Reverse Output Current through a LT3045

Hello Tony,

tonyr said:

Particularly, how much reverse current will the LT3045 output tolerate and where does that current go?

my answer for a standard LDO would be: "it will not work, the LDO output can source current, but it cannot sink any current." But the LT3045 is not standard, because it has an "output overshoot recovery" included. This will indeed sink some current when the output voltage rises too high. If you look at the block diagram on page 14 of the datasheet you can find an npn-transistor close to the OUT-pin which conducts the sinked current to GND.

Regarding the size of the sink current the datasheet says: "As illustrated in the Block Diagram, the LT3045 incorporates an overshoot recovery circuitry that turns on a current sink to discharge the output capacitor in the event OUTS is higher than SET. This current is typically about 4mA." On page 9 of the datasheet you can find a diagramm that shows, how this current depends on the magnitude of the overvoltage and on temperature.

So for currents in the single-digit mA range your idea might work. But I don't see any specification for the "switching speed" of this recovery circuit, and for currents >10mA it surely doesn't work.

best regards

Achim

Hello Achim,

Thank you very much for your answer. It's amazing what one can learn from actually reading the datasheet!

If I want to limit the input current into the LT3045 to say 5 mA I can place a resistor between the Trig-Sync In and the diodes. This resistor drops the voltages seen by the input buffer (that I am trying to protect) as current starts to flow through the diode, but it also flattens the input edge because it forms a RC filter with the input capacitance of the buffer and diode. Ultimately this resistor raises the the input voltage at which damage, to both the buffer and the LT3045, will occur but still keeps the LT3045 in the picture and exposes it to potential damage.

So I decided on something simpler: Use a Zener diode to to ground protect the input as shown below.

This presents it's own set of problems (the diode clamping voltage is very close to the absolute maximum rating of the buffer, the diode can't sustain a DC fault, etc.) but at least it takes the LT3045, and hence all the other devices on the +5V rail, out of the picture. If the buffer blows I can replace it easier than the LT3045 and possible subsequent damage to the other parts on the 5V rail. That was the high-level math behind my decision. 

All the best and thank you again for your very good answer,

Tony

Hello Achim,

Thank you very much for your answer. It's amazing what one can learn from actually reading the datasheet!

If I want to limit the input current into the LT3045 to say 5 mA I can place a resistor between the Trig-Sync In and the diodes. This resistor drops the voltages seen by the input buffer (that I am trying to protect) as current starts to flow through the diode, but it also flattens the input edge because it forms a RC filter with the input capacitance of the buffer and diode. Ultimately this resistor raises the the input voltage at which damage, to both the buffer and the LT3045, will occur but still keeps the LT3045 in the picture and exposes it to potential damage.

So I decided on something simpler: Use a Zener diode to to ground protect the input as shown below.

This presents it's own set of problems (the diode clamping voltage is very close to the absolute maximum rating of the buffer, the diode can't sustain a DC fault, etc.) but at least it takes the LT3045, and hence all the other devices on the +5V rail, out of the picture. If the buffer blows I can replace it easier than the LT3045 and possible subsequent damage to the other parts on the 5V rail. That was the high-level math behind my decision. 

All the best and thank you again for your very good answer,

Tony