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.

UCC3895: UCC3895, unsimmetrical pulses in case of CS pin12

Part Number: UCC3895

Dear Colleagues,

 

we use for a longer time the device UCC3895DW in some power supplies in

phase shift and current mode topology.

Sometimes we have problems with the output pulses and therefore destroyed full power bridge transistors.

We recognize unsymmetrical pulses at the OUTC/OUTD side and not a

duty-cycle at 50% constant. Often actually longer pulses with double time!

Particularly in case of using the current sense Pin12 at 2V limiting and 2.5V disabling.

I constructed a test-setup in order to investigate this behavior.

 

Conditions:

                        Pin15 VDD 12V bypassed with 10µF and 2x 220nF

                        Pin4   VRef 5V bypassed with 1µF and 220nF

                        Pin9   DELAB R=4k7

                        Pin10 DELCD R=4k7

                        Pin 8 RT R=110k (further test with 70k and 700pF at Ct)

                        Pin7   CT C= 370pF

                        Pin1/2 EAN/EAOUT connected

                        Pin20 EAP control signal PWM 0-3V

                        Pin11 ADS to ground

                        Pin5/16 GND/PGND connected the ground plane directly under the IC.

Frequency about 100kHz (200kHz Pin7 CT).

Operation in voltage mode.

            Operation with output power DC link about 100V.

            Operation without output power DC link 0V.

I set a signal rectangular to the input of CS Pin12 with a frequency generator.

First step from 0V to 2.2V (limiting) [00016;00017]

or 0V to 2.6V (disabling) [00023;00024].

Another example with triangle signal 0V to 2,6V (disabling) [00029;00035].

In any kind we can produce such unsymmetrical and longer pulses.

 

Next experiment was to switch the shut down directly with a mosfet transistor and

frequency generator. In this case there wasn´t visible an unsymmetrical pulse.

[00056;00057]

See pictures at the attachment.[….]

 

Our question, how can we prevent such behavior and can ensure that we have no

longer pulses and a duty-cycle of 50%?

 

Thank you very much for your effort.

 

Best regards

Matthäus Gottenöf

Development

 

  • Hello Matthäus

    This may be the issue described in the App note www.tij.co.jp/.../slua275.pdf

    Could you please have a look at this and try the solution suggested there - although I do see your note about using Voltage Mode Control.

    If that does not work then please post a message here. Many instabilities are due to noise coupling into the controller and there may be something in the PCB layout or schematic which we need to look at.

    Regards
    Colin
  • Dear Colin Gillmor,

     

    thank you for your Answer.

    We know this description from TI but this is not our

    problem at very low signal and small PWM modulation, we work at

    higher signal and big PWM modulation. Therfore is it not an issue

    of the offset of the ramp signal.

    In the test stage we switched to the voltage mode, because we want to

    eliminate the current influence at the ramp input pin 3 in the normal operation.

    Our interest is the behavior of the current limiting and shut down at the input CS

    on pin12. In this case we have the unsymmetrical pulses.

     

    We do another tests:

    -       We controlled the correctness of the current measurement-> OK

    -       We controlled all components and values in refer to the data sheet -> OK

    -       We controlled all influences on interferences -> No effect

    -       We tested only the board with UCC3895 without power electronics and

    power switching effects. -> Always the same behavior

    Only the tests with the control board and feeding with a frequency generator

    at pin12 we can produce this effects of unsymmetrical pulses.

    At all times if we pass the trigger level of 2V and 2.5V at this input.

    In this case of over current protection we have a problem in our power

    circuits with unsymmetrical pulses.

    Sometimes get the magnetics in saturation and the result is the destruction

    of the semiconductors .

    What can happen or how we can prevent the false pulse?

     

    Thank you very much.

     

    Best regards

    Matthäus Gottenöf

    Development

     

  • Hello Matthaus

    I've not been able to look at this issue for the past few days - I hope to do so tomorrow.
    One thing we do see from time to time is the controller being upset by interference from noisy signal lines running under the device body - I don't have your layout to hand but if you wished to send it to me at colingillmor@ti.com I'd be happy to look at it. Gerber or native files are preferred to .pdf images.

    Regards
    Colin
  • Hello Matthaus

    Please let me know if this is still an open issue -

    I re-read the posts and one thing that I had missed before was your use of a signal generator to drive the CS pin. There are a couple of potential problems in doing this. The signal generator would need to be synchronised to the UCC3895 controller and the signal should be either a sawtooth or trapezoid so that it mimics the current in the application.

    One method to satisfy both of these conditions would be to use an external signal generator (0/VREF) to synchronise the controller to the generator. Then use a RC circuit to simulate a current sense waveform - reset at the end of each cycle by the rising edge of the SYNC signal. The new cycle starts at the falling edge of the SYNC signal.  I don't have exact values for SYNC pulse width - probably 300ns or so and the UCC3895 will only synchronise to an external source which is faster than its free running frequency.

    Here is a sketch of what I suggest.

  • Hi Matthaus
    I haven't seen a response from you on this so I'll assume that you have overcome the issues you had. Please feel free to open a new thread if you need further help - or email me directly if you prefer.
    Regards
    Colin