markmck's Tang Band mods

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
LEGAL NOTICES

This modification design is my intellectual property. I am offering it to the members and guests of www.diyaudio.com for their personal use only. Any commercial use of this design without prior written permission is an infringement of my property rights.

There are no explicit or implied warranties or guarantees about the suitability of this modification design for any application or use. The modifications cannot be undone. The modification will void the manufacturers warranty. Proceed at your own risk.

Modifying the W4-654S

As I have stated previously, I like Tang Band. They make small diameter drivers that have the possibility of nearly full-range operation. Indeed, a couple of their products have promise. Here is the promised modification to the Tang Band W4-654S driver.

The goal of this modification is to produce a driver with near high fidelity levels of performance. At a minimum, this includes a frequency response that does not exceed plus or minus 3-db. Any resonant structures or performance anomalies that prevent achieving that plus or minus 3-db rating are considered problems.

In stock condition, the Tang Band W4-654S driver has problems. As those problems are complex, the modification is complex. The modification is made more complex because the cone material is paper. The chaotic vibrations of this cone in resonance cannot be controlled by conventional means of applying materials of differing characteristics across the surface. Instead, I have taken an approach where you isolate important cone surface regions for the chaotic vibration behavior in resonance.

Eight narrow slits are made in the cone, forming four Vs. The placement of the slits is shown in the cutting guide. The slits are then sealed with a synthetic rubber compound applied with a toothpick. Very little of the rubber compound is used. Its purpose is to seal the cone and allow it to act as a coherent surface while preventing the origin conditions of the resonant structures. Only the specified rubber compound precisely mixed to manufacturer recommendations will achieve the results shown.

After modification, when mounted in an infinite baffle enclosure, the driver performance is from Fs to 15 kHz, plus or minus three db. Over the critical region the response varies by only plus or minus 2.125 db. The driver now comes very close to high fidelity performance criteria. The design fails to reach 20 kHz by only one-half octave. The roll-off in the high frequency stop band is smooth and well controlled. If you try to crossover to a “super” tweeter, no additional filtering is required for the W4-654S.

There are three complications to adding a super tweeter. The high pass filter on the super tweeter will have to be a high order design to match the W4-654S’ roll-off. The second difficulty is finding a tweeter with significantly better high frequency performance than that of the W4-654S. I have a number of tweeter evaluations posted on my Web site (http://madspeaker.com). When examining onset response, they are rolling off in the same region as the W4-654S. Lastly, because they will not perfectly sum transiently, adding a tweeter will reduce the coherence of the W4-654S.

Over the next couple of weeks I will post additional modification details to my Web site. As other Tang Band modifications are completed, I will post them to this thread.

The attachment to this message is the stock performance. The second posting will contain the modified performance and modification guides.

Good luck and have fun,

Mark
 

Attachments

  • picture1.gif
    picture1.gif
    21.6 KB · Views: 5,428
As promised (and to hopefully save this forum's bandwidth) I have added the modification page to my Web site. I have included a photo of a modified driver to show that extreme precision is not needed in applying the synthetic rubber compound. I have added a source address for the rubber compound. I have also added a before and after comparison graph of frequency and decay response.

You can use this address (http://madspeaker.com/Projects/MADMODfour.htm) to reach the modification page while avoiding all of the commerical aspects of the Web site.

Now, I really do not want to get into baffle step correction. Such a discussion may be of limited value. I am not trying to challenge anyone's belief system or anything else. Let me just say that within certain testing regimes Olsen's work cannot be replicated (if you are aware of how science is supposed to work, this is important). There may be reasons that people are reporting preferences for loudspeakers utilizing what is called baffle step correction that have nothing to do the baffle stepping. If you examine the acoustic driver tests posted to this forum (including my own), you may notice a tendency for drivers to show a rise in output with increasing frequency until major cone resonance. You may be creating an improved loudspeaker by the use of a stepped pre-filter, but it may not be because of baffle step compensation. A loudspeaker with a rising frequency response will sound lacking in bass not because of improper box alignment, but because the bass region level is just down in reference to the higher frequency levels.

This is just an alternative way of looking at loudspeaker reproduction performance. If it will help anyone, I can post narrow and wide baffle acoustic performance comparisons of the modified W4-654S to this thread. If you prefer the sound of loudspeakers with baffle step correction, then please go ahead and use it. Whether you are correcting for baffle step or not, you are probably creating a loudspeaker with a better balance of low to high frequencies. It may also be that loudspeakers already "flat" in frequency response will not benefit as much from baffle step correction.

Mark
 
baffle comparison

I have attached the large (nearly infinite) baffle to very small baffle frequency response comparison graph. The small baffle box measures 5.875 inches wide by 9.875 inches high by 6.625 inches deep. All box walls are 0.5 inches thick. Both measurements were taken from one meter. Drive voltage and microphone gain were held constant. The small baffle box was tightly packed with high-density foam. The small baffle box was elevated on a stand in the center of the room. The data window was truncated to eliminate boundary reflections from walls, floor, and ceiling. The driver had stopped voicing before the end of the 6 ms window.

The small box frequency response of the same W4-654S modified driver differs from its response in the infinite baffle. From 800 Hz to 3 kHz the small box response is up. From 4 kHz to 8 kHz it is down. From 8 kHz to 12 kHz it is up, and from 13 kHz on it is down in comparison to the nearly infinite baffle response. Fs is also higher in frequency in the small box (about 90 Hz).

The small boxes pictured here were artifacts from my high school days. I bought them from Radio Shack. They came with four-inch drivers that were not extended range. I was still using them for producing background noise to mask my computer’s fan noise. After installing the modified W4’s I now have a pair of unusually high fidelity background noise sources. If I were going to use them for serious listening in this small of a box, I would increase the Q on the specified pre-filter and add a second notch filter at 12 kHz.

It is difficult to determine cause for the measurement differences. My hypothesis of cause is diaphragm loading by the internal small box air mass and internal reflections.

Mark
 

Attachments

  • 11w4-654s.gif
    11w4-654s.gif
    17.6 KB · Views: 4,659
In answer to Dave's question:

No, I am much too old to have bought any of the high tech RS enclosures. Just thin wood with a chrome edge decoration around the recessed (held in with screws) baffle.

Just as an aside, I have been playing with the drivers in the tiny boxes for serious listening and have worked out the necessary compensation prefilter values. I am, however, wondering why I am doing it. Fs is 160 to 170 Hz. With the compensation pre-filter, I can get F3 to about 93 Hz but the reason for going with the larger 3 inch cone driver (advertised as 4 inch) with the greater Xmax is to get lower in frequency without bass augmentation. With F3 closer to 100 Hz, it seems you might as well choose the 2 inch cone (advertised as 3 inch).

For a question of my own:

Has there ever been a thread about what reference is used when talking about F3? The world of experience is messier than simplistic algorithmic simulations. Drivers are never flat. Depending on what reference point you choose, actual measured F3 can vary significantly. Of course all other T/S points vary too, so what is the point of using the simulators? Suppose a simulation program specified a F3 of 65 Hz (not for the W4-654S). Would you build it? What if the set-up specified by the program actually produced a system with an F3 of 42 Hz and with no ripple? Would you be more or less likely to build it?

Mark
 
Hi 454,

Can the same modification be done to this woofer with similar results?

From the information Tang Band supplies, I cannot see why anyone would want to modify this cone. Do you have data to indicate that Tang Band's published response graph does not accurately reflect the performance of the driver?

Also, the cone of this driver is polypropylene and not paper. The modification I describe in this thread is for a paper cone. Also, the cone shape may be different. I have a sample of a Tang Band polypropylene cone for a 6.5 inch driver and it does not have the same shape as the smaller paper and polypropylene cones.

Considering these criteria, even if this driver has some undocumented flaws that should be compensated for, it would require a different set of modifications. I am curious about what this driver does that would make it a worthwhile candidate for modification. Does anyone know why this driver should be redesigned? I don't own any samples of this driver, so I just cannot comment.

Mark
 
W4-654S in a small box: A design exercise.

The small box pictured above, measuring 15 by 17 by 25 centimeters with an internal volume of .15 cubic feet, is not the optimal enclosure. If, however, you have a need for a loudspeaker in a very small box, this is a possible solution with fairly good fidelity. Compared to a larger enclosure, the required pre-filter is more complex and more expensive. This small loudspeaker is most lacking in bass extension. Fs is about 160 Hz. Yet, because of rear loading, without the pre-filter, the output is still rising even below Fs. With the pre-filter, F3 is below 100 Hz. There is no ripple or other anomalies in the response. You cannot, however, beat physics. Distortion levels below Fs are high.

The attached graph shows the near field response (microphone placed 12 cm away from driver, data truncated at 25 ms.

With the small box pre-filter, I can listen to these loudspeakers. I can even seriously listen to them. I notice the lack of bass extension, the higher distortion, and coloration of some mid bass sounds. I also notice better than average clarity and detail in the midrange and treble, and a defined soundstage dependent upon the musical selection.

Mark
 

Attachments

  • 12w4-654s.gif
    12w4-654s.gif
    9.8 KB · Views: 3,193
Hi all,


Here is another modification for a different Tang Band four-inch driver. This time it is a PPM cone and a lovely little neodymium motor assembly.

I have posted details of the W4-1052SA modifications to my Web site.

You can only get there by following the links I post to this forum. I will duplicate this message on the "Has anybody heard these" thread.

The modification is simple, inexpensive, and can be undone. This will not void warranties or anything else. The sensivity of the driver after modification is unchanged. In its stock state the rise in upper mid to treble response is greater than specified sensitity. Also, so little mass is added to the cone that T/S specs change no more than the difference between Tang Band's T/S published spec and actual measured specs. In other words, it is unimportant should you want to listen to the sound of the box air mass in resonance and port the enclosure.

After modification, the driver sounds accurate. The improvement is easily heard and brings the performance and accuracy to a level that is equal to any four inch on the market.

Here is the link:

http://madspeaker.com/Projects/TBW4-1052SA.htm
 
TB881S glue ring mofication

My Tang Band modifications posted to this thread are out of order. The W3-881 modification is the first modification I talked about on this forum. You can find performance results in the system picture thread with additional explanations of why the W3-881 benefits from modification. In addition, to really mix up the order of this thread I will be adding posts of one update and one modification that are already part of different threads on this forum.

This posting, however, details how to make one series of modifications to the W3-881. This modification series is different from the modifications pictured in the system pictures thread or the upcoming brick and mortar publication. This modification uses a glue ring addition to the cone and a plug replacement for the dust cap.

While this modification is different from the brick and mortar modification, it is still valuable in that it corrects a number of problems in the stock unit and does bring performance to hi-fi levels. This modification is also easier to complete with less risk of a mistake ruining a driver. Indeed, the only difficult task is removing the dust cap, and that task is fairly easy to accomplish.

The dust cap shape consists of one piece with two shapes. One shape is the central dome; the second is the flat ring or rim used to glue the dust cap to the cone. The preferred method for removing the dust cap is to use an Exacto knife to cut away the dome portion of the dust cap, leaving just the rim ring. Then, with the blade edge facing away from the cone, slide the tip between the dust cap rim and the cone. Lift upward to cut the remaining ring of the dust cap rim. Using needle nose pliers, pull on one of the cut ends and peel the remaining rim away from the cone.

Next apply a thin bead of GemTac glue to the cone as shown in the picture. The glue ring is made from a bead of glue .75 to 1 mm in diameter and three mm in from the inside diameter of the surround.

Last, center and secure a plug to the pole piece to replace the dust cap. A photo is included with a US nickel for comparison of size.

Good designing and good building,

Mark
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.

An externally hosted image should be here but it was not working when we last tested it.
 
Mark,

Bravo! Great work! And thanks for making it so accessible to DIY laymen.

Would you care to describe for us how you go about hunting down the resonant culprits in these various drivers, how you read response/decay graphs of stock units and devise a plan of attack? Or is it largely random cut and try?

I'd be very interested in the processes and reasoning you use to arrive at improvements.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.