TPS66020: fault triggering by RCP or OVP?

Hi Fred,

I do not think this is a RCP because if a RCP event occur on PPHV path the /FLT will not assert.

Since /FLT is asserted, the possible fault conditions what will disable PPHV are overtemperature or overvoltage. /FLT will deassert once the fault condition is removed.

Looking at the oscilloscope capture, it seems like the /FLT line is still asserted after the voltage dropped so I think there might be an overtemperature event.

Regards

HI Tommy,

can you confirm our TPS66020 can support higher OVP threshold to 25V, right?

if yes, then I think it is overtemperature event.

BTW,

datasheet mentioned the LDO output will drop when over temperature happen, 

but we didn't see that happen, is there anything wrong ?

they're in a urgent and we need to find what trigger the fault.

Hi Fred,

Is TPS66021 used with a PD controller or it is just being used as a switch?

What is the EN0 and EN1 setting when this occurred?

Regards

Used with PD,

EN1 is high EN0 is low which is sink state.

customer is not asking for debugging , they already have solution.

they just want us to tell them how can they make sure it's OTP that trigger fault,

please sync with internal SE to see if there's a fast way to identify the fault reason

thanks

Hi Fred,

I will reach out to system regarding this inquiry.

Regards.

Hi Tommy

thanks for understanding the situation , since they’re urgent

looking forward to your feedback

Is there a way other than elimination ?

1.customer want to know why they probe LDO but it didn't drop, we should capture  dropping during OTP fault, right?

2. and what is the temperature trigger point  of OTP inside the IC?   100C or 125C?

sorry I have to push on this since they're really in a urgent case.

thanks

Hi Fred,

Yes if this is a OTP VBUS LDO should drop.

Here is the table for the OTP threshold. Please note that the recommended operating junction temperature is -10 to 125C.

Regards

Hi Tommy,

it's weird now, 

customer pulled the OVP pin to ground now,

 but LDO looks stable as far as I can see,

they doesn't connect LDO to any load, 

do you know why this happen if this is not OVP?

it's this because OTP keeps on and off back and forth or is there any other cause?

Regards,

Fred

Hi Fred,

I am seeing PPHV spikes with VBUS drop when SINK_EN is high. I assume at that point of time the fault pin also asserted. 

Can you zoom in the part where the spike occur and get the actual measurement of the PPHV spike and VBUS dip?

Regards 

Hi Tommy,

the picture you ask is already at the very top of this forum,

I'm not sure what are you implying here,

but can we get back to LDO output? how can't they capture the drop at LDO output ,

is this because there's no loading at LDO, so the dropping is not obvious?

please consult internal SE,  they really want to know this answer.

Hi Fred,

To clarify on the overtemperature protection, there are multiple temperature sensor in the device. The local PPHV, PP5V temperature sensor will disable the power path only and assert fault pin if an overtemperature event occur on the power path. The main TSD disable all power and VBUS LDO if an overtemperature event was sensed by the main temperature sensor.

Please also keep in mind that there is a PPHV soft-start that may lead to a PPHV overtemperature protection event.

Regards

So why can’t we capture LDO dropping during fault?

Hi Fred,

The local PPHV, PP5V temperature sensor will disable the power path only and assert fault pin if an overtemperature event occur on the power path. This does not disable LDO therefore you will not see LDO dropping during power path overtemperature event.

Regards

Hi Tommy, 

I see, so the key point is "MAIN" thermal sensor is not being triggered here.

what's the difference between MAIN and local? where or which spot does the MAIN sensor sense?

Hi Tommy,

customer sink heavy load during PPHV soft start,

but they couldn’t believe local sensor could be triggered within 3~5 ms because there’s no way to check local sensor’s status.

would you explain to them how could this is actually possible that the inrush current could generate heat within 3ms?

Hi Fred,

As stated in the datasheet DC loading should be minimized during soft start otherwise the heat generated due to high power cannot be dissipated fast enough.

How much current were they loading during the soft start?

Regards

Hi Tommy,

one question,

if we have soft start function for PPHV which mean it control the current during power up,

but why would OTP still happen if we control the rising current?

Hi Tommy,

as you can see, "when VBUS is already 20V",

the current (green one)  flow through PPHV is around 4.2A at most,

but PPHV still trigger Fault (over temperature?), while p2p competitor can pass this without fault.

do you know why this happen , is 4.2A too high or we have bad thermal layout ?

and how can we fix this? 

Regards,

Fred

Hi Fred,

Yes layout can affect this. 4.2A should have no problem.

Regards

Hi Tommy,

how could we improve it without changing layout since this is customer last build,

how about we enable this right before when VBUS is 5V,

so the voltage difference is smaller than now?

Hi Fred,

That may help but still if there is DC loading while ramping up the voltage it can still trigger overtemperature.

Another way that might work is limit the current draw to 0 during voltage ramp. I cannot tell what is the voltage value from the scope shot and in what stage that is during the PDO contract negotiation. 

You can also test this on EVM first and if this issue does not occur on EVM.

Regards

Hi Tommy,

I’m confused

there will always be current draw during voltage ramp since we need to charge up the capacitor at PPHV

Hi Fred,

Maybe 0 current draw is not the correct term, what I meant is limit the current draw during voltage ramp. The system should not have other active component that is drawing that much current.

What is the voltage ramping up from? Is this during a PDO change? Like 5V PDO to 20V PDO?

Regards

Hi  Tommy.

Yes, the Vbus is already change to 20V Vbus PDO,

but when enable pull high PPHV just can’t rise up successfully as you can see above waveform,

and since the PPHV voltage is smaller than battery voltage, so system voltage is still dominated by higher battery voltage.

so here we can conclude it’s pure inrush current that trigger OTP during initial charging up

can we remove some of the cap on PPHV or charger to reduce inrush current?

Hi Fred,

Yes you can try removing some caps to limit the inrush current but that comes with the trade off of voltage rail stability during transient.

Here is a document that provide some insight on managing inrush currents (https://www.ti.com/lit/an/slva670a/slva670a.pdf?ts=1688996047842&ref_url=https%253A%252F%252Fwww.google.com%252F)

Regards

Hi Tommy,

1. They didn’t see any obvious peak current , does that mean that’s not inrush current or it’s because of our soft start ?

2. They still question why 4.2A output current at PPHV would cause OTP , which is not even reach close to type-C limit “5A” , do you have any idea?

Hi Fred,

If there are obvious peak current then it is most likely not related to inrush current.

Fred Cheng said:

2. They still question why 4.2A output current at PPHV would cause OTP , which is not even reach close to type-C limit “5A” , do you have any idea?

I did not see the layout but I do think this is very likely due to the layout.

Regards

Hi Tommy,

they're asking why they can't see PPHV(yellow) shut off ( drop to 0V) when Fault assert,

since even local temp sensor triggered is supposed to disable PPHV path.

would you help answer customer regarding this case waveform?

BTW, update layout here 

please give some recommendation

 tuffy.7z

you may search U7202 and U7205 at rightmost side of board

Hi Fred,

Can you provide the test parameter/condition of the oscilloscope capture?

In terms of the layout, the PPHV plane lacks vias to dissipate heat to the other side.

My concern with the VBUS trace is that only one side have large enough plane to dissipate heat but the other side might not dissipate fast enough due to single via.

Regards 

Hi Tommy,

 the condition is same as all previous waveform,

the pure question is why we see type-C current on and off which is OTP we knew, but why PPHV doesn't drop to 0V it shut down PPHV due to OTP? instead it just keep at around 7~8V but can't rise up successfully eventually.

BTW, I agree that the other side of VBUS trace is lacking Vias and plane,

I'll ask them to change if they're willing to.

Hi Fred,

The PPHV power path will turn off but if there are large capacitance on the PPHV side the voltage might not be discharge fully.

Regards

Hi Tommy,

I think of a solution,

can we add a 5ohm at the output of PPHV to reduce the inrush current?

since I=Cdv/dt, if we add this R, the charging current slope would reduce maybe help prevent OTP?

Hi Fred,

I don't think that will help but you can run a simulation to see if the load voltage system side will stay at a voltage good level after adding the 5 ohm.

Regards

Hi Tommy 

may I know why you don’t think this is good idea?

Adding this 5ohm also means we reduce the system load current what ever it’s sinking,

if we can reduce it from system then I thought adding this resistor can help,

and it can slow down the inrush current slope

Hi Fred,

You need to consider the IR drop induced by the resistor. If the voltage drop is too much you might brown out the system side.

Regards

Hi Tommy, 

you're right, 

they remove 33uF Cout and increase reisotr to 20mohm  but still fail, 

looks like inrush is not the root cause,

do you think the layout and thermal resistance are the final improvement we can think of?

since once we fail again after re-layout, we will be design out forever,

please check with system engineer on this ,since this topic has gone so long, we need to converge this issue .

Hi Fred,

This part is essentially a power switch and layout is important for power switch performance.

I will double check with system team on other possible improvements. I think it is also a good practice to test out the setup using two EVM first.

May I know who is the customer?

Regads

Tommy,

HP is the customer .

would you help us try from our EVM if you happen to have one on your side since I don’t.

I would like to know what loading will cause OTP when we power up at EVM. 

Hi Fred,

I will let Deepak know about this also since he is the main contact for HP.

Regards