From arizona!noao!asuvax!cs.utexas.edu!uwm.edu!kodak.com!nobody Fri Apr 6 14:13:29 MST 1990 Article 220 of rec.audio.high-end: Path: arizona!noao!asuvax!cs.utexas.edu!uwm.edu!kodak.com!nobody >From: nobody@Kodak.COM (Barry Ornitz) Newsgroups: rec.audio.high-end Subject: Science behind Armor-All, Improvements over Armor-All Message-ID: <9004060344.AA08879@kodak.UUCP> Date: 6 Apr 90 13:20:17 GMT Sender: news@uwm.edu Organization: Eastman Kodak Company Lines: 566 Approved: tjk@csd4.csd.uwm.edu I think the readers of this moderated group might appreciate learning about this research. Due to a problem with the interconnectivity of our internal corporate network, the article was posted internally last weekend. I do not think it went out over the full UUCP/INTERNET, however. The following discussion is a narrative relating the research Dr. Edward D. Kingsley and I have done on the effects of Armor-All (tm) Protectant on digital compact audio disks. This narrative is rather long and contains a good deal of chemistry and physics. It is being posted in both the "sci.chemistry" and "rec.audio" newsgroups on USENET because of its general interest to subscribers in both areas. ALL future discussion about this posting in general, or about our specific findings, should be directed to the "rec.audio" newsgroup unless your site subscribes only to the science groups. This discussion is _VERY_ long; if you reply, please DO NOT WASTE NETWORK BANDWIDTH by quoting my entire discussion. If you must quote, be selective and use your editor to keep only pertinent parts. Remember, some mailers insist that reply comments must total more lines than the original, and there sure are a lot of lines in the original! If you have comments of interest to the entire net, post them; otherwise specific questions should be sent directly to me. Thanks. Dr. Barry L. Ornitz ornitz@kodak.kodak.com ...rochester!kodak!ornitz ************************************************************************** ARMOR-ALL COMPACT DISK TREATMENT EXPLAINED AND IMPROVED Barry L. Ornitz, Ph.D. ************************************************************************** INTRODUCTION In recent months considerable discussion has taken place in the audio community about various techniques to improve the sound of digital audio compact disks (CD's). Numerous claims have been made about everything from dampeners to toothpaste polishing, Magic Marker stripes to Armor-All (tm) Protectant. At first, I dismissed many of these claims as being ridiculous and having no scientific basis. Over a period of time, however, I began to see a pattern in the claims - especially with the claims about Armor-All. Well-respected audio enthusiasts began to tout the virtues of Armor-All. Tom Krueger's moderated high-end group devoted a number of postings to this material, and even people like Charlie Thompson of Motorola suggested putting Armor-All to a scientific test by counting the number of detected, but corrected, errors that were reduced by using Armor-All. Little did Charlie realize that I have been conducting such tests for several months and have gotten quite interesting results - to say the least! I did not reply to his posting until now, however. This is because I believe I now have an adequate scientific, or should I say engineering, explanation for the effects of Armor-All on CDs. The reason I emphasize an XXXX ENGINEERING explanation is that I do not know ALL of the reasons why Armor- All works. But, I do know enough from my research, however, to explain 99+ percent of the effects, and I have learned enough to make significant improvements over the performance of Armor-All. To probe deeper into the science of how Armor-All can affect the sound from a CD, one needs a thorough knowledge of psychoacoustics, an area in which I have absolutely no qualifications. Like a good engineer, however, this has not stopped me from making useful progress. BACKGROUND I believe the initial idea of using Armor-All originated after people observed that severe scratches could be removed from CDs by polishing with toothpaste. It is logical to think that if Armor-All can hide scratches in automotive plastics and can protect vinyl from the effects of sunlight, it might be useful to coat CDs with the same material. This was also my first impression about why Armor-All might work - it simply reduces scattered light from surface scratches. While most people with severely scratched CDs found improvement with Armor-All, some venturesome audio enthusiasts tried coating new, unscratched CDs with mixed results. Some people claimed amazing improvements in sound quality while others reported none. This disparity sparked my interest in its effects. I must say at this point that my interest was mainly monetary. Like commercial audio manufacturers, I want to make a buck. I am interested in better sound reproduction and in making people happy with their audio systems, of course, but I also have to worry about paying my bills and supporting my hobbies. I figured that if I could understand how Armor-All worked to improve fidelity, I might be able to develop a better material that I could sell. I initially approached my management with this idea, but I was turned down immediately. I was told that the company quit the audio recording tape business many years ago and that the company was currently selling off the Verbatim subsidiary. In the chemicals end of the business where I work, we have only had one product sold directly to consumer markets, 910 adhesive. A product such as this would be difficult for the company to market. I was told that I was free to pursue this project on my own if I wished, but I could not even mention the company name. At this point I approached a close friend, Dr. Edward D. Kingsley, now of Polaroid, to help me with this work. Ed is an excellent analytical and organic chemist, while my background is in chemical and electrical engineering and automatic control theory (including digital signal processing). At the time we started working on this project, Ed and I were working in the development of instrumentation. Ed was enthusiastic about the project and we began working in our spare time, mainly nights and weekends. ARMOR-ALL (tm) PROTECTANT As background information, I should first say something about Armor-All. This is a liquid material manufactured by Armor-All Products of Irvine, CA. It is protected by at least two US patents: 3956174 and 4133921. Armor-All maintains that their chemical composition is a closely guarded secret, but a quick telephone call to a friend at Dow Corning suggested that the major ingredient of Armor-All was a silicone oil in a water emulsion. No other information was obtained on the chemical composition of Armor-All from literature studies. XXXX To see if the scratch hiding properties were the cause of fidelity improvements, Ed and I coated a large number of CDs with a low molecular weight silicone oil (viscosity of about 20 centipoise); a number of other CDs were kept as controls, and others were coated with Armor-All. We started with a combination of new, used, and deliberately scratched CDs. Since I have never been known as being golden-eared, I modified both Ed's and my CD players to obtain error counts. I do not wish to give details of the modifications here because I do not have the time to write up the modifications; furthermore, most readers would not have the technical competence to make these modifications to their home CD players. It is fortunate that my player was quite different in key aspects from Ed's, the importance of which we will explain later. We have since used five more players to confirm our results. We postulated that the silicone oil should work as well as Armor-All in hiding surface scratches. We began by noting the differences in error counts between the various CDs on the different players. I hesitate to give actual numbers because we had far too few members in our populations to give meaningful statistics. We did attempt a crude factorial design experiment, however. Actual error counts are quite player specific, so we normalized our data to 1000 error counts/minute on the untreated CDs. We obtained the following crude data: -------------------------------------------------------------------------- TABLE I Ornitz Player Kingsley Player Untreated Armor-All Silicone Oil Un. A-A Silicone New CDs 1000 130 470 1000 450 495 Used CDs 1000 80 265 1000 285 310 Highly Scratched 1000 45 110 1000 100 130 -------------------------------------------------------------------------- Before anyone reads too much into this, I should say that the actual error counts increased considerably between the new and used CDs and the highly scratched CDs had significantly higher error counts than either of the other two categories. To try to control the amount of scratches present on the highly scratched CDs, we used a combination of 600 grit abrasive followed by glass bead-blasting. As mentioned above, the normalization was used merely to demonstrate the relative improvements obtainable. It should also be noted that with new CDs, the error rate was small enough that the normalization process increased the scatter in our measurements. The importance of the above data really lies in the fact that, at least with my player, the Armor-All has an effect above and beyond merely hiding surface roughness. I was consistently able to differentiate a two to four- fold improvement of Armor-All over the silicone oil with my player. Ed, however, was able to see only a marginal improvement with his CD player. At this point, we had confirmed what numerous people had been saying: "Armor-All works!" and "Armor-All doesn't work!" We were able to clean the silicone oil off our CDs using successive rinses of Freon-TF and very dilute potassium hydroxide. Error counts with the cleaned CDs were slightly higher than they were before initial coating, probably due to abrasion and minor chemical attack in the cleaning process. Coating these cleaned CDs with Armor-All reduced the normalized error XXXX counts to values approaching those shown above. We did not try to clean off the Armor-All coated CDs. At this point, we were confident that there actually was an effect due to the Armor-All, but that this effect was not seen in all CD players. To help understand why, we turned to some chemical analysis techniques. CHEMICAL ANALYSIS OF ARMOR-ALL Obviously, we were seeing an effect with the Armor-All that was due to more than the hiding of surface scratches. We suspected that Armor-All was indeed a more complex mixture than just an emulsion of silicone oil in water. To confirm this, Ed solicited the help of some additional analytical chemists by letting use use their laboratory analytical equipment. The first analytical technique we tried to separate the Armor- All into its constituents was gas chromatography using a capillary column. This was not very successful because of the low volatility of what we suspect was the silicone oil. Next we tried HPLC (high performance liquid chromatography); using this technique we were able to obtain several "cuts" from the original mixture. Due to the small quantities of material available from several of these cuts, we decided to use conventional Fourier transform infrared spectroscopy and Raman spectroscopy to analyze the composition of each "cut". We were able to identify several additional components in Armor-All such as the emulsifying agent, but a few of the components eluded us. With one of the cuts, however, Ed noticed some strong transmission characteristics in the near-infrared region near 1000 nanometers using the FT-NIR instrument (I like wavelength nomenclature - Ed likes wave numbers). It was at this point that the proverbial light bulb, or rather laser, lit up above my head. I remembered that the wavelength of the lasers in the CD players operated in this same near-infrared region. Quickly checking the components in each of our players, I found that my player used as its laser an Aluminum Gallium Arsenide source while Ed's player used an Indium Gallium Phosphide source. Each of these devices operates at a slightly different wavelength - different enough that the absorption characteristics of Armor-All would affect each player differently. This could easily explain why some people noted sizable differences when Armor-All was used while others say only small effects. We continued our analyses using the Raman instrument. In this case we were concerned less with identifying the material than we were in learning how the coating affected the reflection characteristics of the CDs. With this instrument we learned another vital piece of information. In Raman spectroscopy, the surface is excited by a high intensity laser source; in addition to the Raleigh scattered return light, there are other wavelengths returned due to the Stokes and anti-Stokes scattering. We used a green argon laser as our excitation source. We learned, to our amazement, that the polymer used in most CDs can produce a broad spectrum fluorescence in the near-infrared regime if excited with visible light, or especially with ultraviolet light. This phenomena is quite common with many organic polymers. At this point a second key piece of the puzzle had been revealed. This fluorescence phenomena is also a good explanation of why marking the edges of a CD with a permanent ink marker can sometimes help the fidelity. XXXX With these two key pieces of information, I was able to postulate a theory of why Armor-All could improve the fidelity of CD playback. STRAY LIGHT AND PHASE JITTER My theory basically involves the interaction of stray light at the photodetector with the reflected light from the laser beam and how the stray light can cause phase jitter. To understand this, the reader must realize that the optical detection ina CD player is an analog rather than a digital process. The reflected light from the CD either strikes or misses the detector diode based on how the excitation laser light intersects the micro-pits on the CD. The detector diode signal is amplified and then passed to a comparator where the magnitude of the detector signal is compared to a reference signal. The comparator outputs a digital signal, or a series of "1" and "0" signals, to the digital signal processing section of the player. The comparator performs a threshold function. If the signal is too weak, it outputs a "0". If the signal is strong enough, it outputs a "1". There is normally a noise component riding on top of the major signal, but the change in the amplitude of the major signal will normally be so large that the comparator can ignore the noise. If the risetime of the input signal is high enough, the threshold effect of the comparator completely ignores the noise signal. However, in the _real_ world of CD players, the reflected signals from the CD have finite risetimes. Noise riding on top of the signal can cause the comparator to switch in either a premature or tardy fashion depending on whether the noise is in or out of phase with the major signal. The time difference shows up as phase jitter on the digital signal. The magnitude of the jitter is dependent on the magnitude of the noise riding on top of the original signal. Phase jitter on a digital signal behaves much as phase jitter on an analog signal. That is, the signal acts as if it were frequency modulated (or more correctly, phase modulated). Were a spectral analysis to be performed on a single frequency signal with phase jitter, it would be seen that instead of a pure single frequency output, the signal with phase jitter would actually be composed of a large set of related frequencies. A spectra of frequencies would actually be present. Mathematically, this spectra can be described by an infinite series called a Bessel series. I do not purport to know all of the effects phase jitter can have on the digital signal processing within the various CD players (especially phase- locked loops for clock regeneration), but I can say with confidence that phase jitter is definitely not conducive to good audio fidelity. STRAY LIGHT AND ARMOR-ALL My working hypothesis was basically that components of Armor-All reduce the amount of stray light hitting the detector in the CD player. Three effects are seen. First, the near-infrared transmission characteristics of Armor- All, if they are centered on the correct wavelength of the laser used, can reduce the amount of out-of-band energy hitting the photodetector. This is because the spectral response of the photodetector is quite wide, even if it is optimized for the laser used. A narrow bandpass filter on the detector designed to pass only the laser wavelength can perform much the same function. Fabry-Perot or multilayered dielectric filters would be suitable for this application, but they would add considerably to the cost of CD players. The Armor-All coating is probably less effective than such XXXX a filter, but it has a hidden advantage too. The second effect is that an Armor-All coating will block the entry of all of the stray light into the CD except for light in its narrow transmission band. Stray light can enter the edge of a CD and be transmitted within the CD by a mechanism of both attenuated total internal reflection and an evanescent wave. This mechanism is similar to the principle on which fiber optics work. When the stray light within the CD hits scattering centers like the micro pits where the audio information is encoded, some of it is emitted outside the CD where it can be detected by the photodetector. This is why the Magic Marker technique can help by blocking light entering the CD from the edge. The third effect is the suppression of fluorescence in the near-infrared region by the polymer by eliminating the absorption of the exciting wavelengths in the ultraviolet/visible spectrum. The absorption of Armor- All in the UV/VIS region is small, however, and Armor-All actually produces little improvement in fluorescence suppression. At this point, one of my friends while reading a preliminary version of this article asked why the Armor-All coating did not show more of an improvement with Ed's CD player. After all, the second and third effects are still present even if the transmission band was not optimally centered for the laser type in Ed's player. Looking back at Table I, it can be seen that the Armor-All was only marginally better than the silicone oil with CDs in Ed's player. What is happening here is actually a more complex situation than I have just described. It is true that much of the sources for stray light have been eliminated by the Armor-All. However, the absorption at the laser wavelength also reduces the desired reflected laser signal. The net effect is that while the noise has been reduced, the desired signal is also reduced. This led to a situation where the signal to noise ratio did not change significantly in Ed's player. Most of the fidelity improvement in Ed's CD player was due to the reduced effect of surface scratches. IMPROVED COATING MATERIALS At this point, Ed and I felt that we understood the phenomena enough to try to improve upon the properties of Armor-All. Consulting a number of references listing absorption and transmission spectra of organic compounds, Ed was able to generalize on the types of molecular structures that would produce transmission peaks in the desired wavelength ranges while simultaneously producing strong absorption of wavelengths shorter than about 600 nanometers. Ed is an expert in molecular synthesis software, and with this he was able to predict an optimum compound for our uses: L-monomethyl,lirpanoic butryate which we quickly code-named _MMLB_. This compound shows transmission peaks at not only the wavelengths corresponding to the aluminum gallium arsenide and indium gallium phosphide, but also to straight gallium arsenide lasers too. It appears red because it strongly absorbs wavelengths shorter than about 650 nanometers. Additional absorption occurs at longer wavelengths into the far infrared region. Combined with the wavelength characteristics of silicon photodetectors, the effective sensitivity curve includes ONLY the desired laser frequencies. We found that the coating process needed to get a uniform thickness of MMLB on the compact disk was quite critical. This is obvious if you consider that while the material transmits light at the desired wavelengths much better than at undesired wavelengths, there is still some loss associated with the laser light passing through the material. If the thickness of the XXXX material varies, this transmission will very, once again producing a form of noise that rides on top of the desired signal. Ed and I attempted several methods of coating CDs. A fast, easily controllable process involved vacuum vapor deposition of the MMLB on CDs. This would be fine for production coating of CDs but would not be practical for home audio enthusiasts who wanted an after-market product. Ed suggested Langmuir-Blodgett films. In this case the MMLB would be added to a solution of surfactant and water where it would form a monolayer film on the surface of the water. The effect is much like the thin colored films of oil on the surface of a highway after a rain. Dipping the CD into the film would produce a single molecular layer of the MMLB on the CD, an ideal situation. This technique is also practical to use at home. Ed and I found, however, that the MMLB surface was somewhat hydroscopic and could easily be damaged by handling. At this point, I had the BRILLIANT (I AM proud of this, you know!) idea of chemically bonding the MMLB to an ultra-low molecular weight polyethylene. These low molecular weight polyethylenes are really oligomers rather than polymers; they are much like paraffin wax in consistency. The resulting molecule has an ionic end, the MMLB, and an "oily" organic end. This structure itself forms a surfactant compound much like many soaps. By adding a small amount of the MMLB-PE to water, an emulsion is obtained. When the CD is dipped in the emulsion, the MMLB end sticks to the CD while the polyethylene "tail" sticks out. This has several interesting effects: First, the surface of the CD is no longer hydroscopic. Second, the polyethylene surface is much more resistant to handling. And third, the surfactant properties cause a very uniform coating to be produced. This uniform coating is formed because the MMLB-PE molecule looks much like a molecular sized sperm cell with a fat "head" and a long "tail". When the CD is dipped into the emulsion, the MMLB ends bond to the CD leaving the PE "tails" sticking out. Electrical polarization in the solution causes the CD surface to become uniformly coated. Excess material may be simply rinsed away. A crude graphic representation of the CD surface is shown below: Figure 1 | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | @ @ @ @ @ @ @ @ @ @ @ @ ---------------------------------------- ######################################## ######################################## Testing of the MMLB-PE material showed it to be an excellent performer, eliminating the hydroscopic problems and handling problems. Unfortunately it produced a surface on the CD that was very static prone. The surface would readily collect dust and dirt (and at my house, dog and cat hair). Ed came up with a simple solution. By reacting the free end of the polyethylene with an amine, we could produce something like the additive used to make "pink polyethylene", the antistatic packing material so common in the electronics industry. This technique quite effectively eliminated XXXX the static electricity problem. A recent article (Feb/Mar '90) in the trade journal EOS/ESD Technology (Electrical Overstress/Electrostatic Discharge) discussed the chemistry of pink polyethylene. In this article, it was learned that there is a stress-corrosion materials compatibility problem with "pink poly" and polycarbonate plastics. We have not found this to be the case with our MMLB-PE compound because it is the MMLB end that contacts the substrate, but we are still testing. If this turns out to be a problem, Ed feels he can make some other simple chemical modifications to our material that will eliminate the compatibility problem. FURTHER IMPROVEMENTS In the last few days, Ed and I have discovered an additional improvement in our coating material. By adjusting the molecular weight distribution of the ultra-low molecular weight polyethylene, we can control the thickness of the coating quite accurately. In other words, we can adjust the length of the "tail". By making the overall coating in precisely the right thickness, we can enhance its properties as an anti-reflection layer in the desired wavelength bands of the solid-state lasers. A summary of the properties of our material are shown in Table II. --------------------------------------------------------------------------- TABLE II MMLB-PE Properties * Passes the wavelengths of laser light used in CD players. * Rejects other wavelengths of light in the near-infrared region. * Blocks fluorescence phenomena by blocking the excitation source. * Is easily applied using "kitchen sink" technology. * Is tough and resistant to handling. * Reduces static build-up on the surface of CDs. * Reduces surface reflections from CDs. --------------------------------------------------------------------------- The important thing though is that this material GREATLY enhances the fidelity of CDs, both in our laboratory measurements and in our listening tests. I can readily discern the improvements in audio fidelity with the MMLB-PE coatings, even with my relatively "non-golden" hearing. Some of our "golden-eared" friends who have listened to a batch of our treated CDs claimed they did not wish to listen to ordinary CDs again. In actual measurements of error counts in our modified CD players, the results of the MMLB-PE are equally dramatic. The results with both my player and Ed's are now almost identical as are the results with the five other CD players we have tried. In fact, with the new and used CDs, the normalized error counts are now so low as to make our measurements quite XXXX difficult. To get meaningful statistics, our sample size will have to be increased considerably. HELP WANTED Ed and I are now at the point in our research where we need to do the so- called Beta Testing. This is where the rec.audio group on USENET comes into the picture. We are asking for volunteers from the net to help us test this concept. We would like to run double blind tests, but because our coating shows up as pink (remember the coating absorbs all visible wavelengths below approximately 650 nm) on the surface of the CDs, knowing which CD has our coating is obvious. However, we feel that if the listener is kept from knowing which CD he is listening to, our results will still be meaningful. Ed and I were given several cases of a particular classical CD from a well known company who is interested in licensing our technology. From this large batch, we have selected groups of three CDs based on similar error count performance. In each group, one CD will be left uncoated. Another will be coated with Armor-All, and the third will be coated by our MMLB-PE. Volunteers will be sent a group each for evaluation. To show the least bias, we ask that you solicit someone to insert one CD at random into your player doing everything they can to hide its true identity from you. This person should record your responses noting which CD was being played at the time. The CDs should be swapped randomly by your assistant to build up a large database of information. While we certainly look forward to your emotional comments about the various changes in fidelity (more liquid, less harsh, good color, mushy, or whatever), we really want data that shows statistical significance. Thus a simple ranking of 1 to 3 with 1 sounding the best is what we need. With at least 50 samples in the field, we should have plenty of data to work with if our volunteers repeat the test numerous times. We encourage you to have your friends join you in the testing as the more data we get, the more statistically significant our conclusions become. We are asking for volunteers to help us in this testing. We have already approached several key individuals who post to this group regularly because we value their opinions highly. However, don't be offended if we skipped you. We are looking for people who understand what we are trying to measure and quantify with our statistics. We do not want listeners who, for whatever reason, deny the significance of blind testing and scientific methods. The differences between the coated and uncoated CDs are plainly measurable by means of error counts; they are easily seen on an oscilloscope trace of the photodetector output of the players. We KNOW differences exist - we can measure them! At this time, we want statistical evidence that our coating is plainly superior in audio performance. You other folks will get your chance to do testimonials later! I mentioned earlier that at least one CD manufacturer is interested in licensing our coating material. Ed and I are also considering producing the material as an after-market product. We are looking for assistance in this commercialization. Finally, before anyone tries to beat us to market with a similar product, we would like to state that we have already applied for multiple patents on our coatings. One patent is a composition-of-matter patent on our specific coating material, another is on the method of application, and the final XXXX covers the concepts of the combination of properties listed in Table II. FINAL REMARKS This has been a LONG article. Our sincere compliments go out to anyone who has worked his way through it! Ed and I have spent many difficult hours on this project, but we are proud of our results. We hope this article has been informative to USENET readers. We thought you might like to hear about how the research and development process actually takes place before a product is actually brought to market. We welcome serious inquiries and discussion. However, we do not want questions like "Where can I buy it?" at this time. My net addresses is given at the beginning of the article. Ed is in the process of closing on a new house so he does not wish to list a home telephone number; I can be reached after 7:30PM EDST at 615/288-7803 if you wish to speak with me personally. Like they say in the B&J wine cooler commercials, "Thank you for your support!" Happy listening......... Barry