the capacitor coupled output stage. And the cathode follower's cathode works better at keeping any positive grid draw under control, because of the degenerative feedback at its cathode; the directly coupled plate resistor is much more easily is pushed around. Below is a Loftin-White amplifier benefiting from the addition of a cathode follower. Once again, the lesson here is to consider what will happen when a tube fails or is cold, jiggled, or even removed.

Safety Issue two:
What is a safe rating for a part?
     Given a 300 B+ power supply voltage and a 5687 running with 200 volts on its plate and  with 10 mA of idle current, what is the safe wattage value for the 10k plate resistor? The math seem simple enough: 100 volts against the 10 mA equals 1 watt. But what if? What if the wall voltage climbs 10%, increasing the B+ to 330 volts, or what if the resistor becomes covered in a wool blanket of dust, or what if the tube becomes gassy and greatly increases its conduction? Since we had been running the resistor at 100% of its capacity, how much more can it take? Some are thinking, "Overate by a factor of two and all will be well." Not quite, unfortunately. For example, the 5687 does not share the same pin out with the more common twin triodes, such as the 12AU7 or 12BH7. What if this fact were forgotten and a one of these tubes were plugged into the 5687's socket? The worst case would be the connecting of the plate resistor to ground via the heater element or internal shield. Now let's do the simple math again: 300 volts against 30 mA of current equals 9 watts! The twofold over-specifying was 4.5 times low. If the power supply voltage had been only 200 volts, then the worst case power dissipation would be 4 watts.