Bi-amping the AR-3a … IMPRESSIVE, but … WHY??

[onplane]

It’s going on 3 days now that I bi-amped my AR3a’s and everything is still running cool as can be and the 3a’s never sounded so good. The difference in sound is really impressive!! (The mechanics of doing this are in the thread titled "Bi-amping the AR-3a cheaply and non-invasively”)

That thread is getting so long and hard to navigate that I’m starting this thread to discuss WHY/WHAT causes those significant changes. To begin, thought I’d share the well thought out comments from others:

Roy said:

“I was simply suggesting that you have altered the tonal balance of your speakers, which is the same thing equalizers (of various types), and other effects devices do. I highly doubt the "dramatic" difference you hear is strictly due to the elimination of intermodulation distortion. You have simply changed the frequency response of your speakers.”

My response was (Jerry said):

“Frankly, I really believe the differences I hear has more to do with the electronics than the speakers. I seriously doubt that bi-amping has changed the frequency response of the speakers.”

Richard said:

“Roy,

I can't disprove your claim but I can offer anecdotal evidence TIM might be at play.

Unlike Jerry, I opened my speakers and made a modification to the crossover while I was in the process of replacing the old caps. unlike the AR9, the AR90 was not designed to be bi-amped. But at the same time, both speakers use a similar design 4 way crossover with the bass crossover point at 200 hz. Modifying the crossover to break out the bass drivers was extremely easy as the bass section had its own wires running directly from the speaker terminals. A careful layout job with masking tape, a good router bit and patience allowed me to install a 4 post terminal cup allowing the speaker to be bi-amped.

The initial results bi-amping with identical amps (Heathkit AA-1800) proved to be of value. There was an improvement, but not that much, and the sound stage was a bit disjointed. Each amp was still reproducing everything and both amps would still run hot with heavy bass. (Ray Montford; Shed your Skin)

The next step.

I ordered a Marchand XM24 crossover as a kit. The high freq crossover point at 100 hz, the bass at 400hz. This overlap (one octave over and under) the passive crossover prevents the active crossover from interfering with the passive crossover.

The result was absolutely stunning. Everything Jerry listed as an improvement and more was noted. Improved mid clarity, cleaner, better defined bass that actually sounded like it went lower. It doesn't, its just better defined. We hear faint high freq instruments that previously weren't audible, such as triangles. Add to the list an extremely high WAF and it was a winning idea.

Why the dramatic change? I can only guess it was a reduction in TIM.”

Ok, so that’s what’s happen so far. If I missed anything, please feel free to jump in.

Today I ran some more “experiments”. If you remember, the bass tone control on the mid-range/tweeter amp (Pioneer) is adjusted so that bass is was completely off (i.e. little to no bass frequencies going to the power amp in the Pioneer). Theory is that IM increases significantly when we attempt to amplify different frequencies that are far apart … simultaneously. Direct quote from that bi-amp website:

“The effects of intermod are most noticeable when one of the frequencies is much lower than the other, and the high frequency signal is actually modulated by the low frequency.”

Well today, I turned the bass control fully ON! End result … a very slight loss in clarity!! The music was a jazz trio with heavy double bass. When I say a “slight loss in clarity”, you really have to listen close to hear it. Sooo, what’s going on here? Clearly the improvement is NOT simply due to IM.

Well, I have a new idea. I’m now wondering whether under the single amp configuration if the rail voltages become somewhat “unstable” due to the heavy power drain caused by the “power hungry woofers”. If you think about this from the perspective of the last stage in the power amp, on a periodic basis we must send out massive amounts of low frequency power (whenever a double bass string is plucked). This must stress the rails and I assume shortly thereafter, they’ll recover with the addition of more current through the bridge rectifier. My point is during this period of stress, it must create havoc with the high frequencies. These fluctuating rail voltages must introduce significant distortion in the high frequencies.

Well, if this is the case, how do we explain that mixing the lows and highs in the Pioneer amp doesn’t create the same distortion? My answer is the Pioneer, under the bi-amp configuration, never sees the loads of the woofers. Further, if you recall, I never opened the boxes so the xover circuits are intact. With the xovers intact, what the Pioneer sees is the mid-range and tweeter capacitively coupled to the output stage. In short, at low frequencies the impedance could be in the 1000’s of ohms range … no load whatsoever and no impact on the rail voltages!

Further, this might also explain why some people see dramatic results from bi-amping while others don’t. Specifically, those folks with “monster” power amps, might not see as dramatic a difference, because their single amps are better able to handle the stress.

Now the AR1500 is NOT a monster, but it is rated conservatively at 100 ohms RMS into 4 ohms. (I suspect the real differentiator will be the size of the filter capacitors on the rail voltages in the last stage. Bigger filter condensers allow the rails to handle more stress, assuming of course that the power transformer can re-supply quickly.)

In any event, one way to test this new theory is for someone, who normally uses a “monster” amp with their AR3a’s, to attempt my cheap bi-amp scheme and report back what differences, if any, you hear.

Does this make sense or am I off in the weeds again??

Regards,

Jerry

[rrcrain]

You proposed some decent ideas, but here's a few reasons they may not be totally correct.

Without a crossover or another means of restricting frequency distribution to the high freq amp, it's still reproducing everything sent its way. It may not be driving a woofer, but the tweeter crossover is having to disipate the bass energy the amp produces. What the amp is not seeing is the counter EMF produced by the woofer motors. Perhaps a reduction in CEMF and TIM both come into play here?

My two amps are conservatively rated at 250 watts per channel. I traded emails with a retired heathkit engineer a few years ago who was involved with the Pro Series line and according to him, the AA-1800 was rated at 250 watts to meet FTC burn in requirements. Since the amp was a kit rather than a finished product, they downrated it from 300 to 250. It's rails run at plus and minus 100 VDC and have a pair of very healthy caps in the power supply. Design wise, its based very closely to the Leach SuperAmp. Given the amps design, I have serious reservations suggesting my improvement was due to the rail voltage becoming more stable when biamped.

[onplane]

Richard, your suggestion of counter EMF coupled with instability in the rail voltages sounds pretty good to me.

>You proposed some decent ideas, but here's a few reasons they

>may not be totally correct.

>

>Without a crossover or another means of restricting frequency

>distribution to the high freq amp, it's still reproducing

>everything sent its way. It may not be driving a woofer, but

>the tweeter crossover is having to disipate the bass energy

>the amp produces. What the amp is not seeing is the counter

>EMF produced by the woofer motors. Perhaps a reduction in CEMF

>and TIM both come into play here?

Well, I don't agree that there is any significant power dissipation of low frequencies in the tweeter xover. If you do the math, the impedance in the tweeter xover at 30 Hz exceeds 5,000 ohms! The xover is really blocking low frequency current. Consequently, there would be insignificant stress on the rail voltages for the tweeter/mid-range amp.

Now, the counter EMF caused by the woofer is another issue entirely! In most high quality amps, there is a blocking circuit to prevent the CEMF from back feeding. However, it doesn’t have to back feed! The CEMF and all its harmonics are feeding DIRECTLY into the mid-range and tweeter. Clearly, this is a potential source of distortion.

Under the bi-amp scheme, the CEMF goes nowhere! I think I mentioned that another way of looking at bi-amping is that I’ve turned the 12 inch drivers into subwoofers and the mid-range/tweeters into very, very high quality satellites.

>My two amps are conservatively rated at 250 watts per channel.

>I traded emails with a retired heathkit engineer a few years

>ago who was involved with the Pro Series line and according to

>him, the AA-1800 was rated at 250 watts to meet FTC burn in

>requirements. Since the amp was a kit rather than a finished

>product, they downrated it from 300 to 250. It's rails run at

>plus and minus 100 VDC and have a pair of very healthy caps in

>the power supply. Design wise, its based very closely to the

>Leach SuperAmp. Given the amps design, I have serious

>reservations suggesting my improvement was due to the rail

>voltage becoming more stable when biamped.

The rails on my AR1500 are plus or minus 43 volts. I still have this nagging suspicion that even at your 100 volts when suddenly we need to send a large voltage swing to a 4 ohm load, there is serious stress on the rails. For example let’s say we are going to send an 80 volt peak to 4 ohms. That’s a current drain of 20 amps!!

I’ve just got to believe we are going to see some degradation on the rail voltage for a short period and this could cause distortion of just a volt or so on the high frequencies.

In any event, Richard, a combination of IM, rail instability and CEMF are probably all at play here.

Regards,

Jerry

[rrcrain]

I can't argue your last point as the math speaks for itself, and watching the 120VAC swing when the amp bass amp is working is almost entertaining.

[johnieo]

The results you hear, IMO, cannot be uniquely attributed to the cause you suggest, based on the information provided. I hear crystal clear treble in my AR-3a driven with an Adcom 5400, after recapping both mid- and high-range x-o with Zen polypropylene caps and recapping the woofer x-o with polpropylene capacitors containing a half-ohm series resistor to simulate the ESR of the original NPR. The crossover point for the low frequency driver is 525 Hz; it does decay at a slope of -12 dB/oct. The very flat response of the speaker depends on those slopes. However, if your woofer x-o contains an old or poor electrolytic capacitor -- one with a very high ESR -- it will prevent the woofer signal from decaying at its design slope. Perhaps it is that signal which is being attenuated by an active crossover or removed by biamping?

It is possible that you are hearing differences in how old amplifers are being loaded. Would you hear the same differences with two new amplifers that were known to be operating within their specifications?

[ninohernes]

>The results you hear, IMO, cannot be uniquely attributed to

>the cause you suggest, based on the information provided. I

>hear crystal clear treble in my AR-3a driven with an Adcom

>5400, after recapping both mid- and high-range x-o with Zen

>polypropylene caps and recapping the woofer x-o with

>polpropylene capacitors containing a half-ohm series resistor

>to simulate the ESR of the original NPR. The crossover point

>for the low frequency driver is 525 Hz; it does decay at a

>slope of -12 dB/oct. The very flat response of the speaker

>depends on those slopes. However, if your woofer x-o contains

>an old or poor electrolytic capacitor -- one with a very high

>ESR -- it will prevent the woofer signal from decaying at its

>design slope. Perhaps it is that signal which is being

>attenuated by an active crossover or removed by biamping?

>

>It is possible that you are hearing differences in how old

>amplifers are being loaded. Would you hear the same

>differences with two new amplifers that were known to be

>operating within their specifications?

Slightly off topic, what value cap did you use in the woofer x-o? If I remember correctly, the original value was 150 uF which is an extremely difficult value to find. This is the one crossover component that I left original in my 3a's. I could not find the correct value. I would like to replace those ancinet bricks! I am sure that they are hurting the response of the speaker, and wasting power.

[johnieo]

> what value cap did you use in the woofer

>x-o? If I remember correctly, the original value was 150 uF

>which is an extremely difficult value to find.

I believe we paralleled a 120- and a 30-uF; probably Solen. Dayton has several possible combinations; 100+50 seems to be the cheapest.

[Rlowe]

Try this URL for the 150uF poly.

http://www.madisound.com/cgi-bin/index.cgi...d=2446461.15313

[onplane]

John, you might be just the guy we are looking for!! I’ll return to that thought in a minute. In the meantime, just to recap the differences:

1. far better transient response

2. far better stereo imaging

3. far more volume at the same power output

4. amps running much cooler than before

Now, my current theory is that all of the above is due to three things:

1. significantly reduced instability on the rail voltages on the mid-range/tweeter amp

2. NO counter EMF from the woofer feeding directly the to mid-range/tweeter (actually this is Richard's idea)

3. slight reduction in IM by focusing the amps on a narrower audio range (this is really a very small improvement)

Now, as for the condition of my xovers … they are original. However, I ran two experiments this afternoon. On the AR1500 (amp driving just the woofers) I turned the treble tone control to full on and NOT a hint of addition high frequencies was heard. In short, the low pass woofer xover appears to be working perfectly. I mean, I sent one ton of high frequency signals to the woofer and it rejected everything!

Then I did the opposite. I cranked the bass tone control full on for the amp driving the mid-range/tweeter. Result … no hint of additional base, so once again the high pass xover appears to be working on the mid-range and tweeter. Both of these experiments lead me to believe that I’ve made no fundamental change to the frequency response of the 3a’s and that the xovers are doing their job as intended.

Now, John, the amp issue you raise is something we need to explore. What are your rail voltages on the final stage of your Adcom 5400??

One my of theories is that the rail voltages are “unstable” for the short time after we send out massive low frequency power. What we need, John, is someone like you, who is accustomed to listening to your AR3a’s through a modern, high power amp, to try my cheap, non-invasive, bi-amp scheme and let us know what you hear.

John, do you have an old, low power amp/receiver lying around?

Regards,

Jerry

[ninohernes]

>Try this URL for the 150uF poly.

>

>http://www.madisound.com/cgi-bin/index.cgi...d=2446461.15313

>

>

Thanks!