• State Resolved
  • Locked Locked
  • Replies 13 replies
  • Subscribers 65 subscribers
  • Views 3205 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.

TPS40200

YILMAZ KIRÇİÇEK
Intellectual 905 points
Other Parts Discussed in Thread: TPS40200

Hi,

We have designed a cascaded power suply module including TPS40200 and have a little problem with it.

When we assemble the circuit with Switcher Pro calculated components it did not work as expected. When I incerese the Css, say 1uF, it works but another problem arises.

Could you please help me to solve my problem?

 Here is the details;

1 -  Desired function ; 34V to 49V input ==> 27V@1.5A output

2 - First off all I have only assembled the TPS40200 circuit components to isolate problem. (No cascading)

3 - I have designed PCB before Switcher Pro update, so R10 is placeed where the previoud version advice.  Location of the R10 can be real problem ?

4 - Started with Switcher Pro calculated components and there was no expected output; 1.5V with 220 Ohm load, unstable approx 21V with no load.

     Wave form on the SS pin was the same with overcurrent situation as described in datasheet. (1.7Vpeak when unloadaed, 1Vpeak with 220 Ohm load)

     When I increase the Css value to approx 1uF it works fine with resistive load as expected. It is also works with cascaded LDO (24V, no soft start)

     But it does not work with intended load neither single nor cascaded. Intended load can be any type device with switching regulator. 27V output can be

     generated but not stable. 

Soft Start Pin - YILMAZ KIRÇİÇEK

5 -  I have added a added a resistor parallel with the Css to decrease discharge time. Played with the resistor value but it does not solve problem.

6 - Played with  R8 resistor value, compansation components (accorging to Switcher Pro), no way.

7 - I have removed R9 and shorted C4 to disable OCP but it did not worked even if unloaded.

8- I have suspected with PCB and components and try with 6 different setup, but all the same. 

9- Can PCB layout be a problem ?

 Switcher Pro schematic;

 TPS40200 Switcher Pro schematic - YILMAZ KIRÇİÇEK

Applied schematic;

TPS40200 schematic - YILMAZ KIRÇİÇEK

 PCB layout; (Two layer, bottom for power gorund)

TPS40200 PCB layout - YILMAZ KIRÇİÇEK

 

 

 

 

  • 0
    TI__Guru 56985 points

    I have designed PCB before Switcher Pro update, so R10 is placeed where the previoud version advice.  Location of the R10 can be real problem ?

    The location of the capacitor in your schematic is correct.  This is the post-update location.  If it was in the pre-update location (in series with the power stage) it would be causing a problem as the TPS40200 supply would be trying to draw 27/34 * 1.5A = 1.2A of current through a 10ohm resistor (14W power dissipation)

    Started with Switcher Pro calculated components and there was no expected output; 1.5V with 220 Ohm load, unstable approx 21V with no load.

         Wave form on the SS pin was the same with overcurrent situation as described in datasheet. (1.7Vpeak when unloadaed, 1Vpeak with 220 Ohm load)

         When I increase the Css value to approx 1uF it works fine with resistive load as expected. It is also works with cascaded LDO (24V, no soft start)

         But it does not work with intended load neither single nor cascaded. Intended load can be any type device with switching regulator. 27V output can be

         generated but not stable.

    This suggests 2 things:

    1) Insufficient Soft-Start time to charge output capacitor to 27V within current limit.  Your schematic shows 333uF of capacitance, charged to 27V is about 9mC of total charge.  The soft-start time programmed with a 47nF capacitor is about 470us for about 19A of soft-start charge current, which exceeds the current limit programmed with a 20mOhm resistor (which is 3-6A).  Changing the soft-start capacitor from 0.047uF to 1.0uF decreased this charge current to about 0.9uA.

    2) The fact that  it is stable with an LDO and resistive load but not with a switcher suggests the control loop is not stable with a negative impedance.  Switch Mode power supplies, which are typically constant power loads, have negtive dynamic impedance (a decrease in voltage produces an increase in current)  You may need to evaluate the loop's phase margin with no damping from the load and make sure there is sufficient margin to support the negative load impedance.

        Switcher Pro does not support negative impedance loads, but the TPS40k Stability Tool (available from the tools & software section of the TPS40200 product folder) will support a negative impedance by forcing a negative current into the Excel design too.  (I checked to make sure the negative impedance is implemented, and it is - however the design is heavily dependant on the ESR of the ouptut capacitor, so check that as well)

    Also, if your cascaded switcher has input capacitance in addition to the TPS40200's output capacitance, you'll need to include that in your stability model for the TPS40200.

     

  • 0
    TI__Expert 4375 points

    Firstly lets see if the grounding of the IC could be the issue. Looking at your layout the IC decoupling capacitor (C11) is not in a good location. IC signal ground current shares copper traces with the power train switching current. C11 should be connected directly to the VDD and GND pins of the IC. The signal traces to the IC would connect to one side while the power traces would connect to the other. See figure 52 in the data sheet for the TPS40200. See how C3 in this figure is connected to the IC and how the signal level parts connect to the left hand side, while the power train connects to the right. You should try to implement this on your PCB. This may require cutting traces.

     

    Regards,

     

    Richard..

  • 0 in reply to Peter James Miller
    Intellectual 905 points

    Mr. Miller,

    I thank you for you answer. I am going try your advices.  I have downloaded the  TPS40k Stability Tool. But I have a question regarding your first advice;

    As I metioned before I have removed R9 and shorted C4 to disable OCP. In this case why does output capacitor cause OCP. Is removing R9 and shorting C4 not enough to disable OCP ?

    Best Regards.

  • 0 in reply to YILMAZ KIRÇİÇEK
    TI__Guru 56985 points

    Yilmaz,

     

    Due to some internal time constant differences between the VDD path and ISNS path, a circuit with a large ripple at the VDD pin can produce an over-current trigger even if ISNS is shorted to VDD.  This would be expecially evident on a design trying to draw 18A of current as even the inductance in the short between VDD and ISNS could produce the requires 100mV drop.

  • 0 in reply to Richard Garvey
    Intellectual 905 points

    Mr. Garvey,

    I thank you for you answer. I am going try your advices. 

    Regards.

  • 0 in reply to Richard Garvey
    Intellectual 905 points

    Hi Mr. Garwey,

    I thank you for your advice. You were right, signal ground was sharing the copper trace with power train. One of the problem has caused by my poor PCB desing.

    Changing the decoupling capacitor loacation did not change anything alone. But when  I moved  the C3 (RC pin) and C5 (SS pin) capacitor ground connection to TPS40200 GND pin with a jumper wire first problem is solved. Now circuit can work healthy with resistive load and linear regulator + resistive load. And also I can able to disable OCP by shorting ISNS and VDD pins without changin anu other components. (compansation, output cap. etc.)  I will never make light of Kelvin connection importance and make such an error inşallah.

    But I there is still a problem: outputs drops when I connect a switcing regulator as a load. Mr. Miller adviced me to recalculate the compasation circuitry with TPS40k Stability Tool. But could not able manage it, instead I have used brute force; shorted ISNS and VDD pins and everything work fine now.

    I thank you for sharing your experience again.

    Best Regards.

     

  • 0 in reply to Peter James Miller
    Intellectual 905 points

    Hi Mr. Miller,

    I thank you for your advices. There was two problem on my circuit. One of them was  sharing of signal ground and the power train the same copper traceb that caused by my poor PCB desing. With some modification (cutting trace and using jumper) first problem is solved. Now circuit can work healthy with resistive load and linear regulator + resistive load.  And also I can able to disable OCP by shorting ISNS and VDD pins without changin any other components.

    But I there is still a problem: outputs drops when I connect a switcing regulator as a load. You adviced me to recalculate the compasation circuitry with TPS40k Stability Tool. But could not able manage it, instead I have used brute force; shorted ISNS and VDD pins and everything work fine now. Now two question arises;

    Acording to datasheet and my calculation 47nF capacitor gives 950us soft start time and this causes approx 9.4A charce current. Then I decrease the output caps 100uF so now charge current is 2.85A but still have OCP problem. But after PCB modification I can able to disable OCP by shorting Vdd and Isns pins and make it works.  I have connected the switching regulator load, powered up the circuit and seen that everyting is OK. Then without powerdown I cut the short between VDD and ISNS pins, it has kept working. Now its clear that switching regulator as a load consuming high current during it's start start up (no soft start) and my improper PCB design increase effect of this.

    I am going to redesign the pcb as described datasheet but want to be sure about stability.  As you mentioned I did not considered the switch mode supplies as a load. I want to use TPS40k Stability Tool but I could not achieve. #AD? appears instead of numeric values. Could you please gelp me to use it ?

    I thank you for sharing your experience again.

    Best Regards.

     

     

  • 0 in reply to YILMAZ KIRÇİÇEK
    TI__Expert 4375 points

    Not sure why you are getting the #AD? error, I do not get this on my sheet. However I do see that the K-factor method isn't able to give a good loop. So I manually entered the poles and zeroes. I used the poles form the tool initial calculation. Then placed the zeroes at the LC filter double pole. The cross over is not as aggressive as your selection but we now have some phase margin, see below.

    Regards,

     

    Richard.

  • 0 in reply to Richard Garvey
    Intellectual 905 points

    Hi Mr Garvey,

    I guess we are not using the same version of the excel worksheet. I can see it from the difference on left bottom side (Select IC section.) When I select TPS40200 a new section appears. Could you please send your xls file to me ? (via e2e or yilmaz.kircicek@teksanjenerator.com.tr) 

    Best Regards.

     

  • 0 in reply to YILMAZ KIRÇİÇEK
    TI__Expert 4375 points

    I ran the same inputs on the version found on the web and the results were the same.

     

  • 0 in reply to Richard Garvey
    Intellectual 905 points

    So I am going to struggle with Excel setting (macro, security, add-ins etc.)  or going to try it in another PC.

    Regards.

     

  • 0 in reply to YILMAZ KIRÇİÇEK
    TI__Guru 56985 points

    What version of Microsoft Excel are you running?

     

    I have seen people having issues with the Excel tool using the European version of the Excel 2003.  This may be that issue.  I will see if I can find the issue.