the low-powered tube user with more likely a self-built or expensive low-power PP or SET. In this arena, Troels DTQWT comes in to it's own where few others can compete. Like drivers, there is overlap of types of end users - where both meet, it would be a 'try it and see' affair. As ScottG said, there are issues with Troels design because it is not perfect, but what speaker, even driver, is?
(No sarcastic replies from DIY Audio members saying Manger drivers are perfect!)
Well I have no desire to slide this thread downwards in frenzied debate over the speaker design itself. I am trying to understand your position, but it does seem to myself and others to be somewhat Troels design bashing.
FWIW, I believe your speakers and Troel's DTQWT have totally different design goals in mind and that Troels succeeded in achieving a high efficiency design (95db/W) to the Elysium Passage 2.0 (87 db/W???). That was his design goal but not necessarily yours.
I think Troel's design goal, as stated, was to achieve a speaker that will have high enough efficiency to be used with fairly low powered single-ended triodes. In this he succeeded. As you probably know, the higher order a crossover is, the more efficiency we will lose, which is why I suspect Troels has the simple first-order low slope crossover he does. This does however let through those impedance aberrations you dislike so much. If he had used a better crossover, or 18db/octave filter, he would have sacrificed the primary design goal (of maintaining efficiency) and possibly had some other measured aberrations to deal with instead.
The JA8008 is, when measured in free-air, an improvement on the well respected SEAS model on which is is based. It happens to be a rare enough type that in costing this project, I couldn't find a substitute driver with similar efficiency that maintained that efficiency as widely as the JA8008. I looked because the JA8008 is too expensive IMO and I sought an alternative - without success.
However, due to the apparent efficiency considerations, I don't think it fair to compare your design to Troels. If I was looking for a speaker for solid-state I *would* choose yours. If I wanted a speaker usable with low-powered SETs, I would choose Troels.
On the crossover resonance element, I have never heard crossover resonance explained in the manner I infer from you ("thats how crossovers work"). Crossovers attenuate frequencies above/below a set frequency at a particular dB rate, or over a range. In the crossover range, the frequency outputs are 'summed' (constructive frequency interference between drivers reproducing the same frequency). There is always a non-flat result that needs fixing or ignoring (a.k.a. 'summing errors'). We can fix them through additional filters, higher order crossovers, or (my choice) a constant-voltage crossover which will ensure a flat in-phase response. In all cases, these destroy efficiency and move us away from supporting low powered valve-amps.
I suspect (but am not sure) that your CRAN filter is just a notch filter for impedance? Perhaps you could explain it a bit more. Your website has a "The business and me" link under "Speaker projects" so is it just a marketing term for what is nothing more than a notch filter on impedance rather than frequency?
Either way, I don't see it fair to compare your speaker with Troels. You do mention towards the end of the Elysium Passage 2 page that the impedance has been flattened and "your tube amp will love you for it", but I cannot see how if the efficiency is anything like 87db/W with nominal impedance at 3.5 ohms??
I suspect that your software from the link you posted doesn't apply for QWT transmission lines (http://graph.flexunits.com:8080/index.htm). To simulate/evaluate a QWT you would need Martin King's MathCAD models. If you design all your speakers using the software on that link provided, then I would think it would give you erroneous facts if putting in driver parameters from a QWT design in the "speaker parameter" and "vented box calculations" sections.
From my perspective, I understand the following:
1) Troels fundamental design goal is a transmission line with high efficiency for low-powered tube amps including low powered SETs
2) Efficiency is dictated by the drivers selected
3) Complicated or higher-order crossovers will lower efficiency
4) Efficiency can be maintained in a transmission line (TL)
5) If using a TL, line-length is an issue unless we go QWT but then bass depth may become a problem, but we can tune the opening for a particular frequency, plus, shift enough air to get good in-room response without concern over baffle step diffraction or efficiency losses overcoming it.
The trade-offs in this are:
1) A limited crossover will let through some impedance issues that would appear poor on a more conventional box
2) A DTQWT has resonances in the cabinet, which are alleviated with appropriate damping, much of which is predictable (from math), but experimentation and measurement is required to get this as good as possible
3) There may be slow phase variation over the bandwidth of the drivers.
With respect to (acoustic) phase, my understanding is that it is certainly audible, but phase has to shift markedly over a short period to be so. We don't hear absolute phase, we hear differences of phase between two sources and/or rapid phase shifts from a single source.
Would the DTQWT benefit from additional filtering from something like your CRAN? Possibly, but if it kills the overall speaker efficiency then we fail our primary design goal. What is important is the actual acoustic measurements and their acoustic phase response. In this respect, the DTQWTs measure well. The electrical impedance is always 7 or more ohms (up to what looks like 16 ohms). In the DTQWT this is not flat, but for an amp, it is not sinister enough to require additional work if the acoustic response is already up to par. The Ze does look like a notch that a notch filter could remove. Your Passage 2 has a Zmin of 3.5 ohms and at 87dB/W efficiency is a much harder load period.
Because of these different approaches, requiring different considerations and solutions, resulting in different speakers with different cabinet types and efficiencies, I don't think we can realistically compare them. They are built for different end users. Unless your speakers are similarly efficient to be used with low-powered SETs. Troels speakers are well-designed from a QWT / TL high-efficiency perspective. Yours are designed to appear a fairly constant load to an amp, but nowhere near as efficient to be available to such a wide range of tube amp users. You also use more typical sealed/vented enclosures which require different thought paradigms and math to Voigt tubes/ TLs /QWTs in their design. It is this, which frankly makes discussion over whether an impedance variation should or should not be there, a silly discussion as it is dependant on understanding the enclosure, the maths, the required trade-offs and the design goal.
As I said, for a low-powered tube design I would opt for Troels design. It's very hard to build a 95db/W speaker with good bass depth and to be an easy load for nearly every amp. However, for most solid-state amps, your design would certainly seem better, but I suspect your design would struggle compared to Troels if being driven by low to medium power tube amps (up to 25W) which like to see 4 ohms or more. That's not to say that at minimal volume with these amps yours would not sound as good, because they almost definitely will.
In conclusion, your speakers are good-to-excellent from a typical consumer standpoint (solid-state amps, high powered entry level tubes). Troels DTQWT are not aimed at this user, but the low-powered tube user with more likely a self-built or expensive low-power PP or SET. In this arena, Troels DTQWT comes in to it's own where few others can compete. Like drivers, there is overlap of types of end users - where both meet, it would be a 'try it and see' affair. As ScottG said, there are issues with Troels design because it is not perfect, but what speaker, even driver, is?
(No sarcastic replies from DIY Audio members saying Manger drivers are perfect!)
Last edited:
Zaph Audio SB12.3 or Troels' DTQWT?
I'm wondering if this has become the "elephant in the room"
It was stated that madisound had suggested Zaphs ZRT towers instead of the SB12.3s
usmcjlp, you stated that you'll be using these for both TV sound and Hi-Fi. No one has questioned your av amplifiers power output, or whether you plan on any plans to make changes.
I was wondering if madisound suggested the ZTRs 2.0 or the 2.5s?
Would you make you're own cabinets or purchase them?
I believe that Zaph has a sealed and a ported model, which would you consider?
If you're considering Troels, are all the drivers available here in the USA?
Do you know what the parts prices would be for the three/four speakers currently being considered?
Just to add to the pile, have you looked at the Elsinore's?
Any idea which way you are leaning?
m12ax7
I'm wondering if this has become the "elephant in the room"
It was stated that madisound had suggested Zaphs ZRT towers instead of the SB12.3s
usmcjlp, you stated that you'll be using these for both TV sound and Hi-Fi. No one has questioned your av amplifiers power output, or whether you plan on any plans to make changes.
I was wondering if madisound suggested the ZTRs 2.0 or the 2.5s?
Would you make you're own cabinets or purchase them?
I believe that Zaph has a sealed and a ported model, which would you consider?
If you're considering Troels, are all the drivers available here in the USA?
Do you know what the parts prices would be for the three/four speakers currently being considered?
Just to add to the pile, have you looked at the Elsinore's?
Any idea which way you are leaning?
m12ax7
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.