CENTER CHANNEL (PHANTOM)

[DRBILL]

I have a hand drawn diagram by Paul, himself, describing how a phantom center channel could be had using the two 4 ohm taps of a vacuum tube stereo amplifier to a center speaker --one wire from each 4 ohm tap to the two speaker connections (4 + 4 = 8 Ohms).

What is the current wisdom on this? Why is it never mentioned? Were there problems with this?

This is not a philosophical question. I have a vintage model H that I would like to use with my ST70 and Klipschorns without having to buy another mono amplifier.

Father Bill+

[WMcD]

Yes, I have the literature.

PWK was a big fan of center channel. This followed from experiments performed by Bell Labs showing that two channels were better than mono. Further, three discrete channels were even superior. The problem was that the initial research was in the 1930s when recording was only possible with a mono signal.

One demonstration by Bell Labs of three channels was a transmission between two concert halls. One in DC, the other in Baltimore, over phone lines. The transmission was sophisticated and looks like single sideband transmission, to my recall. The orchestra was in one hall. There were three microphones. This was transmitted to a three channel playback system in the other. Yup, all horn loaded.

This is all described in a Bell Labs monograph which was part of the Klipsch Papers.

PWK became involved in "stereo" in the early 1960s. (Don't hold me to that date, it is from memory.) Two channel recording on tape was just becoming available and the stereo Microgroove records coming on line. My reading indicates that a large amount of PWK's work was getting the very best program sources for his K-Horns. He was into everything.

The use of a phantom center channel had been discussed in Bell Lab papers. It is basically just the sum of the left and right channel.

It seems that early on, PWK was quite convinced that Left plus Right as compaired to Left minus Right did not make much of a difference. This might have been from his experience in making stereo recordings on tape. It is difficult to say how long he held on to this belief. It may be that with better recording techniques, and more experience, the superiority of maintaining the addition, came to the fore.

You may realize that connecting the center channel speaker across the 4 ohm transformers outputs is actually feeding the center speaker with Left minus Right. For example, if the outputs on both side of the amp (left and right) are identical, there is no difference in the voltage at the points of connection. Hence the speaker has no signal input.

Remember, the output of the center channel is actually the difference in voltage at its input terminals. Same as a light bulb. If both input leads are at plus one volt, there is no voltage difference, and therefore, no current flowing. I.e. no power.

Let's step aside a moment and talk about the techniques of "bridging amps". There a two channel amp is used to drive one speaker in mono. In that technique, one of the two channels of a stereo amp is fed out of phase, or reverse polarity. The the speaker is connected across the output "hot" outputs. There, because of the reverse polarity, one amp out put is going positive with the mono signal, while the other is going negative with the mono signal.

The power output is often described as potentially double the rated 8 ohm output of the two amps. That is true, if each amp delivers twice the power into a 4 ohm load, as into an 8 ohm load.

We can understand how this works by imagining there is a "virtual ground" within the bridged speaker. A virtual ground means that the voltage is zero, for some reason, even if it is not connected to a reference zero ground.

Let's assume the bridged speaker is 8 ohms, nominally. We know that two 4 ohm speakers in series add up to an 8 ohm speaker. But let us also assume we can consider the point where the two 4 ohm speakers are connected. Therefore, in the model:

Left mono amp output to ( +4 ohm speaker-) to (connection between speakers) to (+4ohm speaker-) to Right mono amp .

You have to take a good look at the model. The connection between the speakers is at zero volts. A virtual ground. Therefore, each amp is looking at something which looks like a 4 ohm load. Each speaker is seeing a 1 volt input. Also, the speakers are fed in phase relative to each other. Therefore they are outputting a mono signal, which is the input to the amps before the inversion of one channel. Essentially, Left plus Right (where here Left and Right are the same, mono input.)

Now getting back to the problem.

PWK's connection across the 4 ohm leads lead to the difference of Left and Right. Or L minus R.

Another article in the Klipsch Papers was one by John Eargle, a well respected writer at the time, and always.

He had a modification to a two channel tube amp which allowed bridging by the center channel speaker across the 4 ohms transformer output taps and yield Left plus Right. To make this bridging work, it isn't stictly necessary to invert (reverse polarity) the input to one of the stereo amps. Rather, things are fine as long as the output is inverted.

The solution was to reverse the input to the output transformer for one channel, just switch the wires in the amp feeding the input to the transformer. Let's say that it is the Left channel transformer. This meant that the Left channel output was actually minus Left. (No "mono issue here, the amps are fed nomally.) Therefore, the center channel speaker was hooked across + R and -L. But recall that the speaker output is the difference between the inputs. So the output was R minus, minus L. A minus times a minus is a plus. So we got L + R in the bridged center.

Errr. What about this problem we've created with the Left channel output. It is reversed in polarity. Very simple. Just reverse the main Left speaker connections so that it is reverse in polarity relative to the amp output. Now it is back in polarity relative to the Left input. There are two inversions, or no inversion.

Why didn't this catch on? Probably because no one wanted to get out a soldering gun and rewire their amp. Also, transformerless transistor amps were becoming popular. Further, some people may have gotten nervous about the fact that the center, bridged speaker presents an additional load to the amp, which may already be taxed by the 4 ohm load (in the bass) of the K-Horn.

I've set up an adding circuit using three Radio Shack 70 volt line transformers. The transitorized amp feeds the inputs (primary) of two. The secondary windings, in series, feed a third transformer, and the output of that feeds the center channel. Gain is adjusted by trying out different taps.

I can send you some info is you give me your postal address. It may take until after New Years to get to it.

Gil

[DRBILL]

Thanks, Gil, for the thorough and clear explanation which jogged my memory and caused me to unpack my large box of Klipsch papers. Sure enough, I found his "Stereophonic Sound with Two Tracks, Three Channels By Means of a Phantom Circuit (2PH3)" (Journal of the Audio Engineering Society, April 1958 v.6 #2) in which he proposed the same circuit that he drew for me on the back of a church service leaflet! There was a slight difference in that the left channel went to the 16 ohm tap, the right channel to the 8 ohm tap, and the center channel to the two 4 ohm taps. In my drawing, he only used the 8 and 4 ohm taps. Interestingly, there is an Appendix to this paper dated June 1960. Paul wrote:

"...(I) disagree with the author of my former paper." (!) "...the "difference" derived center channel signal may fall short of perfection in some instances, although for most stereophonic program material it is indistinguishable from the performance of a circuit with a "sum" derived center channel. Certainly for monophonic signals cancellation takes place so that at least for this special case, the sum derivation is better."

I haven't passed my eyes over this material in over thirty years. It is time I did!

Now that I know that I wont burn up a transformer or suck the brains out of my head, I may try the original idea even though I know that I will settle for the "additive" circuit with a third amplifier.

Many thanks, again, and especially for reminding me to re-read the original papers which had grown dusty.

Father Bill

[WMcD]

I should dig out the papers too.

Let me first say that I usually explain things with Left plus Right as a private convention. Then the Right gets inverted. We wind up with Left minus a minus Right. Of course there is no difference in the end.

For the record, the Eargle circuit shows an adjustable L pad in the feed to the center speaker. It seems to be necessary because PWK had a calculation about the level of the center speaker relative to the flanking speakers. My recall is that the center should be 6 dB down. I'll have to look more closely and try determine whether that is 6 dB down from one channel, or 6 dB down from the combination of the two.

I have a lot of respect for PWK because of his, ocasional, humerous, retractions. Plus sticking to his guns on the important stuff. It is tough be a pioneer, and outspoken, and be right all the time. If you're not in error occasionally, you're being too cautious.

It seems to me that his original thoughts on the lack of distinction in L plus R versus L minus R really came to the fore when stereo mixing on a board became possible for pop recordings.

The difference may have been small when using two mikes in front of a music group. However, with pop groups in a studio, there were a number of pick up mikes. So, typically, lead voice was mixed mono center (that is to say, equal in left and right channels in the recording) and bass might be right only and drums left. Naturally, singer in the center is still the norm.

Therefore, if the phantom channel speaker was Left minus Right, the singer, in playback, would not appear in the center channel at all.

I like my center channel mixer using RS parts. It doesn't add much of a load to the amp, in my rough calculations. In many old and new recordings, the center channel singer is right there, in the middle. Sometimes it is spooky.

Gil

[djk]

A modest proposal:

Use a 15K 1:1 transformer on the input to one channel to reverse the phase. Hook the main speaker to that channel out of phase (net result being that the acoustic output of both main speakers is in phase). Connect the Klipschorns to the 8 ohm taps. Connect the Heresy between the 4 ohm taps. This will give the Heresy a true L+R signal producing an acoustic output within 1dB of the desired level.

A suitible transformer would be the Altec 15335A line bridging transformer. An octal plug in, fully shielded device, these show up regularly on eBay for about $20. Better yet would be to buy two and use one on the input to the other channel too (non inverted). These transformers were designed to plug in to the octal sockets on Altec amps to give them balanced inputs. By breaking the ground connection between the preamp and the power amp most noise and hum are greatly reduced.

Using an input transformer in an unbalanced system is QUIETER than a full balanced system using an electronically balanced preamp driving an electronically balanced amplifier.

http://www.jensen-transformers.com/an/an004.pdf

[DRBILL]

Gil,

From the number of hits on this subject I assume that there is considerable interest --probably because of the new-found admiration of vacuum tube equipment and the possibilities they present.

I found John Eargle's "Stereophonic Localization: an Analysis of Listener Reactions to Current Techniques" in IRE Transactions on Audio (V.AU-8 #5 September-October 1960) to which you referred. On page 170, fig. 3, there is a detail that would be easy to miss in casual observation. The four ohm tap on amplifier "A" is grounded and what normally would be the common is the output. Amplifier "B" is wired in the usual fashion. The center channel (with the L-Pad) is taken from the (ungrounded) common of amplifier "A" and the 4 ohn tap of amplifier "B". The left and right speakers are connected to the common and 4 ohm taps in the usual fashion. This would seem to solve the "additive" problem.

You could not do this with a typical stereo amplifier because the commons are almost always tied together internally and grounded. If you are using separate mono amplifiers for left and right, there should be no problem and it would be a solderless phantom "additive" center solution.

Kids, don't even think about trying this with solid state!

The next thing you hear will not be music!

Father Bill+

[DRBILL]

Dear PJK,

I am certain that your proposal would work. But you must understand that "Old F---s" are driven by the principle of parsimony --the most simple solution rules. PWK himself, when proposing the 4 ohm tap (2PH3) center channel called it "The simplest and most efficient way---." "This is self balancing, (and) uses the least equipment." See where we are coming from?

(Anybody out there remember Rube Goldburg?)

Thanks for your interest in this subject and your interesting contribution.

Father Bill+

[djk]

I tried both of PWKs methods 25 years ago.

L-R does not sound good.

L+R sounds great.

I used a resistive combiner as described in the 'Dope from Hope'. The center speaker was a Belle driven by a Marantz 8. The Klipschorns were driven by an Audio Reasearch D76.

The D76 has grounded 4 ohm taps otherwise I would have used my one channel inverted method.