Power supply design with the LM723.

  We had a big problem in this country with designs using the universal LM723 regulator.   Some 25 or so years ago, one of the popular electronics magazines published a design in which the compensation capacitor which should never have been more than 1nF (nanofarad) in "Miller" mode, was quoted as 100nF, TWO HUNDRED TIMES the correct value.

  The design would work, after a fashion, and was blindly followed by hobbyist and "green" (No! Not "enviro") designers in similar magazines over the next 15 years or more with the result that 723s gained a very unfortunate reputation; "hobbyist", but not what you'd use for a lab supply.

  Apart from very poor line interference rejection, the major problem was that the resulting inability to respond to transients (shorted output in particular) tended to defeat the current limit circuitry, feeding massive currents into the sense transistor (within the IC) whose E-B junction was directly across the (fractional-ohm in many cases) sense resistor.   Sooner or later, this gave up the ghost and of course there was now no short-circuit protection so the pass transistor, current sense resistor, rectifier and transformer suffered in turn.   Lots of smoke!

   The correct fix, to those who knew and I had absolutely no success getting any editors to admit to the cock-up, was to use the correct capacitor value (470pF) as in the professional designs.   Voilá!   A well- behaved, rugged, reliable supply.   Using commonsense and including resistance in series with the current sense transistor (base) also helped reliability.   Since this transistor had all three terminals appearing as external pins, it was actually quite easy to substitute, or indeed to add such a protective resistor.

   My point?   Who knows who, and of what competence, designed your bench supply?   If you are using something other than a H-P or Beckman and it is generating turn-on transients, regulating poorly, generating excessive smoke or any other symptoms suggested by the above why not OPEN it, examine and FIX the problem?   The above comments may well apply to implementations other than the LM723.

   Sticking my neck out a bit here but...   Send me the circuit if you like!