To Whom It May Concern,
Part of my design incorporates an LTC2977 PMBus manager where one DAC outputs feeds the SET node (nominally -5V) of LT3090, a negative voltage regulator. I would like to servo the DAC to maintain/margin the SET node at or near -5V. I'm confused how to assign register values in LTPowerPlay for negative VOUT_COMMAND, as I have not seen much documentation on this front.
Any help is much appreciated.
There are a few things that you need to do. The 2977 does not have the ability to natively sense and servo a negative supply.
To sense a negative rail, you can follow the datasheet recommendation (Fig. 29) that uses a voltage divider and connect the divider midpoint to the Vsensem pin. The Vsensep pin is tied to REFP (1.23V). The bottom of the divider is tied to the negative rail. In this configuration, the 2977 is sensing a small positive voltage that needs to be translated if you are viewing the telemetry in LTpowerPlay. The actual voltage sensed will be less than 1V (and it will be positive) and a scale factor and offset needs to be applied in the GUI in order to see a negative value. The DC2518 demo board shows an example of this. On this board, CH1 is a -12V supply that is powered from the LT3090. I used a simple divider for the sensing. A more detailed description can be found in the DC2518's demo manual on page 14 and 15. Click here
For a -5V output, the resistor values will need to change. I suggest these resistor values in Fig 13: R1=10k, R2=68.1k. This will make the sensed voltage (Vsensep-Vsensem) equal to ~0.8V. This is the VOUT_COMMAND value. This provides a divide-by-7.81 of the voltage (including ref 1.23V). Therefore the Vout Scale Factor is -7.81, be sure to include the negative sign. The Vout Offset is 1.2V. The READ_VOUT register will now read -5V.
To servo a negative LDO, the DAC output needs to be translated to a node that sits below GND. The DC2518 also shows an example servo circuit. A PNP is used to translate the DAC voltage to a current into the SET resistors. See page 16.
I have a question about maximum potential of VSENSEM. In my design I could not use REFP as VSENSEP pull-up for negative voltage measurement, I had to use locally available 3.3V reference. On some channels VSENSEM voltage is higher than REFP (but always under 3.3V). It looks like channels where VSENSM is below REFP work correctly (ADC and supervisor), but in case of channels which have VSENSEM higher than REFP voltage only ADC works correctly. I get correct voltage telemetry and TON_MAX_FAULT_LIMIT condition (which is UV derivative) although I can measure with oscilloscope that voltage rises to correct level way before TON_MAX_FAULT_LIMIT delay (all UV/OV Warning / Fault / Margin thresholds are set correctly). On the page 5 and 6 in the ADC and supervisor section there is information that "Single-Ended Voltage: VSENSEMn" should be between -1.V and 0.1V is it not contradictory with section "Sensing Negative Voltages" which shows clearly that VSENSEMn can be outside of this range?Regards