Jump to content

AR-9 bass crossover modification


Mach3

Recommended Posts

Some time back - maybe a year or two - I discussed the idea of removing the big cap (2500 uF) from the AR-9 crossover. My thinking was that this cap had been placed in the crossover mostly to protect the amplifiers of the late '70s. Without that cap the amp driving the speaker would have seen a load of < 2.0 ohms (two woofers in parallel). That would generally fry an '70s era receiver.

So I cut the cap loose (no other mods) and promised to the forum a F/R plot of how the speaker performed without that large cap. Please find attached the "before/after" plots for my Nines. The black/green plot is the original F/R with a rebuilt crossover (all film caps AND a mistake in the crossover - the 6uF versus 8uF error due to a fold in the original factory drawing). This plot was generated using an Apple Phone app and a pink noise CD. The crossover points are approximately marked in this plot by red vertical lines.

The second plot shows the F/R resolution without the cap in place. This plot was taken with an elaborate dedicated F/R measurement system including specially calibrated mike and a stand to hold it at my listening seat. Nice stuff (about $500 complete). Again a pink noise CD was used to generate the sweeps.

Notice that fixing the one cap in the crossover smoothed out the high frequency response. But do notice how the low frequence area has been extended. Those peaks and valleys are more about room modes than the speaker - because as we moved the mike around the room those peaks and valleys would disappear - in fact at some places in my room the speaker was flat into the mid-20 Hz region.

If you have an amplifier with sufficient drive (modern style - no tubes) you can successfully remove that big cap and actually realize an extention in your bass response - both subjective and objective.

post-111787-0-37388600-1429212245_thumb.

post-111787-0-78605200-1429212266_thumb.

Link to comment
Share on other sites

The 2500uf and 10 mh in AR9’s crossover constitute a LC resonant circuit which has a minimal impedance around 32 Hz. Similar circuits are used in some classic Infinity speakers using Waltkins woofers having 2 ohm voice coils. Do you have frequency response curves tested under the same condition, with and without 2500uf? I suspect the one with 2500uf would have some rolloff in the subsonic region which protects the dual woofers from overdriven.

Link to comment
Share on other sites

Some time back - maybe a year or two - I discussed the idea of removing the big cap (2500 uF) from the AR-9 crossover. My thinking was that this cap had been placed in the crossover mostly to protect the amplifiers of the late '70s. Without that cap the amp driving the speaker would have seen a load of < 2.0 ohms (two woofers in parallel). That would generally fry an '70s era receiver.

So I cut the cap loose (no other mods) and promised to the forum a F/R plot of how the speaker performed without that large cap. Please find attached the "before/after" plots for my Nines. The black/green plot is the original F/R with a rebuilt crossover (all film caps AND a mistake in the crossover - the 6uF versus 8uF error due to a fold in the original factory drawing). This plot was generated using an Apple Phone app and a pink noise CD. The crossover points are approximately marked in this plot by red vertical lines.

The second plot shows the F/R resolution without the cap in place. This plot was taken with an elaborate dedicated F/R measurement system including specially calibrated mike and a stand to hold it at my listening seat. Nice stuff (about $500 complete). Again a pink noise CD was used to generate the sweeps.

Notice that fixing the one cap in the crossover smoothed out the high frequency response. But do notice how the low frequence area has been extended. Those peaks and valleys are more about room modes than the speaker - because as we moved the mike around the room those peaks and valleys would disappear - in fact at some places in my room the speaker was flat into the mid-20 Hz region.

If you have an amplifier with sufficient drive (modern style - no tubes) you can successfully remove that big cap and actually realize an extention in your bass response - both subjective and objective.

Those frequency-response curves are nearly meaningless; besides, they were from different systems. Were they done anechoically or with gaited-measurement equipment? Go back and change the position of the microphone a few times, and you will get a few different frequency-response measurements each time those changes are made. At the very least, you should do your measurements outdoors to try to avoid some reflections, room modes, etc., but the most important thing is that you have changed the crossover, and to determine what has happened, you need to measure the woofers with the microphone about .5-inches away, and measure only up to around 200 Hz. Measure the impedance curve, too, and then you will see that you should not have removed that capacitor, as it has a specific task in the crossover.

Link to comment
Share on other sites

I'm confused on your crossover change comment. The AR9 and 90 both have a 6uF and an 8uF in the xo.

The 6uf is a series cap on the tweeter in series with a 4uF.

The 8uF is a shunt cap (in parallel) on the UMR.

What mistake and change are you referring to?

Link to comment
Share on other sites

First; thank to Tom Tyson. Your information and insight is greatly appreciated.

The "mistake" is in the schematic for the AR-9 crossover that I found on this site. Apparently in the band pass crossover for the LMR (1.5" dome) there is a 6 uF cap specified - but it should be an 8 uF - the first attached image shows - as far as I know - the original factory schematic - note that there is a fold in the paper that makes an 8 appear to be a 6. I originally installed the 6 uF and the resultant was a lump in the F/R.

All the measurements were taken at my listening seat - both F/R plots. I was interested in some kind of objective measurement of what I was hearing - not that such makes a whole lot of difference - I let my ears lead me, not a cloud of technical mumbo-jumbo. The only reason I have these measurements is that a friend of mine is always going on about his F/R plots and he has all the equipment needed to do such measurements.

Yes the measurements were "gated" (as I recall 6 mS) otherwise the measurement is at best confused. The measurements are NOT FROM DIFFERENT SYSTEMS. Same speakers, same room, same CD player, same wiring (Cardas Golden Reference speaker wire, Kimber KCAG ICs). The amps have changed - the original was with Odyssey Audio Khartago "mono-blocks", the second was using a Pass X150.5 (best amp I have ever heard). The only thing that changed from one measurement to the other were the modifications to the crossover.

Since the measurement equipment is not mine I cannot say when I will get the chance to measure the woofers per Tom Tyson's recommendations. But I will as soon as I get the chance. I do not think the second plot is very accurate below 300 Hz or so. We did move the mike around the room and in some places the speaker measured FLAT down to 25 Hz - pretty amazing for a 35 year old speaker. My measurement friend perceived that the dips and bounces in the lower range were more about "room" modes.

From a subjective point of view, listening to "Dark Knight" (sound track from Hans Zimmerman) the performance with the 2500 uF cap removed is absolutely spectacular - incredibly powerful and wall rattling deep. The only reason I made the measurements was to demonstrate to the critics of this modification that the result of removing the cap was in fact demonstrably better - at least objectively.

I really believe, based on Tim Holl's notes (on this site) that the 2500 uF cap was placed in circuit to preserve amplifiers of that period. Modern high quality SOLID STATE amplifiers are not fazed by low impedance dips (which I believe drops below 2 ohms with the two woofers in parallel). Given this analysis there is NO NEED for that enormous cap or the "automatic transmission" capability designed into original circuit. It only limits the lower range performance.

One last note. I had occasion to hear a set of Wilson Alexia speakers ($48,000) driven by Momentum amplifiers ($30,000) in a specially treated, large, dedicated room at Audio Concepts (Dallas). The setup was done by Peter McGrath a very competent and capable employee of Wilson Audio. We are talking a $100,000 system. They played a piece from the SFO that had the organ hitting the low C (32 Hz). Some time before I had heard the Dallas Wind Symphony play a similar piece that also included the same "low C" passage.

To my ears the Mighty AR-9 EASILY OUTPLAYED THE WILSON "uber" speaker. Easily. The Nine was tighter, far more articulate, and far more controlled than the "modern" speaker - and its reach into the "nether regions" was demonstrably better. So hooray for Acoustic Research - they built incredibly fine gear that even in this era stands up as an outstanding example of speakers.

Link to comment
Share on other sites

The Nine was tighter, far more articulate, and far more controlled than the "modern" speaker - and its reach into the "nether regions" was demonstrably better. So hooray for Acoustic Research - they built incredibly fine gear that even in this era stands up as an outstanding example of speakers.

I've also had the opportunity to audition the Wilson loudspeakers, as well as a few in the same "high-end" price range from other manufacturers, and I think what you're hearing is a function of acoustic-suspension vs. vented-port design.

My experience has been that ported loudspeakers tend to be more difficult to properly set up in a given room than a-s designs, perhaps due to their nature of delivering LF sound from more than one speaker surface (as in rear-ported systems), or because ported speakers seem to always find that one note that they really like, and play the hell out of it (again, this is just personal opinion).

While the LF reproduction of the AR-9 is prodigious & musical, it might not represent what younger listeners expect from a high-end design, these days.

Given the virtual absence of a-s drivers in current high-end speaker production, it's conceivable that an a-s design simply might not sound right to those with a limited frame of reference.

NHT manufactures a 12" active subwoofer of a-s design, and it's the only current loudspeaker product that I'd be able to cite as providing LF response similar to that of an AR-9; vented subs just don't sound the same to me, and they do not seem to integrate within a room in the same manner.

Link to comment
Share on other sites

It's hard to tell from your measurements that include room modes if bass extension was actually acheived with removal of the high value cap. Did you measure Fc? That's the cabinet resonance. It could be an indicator of bass extension if before and after measurements were done.

Or, did you do a near field measurement on the woofers before and after? Actual -3dB measurements could be gleaned from those.

Link to comment
Share on other sites

I have hesitated to jump into this from the start but here goes ...

Mach3, your assumptions about how the woofer circuit works and what that capacitor does are significantly off.

The 2500 uF capacitor in series with the 10 mH inductor forms a series resonant circuit around the impedance

peak of the system where the system impedance is high due to the fundamental resonance. Plug the numbers

in here, plug in 2500 uF and .01 Henries and the resonant frequency is 31.8 Hz:

http://www.1728.org/resfreq.htm

Tutorial on series resonance:

http://www.electronics-tutorials.ws/accircuits/series-resonance.html

The LC impedance is a minimum at resonance and it is equal to the inductor resistance plus the ESR of the cap

at resonance. The inductor is 1.84 ohms so lets call the total about 2 ohms, note that the capacitor is a low

ESR type by design with many attachment points to the plates.

The two woofers in parallel do result in a low impedance load, but the large inductor in the crossover

is air core and has a fairly high resistance of 1.84 ohms bringing the woofer pair plus crossover back up around

3 to 4 ohms making the systems safe as a 4 ohm load even without the series resonance circuit. However,

the woofer impedance rises, often 3 times or more at resonance so it is safe to shunt the DC resistance of the

crossover inductor by the series resonant circuit. Note also that the inductor is very large at 10 mH due to the

very low crossover frequency and we can calculate the impedance at 31.8 Hz by entering 10 in the mH field, and

.0318 in the kHz field here:

http://www.qsl.net/pa2ohh/jslcimp.htm

It has 2 ohms of inductive impedance that must be added to the 1.84 ohms of DC resistance for a total of 3.84 ohms

which is significant as compared to the approximate rise to 6 ohms at resonance. The series circuit shunts the

3.84 ohms with about 2 ohms for a total (parallel) of 1.31 ohms. This significantly increases the output at and around 32 Hz.

It makes the impedance peak lower providing an easier, closer to resistive, load to the amp.

These are rough calculations because I don't have a working AR-9 to measure and simulate but someday I will do it.

I would not remove that circuit.

The thing to measure is the electrical frequency response of the crossover as loaded by the woofers, with the box

sealed as they actually operate with and without the LC circuit.

Link to comment
Share on other sites

I have hesitated to jump into this from the start but here goes ...

Mach3, your assumptions about how the woofer circuit works and what that capacitor does are significantly off.

The 2500 uF capacitor in series with the 10 mH inductor forms a series resonant circuit around the impedance

peak of the system where the system impedance is high due to the fundamental resonance. Plug the numbers

in here, plug in 2500 uF and .01 Henries and the resonant frequency is 31.8 Hz:

http://www.1728.org/resfreq.htm

Tutorial on series resonance:

http://www.electronics-tutorials.ws/accircuits/series-resonance.html

The LC impedance is a minimum at resonance and it is equal to the inductor resistance plus the ESR of the cap

at resonance. The inductor is 1.84 ohms so lets call the total about 2 ohms, note that the capacitor is a low

ESR type by design with many attachment points to the plates.

The two woofers in parallel do result in a low impedance load, but the large inductor in the crossover

is air core and has a fairly high resistance of 1.84 ohms bringing the woofer pair plus crossover back up around

3 to 4 ohms making the systems safe as a 4 ohm load even without the series resonance circuit. However,

the woofer impedance rises, often 3 times or more at resonance so it is safe to shunt the DC resistance of the

crossover inductor by the series resonant circuit. Note also that the inductor is very large at 10 mH due to the

very low crossover frequency and we can calculate the impedance at 31.8 Hz by entering 10 in the mH field, and

.0318 in the kHz field here:

http://www.qsl.net/pa2ohh/jslcimp.htm

It has 2 ohms of inductive impedance that must be added to the 1.84 ohms of DC resistance for a total of 3.84 ohms

which is significant as compared to the approximate rise to 6 ohms at resonance. The series circuit shunts the

3.84 ohms with about 2 ohms for a total (parallel) of 1.31 ohms. This significantly increases the output at and around 32 Hz.

It makes the impedance peak lower providing an easier, closer to resistive, load to the amp.

These are rough calculations because I don't have a working AR-9 to measure and simulate but someday I will do it.

I would not remove that circuit.

The thing to measure is the electrical frequency response of the crossover as loaded by the woofers, with the box

sealed as they actually operate with and without the LC circuit.

A Description of The AR9's Low-Frequency Section:

Pete's description in his message (above) is thorough and excellent, and I can't really add much to it other than to give some more specific details.

In the AR9, the two 12-inch acoustic-suspension woofers have a full-section crossover that provides both bass extension and a Q-optimizing circuit. The components used in the AR9 crossover are large air-core chokes (wound on 17 ga. Wire) and high-grade, bipolar electrolytic capacitors (usually Sprague) and premium, high-power ceramic resistors.

The internal (sealed) volume of the AR9 is 4.24 cu. Ft (120 liters), or roughly 2.5 times greater than a single AR-3a or AR-11. The woofers are AR's famous 1200003-series 12-inch drivers, as used in the AR-3a, AR-11, etc., with approximately 10 lbs. ceramic-ferrite magnet and Armco iron magnetic circuit. The heavy-duty voice coils are 2-inches in diameter with approximately 5/8-inches linear excursion and 1.1-inches maximum excursion before mechanical intervention. The woofer's fs (free-air resonance) is 17-18 Hz. The AR9's fc (system resonance) is 31.8 Hz., with the -3dB point of 28 Hz. This all points to an extremely potent low-frequency reproducer, essentially unsurpassed by any production, commercial loudspeaker (before or since) in terms of low-frequency extension, low harmonic distortion and low-frequency smoothness.

There are a few subwoofers with slightly greater low-frequency output, but no subwoofer is able to integrate the low frequencies with mid-bass frequencies nearly as smoothly as the AR9. There is nothing to identify the crossover from the woofers through the lower midrange driver, as it is seems like the output is only coming from the 8-inch driver itself, itself quite an accomplishment by the AR9's engineers. The frequency response of the AR9 from its two woofers up through the 8-inch lower midrange driver is completely seamless, a clear advantage the AR9 has over separate subwoofer/satellite speakers.

AR describes using two AR-11 woofers in a cabinet of twice the volume as a single AR-11, and wiring the woofers in series. Two wired in series this way will give exactly the same resonance, level and response as a single AR-11 but will take ½ the power from the amplifier with the same voltage-input level. So, efficiency rises. Wiring the two in parallel, aside from resulting in an impedance below 2 ohms, will result in exactly the same response and the same resonance, but 6 dB more output for the same electrical input to the amplifier.

If the box with the two woofers is enlarged—as with the AR9—the system resonance drops to the desired subsonic 31 Hz (from the AR-11's 43 Hz), but the "Q" of the system, or damping at resonance, also drops from 0.75 in the AR-11 to less than 0.50, having the effect of actually reducing the low-frequency output by virtue of extreme over-damping. AR could actually have used different woofers with higher impedance, or the engineers could have reduced the magnetic strength in the voice coil (Allison did this with the IC20), but these are compromises that AR was not willing to make with the AR9. The existing AR-3a/AR-11 12-inch woofers were known excellent performers, and it only made sense to use them in the new tower speaker.

To correct this "over-damping" problem in the AR9, AR's engineers designed a full-section bass crossover that would restore the "Q" to the desired level by using a resistor (actually the internal resistance of a choke) in series with the woofer at frequencies above and below the resonance frequency. Since the woofers have maximum impedance at resonance, the woofers can safely be used in parallel in this frequency region—and the resistance is out of the circuit at resonance—though not above or below this region; therefore, AR added the resonant circuit, thus permitting the 6 dB greater output (parallel) while adding resistance above and below resonance to raise the impedance of the parallel 4-ohm woofers to a desired level. This crossover circuit also flattens the output somewhat (to compensate for the 6 dB increase in output) to give the AR9 very uniform and smooth response up to the 200 Hz crossover into the 8-inch driver.

In short, this circuit works, and it works extremely well. So, to preserve the AR9's stellar bass performance, please don't change the crossover network!

—Tom Tyson

Link to comment
Share on other sites

My goodness I think I have a much better appreciation of AR9 woofer crossover than before thanks to the elaborations from Pete and Tom. Here is a reference on capacitor-coupled-bass showing a properly chosen capacitor can lower the f3 and flatten the rise before the resonant frequency.

http://www.diyaudio.com/forums/subwoofers/187634-capacitor-coupled-bass-can-sound-tighter-go-lower-frequency.html

post-119422-0-87441800-1429351189_thumb.

post-119422-0-99986700-1429351204_thumb.

Link to comment
Share on other sites

Many thanks to Tom Tyson and Pete B for their closely reasoned and strongly subtantiated replies. I think we all learned a LOT from these efforts - I know I did. WOW!!!

I think my next move is to measure the bass response alone - as per Tom Tyson's suggestion - and get a better idea what is going on at the speaker (as opposed to the listening seat which is in a corner of the room (has a fichus tree behind it to break up the inevitable mode - but that is probably not enough)).

What I would ultimately like to do is move the crossover to an external box. Going in and out through the woofer "hole" is difficult and wearing on the cabinet. Then I could install a temporary switch to evaluate the difference in response between the two circuit arrangements more easily.

The bottom line is that I "hear" a subjective difference with the capacitor out of circuit. That subjective appraisal is that there is "more bass" - a LOT more bass especially in the 25 to 35 Hz region. Of course I know that such a subjective measure is open to all kinds of error and could be misleading (I have been around enough so-called "audiophiles" to know that people can hear all kinds of phantom events - and of course I may be hearing what I want to hear.) Alas!!! Such is a human condition.

Again many thanks to the contributors, especially Tom Tyson and Pete B - they have provided me with a great deal of food for thought. I will return to this topic when I have better and more thorough measurements to offer.

Link to comment
Share on other sites

The electrical measurement will be far easier and more consistent, other than figuring out how to run

wires out of the box from across the woofer. Really ground is common so you just need one small wire,

remove a woofer screw and feed it out that way.

The woofer(s) in box electrical to acoustical transfer function does not change due to electrical network

changes so this is completely valid.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...