LM/TL431 calculation question

[iXod]

I'm using the TO-92 package LM431.

Vin 12-15 vdc. Vout 9.0 vdc. Load current 5-35mA, although this load may not always be connected when the power is applied to the LM431 regulator.

This calculator does the series resistor determination:

http://solderer.tv/tl431-calculator/

I know all the parameters except Idiv/Iref. (This default value in the calculator is 200, if you don't enter a value.) Searching on the ratio calculation gets me in a little deeper water than I can swim in.

What is this value or how do I calculate it?

Thanks,
iX

[unitedatoms]

I tried this calculator. Nice site. The ratio is simply to supply sense pin with sufficient tiny current. Value 200 is good, it means that sense pin is not starving on current and receives more than say one microampere.

[iXod]

The calculator presumes R1 8.2K and R2 3K. Will using other values (as long as the ratio is the same), like 50K total (I have a pot of this value) require a different series R?

[unitedatoms]

If you increasing divider, then, no, series 82 resistor will not need to be changed. Because it will have little less of sense current supply, so little less of voltage drop, so no target voltage change.

[David Hess]

Idiv/Iref is the ratio between the divider current and the adjustment pin current.  At low divider currents, the adjustment pin current contributes greater error with the higher impedance divider.  The adjustment pin current could be as high as 4 microamps and vary about half of that over temperature.  I do my calculations based on parallel divider resistance rather than Idiv/Iref ratio.

[iXod]

Thanks for your reply.

So you would recommend against the use of a 50K (total) divider vs the 10K?

[T3sl4co1l]

If you don't mind that it's off by, uh, a few low-100s mV at the output I think?  Then that would be fine.

Tim

[Zero999]

Why use a shunt regulator?

The LM317L or the LM78L09 would be better options.

[David Hess]

So you would recommend against the use of a 50K (total) divider vs the 10K?

Just roughly, tolerance of a TL431 is about 15 millivolts of offset plus 17 millivolts from temperature coefficient so 32 millivolts.  4 microamps of bias plus another 1.25 microamps of temperature coefficient is 5.25 microamps.  So I would not make the parallel resistance any greater than 32mV/5.25uA or 6.1 kilohms which would double the voltage error which as it ends up, it about the value TI used for their test circuits.

So 50K is too high but 10K total is reasonable.  In practice there is not much reason to use such high values.