LM3671MF-1.8EV: New news, output is getting voltage before the input. What can I expect? This is for the 3.3V part.

Hi Anthony,

Are you getting some active signals on the input of the regulator?  is a device active and regulating at 1.7V?

If it's not from the device is not active than it might damage the device. Normally, device regulation might be affected and can see some effect on the performance and for that, I would like to know more detail here.

Best regards,

Tanvee

Hi Tanvee!

We are plugging our electrical-to-optical dongle via an HDMI cable into a customer's video source equipment HDMI interface, which is supposed to be inactive except for its 5V supply pin. Once we see the 5V pin, we are supposed to return a hot plug detect signal (HPD), at which point the source is supposed to enable the rest of its HDMI interface.

They designed their source equipment to have ALL HDMI OUTPUT SIGNALS ACTIVE ALL THE TIME.

The other stupid thing is that the HDMI connector connects the 5V signal last when a a cable is plugged in.

So the HDMI high-speed video data signals, which are pulled up by the source to 2.7V, are charging our internal 3.3V rail to 1.7V for anywhere from 10ms to 200ms prior to the 5V supply getting to the regulator.

I'm wondering if it's possible to damage the regulator in that amount of time, especially if the equipment is powered up many times over the course of a year.

We have found that over 99% of the units we ship work in that environment indefinitely, but that a statistically significant number of units fail *after* ~1 year of service.

We have traced the problems to 2 chips: a driver chip and this regulator.

Often, the regulator will act funny but recover. Often, it will start putting out 3.7 to 4.0V on its output. Occasionally, it will actually short its input (VIN) to ground internally and smoke.

Thanks!
- Tones

Hi Anthony,

Thanks for the detail explanation. 

It will not give critical issues on buck converter If the active signal on the regulator is not above the absolute voltage of the device. According to me 3.3V supply will inside device max supply value and as you say it will behave bit different but it starts to regulate the voltage.Is there any high current draw during the power on /off timing? Is short circuit happens during the startup of a device?

Best regards,

Tanvee

Hi Tanvee,

I'm not sure I follow your wording here: "According to me 3.3V supply will inside device max supply value..."

To be clear, VOUT (and FB) are at 1.7V, whereas VIN is ≤0.2V. The EN pin is even lower than that (and ramps slowly to ~2.4V over ~0.5 seconds from an RC circuit).

We have one of our customer's boxes in our lab now, and have been able to see the startup conditions on a scope, but have not been able to reproduce an actual failure.

In fact, because the 5V pin of the HDMI connector is more recessed, I can make the 1.7V condition on the 3.3V rail persist arbitrarily long. That's not likely in real life, but by plugging the connector in slowly, I was able to see the rail at 1.7V for 2.5s before 5V made contact. Typically, it's 10ms to 200ms.

Since this problem is only happening in the field after approximately a year, it may be difficult for us to reproduce conditions that kill the regulator. We are thinking that it may be "death by 1000 cuts" rather than a single power cycle. Is it possible that a VIN protection circuit might be sightly damaged each time?

There shouldn't be much power draw for at least 300ms. However, it may be that because the 3.3V rail is only 1.7V across our entire board, there may be a brown-out condition that could lead to metastability. We're not sure. If any transistors are on when they shouldn't be, there could be low impedance to ground there, which would draw more current.

What we have seen on at least 3 regulators that have been returned to us by our customers is that VIN of the regulator is permanently shorted to GND, and 2 of them have physically smoked. We don't know if the permanent short was created during startup, but that's the only time that VIN is low while other pins are significantly more positive. We also don't know whether or not there are other failure scenarios.

Thanks!
- Tones

Hi Anthony,

Normally, if a device is smoking out than the input voltage is quite higher than the absolute max value of the device. It might be because the device is quite often power turning on/ off. Another reason is the reverse current flow from the output to the input side. as explain this app not: https://www.ti.com/lit/an/slva730/slva730.pdf?ts=1594288191503&ref_url=https%253A%252F%252Fwww.google.com%252F

Is there any ESD protection place before and after the regulator?

Best regards,

Tanvee

Hi Tanvee,

Sorry I missed this when it came in.

We do have ESD protection on the two possible regulator input supply rails (5V_PWR, VBUS).
Since there's nothing connected directly to the outside world on the output, there's no ESD protection there.

While we never expect VIN to be more than 6V above GND, if VIN is at 0V while the output is at 1.7V for any length of time, then the output is certainly higher than VIN+0.2V. How long should it take for that condition to cause damage to the device? If it has that condition for any short length of time (say, 50ms-200ms), is it possible that a repeated condition would cause cumulative damage?

I have actually forced the condition of setting the 3.3V rail to 1.7V (and therefore regulator SW through the inductor) while VIN is at 0V for arbitrarily long times (seconds or minutes), and haven't been able to damage a device this way. If this is a failure mechanism, why wasn't I able to get it to fail that way?

As an experiment, I put a 300 ohm resistor between the regulator VIN and the 3.3V rail, and VIN easily followed within 600mV of the 3.3V rail. Is there any reason not to do this other than power dissipation of the resistor? I could add a diode, but couldn't that possibly still allow a drop greater than 0.2V between output and input?

Thanks!
- Tones

Hi Antony,

I assume you are using LM3671MF-3.3/NOPB? The device in the title of the post is the 1.8V output voltage version. The damage you describe typically is caused by voltages exceeding the absolute maximum ratings of the device. Tanvee already mentioned this.

I assume you can get the too high or too low voltages during plugging in and unplugging the HDMI cable. I would not expect the lower voltages to damage anything. Your experiments did not show problems as well when forcing the unexpected low voltage condition so I would take this as a kind of confirmation.

Best regards,

Juergen