John
Hi!
1. "Why not remove one of the 100K resistors to decrease the resistance to allow the correct feedback level not so close to ground on the wiper of the pot?"
The feedback circuit is (to an approximation that is good enough for this discussion) in the form of a current source. The voltage across the output of the (correct) 6146 produces a current that will be pushed through the combination of the existing three 100k ohm resistors (300k ohms of series resistance); and, then, through the resistance that corresponds to the combination of the 100k ohm potentiometer and the resistance of the branch leading to the 6C4. The voltage that will be produced across the 6C4 (cathode-to-ground) will be: the cathode input impedance of the 6C4 (roughly 400 ohms) x the portion of the total feedback current that flows into the branch leading to the cathode of the 6C4. In the circuit defined by the current schematic,
almost all of the feedback current will flow into the branch that leads to the cathode of the 6C4, unless the wiper of the potentiometer is close enough to the grounded end to produce a resistance between the wiper and ground that is just a few thousand ohms (or less).
To make the feedback adjustable, without having to set the wiper of the potentiometer so close to the grounded end, you need to put additional series resistance between the wiper and the cathode of the 6C4. Changing the existing 2.2k ohm series resistor to a 100k ohm series resistor will produce the desired adjustability.
You should not remove any of the existing 100k ohm series resistors. Doing so will increase the total feedback current... and, therefore, increase the feedback (which is already much too high). If anything, you may have to add a megohm resistor in series with the existing three 100k ohm resistors, in order to reduce the feedback... but you can cross that bridge if you come to it.
The first step is to change the 2.2kohm resistor, between the wiper of the pot and the cathode of the 6C4, to a 100k ohm resistor.
2. "What does 10db mean? At what frequency is this to be measured and how would you measure it? Is it linear across the audio range? If so, when why do it? I'd guess not."
You can look up the definition of dB on line, or in any "ECE-101" textbook. You don't need to know what 10dB means to fix this problem.
The gain around the feedback loop should be constant (but not anywhere near as big as it is now, unless the wiper of the pot is set almost all the way to the grounded end) at audio frequencies of interest (e.g. 50Hz - 5000Hz). Beyond the top of this audio range, the gain around the feedback loop must drop to less than 1 (unity), as the frequency increases, before the phase shift around the feedback loop increases to a total of 360 degrees (i.e. 180 degrees of additional phase shift, v. 180 degrees of phase shift with negative feedback).
After changing the 2.2k ohm resistor (from the wiper of the pot to the cathode of the 6C4) to 100k ohms, you should add a capacitor between the wiper of the pot and ground. The capacitor's value should be chosen so that the magnitude of its impedance at 5000 Hz is approximately 100k ohms. I.e., the capacitor should have a value of approximately 317pF. You could use a standard 270pF capacitor, or a standard 300pF capacitor. This added capacitor will roll off the gain around the feedback loop, by a factor of 2 for every doubling of the frequency, starting at around 5000Hz.
Even with the added capacitor, the gain around the feedback loop is so high (unless the potentiometer's wiper is set almost at the grounded end), that the rolloff in the gain around the loop produced by this capacitor will be insufficient to bring the gain down below unity at higher audio frequencies (beyond 5000Hz) at which the phase shift around the loop increases beyond a total of 360 degrees.
Therefore, when you adjust the potentiometer (after increasing the value of the resistor between the wiper and the cathode of the 6C4 to 100k ohms)... starting with the wiper at the grounded end... there will be some setting that results in positive feedback/oscillation. You need to limit the setting to a value below that.
If, after making the above changes, the setting of the potentiometer's wiper (to avoid positive feedback/oscillation) is still very close to the grounded end... then you should add a 1 megohm resistor in series with the existing three 100k ohm resistors. After this change, if the maximum setting of the pot wiper is still very close to the grounded end, then replace the three 100k ohm resistors with two 1 megohm resistors (for a total of three 1 megohm of resistors in series, including the previously added 1 megohm resistor, replacing the three 100k ohm resistors)
3. "Why are the plate resistors of the driver different? Is it to make the positive portion of the modulation envelope larger? Is it because you're hanging the feedback loop off of one of the plates and need to be rebalanced?"
The phase splitter does not produce exactly equal (and opposite) currents in the two halves of the splitter circuit. The larger plate load resistor on the right side half compensates for the fact that the audio frequency current flowing in the right side half is somewhat less (in amplitude) than the audio frequency current flowing in the left side half.
Jim has pointed out that an even better design (but probably not needed to be done now) would have been to make the left side plate load resistor adjustable... using 50k ohm potentiometer. That way, you could fine tune the value of the adjustable resistor to make the amplitudes of the output voltages, on each side of the splitter, the same. As it stands now, one has to trust that whoever picked the left and right plate load resistor values did so by measuring the amplitudes of the left and right output voltages... and selecting the resistor values to cause them to be close to being equal... using an oscilloscope to make the measurements.
Stu
What I posted about the phase of the feedback in relationship to the cathode to grid voltage of the 6C4 kept bothering me. I worked on that in my head laying in bed last night and it became apparent that I must have been thinking wrong. When I read Stu's comments, the light came on. Yep, I was wrong. I think we'll take another stab at it with a renewed understanding...except I have questions.
1. Why not remove one of the 100K resistors to decrease the resistance to allow the correct feedback level not so close to ground on the wiper of the pot?
2. What does 10db mean? At what frequency is this to be measured and how would you measure it? Is it linear across the audio range? If so, when why do it? I'd guess not.
3. Why are the plate resistors of the driver different? Is it to make the positive portion of the modulation envelope larger? Is it because you're hanging the feedback loop off of one of the plates and need to be rebalanced?
Questions, questions, so many questions and so little time.
John in Tucson
I can see how this might have confused you (or anyone else, including me).
The cathode of the 6C4 is a negative feedback point.
The grid-to-cathode voltage on the 6C4 is causing current to flow. This current flows through both the cathode resistor of the 6C4 and the plate resistor of the 6C4. The resulting cathode-to-ground voltage is in the direction that causes the grid-to-cathode voltage to decrease. I.e. the grid to cathode voltage being applied to the 6C4 is: [the grid-to-ground input voltage - the cathode-to-ground (feedback) voltage]
When the larger loop's feedback signal (from the output of one of the 6146s) is applied to the cathode of the 6C4, it must have the same polarity as the open loop 6C4 cathode-to-ground voltage... in order to produce negative feedback around the larger feedback loop.
Stu
Turns out that when we compared the phase of the feedback with the signal on the cathode of the 6AU6, it appears that the feedback was in-phase. So, we swapped it out. Then everything looked good until we hooked it back in place and then it was really hosed. Now, maybe I'm wrong, but to me, "negative" feedback means that the waveform coming from the plate of the 6146 will be out of phase from the mike audio on the cathode of the tube. Didn't look like it was right. We switched it and it was even worse. So, for whatever reason, the feedback was disconnected altogether and it behaves normally. That may explain why it was disconnected when Dave got the unit from the seller in the first place. Now...here's the issue: The audio is a bit raspy on peaks with Dave's voice without the feedback. Why did it screw up? Don't know. But the good news is that the mod transformer and the audio chain is good. I even figured out why one plate resistor of the phase splitter is a 51K and the other is 39K. What do you think?