Interesting PSU short protection

[iMo]

Long time back (in the era of CB radio) I did a teardown of "NT35 13.8V/3.5A" PSU made by "Avanti".

The PSU is of simplest design, but what was interesting to see was the "output short protection" design. None shunt sensing resistor was used.

Schematics (except the bridge and filter caps) below shows the protection.

The short 50Hz pulses are fed into the base of Q5.

Under normal operation the Q5 is saturated (via R4/R5) thus there are no pulses at its collector (mind its collector resistor R6 is wired at the regulator's input side).

When the PSU's output is hard shorted the Q5 works unbiased (the hot side of R4 is now near gnd level - the base is wired R4||R5 against gnd) and the 50Hz pulses propagate into the base of Q4.

The pass transistor Q1 2n3055 is opened for about 200us at 50Hz when output shorted. Q1's power loss is therefore pretty small. In practice the Q1 wasn't noticeable warm when PSU's output shorted.

[T3sl4co1l]

Short protection, yes; but you need to get there first!

This is a special case of a more generally known circuit, the foldback current limiter.

Obviously, this is not brown-out protection (i.e., against a moderately heavy -- not shorted -- load), and peak output current will still be hFE limited(!).

There are more clever ways to solve this problem -- that is, current sense without shunt.  Monitoring the pass transistor's Vbe for example, usually as part of a dissimilar matched pair.  This is hard to do in a discrete circuit, of course, and notoriously hard even in an IC, as the current limit ratings of many LDOs can attest.

Tim

[David Hess]

That is an awful lot of complexity just to avoid a sense resistor which could be used to make the regulator more stable.  I wonder though if they did it that way so that the regulator could hard start into a difficult load; fold-back current limiting can have problems with this if it is misapplied.  Some early integrated voltage regulators had the same problem when hard starting into a heavy load if their thermal protection was tripped; the solution was to add hysteresis.

Old discrete fixed voltage designs, unlike the 723, often used a separate differential pair to measure the voltage across the current shunt so the voltage drop could be much lower than 1xVbe, roughly 0.6 volts, yet yield high accuracy.  From that, adding fold-back current limiting typically took just one more resistor.

[iMo]

723 PSU with the similar "impulse foldback protection":

https://www.elektronik-kompendium.de/public/schaerer/ntifb.htm