Revised Fig 3.113
Changes to the relevant text on page 211 (lefthand column):
The traditional approach is to put a “bleeder” resistor across the storage capacitor, sized to discharge it in ~10s or so. Good enough. But it is not really satisfactory when you have a large-value capacitor, for example one that you would use to store considerable energy to power a short-duration high-voltage pulse generator. Figure 3.113 shows such an application, in the form of a 100uF storage capacitor that is charged to +400V by a dc-dc converter of modest power capability (say 10W), the latter powered from a low-voltage line-powered dc supply (whose output also powers the rest of the pulse generator circuitry).
What you’d like is a bleeder resistor that is connected only when the external power is removed. Figure 3.113 shows a nice way to do that: depletion-mode MOSFET Q1 is held in the nonconducting state when the supply is powered (VGS=-9V), but is sent into conduction (VGS≈0.6V) when the +12V is absent. The particular part listed is good to 500V, has a minimum IDSS of 3A, and costs about $2. (For higher voltage applications you can get depletion-mode MOSFETs up to 1kV.) In this application we don’t need 3A (that would discharge 100uF in just 13ms); but we do need a MOSFET large enough to absorb the stored energy, here 8 joules. From the datasheet’s plot of Transient Thermal Impedance we estimate that the bare part can absorb a 20J pulse without exceeding TJmax. In this circuit, pnp follower Q2 is needed to boost the discharge current, which otherwise could be as little as a few milliamps (set by bleeder R2).