This is admittedly a very specific project but I felt perhaps the idea could be useful in other contexts too. I have a 1968 Hewlett Packard 334A distortion analyzer (ah, what a beauty!) which is of course in working condition. However, my heart would skip a beat each time for example I flipped from Distortion Mode to Voltmeter mode without resetting the voltmeter scale, causing the needle to swing wildly past the end of the scale and hit the end. Ouch. Needed to protect that piece of Unobtainium. I wanted a method to trap the swing of the needle beyond the end; a simple diode across the meter circuit would work but not quite as well as I wanted. It caused a one-division reduction of the indication and would slow the meter swing but not stop it before it hit the end. After some trial and error I landed up with the attached circuit that does exactly what I want. Even at full scale the loading on the meter is barely noticeable to the eye. Yet, even if it moves just two divisions off the end of the scale, the FET kicks in and sucks away the meter current, causing the meter needle to pull back.
How it works: When the meter is at less than full scale, neither transistor conducts appreciably and the gate-source voltage of the FET is less than 2 volts. At about 280mV (full scale) the BC549 base-emitter junction starts to conduct very very slightly. (Ever seen a silicon transistor used at such low Vbe?)This causes an amplified current to flow into the base of the BC557. That gets further amplified at the collector and causes the BC557's collector voltage to rise abruptly by about 2V. This triggers the FET which conducts and prevents additional current from flowing into the meter. You may need to play with the value of R2 to get the FET to trigger exactly where you want. However, it is the abruptness of the transition which is very interesting and which is what makes the circuit successful.