Ever since my old Magnavox died, I had searched high annd low for a similar amp, in spite of the fact that I had been able to duplicate the sound with solid state. I finally found one with at least a similar power output and output tube compiment (6BQ5). It even used 6EU7's but for the phono stage. I however want to take it to the point where the old Maggie was. I had done many mods and it sounds alot better than originally, but I really want that Magnavox sound.
Every other amp I can find uses either the single tube phase splitter, or the differential pair phase splitter. This causes some harmonic cancellations to occur. This is not necessarily a bad thing, but I would like to maintain a balance of harmonics. I believe that the one I am using as pictured below will do so by the fact that both halves of the output will get strong second harmonic content. In the final stage much of it will be reduced, but I believe that it will be closer to the "ideal" ratios (psychoacoustics and music) that make the sound most musical.
This phase splitter is called a paraphase or anode follower. It takes the signal amplified by the first tube, and inverts it unamplified. It does this by virtue of the 1:1 NFB loop. The signal of that tube is taken from the other side of the capacitor coupling it and the pentode. The value of the resistor is the same as the grid resistor and the coupling resistor. This ensures unity gain. The benefit here is that unlike a cathode follower gain is unity, not less than unity. It also inverts the phase of the signal so the final output will be push pull. I haven't yet figured out the values for the components, but it will use 6EU7's for the pre-drivers, and 6BQ5's for the output.
Funny thing about this design in the
old Magnavox, when I pulled one of the 6BQ5's out the sound was the same
but a little less powerful, but when I pulled the other one out (putting
the other back in, of course) the sound became muddy. Weird.
UPDATE - 9/17/99
I have completed the design and assembly of this amplifier after a lull in activity. I was also waiting for some new tubes from one of my favorite suppliers Triode Electronics. Actually the fault was mine, because I just ordered them last week. I also just ordered the filter choke. The trouble was in selecting the rectifier tube. Most tubers prefer the 5Y3 rectifier. I tried the 5Y3 and the 80 and 5Z3 tubes. All of these dropped too much voltage. I got a maximum of 277 volts before the filter choke and 250 after. This was not a good thing for me. I designed the circuit with 300 volts in mind. Installation of the 5U4 yielded the desired voltage. Actually, I got 315 volts with all tubes installed and hot. The problem is due to the tube's dropping too much voltage for the current. According to two different sources (the RCA tube manual and Nostalgia Air's tube substitution guide), the 5U4 can handle a peak current of 1 amp with transients of up to 4.6 amps. All others go up to about 750 milliamps. Voltage drop at a continuous 245 milliamps is 44 volts. I am continuously drawing about 150 milliamps.
Now for the circuit and values.
As you can see, I seem to have gone nuts with the 470k's. Well, the resistors are there to make the phase inverter triode amplify by a factor of 1. So all it is doing is inverting the signal while keeping it the same level as the first triode. The original Magnavox amplfier called for the second triode amplifier to actually amplify a reduced signal. I felt that this way made for an insured equal but opposite signal. To get a voltage divider to actually divide the signal corrrectly would be a test in accuracy. I wasn't about to go through dozens of resistors to find four exactly matching ones. On the other hand, this one would call for more accurate resistances also. But I figured, what the heck. Besides, I also believe that if I had gone with the divide by and amplify method, the characteristics of the phase inverter will be different. I wanted to maintain the same input impedance as the first triode so as to have near perfect symmetry, save for the likely (very minute) delay the second triode may introduce to the overall signal.
Let's get an idea of its operation by following the signal path. From the input at the left, the signal gets amplified by the first triode at the gain set by inverse feedback. I set it to (15k/1.5k)+1. or 11. So an input signal of 1 volt will come out of that tube as a signal of 11 volts but inverted from the input. It gets fed directly to the top pentode through a capacitor, but it can only amplify most of it, since it is biased as class AB. The pentode amplification is about 50 times more, so the final output is about 550 volts (actually, since the overall amplification is set by global feedback, taken from the 16 ohm tap of the output transformer, the actual output of the tube may be much less. Let's just say that open loop gain, without feedback, could be about 1000). The signal also gets passed on to the second triode through a 470k resistor from the top pentode's grid (after the blocking capacitor) to the input of the second triode, which is set to amplify by 1 only. This tube's amplification factor is set by the 470k resistor from its output (after the blocking cap there). The signal is inverted again and feeds the bottom pentode.
Now, since I maintained a consistent input impedance by using 470K ohms at the grids the signal should be symmetrical. Actually, this did not quite happen because I failed to consider the effects of the other resistors in parallel with the grid resistor. Also the gain of the first triode suffers with the load. So let's tear this apart further.
Let's calculate the open loop gain (without feedback) of the first tube. Its cathode resistor is 1500 ohms. The transconductance of the tube is approximately 1600 uS. The plate resistance is approximately 63500 ohms. Using the transconductance the cathode resistance is 1/Gm or 1/0.000016=625 ohms. this needs to be in series with 1500 ohms. Total cathode resistance, therefore, is 2150 ohms. Now, plate resistance is internal resistance in parallel with plate resistor in parallel with load resistance. The load resistance here is pretty complex, because we have the directly connected grid resistor of the pentode, then the connecting resistor of the phase inverted in series with its grid resistor, in parallel with the feedback resistor in series with the grid resistor of the other pentode. Now, it is possible to consider the effect of the inverted feedback on the input impedance of the second triode, which might be unusually high due to the large amount of feedback. We may be able to neglect this and use the connecting resistor in series with the grid resistor of the second triode in parallel with the gridresistor f the first pentode alone. With all that confusion lets just say that the load resistance is 470K n parallel with 940K, or 313k, in parallel with 220K, or 129k in parallel with 62500 or 42k (whew!). Now we divide this by the cathode resistance to get gain. 42000/2150=19.
The pentode amplifies an additional 60. How do we get this? Same way. The 6BQ5's transconductance is 11,500 uS. The plate resistance is 40k. The load resistance is about 6000. The effective plate resistance therefore is about 5200. The cathode resistance is about 87. So, 5200/87=59.77. Total open loop gain therefore is about 19X59.77=1135. Wow! Of course, it never gets there because the power supply is only good for up to 300 volts. I would need 1200 volts if I wanted to get that much gain. So we get clipping galore. Also the transformer reduces the output voltage due to turns ratio. So output voltage may only be about 20 volts for a 1 volt input. So we use NFB to cut it down to size. But with so much gain and NFB the bandwidth increases to nice proportions and distortion reduces to acceptable levels. Noise also reduces.
OBJECTIVE TESTS
I did none except for voltage measurements. Most of the voltages were as predicted. I say most because at the plates of the triodes, I got different readings. The plates of one of the triodes in each channel are 145 volts exactly. Just what I desired. However, the second ones were slightly off. One was about 149 and the other was about 153. I suspect one of three things. One is the possibility of different values of cathode resistance. While I used 5% tolerance resistors, the resistance value may be such that it causes the difference. The other is that the triodes within the tube may actually be slightly different. The third may be that the resistors at the grid of the second triode (I did not actually take note of which triode was 145 volts and which one was different, but I will tonight) cause it to leak away some current causing the tube to conduct less, which raises the plate voltage.
SUBJECTIVE TESTS
This is the part I always skip to when reading the review rags. I first tested the thing on a pair of cheap speakers just to make sure it worked at all. The sound was astonishingly good through these pieces of crap. From mid-bass to the high end it was clean and revealing. I then hooked up the finished product to my high efficiency mini console speakers. I got good satisfying bass. However, the midrange seems a bit on the sterile side. It is good but I get a strong upper midrange. I have been used to the midrange I get from an SE amp using older 6SL7's. It is fatter. However, the upper mid intensity is actually what I grew up with from the Maggie. On the other hand, I played "It Was a Very Good Year" off of the CD September My Years by Frank Sinatra and got that sound I had with the Maggie, So I was satisfied. It brought back that memory.
I then brought it into the house and attached my medium efficiency speakers to it. There I play LPs, CDs, VCR and DVD movies. I first played a DVD. I do not think that my 100 watt per channel amp had as wide a dynamic range. I was really taken aback by the punch and power this thing put out without breaking up. It has supposedly an 18 watt per channel output. I first played Star Trek VI, from the point where Praxis blows up to the part right after Sulu's ship the Excelsior passes through the wave. WOW! I then played half of Batman (the first one. Jack Nicholson. Both these films on DVD) I heard stuff I never heard before. Sound effects, footsteps, people talking in the background, and even lines I could hardly make out coming right out and identifying themselves. Man. I may never put my SS amp back in! It's 100 watts per channel!! (Just think, even Magnavox needed to replace their 15 WPC tube amp with a 100 WPC solid state. Hmmmm)
The true test however is playing LP's. I have an LP that was made when stereo was new on LP back in 1959. It is on a label called Command Records. It is called Persuasive Percussion Vol. 1. This record will make a Close 'n' Play (remember them?) sound good. My lord, what a sound. All I can say is that the instruments were there in my living room. The bass is perfect, the highs are perfect. The midrange is perfect. I am satisfied. Now to see if I can improve it. ;-)
I made a quick makeshift preamp using the same values as the first triode circuit in the power amp. The exception is that I used a 330 ohm resistor in series with a 100uF capacitor to bypass the cathode. This allows for more gain with some degenerative feedback. It uses a 6EU7 also. I worked in the phono preamp which was already there that also uses 6EU7's. Love those 6EU7's!!
My opinion of the 6EU7 is high because I get a wider range than I do the 12AX7's. The 12AX7 is a good tube, to be sure. They tend to sound warm. But to me they also seem to do so at the expense of deeper bass. Perhaps there is a way to design with them that makes it sound the way I would like. But the 6EU7 does it so readily.
Many say (including the RCA and other tube manuals) that the 6EU7 is a 12AX7 with different pin outs and a strictly 6 volt filament. I say, "Oh yeah? Why does it behave differently with the same circuit components?" I have designed an amp with both and found that the 6EU7 had different voltages with the same exact value components as the 12AX7. So to me they are both different tubes. They even sound different with the same voltage and current levels, even though attaining those requires different bias resistors.
It is truly unfortunate that the 12AX7 became so popular. If the 6EU7 had at least equal popularity, many would be able to see what I mean for themselves. As they stand they are at least double the cost. But to me it is worth it. It is as I always say, the book isn't always correct. Well, I say it mostly to myself, anyway.
UPDATE 9/27/1999
I have recently and thankfully recieved a copy of the original Magnavox chassis schematic kindly provided to me by Jon Staebler. I had it for about two days before deciding to actually redo my amp. That took about an hour. As soon as I finished redoing my amp I got the delivery of 6BQ5's from Ned Carlson at Triode Electronics. Great timing. Maybe an omen? It was meant to be?
I made only three modifications to it to comply with what I currently have. Actually, If I include the power supply differences, then there are four differences. This is the schematic:
I have a full description of both the schematic and the sound quality at this page. All I can say is that it puts my SE amp to shame. The sound is definitely tubes, but has an air and openness that seems to transcend the limits of the output transformer. The bass is deeper than anything one could hear. Bass notes sound like bass notes all the way up the scale, and have such a texture to them that sounds like your ear is right next to the string.
I tested this amp with both CD's and LP's, and true to the original, if there is bass in the music it will be there. If not, it won't. All of the other tube amps I have built so far cannot do this. They all produce bass. Midrange is clear, distinct, there. Voices are true, not fat or colored, like the SE amp I made produces. What I mean is that you get the full sound of the tube, but it is not dominant. Dynamic range is awesome. Imaging is all around. One becomes involved with the music. This is the way the old Magnavox sounded. Truly "magnificent".
To really put this beast to the test, I played that LP my father got with the purchase of the original Magnavox back in 1961, Command Records' Persuasive Percussion. As the name stipulates, percussion is the key here. It was recorded in 1959 through the top of the line equipment of the time. It rivals CD's today. I would say it puts CD's to shame! If it weren't for those darned pops and clicks...
Anyway, there are a variety of percussion instruments in this recording. Bongos, drums, scratchers (those spanish banana shaped things), xylophones, bass drums, blocks, chinese bell trees, and a host of other things. This recording was designed to put a stereo system through the ringer, while providing some good entertainment and sound quality. Some (actually, I think all) of the intros of music were switched from left to right for channel balance. Some were so "bangy", or fast attack for amplifier control testing. The sound of this amp, for reasons still unknown to me, is airy and precise. The kind of sound that brings out the life of the music without the apparent coloration of even order harmonic euphony. While the euphony is nice, realism is even nicer.
I also played a direct metal master recording of Beethoven's Sixth symphony. If ever there was a test, the third movement here is the piece de resistance (or however that is spelled) of tests. Every instrument, just about, could be heard throughout the range that it was played. Unlike many amplifiers I have heard, the tympani was powerful throughout its range. Even some recordings where the tympani is reduced (analog mastered LP's) in amplitude come out audible, making the overall music that much more enjoyable. This was always typical of this amp. Some amps emphasize the lower tympani, while others emphasize the upper. I had my son, the violin prodigy, listen along with me. He says it was the best he had heard yet. It was clean, full, real.
I do not know what it is, the way they designed the paraphase inverter, or the way NFB is employed (at a divider), but this amp is the best sounding tube amp I have ever heard. And I have heard alot of them.
The Magnificent Magnavox. I believe
it! If you are a purist, I strongly suggest trying this design out for
yourself.