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.

TPS92518HV: Strange Behavior Near Saturation

Part Number: TPS92518HV

Please note this waveform:

The signal shown above is the input to the inductor from the PMOS FET. The TOFF setting is 4. The probe is 10X and I forgot to set it as 10X so the scale on the left is /10.

Other relevant settings: L = 390 uH. Rsense = 0.2 Ohm 1%. The dual 1uF output caps have been removed from the evaluation board. IMAX is set to 255. The diode load is roughly 53.6 Volts and Vin is 57.5 Volts. (The input and load voltages are non-negotiable hard specifications that I have to meet, all other parameters and values are fair game.)

The transistor is nearly in saturation and half the time it takes quite a while for the output current to get to IMAX. The other half of the time IMAX is reached very quickly.

That behavior seems abhorrent to me.

Questions:

1. Is this behavior expected? If not, is there a way to fix it?

2. The average frequency is still within the tolerance of the part, but the frequency of the back to back pulses is way too high. Even if it is abhorrent, is this behavior okay from a design perspective? In other words, am I damaging the part running it like this?

This will cause the ripple to go higher than desired (I'm trying to keep the ripple down to 50 mA), I'd really like to be able to get the frequency to around 800 KHz. With this double pulsing, the ripple will be almost double, even if I get the average frequency to 800 KHz.

  • Hi Brian,

    It looks like it may be a minimum on-time pulse. With the off-time that short the current doesn't drop much so it could be re-tripping the peak current sense from leading edge noise or slight ringing. If you look at the current waveform it may show why. Also, if the Toff setting is 8 is the frequency the same as above with only one pulse (or about so)? The other thing to watch is the boot cap will have to have enough time to reset, it's probably okay with the above signal but I'd have to look into this. Guesstimating the current drop will be around 7-14 mA out of 1.27, not much ripple.

    If you have diodes or leds for loads the voltage across them changes significantly with temperature so the first time turning on cold it may take longer to get to full current than if it's been running and hot.

    1) I guess it doesn't surprise me, it's a corner operating case with a very small di/dt on the inductor. Ideal, no, I've seen similar operation on power supplies, double pulsing, is some corner cases. It'll work this way, not sure I'd want it to.

    2) It's probably not desirable but from a switching frequency stand-point the average isn't all that high. I'd try the toff of 8 and see if the fsw is the same, if it has single pulses. Or you'll have to look at the current waveform and see what's causing it though at 7-14 mA out of 1.27A I don't know if you'll be able to.

    I calculate the ripple being much lower than that. If I use toff of 100 nS, which those pulses look smaller than that, Vout at 54V and L at 390 uH I get 14 mA.

    Regards,
  • Thanks for the quick reply.

    Setting the TOFF to 8 did remove the second pulse and kept the single pulses to around the same period as the double pulses. If I let the diodes warm up and settle near 54 Volts, then I can set the TOFF to 5 and not see the double pulses at a frequency of around 400 KHz. Since the ripples are so much lower, 400 KHz should be within spec. I'll check the math on that later... You are right about the ripple. It is much smaller when the voltages are close, so that is not an issue.

    I've convinced myself this is a non-issue.

    But, I'm back to why I started this research. I'll open another ticket to talk about it.