Will the real AR-3a please stand up?

[johnieo]

This spring I purchased an unmatched pair of AR-3a speakers. Although they sound great, one of the speakers (early AR-3a) has always sounded a bit louder in its bass region relative to its mate (1976 factory upgrade 3-to-3a). The reason for this difference has both eluded and bugged me. We just discovered that it is caused by uneaqual woofer inductance values.

With helpful insight from Tom Tyson and Edgar Villchur, any possible woofer problems were eliminated long ago. New surrounds; no air leaks; clean level controls--all thanks to the craftsman from whom they were obtained. I replaced all the capacitors with matched components.

So, what’s left that could cause this difference? Perhaps the inductors? After careful removal and measurement on an HP-4246A LCR meter we found the woofer inductor in the early AR-3a to be 1.9 mH (R-dc = 0.56 Ohm) and the woofer inductor in the factory upgraded AR-3 to 3a to be 2.85 mH (R-dc = 0.788 Ohm). This would produce at least two changes:

-- The crossover frequency would be reduced. The published crossover decrease from 575 to 525 Hz is slightly more than a halftone decrease, but less than a full tone. (On the piano 5C = 523.25 Hz, 5C# = 554.36 Hz and 5D = 587.23 Hz.) This difference likely would not noticeable unless one listened (on an unmatched pair) to a Wagnarian soprano bellering 5C#!

-- The low frequency level would be slightly increased in the early unit, because the coil resistance is part of a voltage divider.

One more data point: Please examine the attached gif image. Its left portion is taken from the human speakers web-site and depicts a completed crossover upgrade of what is stated to be an early AR-3a. The right hand portion of the image illustrates the two woofer coils from my AR-3a units. A careful observer will note one significant difference in the AR inductor numbers shown in the hummanspeakers image. (The shorthand part number is the single digit written on the coil face in black marker.) Sm-mid is #1, lg-mid is #4. The woofer coil is inscribed with #7.

The correct AR part number for the later AR-3a series woofer inductor is #9. (e.g. 805009, or L-1001-9) and it’s inductance is 2.85 mH. The woofer coil pictured in this web-site, #7, is physically identical to the woofer coil in my early AR-3a.

In an earlier post Steve F. reported that AR, in response to his letter, stated the crossover frequency printed in the literature was changed from 575 to 525 Hz “to reflect changes made at an earlier time.” No reason for this change was given. If my observation about the inductor differences were correct, it would explain HOW, but not why this change was made. Either one of my speakers sound great alone, but when paired, the difference was discernable. After replacing the offending inductor with a genuine AR #9 woofer coil, the two units sound absolutely identical. Also, a small resonance near 40 Hz, previously attributed to room reflections, seems to have disappeared. This resonance was discussed in another post, but its source was never identified.

Added note: IMO, it is very important, when restoring AR-3a speakers, to make no changes in the inductor or potentiometer properties. One should keep all their properties *exactly* the same as original, if one is to retain the sound that made these speakers famous. Tom Tyson provided the correct L values in an earlier post and they were the values measured in my upgraded crossover. All are wound with 16-gauge, varnished copper wire. I see many, including Layne Audio, recommending the use of very low resistance coils “to increase the woofer output.” IMO, this would destroy the balance between the woofer and mid/hi driver levels. Also, IMO, one should never replace the 15-Ohm level control potentiometers with 15-Ohm L-pads. Although L-pads would present constant resistances to the crossover network, they would not be the same resistance as when the mid- or hi-pots were set to their design positions. Thus, the upper-mid crossover frequency, and the lower-hi crossover frequency would be constant but forever incorrect. With L-pads the intended crossover frequencies would not be achievable, unless the level controls were set to zero. Should you be new to electronics and not know the distinction between an L-pad attenuator and a potentiometer, please view, for example:

http://www.bcae1.com/lpad.htm

One last comment—the crossover rebuild described on the human speakers website

http://www.humanspeakers.com/

contains numerous technical and component value errors.

[Guest Bret]

>Added note: IMO, it is very important, when restoring AR-3a speakers, to make no changes in the inductor or potentiometer properties.<

But to which value?! You had two AR units with different components!

I imagine that this is probably more likely in the 3a than most since the 3a was in production for so long.

In any case, the detective job you did, the measurements you took, and the repairs you made are all very interesting. Considering all we've learned it seems that restoring these old speakers is a little more involved than polishing the cabinets and getting drivers that work.

Thanks for the explanation of "what happens" when you change the inductor values and use non-original level controls.

Bret

[soundminded]

First of all, congratulations on your remarkable hearing.

Maybe the reason they lowered the crossover frequency from the woofer to the midrange was in recognition that the woofer response at the high end of its range is a little on the rough side and that they could get better overall response by pushing the midrange driver a little harder at its low end.

Thanks for the L-pad calculator. I'm sure I'll use it a lot for choosing fixed resistors for midrange and tweeter attenuators. BTW, while I know it goes against the grain for many people to change the original design even slightly when they do a restoration, do you think an L-Pad on the tweeter and midrange would give better overall performance. After all, doesn't changing the R in the circuit alter the Q and the F of the crossover? If I were to try it, I think it would be better to start by inserting the equivalent R for the circuit at the indicated "flat" setting as a substitute for the potentiomenter. This is undoubtedly the optimal value for overall performance of the crossover network. Then you could use a 4 ohm L-pad on the tweeter itself as an attenuator. Any thoughts?

[johnieo]

Bret and Soundminded:

re: which to replace? I *assume* that the correct woofer coil is 2.85 mH, as that is the number always quoted by ex employees, or measured by a friend who is into these speakers. There are only two reasons that the 1.9 mH could have been installed. Either it was used in early units, or the assembler mistakely grabbed the wrong coil. I surely don't know if AR made a a minor change, or it was an accident--that is why I posted the note!

Who would have imagined that 46 years later, anyone was fussing over such a detail, let alone applying a bit of polish whilst listening to their great sound! The fact that we are is a tribute to their intrinsic quality.

re: good ear. I can assure you that the only reason we could detect any difference was because all other factors, location, etc, were the same--direct comparison. My wife's first degree was in vocal performance. She is a contralto, who also happens to have perfect pitch and is more sensitive to subtle differences than I ever will be.

re: fixed resistors. I would assume that most users would rotate pots only a small degree from their "flat" setting? Thus the crossover f would not be significantly altered. I would not use the L-Pad calculator, but simply on the mid install a non-inductive 3.25 Ohm in series and a (15-3.25 Ohm) in parallel with the driver. Has anyone with modeling software ever analyzed this in detail? Also take note when re-installing pots that there is a little arrow on the end of the shaft. If the little arrow (the flat setting) is to have any meaning relative to the dot on the cabinet rear, then one must first set the wipers to the correct terminal B to terminal 1 resistances, before locking the pots in position! (B-to-1 values are given on the archived AR-3a schematic.) I almost forgot that detail.

Opening loudspeakers is like opening Pandora's box! :-)

Cheers,

[johnieo]

We have uncovered the likely cause for the improper AR-3a woofer inductor coil found in my unit. Stephen at Layne Audio suggested that the smaller size of the offending coil implied that it was destined for one of AR's two-way speakers that had a higher crossover frequency and that it was installed in error. Now, in the Forum Archives, under AR-2x upgrade, the author writes:

"... since I really don't know what the inductor value is at present (they seem to be marked with a "7" in black magic marker) ..."

Voila, the coil I found was intended for the AR-2x not the 3a. However, it is apparent that this happened more than one time during the 3a manufacturing cycle! Problem considered solved!

Also, regarding the issue of replacing level control potentiometers with L-pads--Stephen indicated that he recommendes subsituting an 8-Ohm L-pad for a badly-corroded 15-Ohm potentiometer. A quick calculation says that this is reasonable, as the net resistance between potentiometer terminals 1-2 is reasonably close to 8 Ohms. e.g. 3.25 + (mid-driver in parallel with 16-3.25). My original concern was that it would be inappropriate to subsitute a 16-Ohm L-pad for a 16-Ohm potentiometer.