bjmsam

That review states, "Crossover Frequencies: 400 Hz and 3 kHz." This summary states, "crossover points 250 Hz and 2,500 Hz." Which is correct? Thanks for the schematic! The upper section and lower section appear to be independent (like the AR9's crossover); would the upper section behave differently with the lower section removed? Do the two sections work together like parallel resistors? Req=R1×R2 / (R1+R2)

Welcome to the forum! Anyone logged in should be able to download the copy attached to this post (click the black arrow at right to view): The forum index includes this link:

That's a fine choice. To be clear, I used M.D.L. NPEs for both my AR9C and my AR-3a projects, not Solens. If your AR-3a speakers sound dull with new NPE caps, then rebuilding the 50+ year old tweeters (an option missing from page 11) might be more prudent than altering the xover with MKP (PP) caps to run them "hot" (see page 16). The goal of my AR-3a recap was to restore / maintain performance and protect drivers by ensuring that the xover operates as originally designed. When it comes to upgrading sound quality, I strive to prioritize investments that objectively make a measurable improvement, and in my experience, room treatment tends to offer the most significant return. Good luck!

Yes, I did (more info here), and yes, I am!

Thanks! The AR9 pair is driven by an Adcom GFA-585 while the "AR9C" center speaker is equivalent to the top section of an AR9 mounted flush in the treated front wall behind a 120" acoustically transparent screen driven by the bridged channel of an Adcom GFA-2535 crossed over at 200Hz. A Denon AVR-X4000 configured for 7.0 surround (no need for separate subs!) also feeds Definitive Technology in-ceiling sides and KSI in-ceiling rears. Though some room challenges remain to be addressed, the setup measures extremely well and sounds fantastic.

Great story. Thank you!

It seems realistic to expect people to become enthusiasts about products that best achieve the primary objective(s). Relatively few today are enthusiastic about cylinder players or tube radios because every alternative performs better and is more convenient. During the classic speaker era, performance was valued most, and those products still deliver, but are not particularly convenient. During whatever era this is, it seems convenience is valued most, often at the expense of performance. LPs are a curious exception from the perspectives of both performance (IMHO) and convenience (adding that distortion electronically is easier).

That is wild. Nicely done! How far out of spec were the original BA caps you replaced?

I spotted this on eBay and don't know if it is authentic but figure someone might be interested or have insight to offer. Emile Berliner Award presented to Edgar Villchur acoustic research founder 1974

About lossless audio in Apple Music Samplerates: the higher the better, right?

Wow! Have you posted details and photos of this setup anywhere? That vintage "classic speaker" approach surely rivals high-end "modern" designs. My seven channel HT employs AR9 as front left/right and my custom "AR9C" as center with DefTech UIW BP/A sides and KSI 8081-CSD rears; I have imagined including my AR90 pair and other AR (3a, 3a-Improved, 94) and BA (T830 and A60) pairs so am interested in your experience. What I can't imagine is being satisfied with earbuds, "mono bluetooth speakers and Sound Bars..." 😞

In case you're still looking, I found something close: It's available in different thicknesses.

Sometimes a room will not work well regardless of speaker or placement. For the MLP, equalization (preferably more granular than 12 bands, such as a PEQ or Audyssey, Dirac, MiniDSP, etc.) may help attenuate peaks but cannot eliminate nulls as those frequencies cancel regardless of amplitude. Using something like REW to simulate and measure the room would be most revealing (I am happy to help if you live nearby - can't tell from your profile). Guessing is expensive; I certainly would not divest from the AR9 before confirming that an alternative sounds better! Other important aspects include height, construction, and any acoustic treatment to relieve bass pressure, which can be overwhelming. Example: In my home theater (much larger than your room and treated, though far from perfect), my AR9 sound a bit better than my AR90. In my game room (slightly larger than your room with little treatment), my AR90 sound a bit better than my AR9. In my living room (about the size of my home theater with no treatment) where WAF is more important, my AR-3a sound better in some parts of the room than in others. Anecdote: I drove to Long Island to buy my AR9 from the original owner about 15 years ago. I auditioned them in the untreated living room of his tiny cape-style house and was disappointed by the lack of bass. They had been recapped, all four woofers had significant extension, and he had "audiophile equipment" (fancy turntable, glowing tube amp, "articulate" (i.e.- overly bright) replacement speakers, etc.), so I figured the room was highly modal. Upon returning home, I placed them against the wall, connected them to my GFA-585, and was rewarded with terrific sound.

Thank you for your service and significant sacrifice. I hope you recover soon and continue cultivating your son's interest. I chimed in on your thread. 😉

Thanks! Being happy is all that matters. 😉 Your speakers look great! My woofer surrounds (and dust caps) were bone dry and felt like fragile paper. Now they feel much more robust, and the cone returns much more slowly after being displaced. One coat seemed adequate. Any difference in sound is too subtle for me to hear. Here is a graph of the woofer section measured on-axis in my room before and after sealing: Since a 100Hz HPF was inadvertently engaged on the AVR channel used for the measurements I posted earlier comparing before/after capacitor replacement (inconsequential but annoying), here is a final graph of measurements on-axis in my room with everything done and dialed in: The speakers are well aligned and not responsible for deviations from the red AR tech data trace as acoustic treatment of my room remains a work in progress... 🤔 Thanks again for the outstanding guide and support!

As in my center channel project thread: *** WARNING! Images of crossovers with NPEs rather than MKPs below! Viewer discretion is advised! *** The new capacitor measurements for left and right were virtually identical so those below from one or the other are representative. High readings are typical as discussed in the Cap Value Accuracy thread. According to someone who posted a review of my Proster LCR meter, it measures capacitance using a 300Hz triangular 200mV pk to pk voltage waveform. I have found no such information on my TC1, which reads the same as the BM4070 for the larger caps, though for the small caps, it reads lower individually but higher when in parallel. Vloss sure looks better... 2.7μF + 3.3μF capacitors: 50μF capacitor: 150μF capacitor: Reassembly: Measurements and Impressions Before vs. after graph below. Again, the green and orange traces are RMS averages of measurements on axis aimed between tweeter and midrange at 0.5m and 1.0m from the same position in my room. All pots now work but I did my best to position them the same as before for consistency. Digitized and SPL aligned at 400Hz for reference are the two curves posted in the AR-3a response curve thread. *Edit 2022-03-29: A 100Hz HPF was inadvertently engaged on the AVR channel used for these measurements hence the lack of bass extension. Any difference is too subtle for me to hear - they sound terrific! All that's left to do is apply cloth surround woofer sealant, which is on the way now that Vintage-AR has returned from vacation. I will measure before and after (with the AVR's HPF disabled this time) to see if there is a difference in output. This has been a fun project and I am very pleased with the outcome!

Here is 06064: 6μF capacitor: 50μF capacitor: 150μF capacitor: The potentiometer that I was unable to adjust: After DeoxIT and emory cloth: The other pot looked (and mostly worked) fine so a little DeoxIT was all it needed.

Here is 06052: 6μF capacitor: 50μF capacitor: 150μF capacitor: Both potentiometers (of the type @RoyC described) work and the one I opened looked fine so I left them alone.

"Before" graph below. The green and orange traces are RMS averages of measurements on axis aimed between tweeter and midrange at 0.5m and 1.0m from the same position in my room. One of the pots on S/N 06064 works only near its maximum setting, so I set the other three the same for consistency. Both speakers are as they came from AR in the late 1960s except the pots in S/N 06052 and the midrange driver in S/N 06064 were replaced in the late 1980s. Digitized and SPL aligned at 400Hz for reference are the two curves posted in the AR-3a response curve thread. *Edit 2022-03-29: A 100Hz HPF was inadvertently engaged on the AVR channel used for these measurements hence the lack of bass extension. Has anyone already modeled the AR-3a crossover in XSim? If not, are the FRD and ZMA files available for all three drivers?

But per the Old "Block" Caps thread, you did inhale! 😉 According to the Early Capacitor Manufacturing History thread, AR discontinued use of PCB caps several years before the AR-3a was introduced (but beware of early serial-numbered AR-3 speakers made through the mid-1960s). The wax-paper-impregnated caps that replaced them had an even lower ESR, and both were orders of magnitude less than that of NPE caps. These statements by @johnieo are most relevant to the questions I posed earlier (note that he inadvertently reversed the crossover frequency change, which decreased rather than increased). He and @r_laski also provided some very interesting cap descriptions, measurements, observations, and recommendations here.

Thank you! Very interesting. So AR replaced those wax-paper-impregnated caps with NPE caps of equivalent capacitance (but presumably different Equivalent Series Resistance, Dissipation Factor, Dielectric Absorption, charge migration, etc.) with no other changes to the crossover network than one inductor that most likely was to accommodate a different woofer driver? I do not have a way to measure caps like @Carlspeak did in the HERE'S WHY YOU SHOULD REPLACE THOSE OLD 3A CAPACITORS thread as I have only an LCR tester (and a TC1 tester that does not always show ESR) but will use REW with a calibrated UMIK-1 microphone to measure speaker output (in my room) before and after as @genek requested in that thread.

Thank you! I agree and am thankful to have had the privilege of literally growing up with this pair. After 54 years of continuous use, they still sound terrific and are most deserving of my attention. CSP is a treasure!

Echoing feedback posted over the past fifteen years, the guide is terrific and most appreciated! I am preparing to dig into my pair which Dad bought new in the late 60s (serial numbers 06052 and 06064). I am comfortable with NPEs (recently employed with success in my center channel project)... ...but what are the early caps, exactly? The guide seems somewhat ambiguous on this point: Also, the guide states: Since my woofers have cloth surrounds, the #7 inductor should not be replaced with a #9 coil, correct? Sure glad I saw this - thank you! Perhaps the guide should be updated for those not poring over every post?

The first graph compares the response predicted by XSim for the AR spec crossover (red) and my crossover (green) to the response of my center speaker C (purple) as measured by an Audyssey microphone positioned on-axis aimed between tweeter and UMR from 1.0M, 2.0M, and 3.0M (RMS average). The screen was not in place, Audyssey was off, and a GFA-585 was used for C (swapped with R temporarily). The second graph compares the response of C alone (purple) vs. C with L+R handling frequencies below 200Hz (green) vs. R alone (red) as measured by a calibrated UMIK-1 positioned on-axis aimed between tweeter and UMR from 1.0M and 1.5M (RMS average). The screen was in place, Audyssey was off, and a GFA-2535 was used in bridged mode for C. The third graph compares C with Audyssey off (purple) vs. C with Audyssey on DEQ off (orange) as measured by a calibrated UMIK-1 positioned at three locations left, right, and center of the MLP (RMS average). XSim traces are included at the top for reference. The screen was in place and a GFA-2535 was used in bridged mode for C. The orange trace approximates what I hear when listening to content. Observations: The SS AR-12000840 Acoustic Research Genuine .75" Dome Tweeter seems uneven (but so is that section of the crossover, presumably to match the original 200029 driver - the AR9 R on-axis (red) trace in second graph looks flat); it would be interesting to see how the MW Audio MT-4121: .75 inch Dome AR Copy Tweeter measures in comparison. The MW Audio MM-2044: Acoustic Research Copy Midrange UMR seems to perform well, and arguably is most important. The AR 200027 original driver LMR is not performing to the spec below, which seems unrealistic. From the library: I had overlooked Note 11 during assembly so just now stuffed the LMR enclosure with Roxul AFB to see if it would smooth and extend its frequency response but noted little difference between with wadding (light purple) and without wadding (purple). Included for reference is the AR curve (red) digitized from the AR graph above and the crossover response predicted by XSim seen earlier. What I am measuring is consistent with what XSim predicts for the AR spec crossover (and my crossover is very close, at least when ignoring unknown ESR and DCR values). Even ignoring my room challenges (note similar boost above and suck-out below 300Hz in my AR-3a curve), I really don't see how an AR spec LMR driver (which I am using) could roll off so low or steep without a change to the crossover. Adding the woofer section of the network to XSim did not change the curve significantly above 200Hz. I simulated some straightforward changes to my crossover that might help (comparison below) but probably won't bother since the HPF in my AVR seamlessly routes those frequencies elsewhere (green AR9 C on-axis sub 200 trace in second graph) and all of the speakers working together as a system sound well integrated. I am very pleased! It sure was! Thanks again for the inspiration and input.

Introducing the AR9C! 🙂 For the acoustic blanket I used 1/2" thick F26N wool felt with adhesive backing, which seems like a very good alternative (more here). Here is a quick look before I re-hang the screen. The red protective ring pressed in over the LMR mounting flange (to help form the felt while repairing my cut irregularities) was popped out before use. A quick test confirmed that it works and sounds great! The timbre is consistent across the soundstage, dialogue localization is much improved off-axis, the comb filtering effects are gone, and I can't tell that lower (less directional) frequencies are from left and right rather than center. REW measurements coming soon!