Interesting TV Switch Mode Power Supply

                         Q1 switching/reg
                            transistor
                  (+)            ---
               +---+-------------| |---------------+-----------+
         +--+  |  _|_            ---               |          _|_
   AC >--|  |--+  --- C1          |                |    diode /_\ D2
         |  |      |              |                |           |
   AC >--|  |------+---+    +-----+------+         |           |
         +--+     (-)  |    | Reg. Drive |         C           |
          D1           |    +-----+------+         C choke     +---- gnd
                       |          |                C          _|_
                       |          |                | +135 cap --- C2
                       |          |                |  I--->    |+
                       +----------+----------------+-----------+---> + 135

Although at first this appears to short out the line supply, when drawn like this it turns out to be a valid switching regulator:

• Q1 is driven by a pulse width modulated signal a the horizontal rate.

• Q1 turns on putting 150 V across choke. Current ramps up in choke - more or less linear until saturation which should not occur. This time increases with increasing load.

• Q1 turns off. Since current in an inductor cannot change instantly, current continues to flow, now through D2, C2, and +135 load. LCR (R of load, diode) time constant - charges capacitor and powers load.

It would appear to fail and run away under the following circumstances:

1. Inductance is too low and choke cannot store enough energy even at high duty cycle to supply load. Too high a duty cycle and core saturates at which point transistor blows up.

2. Inductance is too high relative to switching frequency so that choke does not have time to discharge (its current) before next current pulse - DC current will just keep increasing until core saturates. This could only really happen if the switching frequency were too high for some reason unless someone changed core material or something like t.

3. Load is too great due to fault elsewhere.

Unfortunately for reliability and troubleshooting, many modern power supply designs have similar failings.

When attempting to diagnose problems with these types of circuits where the natural outcome of a fault is for one or more expensive parts to fail catastrophically, it is wise to either use a Variac to bring up the input voltage slowly and carefully observe the behavior hopefully before too late or put a load in series with the line such as a 100W light bulb to limit the current (though this will change the behavior in various ways).