LTC2977 DAC_SATURATED

The LTC2977 is fine-tuning the DCDC output.

When the servo is turned on, if the DCDC output voltage is also quite large with the LTC2977 set value, the DAC output will rise (saturate) to 1.38V.
If this happens, the DAC output will saturate for a few seconds and return to normal over a few seconds, regardless of the voltage you are monitoring.

If the DAC output reaches the maximum value, will it become uncontrollable for a few seconds?

Parents
  • There would be a possibility to help you, if you could provide at least a scematic of the circuit.
    And what voltage do you measure at what place.
    Why should a DAC output saturate, if you don't force a current over 2mA into it?
    Try to remove any connection from DAC output and measure again.

  • Thank you for your reply.
    Please refrain from exchanging circuit diagrams on the bulletin board.
    The current value to the terminal is about several hundred μA, and each resistance value was calculated by the tool.
    I think there is no problem

  • 0
    •  Analog Employees 
    on Mar 10, 2021 6:34 PM in reply to yito114

    Hi yito, we understand your concern over posting a schematic.  Can you tell us if this is a standard hookup, i.e. a regulator with a feedback node (at 0.6V or 0.8V), DAC resistor tied to the fb node through a resistor, servo engaged, DAC set to inverting mode? 

    If so, when the DAC goes up toward 1.38V this means the sensed voltage is too high, and the servo loop is trying to reduce Vout, so the DAC goes high and pushes current into the fb node.  This causes the regulator loop to lower Vout but can't go low enough to reach the target, so the DAC saturates.  It will likely stay there at 1.38V indefinitely.  Either the VOUT_COMMAND (target voltage) is set too low or the DAC resistor value is too high and doesn't allow enough current into the fb node.  You could set the DAC range to high range as a test, but I think there's something more fundamental happening.

    If the DAC polarity is set to non-inverting, this could cause a problem as well.  That should be checked.

    You could put the DAC into disconnected mode, and look at the regulator Vout.  This might tell you something.

    The other possibility is that, if this occurs when the channel starts up, the TON_RISE time is set too short.  Try setting TON_RISE to a larger value.  This relates to the time when the DAC attempts to connect to the fb node. If the fb node is not stable (still ramping), the soft-connect operation could get confused and saturate.

    Mike

  • thank you for your answer.
    The DCDC I am using is the LT8648S.
    The FB terminal is a combination of 101.5K and 3.3K and is set to 19V.

    There is 6.8KΩ from the LTC2977 to the FB pin.

    If I turn on the servo and try to make it 15V, the DAC output will rise to 1.38V.
    VOUT does not follow the DAC output (because it was unloaded) and gradually decreases.

    When VOUT drops to about 8V,
    The DAC output drops very slowly and the VOUT rises accordingly (approaches the set voltage).

    This DAC goes down very slowly, taking more than 10 seconds.
    Will it take so long?

Reply
  • thank you for your answer.
    The DCDC I am using is the LT8648S.
    The FB terminal is a combination of 101.5K and 3.3K and is set to 19V.

    There is 6.8KΩ from the LTC2977 to the FB pin.

    If I turn on the servo and try to make it 15V, the DAC output will rise to 1.38V.
    VOUT does not follow the DAC output (because it was unloaded) and gradually decreases.

    When VOUT drops to about 8V,
    The DAC output drops very slowly and the VOUT rises accordingly (approaches the set voltage).

    This DAC goes down very slowly, taking more than 10 seconds.
    Will it take so long?

Children
  • +1
    •  Analog Employees 
    on Mar 11, 2021 2:51 AM in reply to yito114

    Hi yito, the LTC2977 cannot directly sense 19V, hopefully you have divided down the 19V to a level that is less than 6V.  I wonder if you have damaged the chip. If your goal is to move Vout over a large range, this is more involved.  I need to know the total Vout range, and if there's a nominal starting voltage. The 2977 is designed to move an output over a limited range but it can work over a large range if we do a few things.

    To keep it simple, suppose you have a divide-by-4, i.e. 20V is 5V on Vsensep.  And when Vout = 8V, Vsensep =2V.  You might want to start the channel at a low voltage, say 12V (Vsense=3V), and servo from there. The VOUT_COMMAND value sets the target.  For a divide-by-4, VOUT_COMMAND should be initially set to 3V (Vout=12V). You can either change the VOUT_COMMAND and stay in soft-connect mode, or operate the DAC in manual mode and issue DAC values to move Vout. 

    For a large range, the DAC resistor needs to be a lower value.

    The slow DAC response is likely due to the DAC moving 1 LSB for every servo loop (~90ms/loop). 

    Mike

  • thank you for your answer.
    I thought it was caused by DAC moving 1Lsb for every serbo loop

    I understood as follows

    In high-speed servo mode, it changes by 1LSB every tS_VDACP time.
    When the output voltage target value is reached, the normal servo mode is set, and 1 LSB changes for each tCONV_ADC.
    Here, tCONV_ADC is the value described in Note 5 of the data sheet P9.

    Under my terms of use, tCONV_ADC will be 45m

    Since the DAC output is 360LSB from 1.35V to 0.87V
    360 x 45mS = 16.2 seconds, which is consistent with my phenomenon.

    Thank you very much.

  • 0
    •  Analog Employees 
    on Mar 12, 2021 6:27 PM in reply to yito114

    Hi yito,

    Yes, the fast servo mode changes the DAC code 1 LSB every 500us.  The ADC is muxd' and conversion time is 6.15ms (voltage measurement), however the ADC update rate is never faster than 86.1ms. Note 5 indicates the loop time is 36.9ms + (6.15ms x #ch in ADClowres mode) + 24.6ms x #ch in ADC hires mode). Unfortunately the user cannot remove (opt out) any channels from the ADC loop.

    Mike