• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 2 answers
  • Subscribers 33 subscribers
  • Views 650 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

UCC21520: RC filter for INA pin

Brian Wang0928
Genius 11640 points
Part Number: UCC21520

Customer tried to use different R/C values on INA pin. As a result, we saw something is strange on delay time between Input threshold(INA) to the Ouput(OUTA).

1. What's the inside circuitry in the INA pin? 

2. What's the recommendation for the R/C filter at INA pin?

3. Please explain the test results.

Same Test conditions.

configuration

delay time

1000ohm, 1000pF

123.2ns

1000ohm, 100pF

150.4ns

1000ohm, 0pF

144.4ns

499ohm, 1000pF

154.4ns

499ohm, 100pF

230.4ns

499ohm, 0pF

218.4ns

249ohm, 1000pF

220.4ns

249ohm, 100pF

224.4ns

249ohm, 0pF

206.4ns

UCC21520(RC FILTER WITH INPUT A).pdf

  • 0
    TI__Mastermind 28960 points

    Hello Brian,

    1. Inside of INA is a small deglitch filter, set to about 20ns. This is sufficient for many transients such as the common mode transient due to the switch node dv/dt. In noisier systems, an external filter is still recommended, since noise may last beyond the duration of the deglitch filter.
    2. This input filter should be determined by the customer, based on the noise frequencies in their system, the main switching frequency, and what pulses they must be able to reject. If the customer is uncertain, a good starting point is to set the corner frequency to 10x the main switching frequency. This allows the input pulse to retain enough harmonics to stay square with a sharp edge transition, while still providing significant noise attenuation above typical PWM frequencies.
    3. Please take a look at Figure 29 in the UCC21520 datasheet, which defines the propagation delay timings. Observe that the rising edge delay time is given by the time between the input crossing the high threshold, and the output crossing above 10% of the final value. In the attached oscilloscope plots, the customer is measuring to more than 60% of the final value of OUTA. This largely explains the observed difference. Other factors could include probe delay synchronization (are the input probe propagation delays matched?) and output loading (10% of final value driving 1µF load will take longer to reach than 10% of final value driving 1nF load, since I/C = dv/dt).

    Regards,

  • 0 in reply to Derek Payne
    Prodigy 60 points

    Hi, Derek Payne

    3.
    Same condition, "no other factors".
    Just change RC filter for INA pin.
    Confirm the waveform please.

    Regards,

    Ben

  • 0 in reply to user4783191
    TI__Mastermind 28960 points
    Hello Ben,

    The waveforms in the PDF document of the first post do not measure propagation delay according to the same standards that TI measures propagation delay. The output is loaded and delay measurement is captured at a different point in the output waveform. If you would like confirmation, please remove any load from the outputs, synchronize probe delays, and recapture the waveforms measuring rising edge propagation delay from the time the input crosses the high threshold, to the time that the output crosses above 10% of the final value (example: with +12V/-5V supply, measure when the output crosses above -5V + (12V- -5V)/10 = -3.3V). This should produce propagation delay figures that are less than the maximum guaranteed value in the electrical characteristics. If a different point in the waveform is selected, or if the outputs are loaded, the rise time will be extended by an uncertain amount and the measured propagation delay will increase as a result. Furthermore, the measured propagation delay on a loaded output may not be stable, since the load capacitance is a nonlinear function of the drain-source voltage across the MOSFET.

    Regards,
  • 0 in reply to Derek Payne
    Prodigy 60 points
    Hi, Derek Payne

    So, in different RC filter, the ucc21520's drive output current is different, because the load is same condition.
    Maybe ucc21520 have a good propagation delay, but it has a big trouble on other thing.
    just change RC filter for ucc21520, but the output is different "in same load".
    it is very dangerous, if the configuration is not match the point.
    it is not same as si8238, if i want to p2p the ti's driver.

    Regards,
    Ben
  • 0 in reply to user4783191
    TI__Mastermind 28960 points
    Hello Ben,

    The UCC21520 input RC filter does not affect the typical output drive current in any way.

    I have re-examined the waveforms you provided, and it appears that you are measuring from about 2.0V on the input in all cases. This is the maximum threshold which will ever be seen on a functioning production part, at the worst case temperature. Please look at Figure 23 in the datasheet for a more accurate typical representation of the rising edge threshold, and re-evaluate your measurements to this point.

    The load condition may be similar in all cases, but that does not imply the duty cycle is similar in all cases. You are changing the value of the input filter, which changes the time at which rising and falling edges are crossed on the input, in turn changing the pulse width being fed to the driver. If the PWM source is generated open-loop with no correction for the new pulse duty cycle, then the actual load on the output could still be changing. And if the PWM source is generated closed-loop, any oscillation in the control loop (perhaps due to different propagation delay timings or rise/fall times) could be altering the instantaneous loading conditions without affecting the average loading conditions.

    The UCC21520 output structure behavior is very deterministic. The factors which affect rise and fall times are primarily output loading and driver/switching transistor temperature. Were all the waveforms provided with an adequate amount of time for the system to reach a thermal equilibrium in every case?

    Please note that Si8238 datasheet specifies that peak source current is typical 2A, peak sink current is typical 4A. UCC21520 peak source current is typical 4A, and peak sink current is typical 6A. The output behavior of Si8238 will not exactly match that of UCC21520. This is expected. Some adjustment to gate resistor values may be required if matching the behavior of Si8238 is required.

    Regards,