• State TI Thinks Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 434 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.

UCC28950: CS signal differences cause peak value to be lower

Nitish Agrawal
Expert 1595 points
Part Number: UCC28950

Tool/software:

Hello,

We are seeing differences in the peak CS signal being measured across our converter units.

Some converter units show that the CS signal doesn't completely drop to zero about 500ns before start of the next switching cycle, while other show that it drops close to zero soon after OUTC or OUTD goes low.

In the converters units where the CS signal peak reaches the expected (calculated) value at any given load, the CS signal doesn't drop to zero after OUTC/OUTD signals turn off their respective primary side MOSFETs.
While in the units where the CS signal does drop close to zero, the peak value is lower than expected.

The attached waveforms show that. At 27A load, we expect the CS signal to be around 1.84V. The signal to the right side reaches that expected value, however, the CS signal stays above zero after OUTC/OUTD have gone low.
In the signal to the left side, the CS signal is seen dropping close to zero after OUTC/OUTD turn off their respective MOSFETs, however, the peak CS signal is ~32mV lower than expected value of 1.84V.

We are unable to determine why this is happening. We checked the isolation transformer's magnetizing inductance values but they look alike across all units. We have checked the output inductor values and they look alike as well. We have probed the units to check for other differences and don't see any.

We would appreciate any help in figuring out what is causing this difference.

Sincerely,
Nitish Agrawal

  • 0
    TI__Guru* 85616 points

    Hello,

    Your inquiry has been received and is under review.

    Regards,

  • 0 in reply to Mike O'
    TI__Guru* 85616 points

    Hello,

    Did you have the scope probe ground and probe tip at the same connection points when measuring these waveforms?

    The waveform you have with an offset shows quite a bit of noise on it while the one without the offset does not.

    Do you know why this is?

    The differences that are showing are in the middle of your first low pass filter.  Generally designs only use 1.  Why did you need 2?

    You need to be careful when probing a power supply not to pick up stray noise.

    You need to make sure your scope probe does not cross any EMI generators.

    Also the grounding of the scope needs to be reference at the same place.

    I am also curious at what the CS signal looks like across the UCC28950 GND pin and CS pin?

    Are there any differences there?

    Regards,

  • 0 in reply to Mike O'
    Expert 1595 points

    Hi Mike,

    The reason for the noise on one is that a longer ground lead of the oscilloscope probe was used.  The one without noise had a much shorter ground lead of the probe which was soldered to the ground pad of the PCB.  The same ground pad on the pcb was used for both measurements.

    We have two stages, since the UCC28950 controller IC sits on a daughter board.  The first stage is on the main board while the next stage is close to the IC pin 15 on the daughter board.  The signal was measured on the main board since it is easier to measure there, while much more difficult on the daughter board.

    This phenomenon is repeatable and is not dependent on the probe wire locations.  Some units show this and other don't.

    Have you seen this "offset" in any of your customer's or your own measurements?

    We can try measuring the CS signal at the IC pin...but it is a lot more sensitive and difficult to access.

    Regards,

    Nitish

  • 0 in reply to Nitish Agrawal
    TI__Guru* 85616 points

    Hello,

    Your inquiry has been received and is under review.

    Regards,

  • 0 in reply to Mike O'
    TI__Guru* 85616 points

    Hello,

    This is still under investigation and as soon as I have an update, I will respond.

    Regards

    Mike

  • 0 in reply to Mike O'
    TI__Guru* 85616 points

    Hello,

    Are these designs identical?

    Are you checking the CS signal at the same locations and same load condition?

    Regards,

  • 0 in reply to Mike O'
    Expert 1595 points

    Hi Mike,

    The designs are identical.  Many units show such a difference, which is concerning.  CS signal is measured at the same location and same load condition, 27A.

    -  Have you seen this "offset" in any of your customer's or your own measurements?

    -  Could it have something to do with the behavior of the internal 200ohm pull down MOSFET at the CS pin of UCC28950?

    Regards,

    Nitish

  • 0 in reply to Nitish Agrawal
    TI__Guru* 85616 points

    Hello,

    After studying these waveforms both of them are controlling to different peak current values.  

    I think you are just abserving normal capacitor discharge with different peak voltages to strat (Vpeak).  The signals voltage discharge can be defined by the following equation.  

    Vsignal = Vpeak*(1-e^(t/R*C).

    If both signal had the same peak value I would expect them to discharge to the same voltage value. 

    I think this is something you can verify by adjust the load and monitoring and setting the sense signal peak.  Then comparing the waveforms on both designs with the same peak voltage.  I think you will find they discharge to the same voltage level.


    The peak signal in both design may different because the efficiency of the designs might be slightly different.  Even if it is the same design and different boards this could be the case.  The variation in efficiency can be cause in the variation in Rdson, diode forward biasing voltage, etc...  

    Regards,

  • 0 in reply to Mike O'
    Expert 1595 points

    Hi Mike,

    Let me start by saying that the both boards are set to limit the peak current value to about ~1.86V.

    Wouldn't you agree that in both measurements, the voltage falls by ~1.5V in about 1.2us, regardless of the peak value achieved?  Since Vpeak seen is different in the two cases, 1.52V and 1.84V, your formula predicts different discharge amounts, but the measurement shows the same discharge amount of 1.5V.

    The CS signal where Vpeak = 1.84V drops (or discharges) to about 1.84 - 1.5V = 0.34V, and hangs around that value until 500ns before start of the next power transfer cycle.  This abrupt drop in signal from 0.34V to zero doesn't look like a normal capacitor discharge.  According to your formula, this signal should have discharged by 1.8V in 1.2us (since R=1k and C=300pF), but it only discharges by 1.5V.

    Can you explain this discrepancy?

    Sincerely,

    Nitish Agrawal

  • 0 in reply to Nitish Agrawal
    TI__Guru* 85616 points

    Hello,

    Your waveforms show one signal has a peak of 1.52 V when reference to ground while the other has a peak of 1.84 V when referenced to ground.

    These should be the same.  If they are not for the same load the efficiency levels are different.

    You are correct that discharge time is roughly the same.  The RC time constant is the same and you can use 1.2 us.

    Please do the calculation for RC discharge using 1.84 V peak and then again with 1.52 V peak.  With the same RC time constant and the same discharge of 1.2 us.  They should come out to be different voltages.

    One other thing to point out if your RC is 300 nF and 1 K ohm your RC time constant is 300 us.  5*RC = 1.5 us.  The reason why both voltages are not returning to ground is that the peak current sense signals are a different value.

    One other thing worth mentioning is the point that you are testing is in the middle of a cascaded two stage low pass filter.  You might want to study the UCC28950 CS pin of both designs just to see what they are doing.

    Thank you for interest in Texas Instruments (TI) products.  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.

    Regards,