TLC5940, making XERR reliable?

Other Hardware LEDs and Multiplexing
1

Hello all,

I'm trying to get the XERR pin(s) on a series of TLC5940s to behave reliably. In a trivial circuit, with short wires and lower power, the XERR pin behaves correctly, in that it rightly identifies LOD (Led Open Detection) as tested by just pulling out an LED. However, in a higher current (not that high -- 40ma/output), larger circuit (longer wires between TLC5940 sinks and LED cathodes), XERR behaves unreliably -- it prematurely fires on lighting of any LED, the best test case being 0->full. This makes it near impossible to use for, for example, thermal protection -- since it can't be reliably counted on.

This seems to be exactly what's described here, under 'XERR' issues:

(Related/inspiration: TLC5941 XERR - Power management forum - Power management - TI E2E support forums)

... with the theory being: "because at the instant it turns on an LED, wire inductance causes a dip in the LED supply voltage".

This article also proposes increasing the voltage specifically to get a clean XERR, but with that comes a whole range of power dissipation issues, so this really isn't an option for me.

Any advice? Has anyone successfully/cleanly got XERR working on a 5V LED TLC594X supply circuit?

Thanks.

2

Are you using decoupling capacitors across the power and ground of each chip?

3

Yes -- plenty of next-to-chip, and power rail, decoupling. Circuit behaves perfectly otherwise.

4

Then you are not using enough or of the right value or type.
There needs to be a 0.1uF ceramic in the mix and then some high value bulk stuff as well.
Also look to see if there is any instability on the LED lines, if there is a long way between the chip and LED, a small cap of about 47pF might be needed across the LEDs.