TLV431 maximum ratings

On a sunny day (Sun, 25 Aug 2019 14:15:06 -0400) it happened bitrex wrote in :

The standard way to make an uni-directional 'thing' bi-directional is to put it in a bridge rectifier. But that will always add 2 diode drops each way: _______________________________ | | |

--- --- | \ / \ / |

--- --- | | | + |-- AC --| thing | | -

--- --- | \ / \ / |

--- --- | |________|____________________|

'thing' can be any-thing here, your TL431, transistor, MOSFET, LED, etc et c

It won't be very accurate once it is inside a bridge rectifier. However, the TI data sheet gives current/voltage curves and absolute maximum ratings for positive and negative currents.

John

Yes.

Depending on your needs, it's often good to use two parts in parallel. For example,a Schottky diode in parallel with a TLV431, get Vzener in one direction, and low V_Schottky in the other.

For another example, in my 1MHz 1.2kV amplifier(!), I needed a 7.5 to 10V zener to protect the output MOSFET, see Figure 4x.137 in the draft article:

But xeners have 55pF of capacitance (compare to 40pF for the MOSFET) so I used a low-capacitance D1213. It has only 1pF, but wasn't able to handle 150mA in the other direction. Solution, use a BAT54 Schottky, only 5pF at 3V, but conducts 150mA with Vf = 700mV.

Easy, wire two back-to-back like you'd make a TRIAC out of SCRs*, then it'll clamp symmetrically.

(*Except this is a sometimes-repeated lie; to do that properly, you need reverse diodes to clamp the gates in the reverse direction, and string up the trigger (a phototriac or what have you) between the gates.)

Tim

The alternative is to use a two-terminal current source in the DC path of the bridge, and feed the AC drive through that bridge, then to a resistor to GND. While forward voltage isn't necessarily well-controlled, the reverse current leakage of the bridge rectifier can be very small, while the current source (and its resistor) are identical for both polarities of the drive (and output).

On a sunny day (Sun, 25 Aug 2019 19:54:40 -0500) it happened "Tim Williams" wrote in :

Yes, then only 1 diode drop. I see I reversed + and - in the ASCII diagram (oops).

On a sunny day (25 Aug 2019 14:25:26 -0700) it happened Winfield Hill wrote in :

I am glad to see somebody use a trimmer and trimpot :-) Component tolerances make it hard to select just the right C, so trimmers are great here. Especially when it is just a one time project.

Yes. Look at the graphs on top of page 5.

When forward biased you get the normal si diode behavior of the chip's substrate-internal node diodes and a rather sloppy looking knee. When reverse biased you the designed for functionality and a sharp knee.

Either way 20mA isn't going to create enough heat to damage.

piglet

To add to my previous :>

I meant 20mA forward wouldn't create too much heat, 20mA reverse (as zener mode) might depending on programmed shunt voltage.

piglet