Well this is way way way off topic, but SVD used by myself isn't about prediction or trend finding, it is about building robust matrices to study correlations. Doing SVD on a correlation matrix, setting singular values below a certain threshold to zero and reprojecting back into the original space ensures an invertable matrix. Its very much related to common computations of generalized inverses.
There are a lot of really cool non-trivial math problems in finance but there are only a few that are of economic significance.
when you say 3" driver, do you mean 3in full range driver? I don't think any of 3in full rangers can play up to 20k. too much energy storage at higher frequencies in a such large cone.
and those full rangers have puny power handling.
has anyone tried 3in dome midrange in a large waveguide.
and those full rangers have puny power handling.
has anyone tried 3in dome midrange in a large waveguide.
What I did was to reason that in a duct that is curving away from the throat, a tangent to the wall that crosses the axis is the "r" in the kr number.
IF you look at a conical horn this varies linearly with the axis and the tangent crosses at the frustum.
A circular arc that starts at an angle of fourty five degrees also has a linearly reducing curve of this sort although the poit where it intersects the axis is variable.
I reasoned that this is analogious to a conical horn in which the part after a certain kr value is reached is removed, and we might say that the directivity is that of a piston that is continually reducing in diameter as frequency increases, and if this reduction in notional diameter if linear then the device would have constant directivity.
rcw.
IF you look at a conical horn this varies linearly with the axis and the tangent crosses at the frustum.
A circular arc that starts at an angle of fourty five degrees also has a linearly reducing curve of this sort although the poit where it intersects the axis is variable.
I reasoned that this is analogious to a conical horn in which the part after a certain kr value is reached is removed, and we might say that the directivity is that of a piston that is continually reducing in diameter as frequency increases, and if this reduction in notional diameter if linear then the device would have constant directivity.
rcw.
MisterTwister - I'm using Balanced Mode Radiator that is different to the ordinary fullrange driver. When I repair my measurement mic I will post 1" 400Hz JMLC horn with this new driver - it's older generation designed for parametric eq. and wide bandwidth.
Earl - tooling cost is my problem not Yours. At least 10 people interested makes the price attractive. Here you have said Chinese drivers can be as good as let say TAD4001 in a blind test, on your website the opposite. Have you decided?
Earl - tooling cost is my problem not Yours. At least 10 people interested makes the price attractive. Here you have said Chinese drivers can be as good as let say TAD4001 in a blind test, on your website the opposite. Have you decided?
I have decided that drivers are a commodity and that anyone can make good ones, it's all a matter of design. The Chinese are perfectly capable of making great drivers, but not of designing one. They copy well, but not really knowing what they are doing they don't copy well enough. Take the B&C copy of the DE250 that I had about a dozen of back in Thailand. This was a very good driver, but not quite as good as the B&C. It was $17 to B&Cs $80 or more. I knew exactly what made the Chinese driver not as good and exactly what they would need to do to make it as good. They were willing to make these changes for free. I declined to do this since I did not want to get into the business of upgrading a Chinese loudspeaker companies technical capabilities on my time.
From following this and other related threads, its very clear that a throat like this need some careful work
Surely also in line with the latest posts in the matter of crutial flaws in a copy process
Funny thing
If I remove the screwon thread its almost exectly 36mm, which should fit a 1.4" exit driver
Only, those are awfully expencive
Sound like the problems are worse at lower frequencies
Is that a tendency to expect
That problems expands with lower xo point
Surely also in line with the latest posts in the matter of crutial flaws in a copy process
Funny thing
If I remove the screwon thread its almost exectly 36mm, which should fit a 1.4" exit driver
Only, those are awfully expencive
Sound like the problems are worse at lower frequencies
Is that a tendency to expect
That problems expands with lower xo point
Attachments
Last edited:
I wouldn't say that the problems are worse at the lower frequencies, in fact I'd say its the exact opposite. When the "errors" are small compared to a wavelength they are benign, but when they get to be comprable to a wavelength they can be problematic.
Well what if they used your fixes AND added your phase plug for less than B&C.
I can't speak for Dr. Geddes, but it's not uncommon for these manufacturers to consider these sorts of changes in their ownership, and able to be used throughout their product line, with Dr. Geddes losing control of the innovations, all with no pay for his time, work, and research. Even with patents, the international situation seems much less clear as to what is and isn't allowed. On top of that, if they do violate a patent he has, what could he do, the lawsuit would cost a fortune. It seems like many a good man has lost his company fighting patent infringements, even when they are right, due to a lack of funds to take on international parts suppliers.
You Call SVD Math and an engineer is impressed, I talk about Principle Component Analysis as Math and get laughed at by the engineer's and non-social scientists. I'll be showing a paper at an upcoming conference (American Educational Research Association)describing the advantage of applying Principle Component analysis to all theoretical factor sets, as apposed to simply using other peoples developed factors. As I understand it, SVD is just another term for PCA used by other fields, but usually specifically referring to the use of the covariance matrix, i.e. a Least Squares method, instead of an iterative method for convergence.
The point is that I'm not about to help them improve their product unless I can take advantage of it. I can't at the moment, my volumes are just too small to consider sourcing overseas.
This is the real issue. I need to be sure that I can somehow gain some advantage for my work in improving theior products.
You are correct, they are basically equivalent techniques. The significant eigenvalues of the matrix are the "principle components", the ones clustered near zero are noise. If they are near zero then the data has no trends or correlations.
I typed a whole response to this explaining what I was doing with it and then realized it was so far off topic it wouldn't be fair to others to submit it.
We use such methods to develop measures of phenomena we observe, and I would love to see such methods applied to audio equipment reviewing. It will likely never happen, the rigor necessary would not go over well with reviewers, readers would never fully understand it unless they were in the field, and there seems to be preferred methods for this in research designs looking at very specific audible differences.
We use such methods to develop measures of phenomena we observe, and I would love to see such methods applied to audio equipment reviewing. It will likely never happen, the rigor necessary would not go over well with reviewers, readers would never fully understand it unless they were in the field, and there seems to be preferred methods for this in research designs looking at very specific audible differences.
Well, maybe you would, in getting better drivers
I suppose thats how OEM works
I dont think OEM product can be exclusive forever
Eventuall they get available to all customers
And occationally, if theres a good "tricks" used in a OEM product, it will end up in other products as well, later on
I suppose thats the name of the game, if ordering modified drivers
Most companies who develop proprietary technology, and patent it, like to keep it proprietary for as long as possible. I think most companies want to keep it proprietary regardless of patent or not.
If I were in Dr. Geddes shoes, my biggest concern would be one of two things. First that, he does it this way, they introduce the technique, and the drivers he gets are as he wants, so he lets people know. Then they introduce these changes on all their drivers, people find out the changes came from Dr. Geddes, but without someone like dr. Geddes doing QC before they get put in the hands of customers, bad drivers could get out, with his name inadvertently attached to them. His name gets dragged through the mud over a manufacturing situation he had nothing to do with (This happened with a certain woofer designer an Taiwanese manufacturer this person worked with).
The other would be if, again, I was in his shoes, I actually attached my name to the new and improved driver. Maybe for economies of scale I introduce them to the public. They incorrectly introduce these changes in other products, implement changes that ruin the driver without telling me, and again, my name is attached to a now faulty product.
I worked with a modification company working with chinese built tube amps, revamping them and selling them here in the US. They changed the designs on us constantly without telling us, and eventually the only thing being kept was the chassis. I've heard many times this is more the norm than an exception.
This is absolutely the norm.
The bottom line here is that until such time as I can have driver built specifically for me, all of this discussion is moot. The Chinese require a minimum order of 1000 - 10,000 pieces (depends on company, product, price, etc.) and you have to deposit this amount in a bank for holding prior to their moving forward. That is a huge amount of cash to put down. It takes a pretty big volume to be able to do that. As much as I would love to do that, its still quite a ways off.
The whole reason I write software for a living is that I just don't think there's much of a market for excellent loudspeakers. It's fun to geek out on this stuff, it is an interesting intellectual challenge. But I don't have any illusions about any of my ideas being marketable.
This afternoon I was talking to a friend about the crazy stereo in my car, and doing my best to describe why improving intelligibility and power response is something that everyone can appreciate. I mean, the things that you can do with waveguides can be appreciated by anyone, even those with tin ears. So I'm trying to explain what I'm doing, in English.
Her response:
"My friend has a really nice stereo in her car... It's a Bose."
Marketing wins, every f'ing time...
yeah, thankfully my roll was as a solder slinger. No money invested, paid hourly, no real intellectual involvement in the designs. The guy that was doing this basically went out of business as a result of it, he didn't expect the amps to be different, and had planned around the amps coming with certain parts, needing certain parts replaced, and certain parts removed and/or redesigned. His profit margins were small, and he couldn't eat the cost of the mistakes. As I recall, once he was done, around the time I left, he was selling amps we had paid a few hundred dollars for, for almost ten thousand.
I actually have one of these, a bit of a one of a kind too since I bought it, took it to him, he hired me to slind solder in exchange for teaching me how to redesign it (I was paid too). However, it mostly sits in a box, tubes aren't real practical in a theater. 805 transmitter tubes run in Class A tend to heat a room up quick, so maybe I will use it more in the dead of winter.
My own tube amp was a stereo amp built as two monoblocks in one chassis. Each channel had a slightly different circuit design to it, and parts that should have been identical values, typically not only weren't, but weren't even remotely within an acceptable tolerance. It was completely unable to be properly biased as is, and was basically a 120lb door stop.
I actually have one of these, a bit of a one of a kind too since I bought it, took it to him, he hired me to slind solder in exchange for teaching me how to redesign it (I was paid too). However, it mostly sits in a box, tubes aren't real practical in a theater. 805 transmitter tubes run in Class A tend to heat a room up quick, so maybe I will use it more in the dead of winter.
My own tube amp was a stereo amp built as two monoblocks in one chassis. Each channel had a slightly different circuit design to it, and parts that should have been identical values, typically not only weren't, but weren't even remotely within an acceptable tolerance. It was completely unable to be properly biased as is, and was basically a 120lb door stop.
I agree with Gedlee, the Chinese driver is moot. If he wanted to use a less expensive driver in his kits, I am guessing Gedlee could use a driver at half the cost of the B&C250, (maybe Selenium or similar) and modify his crossover to get the same frequency curves and polar responses. He does a good job with his measurements and crossover design.
Let's not underestimate the quality of the B&C, they are a first rate driver, up there with JBL and TAD, etc. - all the best brands. Selenium and Eminence and most of the others are not, I've tried them. Basically I don't see how I can beat the deal I get on B&C for the quality unless I designed the driver myself. Then I could improve the driver AND lower the cost - that would be good thing!! But the investment is very high.
I recently landed a long term consulting job in China (90% confident) and being there just might allow me to do some things that I couldn't do from here. It's certainly in the back of my mind that's for sure.
For the record - on advice from a marketing consultant - I am using the name "Geddes" for my speakers now. GedLee has been confusing for people and the name doesn't connect with all my work in the field. GedLee will remain the consulting business. So its "Geddes Loudspeakers", and Geddes-audio.com, the "Geddes Abbey", etc. I was chastised for using my kids names on the product, but now that's pretty well stuck and there's no undoing it this late in the game. It was actually done as a joke!