Hi Guys,
I’m glad the debate on T/S accuracy and testing is progressing. There’s been a few recent references in other threads. Maybe this thread will become more substantial.
I understand Martin’s position of not trusting published driver information provided by most makers. This is a sad situation and an indictment on some elements of driver manufacturing. At the risk of me being further disliked by some manufacturers (and their supporters), published data should be close to actual driver performance. I am aware of instances where test data has been changed in an effort to make drivers more marketable. Such practices are misleading and shouldn’t be acceptable.
Another challenge is the practice of issuing data without quoting the PUV (Production Unit Variance). The figure is typically +/-20% on all major T/S measurements from the mean or “golden sample”. The common practice is to make a series of test units to a pre-production state, the final sample being the “gold” benchmark unit. A proportion of production drivers are then checked to a specified +/- standard either side of the golden sample. Rejection rates vary but typically is 2% on production runs where +/- 20% PUV is applied. Reject rates increase significantly as the PUV is lowered.
I have given this information in earlier threads but it may still surprise some members that the figures are so wide. A wide PUV allows makers to reduce costs by using simple assembly methods, lower grade components, while acceptable for most commercial and OEM applications, it's not so good for audiophile grade full range driver production. From my engineering standpoint, these tolerances were far too broad. This is one of the major factors as to why I elected to design and build all our own tooling and critical components as part of the effort to reduce production variations to less than 10% PUV.
I have demonstrated that through applied design and engineering practices, it is possible to make drivers perform sensibly close to their published data when independently tested. I’d like to see a time when users can trust all makers published data, knowing it is close to actual performance.
I understand Mightyd from Scotland. His situation is I suspect common for many new DIYers. We should remember that many DIY'ers may not have the time, skills and/or resources to invest in test equipment of adequate quality. I am worried that DIY full range speaker building runs a big risk of becoming a Cinderella hobby. More needs to be done to make building easier for new members and to attract of new generation of DIYers. I urge the senior and long-standing members of this forum to think more on this situation.
It’s not all plain sailing. Both the manufacturer's and independent test hardware and software also have tolerances. Test methods and environmental conditions also vary. It’s not possible to have 100% accuracy and reliability either from the maker’s published data, or independent tests.
All said, it should be possible to gain more convergence and agreement on driver testing methodologies to improve the quality of published data in relation to actual performance. Perhaps the moderators would consider renaming this thread (hoping Christian will agree) to encourage more member to contribute.
I'm flying to Frankfurt this Tuesday where I'm expected to deal with allot of technical work. Hopefully, the debate will progress in this part of the world where full range DIY also has a large fan base.
Cheers,
Mark.
I’m glad the debate on T/S accuracy and testing is progressing. There’s been a few recent references in other threads. Maybe this thread will become more substantial.
I understand Martin’s position of not trusting published driver information provided by most makers. This is a sad situation and an indictment on some elements of driver manufacturing. At the risk of me being further disliked by some manufacturers (and their supporters), published data should be close to actual driver performance. I am aware of instances where test data has been changed in an effort to make drivers more marketable. Such practices are misleading and shouldn’t be acceptable.
Another challenge is the practice of issuing data without quoting the PUV (Production Unit Variance). The figure is typically +/-20% on all major T/S measurements from the mean or “golden sample”. The common practice is to make a series of test units to a pre-production state, the final sample being the “gold” benchmark unit. A proportion of production drivers are then checked to a specified +/- standard either side of the golden sample. Rejection rates vary but typically is 2% on production runs where +/- 20% PUV is applied. Reject rates increase significantly as the PUV is lowered.
I have given this information in earlier threads but it may still surprise some members that the figures are so wide. A wide PUV allows makers to reduce costs by using simple assembly methods, lower grade components, while acceptable for most commercial and OEM applications, it's not so good for audiophile grade full range driver production. From my engineering standpoint, these tolerances were far too broad. This is one of the major factors as to why I elected to design and build all our own tooling and critical components as part of the effort to reduce production variations to less than 10% PUV.
I have demonstrated that through applied design and engineering practices, it is possible to make drivers perform sensibly close to their published data when independently tested. I’d like to see a time when users can trust all makers published data, knowing it is close to actual performance.
I understand Mightyd from Scotland. His situation is I suspect common for many new DIYers. We should remember that many DIY'ers may not have the time, skills and/or resources to invest in test equipment of adequate quality. I am worried that DIY full range speaker building runs a big risk of becoming a Cinderella hobby. More needs to be done to make building easier for new members and to attract of new generation of DIYers. I urge the senior and long-standing members of this forum to think more on this situation.
It’s not all plain sailing. Both the manufacturer's and independent test hardware and software also have tolerances. Test methods and environmental conditions also vary. It’s not possible to have 100% accuracy and reliability either from the maker’s published data, or independent tests.
All said, it should be possible to gain more convergence and agreement on driver testing methodologies to improve the quality of published data in relation to actual performance. Perhaps the moderators would consider renaming this thread (hoping Christian will agree) to encourage more member to contribute.
I'm flying to Frankfurt this Tuesday where I'm expected to deal with allot of technical work. Hopefully, the debate will progress in this part of the world where full range DIY also has a large fan base.
Cheers,
Mark.
Hi Mark (Martin, Dave and everybody...),
It's a pleasure to see you here, this thread have took a very interesting and instructive way!
Really no problem for me, I think the issue you are discussing is far more important than my original and individual request.
Please go on...!
Thanks and bye,
Christian
It's a pleasure to see you here, this thread have took a very interesting and instructive way!
Really no problem for me, I think the issue you are discussing is far more important than my original and individual request.
Please go on...!
Thanks and bye,
Christian
Ted wrote an article on variable measured parameters some years ago. I don't know if it is relevant to the present discussion but you can find the PDF here.
And looking for it, I came across these rather exotic curved wood enclosures from a UK company.
FWIW If going for the Jordans (I haven't heard the Alpairs), I've built both the VTL and the 48" MLTL and the MLTL goes lower and has more weight. However it needs BSC to balance the sound so you end up with a lower sensitivity system.
I'm in the middle of building the 31" MLTL to see how that compares.
And looking for it, I came across these rather exotic curved wood enclosures from a UK company.
FWIW If going for the Jordans (I haven't heard the Alpairs), I've built both the VTL and the 48" MLTL and the MLTL goes lower and has more weight. However it needs BSC to balance the sound so you end up with a lower sensitivity system.
I'm in the middle of building the 31" MLTL to see how that compares.
Colin said:
The wood waste is a bother here to, and that i have questions about whether it is the best "direction" to use the wood.
Translam can produce some stellar looking shapes, IMHO this isn't one of them.
If this is a TL, looks like a guess & by golly design... no attempt to take advantage of off-set.
dave
Hi Colin (and everybody...),
Thanks for your reply. What I'm looking for is an enclosure design that "sounds" with no compromise in the mid frequencies and just clear and not lacking in the highs. Of course, as much bass as possible but... not at discapit of the maximum mid quality that the speaker can delivery. Jordan or Mark Audio with no preferences or foreclosures, I think to have understood right that both are just great. About the design, is it possible, in some way, to realize an MLTL with the aspect of the VTL (Konus Essence alike)? I found a previous post here on diyaudio inherent to this but I don't understand so well what to do. The Mathcad worksheet by Martin is not yet available and, most important, I'm a total newbie in speaker design (I think, and hope, I'm just able to assemble a project already designed), so any help (please with patience and comprehension) are really welcome, in the meantime I'm studying a lot but I know I still have a long way to go...
Thanks a lot and bye,
Christian
Thanks for your reply. What I'm looking for is an enclosure design that "sounds" with no compromise in the mid frequencies and just clear and not lacking in the highs. Of course, as much bass as possible but... not at discapit of the maximum mid quality that the speaker can delivery. Jordan or Mark Audio with no preferences or foreclosures, I think to have understood right that both are just great. About the design, is it possible, in some way, to realize an MLTL with the aspect of the VTL (Konus Essence alike)? I found a previous post here on diyaudio inherent to this but I don't understand so well what to do. The Mathcad worksheet by Martin is not yet available and, most important, I'm a total newbie in speaker design (I think, and hope, I'm just able to assemble a project already designed), so any help (please with patience and comprehension) are really welcome, in the meantime I'm studying a lot but I know I still have a long way to go...
Thanks a lot and bye,
Christian
Hi everybody,
every single TL design on diyaudio (and on the entire web) is done with the Martin J. King Mathcad worksheets. Since this is not yet available... please, is there someone who can kindly indicate me (if exists) an alternative way to try to project my own speaker enclosure?
Thanks and bye,
Christian
every single TL design on diyaudio (and on the entire web) is done with the Martin J. King Mathcad worksheets. Since this is not yet available... please, is there someone who can kindly indicate me (if exists) an alternative way to try to project my own speaker enclosure?
Thanks and bye,
Christian
Hi ilwooz, why not use MJK's TL alignment tables? No software required: http://www.quarter-wave.com/TLs/TL_Theory.html
Hi rjbond3rd, Henkjan and everybody,
thank you very much for yours replies. I will start to study and then I'll try to "project" something following yours suggestions. From a first view, both methods seem a bit complicated (at least for a newbie like me) but... I have no alternatives. My actual goal is to modify, I suppose a little, the "classic" VTL design for the Jordan JX92S (published on the Jordan website) to fit the corrects parameters for use the enclosure with the Alpair 10. I hope this may have sense...
Thanks again and bye,
Christian
thank you very much for yours replies. I will start to study and then I'll try to "project" something following yours suggestions. From a first view, both methods seem a bit complicated (at least for a newbie like me) but... I have no alternatives. My actual goal is to modify, I suppose a little, the "classic" VTL design for the Jordan JX92S (published on the Jordan website) to fit the corrects parameters for use the enclosure with the Alpair 10. I hope this may have sense...
Thanks again and bye,
Christian
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.