• State Answer Suggested
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 1322 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

UCD9244+UCD94120+UCD94106 strange behavior

Gergely Janosi
Prodigy 30 points
Other Parts Discussed in Thread: UCD74120, TMS320C6678, UCD9244

Hello,


I am using an UCD9244 to generate power for a TMS320C6678 DSP. For the core voltage a UCD74120 is used, for 1V, 1.5V and 1.8V UCD74106s are used. The input voltage is 5V for all the converters, the 3.3V for UCD9244 is generated by a separate power module (and also used elsewhere on the board).

I set the switching frequency to 500kHz, but when I measure any switch node I never get more than 450kHz and this frequency is depending on the load currents of power stages, under heavy load conditions it drops down to 380kHz. I measured 3.3V, it remains stable, but the internal 1.8V measured on BPCAP pin of UCD9244 is not (1.7V when the system is unloaded, and drops down to 1.6V when loaded). I also noticed that the temperature readings of the controller are depending on the load currents, even the internal temperature of the controller goes up and down with the load currents (of course the current flowing through the power stages will heat them up, but not in a fraction of a second). It seems like there is a crosstalk between the input channels of the auxiliary ADC.

Voltages at the outputs are not even close to stable, loading the 1.5V rail with ~3A causes about 200mV drop and few mV increase at the other outputs, but UCD9244 displays far more than that, see attached figures.


What is going on here? Please help!

Thank you,
Gergely

Relcom Ltd.

~8A load step on CVDD rail (DSP disconnected):

~4A load step on 1.5V rail (note the internal temperature!):

  • 0
    TI__Intellectual 815 points
    Hi Gergely,

    Could you help to check below things:

    [1] Check the ripple on V33 rail, try adding a RC filter on V33 rail, recommended value: 5ohm+4.7uF. Because the internal reference of UCD9244 may be impacted by ripple on V33 rail.

    [2] Ground potentials may be different near UCD device and near load. Use ground node near load as reference, check if load current can change ground potential near UCD.

    Thanks,
    Luke
  • 0 in reply to Luke Lu
    Prodigy 30 points

    Hello Luke,

    Thank you for the answer, I soldered an additional 4.7uF capacitor to pin 46 (V33A) and it reduced the effects condiderably, so I hope that adding the RC filter will eliminate them completely. However the frequency of the PWM signals, are still far away from the setting and heavily influenced by load currents. According to the datasheet they should be within +-5%. Is this still due to the noise at the analog supply? Internal temperature reading of UCD9244 is also influenced by the load current. See the previously attached figures.

    I suspect there should not be significant difference in ground potentials since I am using a 18-layer board with 4 1.3mil thick solid ground planes plus additional ground polygons near the power supply. Ground feedback point for each power rail is taken from the output bypass capacitors.

    Thanks,

    Gergely

  • 0 in reply to Gergely Janosi
    TI__Intellectual 815 points
    Hi Gergely,

    As your description, adding a Capacitor to V33A rail can improve the performance. I think maybe the load change somehow affects the V33A.
    I will suggest to improve the V33A power quanlity, such as tuning the feedback loop to make it more stable. I believe the V33A instability is the casue. Could you help to check it?

    Thanks,
    Luke
  • 0 in reply to Luke Lu
    Prodigy 30 points
    Hi Luke,

    The 3.3V rail on the board is absolutely stable, reads 3.338V and is not affected by load currents. Anyways, I rearranged the components on the board, so that the analog side of UCD9244 gets as far from the power stages as possible, also added the filtering you mentioned and some more decoupling capacitors. Let us hope this will solve the problem. Thank you for your help.

    Regards,
    Gergely