I posted first to the C5000 forums but was directed to try the power management forums. This is the situation I'm looking at:
The specification for the C5509A core voltage running at 200 MHz is 1.55V min, 1.6V nom, 1.65V max. I've very simply been told that if the operating voltage is below 1.55V the part's operation is unpredictable.
Is there a recommended way to set up voltage supervision for the core voltage? The power supply being used (TPS62302) has an output voltage load regulation spec of .002% / mA max as well as a voltage line regulation spec of .2% / V max and an accuracy of 2%. The C5509A appears to draw .6mA / MHz at 200 MHz for a total of 120mA. The "TMS320VC5503/C5506/C5507/C5509A Power Consumption Summary" application report has a link to a power calculator that suggests a total core current of about 237mW (148mA). I wouldn't expect to see peak currents more than maybe 50% higher. That would mean a load current caused output voltage change of 75*.00002 = 1.5mV. The input shouldn't change more than maybe .2V resulting in maybe a 400uV change. The accuracy of the TPS62302 could put it at .98 * 1.6 = 1.568V. Adding everything together the output could droop to about 1.56V. I'm looking at the TPS3808 adjustable voltage monitor and the negative-going input threshold accuracy is 1.5% max. This means if we were to set the threshold for 1.56V it might not trip until 1.54V.
What's the best strategy for making sure the part is operating correctly?
Hi Mike,
You are in a tough situation. Ideally, as you know, you would want the Supervisor trip voltage to be under lowest possible output voltage of the TPS62302 yet above the worst case low voltage acceptable to the C5000. By my own (back of the napkin) calculation the lowest output voltage from the TPS62302 is in the range of 2.7% off of 1.6V while the lowest acceptable input to the C5000 is 3.125% off of the 1.6V (1.55V). That means that the perfect supervisor would reset the ucontroller between 2.7% and 3.125% off of the 1.6V (or between 1.5568 & 1.55V). And this would require roughly a +/-0.2% accurate supervisor threshold/trip voltage - which does not exist.
The compromise that needs to be made here is made more often than you likely think. [The best solution might be to use a more accurate regulator, perhaps an LDO. ] Either you set the threshold trip voltage of the supervisor to trip on possible good voltages from the TPS62302 or you set it so that the trip point is actually lower than the worst case allowable voltage of the ucontroller. Neither solution is disirable but often necessary.
The most accurate adjustable supervisor in TI's is the TPS389x. This supervisor will be nearly 1% accurate after consideration of the 0.1% divider resistors that would need to be used.
I will suggest an LDO for you before the end of the day.
Regards
Bill
Thanks for the response Bill. I just wanted to make sure I wasn't over complicating something. I'll consider your thoughts and see if I can figure out something that will work.