• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 833 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.

UCC29002: Load share IC used for current sharing among 3 AC-DC converters. Current sharing failing randomly and behaving differently from module to module (among different PCB s).

Shamsad P K
Prodigy 90 points
Part Number: UCC29002
Other Parts Discussed in Thread: UCC39002

UCC29002 is used for load sharing among 3  AC-DC converters in our design. 3 No of TDK Lambda converter PFE300SA-12 outputs are connected via 3 No. of oring controller IC LTC4357 for final output loading of +12V @ 60A. When module is loaded for current more then 21A (ie 7A per converter), load sharing is failing randomly.

It is noted that the failure is happening always for higher output currents  and load sharing failure 'current values' changes over testing, but nearly greater than 21A.

We have developed 3 modules, and each module load sharing section is behaving differently.  First module is failing very rare where the second and third module is failing very often -  and there is no hardware change among the modules.

Our input to the module is AC 200V 400Hz, which is convertered into +12V DC , which is the VDD for UCC29002. Switching frequency of the converter is 232KHz for the output side.

We have got better result (Load sharing failed for higher out put current) while configuring the EAO pin compensation. 10uF and 220 Ohm resistor is used in First module gives better current sharing, where same implimented in second and third modules not helping for load sharing.

While probing the IC pin functioning, 400Hz noise is observed in the 'EAO pin' for an amplitude of 50mV (Approx) in the sections where the load sharing is failing. For Example, In second module there are three UCC29002 located at different places of PCB. In that, UCC29002 at Location 2 is having 400Hz component noise in its EAO pin, comparatively higher than that of location 1 & 3. While increasing module output current, Location 2 EAO is saturating 3.4V (Approx) and CSO measures 0V. So always the current sharing section with higher '400Hz noise coponent' is failing in load sharing.

Once the load sharing is failed, oring mosfet corresponding to failed Load sharing IC  found as OFF (Vgs < Vgs th). Also individual output voltage of the converter corresponding to the failed Load sharing IC section, output voltage is measured less than the ored voltage (ie other converter voltages). Example, For Vadj set as 450mV - Load sharing failed Converter output measures as 12.4V where the remaining two converters are at 12.45V. So oring section is not turning ON  the mosfet since it senses a negative voltage from OUT to IN.

Kindly advice for any further check to be done. Pls advice the correction.

Pls advice the EAO compensation -  capacitance and resistor  values for Load sharing LOOP to not to respond for the 400Hz noise observed in the IC. 

Regards

Shamsad.P.KDP-PSU-6059-V1-SCH-PDF-R2 oring issue.pdf

Engineer HDD

  • 0
    TI__Expert 3095 points

    UCC29002_Loadshare.pptHi Shamsad.P.K,

    Can I suggest that you try testing the load sharing without the oring circuits fitted to see if this resolves the issue.

    For the EAO compensation please follow the procedure given in the Mathcad sheet in the product folder. I have given a link to this file below.

    Finally please check that each controller has sufficient headroom to adjust the output voltage of its corresponding module. If you notice that the EAO output is saturating this suggests that the load share controller is unable to adjust the output voltage of the corresponding module sufficiently to achieve current load share. In this case load current sharing will be rapidly lost.

    Thanks

    Joe Leisten

  • 0 in reply to Joe Leisten
    Prodigy 90 points
    1. Testing the load sharing without the oring circuits
    - Design uses 3 No of power conversion modules (AC-DC Converters, PFE300SA-12) which can sink and source. Powering multiple converters without oring-FET circuit may damage the modules due to current sinking among converters.
    - For using decoupling diodes instead of oring FET, a high current of 20A per path causes huge power dissipation and efficiency drop in the module. So that also cannot be done.
    - Reference for Oring FET and UCC39002 is referred from ' SLUA550–March 2010' document.

    2. For the EAO compensation
    - We do not have the bode plot of the power converters using in the board. It is requested with vendor and same shall be corrected once the power module gain Vs Freq response is available.

    3. Sufficient headroom to adjust the output voltage
    - Testing trials has been done with ΔVadj of different value and result found same. For +12V output, with 4% and 10% ΔVadj trials has been made.
    -For different Radj values, output voltage maximum value is changing, where EAO voltage finally getting to saturation and load share failing.

    Kindly advice.
  • 0 in reply to Shamsad P K
    TI__Expert 3095 points
    Hi Shamsad.P.K,
    The load share controller can only achieve sharing if it is able to adjust the output voltage of the module over the required range. If the EAO output on any channel becomes saturated then it means full adjustment of the ADJ pin is not able to make the sensed current on that unit match the LS bus. If this is happening in the steady-state it means that you do not have sufficient control of output voltage to achieve sharing.
    Can I suggest that as an experiment you disconnect the ADJ terminal of the UCC29002 for each channel and replace it with a multi-turn pot in series with a 500R resistor to GND. Then try adjusting the pots for each channel manually to achieve load current share. If you are able to achieve this at full current then we can conclude that the load share controller circuit is not working correctly. I suspect that you will find at some level of load you are no-longer able to achieve current share because one of the pots is set to zero but the corresponding channel is still supplying less than its share of current.
    I hope this helps
    Joe Leisten
  • 0 in reply to Joe Leisten
    Prodigy 90 points

    Hi Mr. Joe Leisten,

    Bode plot of power converter is recieved. Unity gain crossover frequency is 14KHz. Considering Fco for the load share controller as 1/1000 times ie 140Hz (Ref. SLUA311 - June 2004 and SLUU166A - December 2003 - Revised February 2004 ), CEAO is calculated as 80uF.

    We are using Ceao 44uF with Reao 26 Ohm - Is this correct value..? Weather only final crossover frequency is important or used capacitance and resistor value is related..?

    For the ADJ resistor - Currently we are using resistor of  60.4Ohms across the SENSE+ (ADJ pin) and V+,  corresponding to 420mV Vadj. Potentiometer connection you have suggested, didnt understood connecting series resistor 500 Ohm to GND. There is no ADJ pin to GND connection in our schematic. Kindly clarify.

    500 Ohm in series will give as Vadj equal to 3V in our dsesign. It is nearly 25% of the output voltage. Kindly clarify 500Ohm connection.

  • 0 in reply to Shamsad P K
    TI__Expert 3095 points
    Hi Shamsad.P.K,
    Are you able to follow the Mathcad Document that I sent you in a previous post?
    This gives a very detailed description of the design process you should follow for the load share circuit.
    If you cannot access the Mathcad Document please let me know and I will send you a PDF version.
    I hope this helps.
    Joe Leisten
  • 0 in reply to Joe Leisten
    Prodigy 90 points

    Hi Mr. Joe Leisten, 

    Kindly send me the PDF version. Mathcad document i am not able to access here.

    As per your recomendation in the previous post,  500Ohm is mounted at Radj to increase the ΔVadj to provide sufficient control output voltage. Corresponding to 500Ohm,  ΔVadj is 40%. But EAO output is getting saturated for output current lesser than the previos test results with 60 Ohm (4% ΔVadj ).

  • 0 in reply to Shamsad P K
    TI__Expert 3095 points

    Hi Shamsad.P.K,

    I am confident if you follow the attached PDF process all issues will be resolved.

    Thanks

    Joe Leisten

    4718.Mathcad - loadshare_worksheet_web.pdf