Other Parts Discussed in Thread: , TPS65910, TPS650250
what is the maximum deglitch time for the PWR_EN pin (pin 9)? The datasheet mentions 50 mS TYP. ("not tested in production"). (page 14/91).
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.
Catalin,
If the datasheet does not give a minimum or maximum value, then I am not able to speculate what this value could be. Most timing parameters are not testing in production because the entire test could be shorter than a 5 or 8-second timer (the TPS65217 has both!)
When PWR_EN is transitioning from Low to High, it needs to be high and stable withing 5 seconds from the beginning of the power-up sequence (as shown in Figure 3 on page 19 of the datasheet). Compared to 5 seconds, 50ms is a negligible amount of time and the maximum will not be 2 orders of magnitude higher than the typical (5s/50ms = 100).
On a falling edge of PWR_EN (High to Low), the power-down sequence is executed immediately after the de-glitch time expires, but generally speaking the falling edge is not as critical for de-glitching. A High-to-Low transition is usually fast and deliberate.
Are you having any specific issues with the TPS65217 device that are related to the de-glitch time on the PWR_EN pin?
Thanks for all of the detailed info Catalin. It may take me a couple days to analyze your data and try re-create it on my lab bench.
I should be able to reply by Monday at the latest.
Hi again Brian,
As I mentioned, the PWR_EN deglitch time variations can be very easy replicated on a TPS65217EVM module, just by using the PWR_EN pin (pin2 on the JP9) and taking it LOW while monitoring the DCDCx
For the over-shoots on the DCDC3, it become quite obvious that the slower the Vin decays in the UVLO (3.3V) area the easier it is to replicate the overshoots. Basically a slower decaying Vin (5.2V) reaching the UVLO will dramatically increase the chances that an overshoot on DCDC3 (and DCDC1) occurs.....I got to a point where I can replicate it very easy (1 in 3 power-down...) with a very slow decaying Vin (5.2V). Basically I went the other way...so, I made sure that the Vin (5.2V) will go down very sharp at power down and the chances of replicating it are less that 1 in 100 power downs....
regards,
Catalin,
I apologize for the delay, but I am unable reproduce your issue on my lab bench setup using the TPS65217CEVM.
The scope is setup just like your attachment and both rails (VDCDC3 and LDO3) go low approximately 50ms after PWR_EN goes low, and VDCDC3 never spikes up above its setpoint.
Is it possible there is something unique about the way you are testing the EVM?
Are you aware that PWR_EN can toggle an infinite # of times in 5 seconds after going high the 1st time and all toggles will be ignored until the 5 second timer expires?After the 5s ends, the next low edge causes the power-down sequence. This timer runs regardless of whether or not the input voltage has recently powered-on.
The same thing goes for when PWR_EN toggles low: PWR_EN can toggle many time until the 1s delay expires before the next low-to-high edge is recognized and the power-on sequence occurs.
Trying to run tests quickly on the EVM can lead to assumptions that the device is not working, but that is only because there are a lot of delays in the part. Edges are captured while the delay is running, but nothing will occur until the delay expires. Trust me, I have been there before and made many of these assumptions myself.
I hope this resolves the question related to PWR_EN.
However, I still firmly believe the spike on DCDC3 is related to the processor and improper sequencing. It is possible the PWR_EN issues were causing this improper sequencing, but I cannot say for sure.
Any rail rising unexpectedly on the PMIC is usually traced back to a leakage path through the processor and is not the PMIC spontaneously changing voltages.
AA,
The PWR_EN pin is "level-sensitive" in the sense that the pin can toggle during a delay timer (1 s or 5 s) and the PMIC will recall that the toggle on this pin has already occurred when the timer has expired. The next State in the state machine will be entered as long as the pin is held in the new state.
If the pin returns to its previous state, a new edge will be required to transition to a new state in the state machine.
Catalin,
We have observed situations like this before where there is no battery in the system and the input power supply can accidentally be removed/re-applied multiple times in a short period of time.
These scenarios have all been resolved with an inconvenient but highly effective external circuit placed in front of the TPS65217 AC pin.
In the Application Note (link below) that addresses a different but similar topic (known as VIN brown-out), this circuit is shown in "Figure 10. Solution Circuit Number 3" on page 10.
Another option would be to add a lot of bulk capacitance at the AC pin, but this will slow down the ramp rate which could cause a different problem: the datasheet recommends a rise time of <50ms on the AC pin.