fedeleluigi

Jay, AR was very reluctant to change item part numbers. When it did, in general something really important had changed! On the AR drawings you can see that the on axis frequency responses of the 200040-0 and 200004-2 woofers are rather similar but the former has a 2-3 dB higher sensitivity than the latter. The Tonegen 1210040-xx woofer you posted was the replacement part for the 200040-0 woofer. When it was used as replacement part for the 200004-2 woofer it had a big resistor (probably to be connected in series to "tame" its greater sensitivity). Unfortunately I have never seen an official AR or ABtech document showing how to connect this Tonegen replacement part in the AR2ax, 5, LST2, 14 and 12.

Hi Jay, I think that around 2000 the stereo repair shop reconed your 200040-0 woofers instead of just replacing the foam surrounds! The part # of the 200040-0 original paper cone is 203043x and is printed on the back of the paper cone itself.

So sorry to hear this. I'll miss his very good contributions. Although late, my deepest condolences to all his family. RIP Robert

If that is the only problem, the tweeter is repairable but it is a very delicate procedure and should be done by very experienced hands. The 200014-0 / -1 tweeters are very rare and very hard to find in the used market. They are different from the They differ both from the 200005 tweeter (used in the AR-6, 7, 8 and 4xa) and from the 200014-3 tweeter (used in the AR-18, 17 and 25) and cannot be interchanged with them. The AR-MST / 1 (USA) and the AR-MST improved (Europe) have 3 tweeters in parallel and, for this reason, do require high impedance tweeters. So, differently from 200005 and 200014-3 tweeters, the 200014-0 and 200014-1 units are high impedance tweeters and have a DC resistance of about 10 Ohms (from 9.5 to 10.5). Unfortunately, in some old threads I have read about people who changed the first European version of the AR-MST (the one with 4 tweeters connected in series-parallel) to the "improved" version by using the MST improved crossover and connecting three 200005 tweeters in parallel. By doing so, the impedance at 9000 Hz is about 1.7 Ohm! Without considering the phase!

The 0.5 Ohm resistors change the damping of the filter cells and must not be removed! The polarity reversal for the tweeter connections depends on the fact that the high pass filter is of the third electrical order for the AR-9LS while it is of the second electrical order for the AR-9LSI. So, if you change the filter you must also change the polarity unless you can verify the result with appropriate FR measurements. The crossover changes made by the AR engineers in the AR-9LSI crossover are very small. The AR-9LSI used the "new" Tonegen drivers and the small changes made in the LSI crossover were most likely made primarily to handle the small sensitivity an FR differences of the new Japanese drivers compared to the US-made ones. So, if your speakers have the original (made in USA) drivers (tweeter-midrange and lower midrange), my personal advice is not to modify the original crossover. If these small changes had led to any improvements, you can rest assured that AR engineers would have done them long before the introduction of the 9LSI as any advanced hobbyist would! In any case, I would not be surprised at all if most audiophiles would say, due to the "improved" suggestion, that the 9LS sounds better with the 9LSI crossover...😀!

Adriano, for those who know a little about acoustics, those measurements are not comparable. AES paper AR-3a 2Pi frequency response measured on axis in AR "particular" anechoic chamber (please, see my previous post for details). I do hope you can appreciate the great difference between Fig.9 (4Pi anechoic response on axis) and Fig. 5 (2Pi frequency response measured in AR "particular" anechoic chamber) graphs shown in the AES paper for the same AR-3a speaker. So, as I wrote before, you cannot compare apples with oranges. Moreover, it is not a question of whether the AR-3 is better than the AR-3a or vice versa. The problem is that what you say does not correspond to the truth. For many years, you have had the theory that the AR-3 and AR-3a were speakers conceived and designed differently and consequently their sound was very different. However you have never been able to prove this theory of yours. Furthermore, you have never taken into account a lot of data that has been provided to you over the years. In summary: -First of all, no company in the world would replace a best-selling loudspeaker such as AR-3 with one that is very, or even totally, different in sound. Moreover at AR they pursued the accuracy of reproduction and not a particular sound philosophy! Or, if you prefer, the sound philosophy of Villchur, Allison and many other talented people who worked at AR was exactly the same: to make a very accurate speaker! -Second, all the reviews from that period claimed the differences in AR-3 and AR-3a sound were rather subtle and not earthshacking. -Finally, even the testimonies of very experienced people who were able to compare them when both AR-3 and AR-3a were new, state the same thing: the sound of the two speaker systems was ultimately very similar. For all these reasons, I previously quoted an old Tom Tyson's post where he says: “The AR-3a certainly improved upon the same formula that made the AR-3 such a good speaker, but the refinements were quite subtle. For example, there are no pronounced differences in the measured response of the two speakers - there are more similarities than differences. I heard a "new" AR-3 and a "new" AR-3a compared in the AR Music Room in 1967, and the "new" speakers sounded much more alike than have been described by some people on these pages. The biggest problem I see is that it is difficult to find AR-3s today, after so many years, that perform as they did when new. This is also applies to the AR-3a's 3/4-inch tweeter, as well.” You seem to want to convince yourself that your AR-3s still sound like they did when they were brand new! Unfortunately, it is well known that the AR-3's midranges very often age poorly and badly as the whitish material around the dome unfortunately hardens. Until it completely hardens, this causes a progressive increase in the unit resonant frequency which progressively results in a low output in the mid-range band and an increased output in the high-range band (just like when you tighten the strings of a guitar more and more: the "basses" progressively decrease and the "highs" conversely increase). That's why "no midrange” and “sparkling highs" as you say in the first post! That’s correct! That’s the result of AR-3 midrange aging until, ultimately, the output will be very, very low at all frequencies like some members described in this forum. Reviews that compared AR-3 and AR-3a - - Electronics World enterely reprinted Julian Hirsch's review.

As I mentioned, the graphs of the frequency response curves you posted are not comparable for several reasons. In summary, you can correctly compare measurements of two or more speaker systems only if they are acquired according to the same standards. We are sufficiently familiar with the "standards" used by AR through the AES paper (https://www.aes.org/e-lib/browse.cfm?elib=2058) from which the AR-3a graph you posted comes from but we know nothing about the "standards" adopted by High Fidelity Magazine. In the AR-3 article they just say: “The effects at 150 to 250 Hz cps are ground reflections and diffraction effects (the speaker was lying on its back, facing upwards for these measurements)”, and for the higher frequencies : “These measurements were made with both midrange and tweeter level controls at the indicated flat positions. But nothing is specified about the acoustic characteristics of the chamber in which the measurements were made, the positioning of the microphone, the equipment used, etc. It is not even specified whether the speaker system was flush with the floor as AR did in some of its measurements shown in the AES paper I mentioned. In the AES paper, as regards the 2Pi measurements, AR says: “Our main anechoic chamber has non-reflective wedges on only 5 of its interior surfaces. The sixth is smooth concrete, with an opening in its center; speakers to be tested are placed at this opening with suitable adapter baffles so that they are flush with the inside chamber wall. In this manner a 2Pi radiation angle is obtained with minimum discontinuity”. But what is even more important is that the AR-3a frequency response graph you posted is not the correct one. In fact, it refers to the AR-3a system measured in 4Pi anechoic environment: Fig.9 in the AES paper (please see the attached graph). In the AES paper you can read: “The picture becomes far more complex when the same system (AR-3a, Ed.) is measured in 4Pi environment. We took this AR-3a to the large walk-in anechoic chamber at Harvard Acoustics Laboratory, and repeated the measurements there. Fig. 8 shows on-axis response of the individual speakers of the system, in the cabinet and with molding in place. Mid-range and tweeter curves are shown for both the normal and maximum level control settings. The most obvious feature of Fig. 8 is the woofer response. There is a continuous down-hill slide from about 400 Hz, at which frequency the cabinet is a reasonably effective 2Pi baffle, to about 170 Hz, and then a flat response below that frequency. At 170 Hz and below the radiation angle is 4Pi steradians and the output, quite predictably, is lower than it was into a 2Pi angle. Molding and cabinet-edge diffraction are clearly at work on the axial response curves of the mid-range unit. Fig. 9 is the on-axis system curve, with molding, level controls at maximum.”

Adriano, No doubt that the charts you posted show two very different curves. But they were acquired in different situations and ways and for this reason they are not comparable each other otherwise you compare apples with oranges.

@AadamsThank you for the link! Tom Tyson also explained the reasons for the sonic differences in the second part of his post (I highlighted it in red). I agree with him. Although many people believe that the AR-3a project is essentially the work of Allison and McShane, personally, I have always believed that Vilchur certainly and actively participated in that project that probably began immediately after the introduction of the AR-3 in 1958. Taking into account Villchur's life after the sale of Acoustic Reseach to Teledyne (in 1967, a few months before AR-3a was introduced) I can't believe that in the years before the sale of his company Villchur only dealt with other issues of the company except speaker design. So, personally, I think there was no discontinuity with Villchur's "sound philosophy" with the introduction of AR-3a and the sound differences that today emerge when comparing AR-3 to AR-3a essentially depend on their different aging.

@Anders Nilsson I do think that the paper collar of the voice coil is damaged or even totally cracked and this generally causes that kind of problems you are hearing. Unfortunately the collar damage is a problem of the 12" ferrite woofer produced before about 1977. Then AR switched to aluminum for the voice coil formers and it is much more stronger than paper. On page 5 (fig. 2.4 and 2.5) of "Restoring the AR-3a" you can see the characteristic final crack of the paper voice coil collar.

I saw on Ebay two "strange" 12" Alnico woofers and I wish to share their pictures here because, as far as I remember, this is the first time I've seen this peculiar type of 12" Alnico woofer. They date Jul-Aug and Sep 1968 respectively. Although the magnet is Alnico, the cone and surround are not typical of the 12" Alnico woofers. In fact they are the same as those used in the "earliest" generation of the ferrite woofer. In other words it is used a smooth paper cone with a damping ring and the surround is made of foam. The baskets have the mesh screens (they were removed in woofer #2). Differently from the "earliest" generation of the ferrite woofers, the "classic" masonite ring for the surround is present (in woofer #1 it was removed). The spider seems particular (unfortunately the photo is not very clear). The peculiar remains of old foam surrounds do seem to come from the original AR foam surrounds (that is both woofers have never been refoamed previously). This important aspect would rule out that these woofers come from some restoration using some parts coming from the "first" generation ferrite woofers. In other words these alnico woofers do seem original! Dust caps are distinctive. #1 woofer: The masonite ring has been removed (see the other woofer) #2 woofer: "First" generation of 12" ferrite woofer (Mar 17 1971):

Ciao Michele, If your AR-LST crossover has a 10 Ohm resistor in parallel with the 2500 uF cap you are basically measuring the resistance of that resistor. You can see some AR-LST measurements here AR_LST impedance at two different switch positions (2 and 6)

It's hard to say. Unfortunately I have never seen and examined the AR-3a Limited. All I know about it derive from some members of this forum especially John O'Hanlon (forum nickname Johnieo @johnieo) and Mihn Luong (forum nickname mluong303 @mluong303). Unfortunately, they have not been writing on these pages for a long time. If they still read the pages of this forum, I take the opportunity to greet them. Some aspects of your AR-3a are similar to those of the AR-3a Limited others to those of the Italian AR-3a Replica. For instance your 3a crossover (components, PCB and wires) is the same as that of the 3a Limited (see the photo posted by Mloung303 in this thread http://www.classicspeakerpages.net/IP.Board/index.php?/topic/5330-ar-3a-reissue/&do=findComment&comment=78855 ) . On the other hand, the fact that the furniture seems to be veneered with real wood is an aspect similar to the AR-3a Replica. However, like the cabinets of modern speaker the AR-3a Replica has a veneer surface much more refined and smoth than that of the Original AR 3a. Johnieo said that: “the Genuine AR-3a Limited had a "picture frame" front similar to the US AR-3a, with the exception that its width was 1-inch instead of 1-1/4 inch” (http://www.classicspeakerpages.net/IP.Board/index.php?/topic/5330-ar-3a-reissue/&do=findComment&comment=78841 ) The AR-3a Replica picture frame molding is 31.8 mm (1.252 in) thick at the “base” and 31.4 mm (1.236 in) thick at the “top” and, moreover, unlike original AR-3a the inner edge is not squared but round (see pictures). So please check the thickness and the edge of you AR-3a picture frame molding and please let us know. In the AR-3a Replica I examined, all the screws are metric but I don’t know if there was some variations during their production. So, please check also your AR 3a screws as this can be a useful information. Today I have taken my brother’s AR-3a Replica apart to take some photos of its crossover. I have also taken some photos in comparision with a genuine old AR-3a that I should restore. These AR 3a Replicas were first bought by a friend that later gave them to my brother since in his listening room they never sounded well. My friend replaced the original caps and wires with “better components” in the hope that the sound could improve but to our ears nothing changed probably because the original electrolyte NP caps were new. The original tweeter cap was a generic polypropylene (or polyester I don’t remember) one but my friend replaced even that. The original connection wires were common type electric wires that my friend replaced with Cardas cables. PS If you don’t mind could you please post also the part number of the other tweeter of your AR-3as? Unfortunately the part number you posted is not clear and these tweeters are rare and their part number variation are still an enigma for me. Thank you the AR-3a Replica upgraded crossover: bobbins, L-pads, PCB are genuine. Original capacitors, resistor and wires were replaced with "better" components. The original crossover Original AR-3a and AR-3a Replica on the left and right respectively differences in the wood veneers: on the left the Original AR-3a, on the right the Italian AR-3a Replica. The wood veneer surface of the AR-3a Replica is much more refined and smoth than that of the original AR-3a but this aspect can't be fully appreciated by the photo. AR-3a Replica picture frame molding. The inner edge is round. The picture frame molding is 31.8 mm (1.252 in) thick at the “base” and 31.4 mm (1.236 in) thick at the “top”. Original 3a picture frame molding. The inner edge is squared. The thicness is the same both at the base and the top (31.75mm, 1.25").

@jomede Thank you Jomede, The driver is a rare paper dome tweeter manufactured by Tonegen. Unfortunately the number printed on the label is not clear (it could be 200084-0). In the early 90s there was a limited manufacturing of Ar-3a and AR-10 Pi Replicas in Italy. A very long time ago ( in 2003 ) I wrote about them here: I asked you for the tweeter part number because I saw several part numbers used for that tweeter. So if you don't mind checking also that of the other tweeter I would appreciate it very much! In 2014 I created a thread on the Italian Ar-3a and Ar-10 Pi Replicas in wich I added some photos: The Tonegen paper dome tweeters were also used in some Cello Amati (Mark Levinson). Member Adriano (nick name Sonnar @Sonnar) has an AR 3a version very similar to yours. The Tonegen woofers used in yours and Adriano's AR 3as as well as in Cello production have larger dust caps in comparison with those that are seen more frequenty in "normal" 1210003-2a woofers .

@jomede Hi Jomede, Congratulations on your AR-3a! If you don't mind, could you please post clear photos (front and rear with part number) of their tweeters. From the first photo they look like the rare Tonegen paper dome tweeters. Thank you PS: if the domes of your 3a tweeters are made of paper be very careful with them because they are very very fragile.

@budney Welcome to this forum! In my opinion you should check that woofers and midranges are the correct ones and in good electrical and mechanical conditions. If you post some pictures we can help you to identify the speakers. All AR-3a (improved) drivers are 4 Ohm. As regards DC resistance it should be around 2.6 Ohm for for woofers and around 3 Ohm for midranges. An identical mid-range was also supplied in an 8-Ω version for the AR-5. Therefore it is imperative to measure the DC resistance for each driver for the proper impedance match. You should check that the crossover has not been modified. Here you can find a schematic. Here you can download the file "Restoring the 3a" it will help you a lot even though you have an AR-3a improved.

when you can , would you please inspect carefully the sides of the ferrite magnets because the manufacturing dates were also printed there? thank you

yes. no, you are not missing any pieces.

@cbolen your woofers didn't originally have the so called "masonite ring" on the frame (see the photo of the woofer I posted as it has the masonite ring). In your woofers the foam suspension was directly glued to the metal frame. If you want to keep your woofers original you don't have to use any masonite ring. Later AR used a masonite ring glued to the woofer frame and the foam suspension was glued to that masonite ring.

@cbolen thank you very much for the woofer photos with dates. The date printed on woofer #1 should be MAR 17 1971. About woofer #2 I can clearly read only the month (MAR). Also if not leagible the year is almost certainly the same as the other woofer (1971) since your 3a have very close serial numbers. Please inspect carefully the sides of ferrite because the dates were also printed there. In the attached pictures you can see a woofer made only one month later (APR 22 1971). You can note that just one month later AR had switched to the so called "long wire" woofer. This woofer had a mounting ring on the frame for the surround and no damping ring on the paper cone. a

@cbolen Yes, all the drivers are original. The woofers are very rare. If the woofer spiders and voice coils are ok, you only need to refoam them. Since this type of woofer is rare, could you please post clear photos of all the dates printed both on the ferrites and the bottom plates ? Thank you.

@lARrybody the DC resistance you measured (3.9 Ohm) is not correct for the 200003 woofer (it should be about 2.6 Ohm +/- 10%). If your digital multimeter is accurate the tinsel wires could have some problem or the voice coil could have been replaced with a not appropriate one. The woofer basket looks like a 1975 one therefore if the the woofers are the stock ones the original tweeters were replaced with later versions since the tweter domes/suspensions were yellow until the first months of 1976 and later black. In any case there was no foam on the tweeter flange for the AR-10 Pi MKI. The original tweeter used in the AR-10 Pi MKI and AR-11 MKI are shown in the file "Restoring the AR-3a" (Fig. A.21 and A.22).

You can find a higher definition drawing in the library (AR drawings). The drawing is the 241th and is named "no image". Luigi

Hi Scoup, Thank you for answering my questions. I think that It's rather difficult to make accurate comparisons among speakers, especially if they are old, because there are many variables that can "distort" the perceived results. That's why I asked you some questions in my previous post. For example, in vintage speakers capacitors are often out of specs randomly. Different types and brands of caps may rapresent other unpredictable variables. Beyond capacitors, the drivers can also age differently depending on many factors. In other words, with aging, even speakers of the same model often sound differently from each other. Similar speakers like AR9 LS and AR9 LSI should be placed in an identical position if you want to make a correct comparison because speaker placement does affect the sound a lot. So, in this regard, I think you should have swapped the speakers position to get more objective results during your comparison. Different model of speakers often require different placements and/or amplifiers to get the best they can give. The SPL at the listeng point should be as similar as possible for all the speakers under test. Obviously our subjectivity is another variable but it's very difficult to eliminate it without resorting to a double blind experiment. Luigi

@BjarneAndersen Hello Bjarne, Unfortunately the coil is out of focus. It seems that you made only one layer winding covering all the former. If so, this is not correct. 1) the winding must be a two layer winding and 2) the former must not be covered completely. In the AR Library I've found the tweeter voice coil drawing of the "Lamba driver" used in the AR 9 LS, 98 LS and 78 LS. It is the 241th drawing. As regards wire, even though the drawing reports: "2 wire: 22 to 24 turns of N° 38 AWG (a diameter of about 0.1 mm) copper……….. in two layers, 11 to 12 turns per layer", actually I have usually found an inner layer of 14 turns and an outer layer of 11 turns. Hope you removed the tweeter diaphragm only mechanically since solvents usually change tweeter suspension stiffness permanentely and consequentely the tweeter FS and Thiele-Small parameters. Luigi