|
|
|
|
|
|
If feedback were not a necessity in this type of amplifier, I would recommend the inductive components. But the phase shifts inherent transformer and inductor use could turn the amplifier into an oscillator. The Cathode Follower has the advantage of reducing the size of the coupling capacitor needed (although, the cathode directly attaches to the input grids of the output tubes, the Cathode Follower itself still sees a coupling capacitor at its grid). It also fights the output tube's grid from pulling up the grid voltage at input overload. So, why not use a Cathode Follower directly to grid 1? You certainly could, but then you would have to make sure not to remove or jiggle any of the Cathode Follower tubes while the amplifier was in use. Furthermore, always try to use as little extra circuitry in the signal path as possible. Besides, the first solution, using a low value grid resistor, could offer some extra advantages.
Instead of using one large valued coupling capacitor (2 µF) and one low valued grid resistor (20k), we could use one smaller coupling capacitor and one higher value grid resistor per output tube. In other words, each output tube would find its own individual capacitor coupling network. This would certainly be more expensive but worth it.
|
|
|
|
|
|
|
|
|
|
The advantage of this arrangement lies in the added protection it offers the output tubes in the event of a tube failure. For example, if an output tube develops a short from its grid to its cathode, then only this tube will conduct wildly and glow red, as this fault cannot spread to the other output tubes because of the DC blocking performed by each coupling capacitor. Additionally, this setup allows for setting the biasing each tube individually, lessening the need for tube matching. (Furthermore, I would prefer to use smaller valued capacitors, as I distrust large valued capacitors and worry about a poorer high frequency limit due to the increased inductance that comes from the additional windings.)
The EL509 data sheet states that its grid 1 sees no more than 100k resistance to the bias voltage source. If this guideline is not followed, we risk having the output tube break into positive grid bias. The 100k limit must be divided by number of output tubes used per bank. For example, if a bank of tubes consists of five EL509s and they all share a common grid resistor, then 20k becomes the limit for this resistance, or if each output tube has its own coupling capacitor network, the total effective paralleled resistance is 20k. This resistance is the load impedance that the phase splitter stage must drive.
|
|
|
|
|
|
