TPS650864: PMIC o/ps are not coming up

Hi Prema,

I see PMIC Buck1 is configured for 5V output but the external FETs are also supplied with 5V, there needs to be some headroom for it to get to 5V output. 

Try connecting U5 supply to same voltage as Vsys and see if you get the 5V output at Buck1.

Sathish

Hi Sathish

Yes we did that and then we are able to see 5V as I have mentioned in my query above.

" 

• Since 5V PMIC output is an issue, we isolated boost 6V by removing R24 and connected VIN_OV/UV_5V to TP41. The input voltage we increased to 5.7V for now , then we are seeing 5V and 0.85V PMIC output. But the other rails are not coming up. We have not used buck 5 o/p and hence left LX5 unconnected and FB5 grounded as per PMIC schematics checklist. But want to know if this 1.2V rail has dependency on the other rails as the power sequence in datasheet appears that 1.2V sequences other power rails. Please let us know is there any way to bring up 1.8V (buck6)without 1.2V rail(buck 5).

Could you please reply to the second part of our query on the unused buck5 port. We are unable to get the rest of the voltages other than 5V and 0.85V from PMIC. Does the unused port required in power sequencing?

Hi Satish

we did some rework for buck5 (by providing inductor and capacitor and connecting feedback) and now the PMIC o/p rails coming up. But when we connect the load , the rail voltages are dropping. Is TPS6508641RSKT capable of supporting XQZU19EG-2FFQB1517I?

We have done PI analysis and ensured our layout is capable of handling the required voltages at the current rating. Following are the readings: Please let us know what could be the issue.

PMIC o/ps: buck 1; 4.87V

buck 2: 0.827V

buck 3: 1.09V

buck 4: 3.03V

buck 6: 1.72V

LDOA1: 1.62V

SWB1/SWB2 : 1.32V

external buck for VCCINIT 0.85V: 0.786V

Also , one more concern, the impedance on the external buck 0.85V rail is only around 1.5 ohms and also PMIC 0.85V rail impedance is around 6 ohms

Hi Prema,

It is good to hear that now you are able to power up PMIC.

Where are you measuring the voltages after load is applied? At the point of load there will be some IR drop.

Sathish

Hi Satish

We are measuring near the PMIC.The supplies are required to be within 3% tolerance.

Prema,

For Buck3, 1.1*0.97=1.067V, you are measuring 1.09V which is above 97%.

I see for external buck VCCINT of 0.85V, this is not the case, but could you please check that with respective Buck regulator team?

For Buck6, the load switch could be causing the additional drop.

Do you have the kelvin connection to the supply to make sure the supply voltage is not dropping? 

Sathish

Hi Satish

we haven't used external switches. I have attached the sch again. 

Now we are doing load testing. We isolated R32 and provided 300mA load and we are Geeting the below waveform. Please let us know what could be the issue.

1665.PMIC1.pdf

Hi Prema,

Is the thermal pad of PMIC is solderdown on to the PCB?

Does increasing input voltage to above 6V help any on the above Buck1 output voltage?

Sathish

Hi Satish

We replaced the PMIC itself and checked but same observation. One more board also we checked and same observation. Increasing Vin didnt help.

When R32(pmic 5V isolation resistor) is not mounted 5V pmic o/p looks fine. But when R32 is mounted the following is the observation, o/p 5V is having same frequency as drvh coupled to it. Due this we are seeing all other rails also the same issue.

5V signal 

SW1:

DRVH:

In layout we have used 4mil trace width on inner layers for DRVH. Could this be the issue? Please let us know

Hi Prema,

I don't it is the problem with Boost converter, i think we need that for VSYS of PMIC to be at 6V. So, i think R24 needs to be populated and VDD_6V should be applied to VSYS pin of PMIC.

The issue of PMIC Buck1 not coming up is because U5 Drain is connected to 5V while Buck1 is trying to output 5V. Try connecting U5 Drain to same voltage as VSYS (6V) and check.

Sathish

Hi Satish

The above scenario is with boost isolated. We are giving 6V to PMIC and also to MOSFET drain. We removed R24 and connected TP40 and TP41.We are providing 6V to VIN_OV/UV_5V. Why is the switching noise coupled to the output? 

Can i share our layout file for your review on PMIC? PLease let us know

Hi Prema,

So, is it okay to apply 5V at VIN of U2 and 6V at TP40?

Please confirm that PMIC powerpad is solderdown to the board.

Buck1 output is the input supply to Bucks345. Please confirm if there is additional load at TP6 or TP9? How do those voltages look like?

Please go ahead and send me the layout. There could be multiple issues, but i am trying to understand why Buck1 output is not stable.

Are you using the default OTP settings of PMIC or anything changes?

Sathish

Hi Satish

We are now giving only one supply 6V,so Vin of U2 is itself 6V. Once we are able to power up FPGA with 6V in,then we will check for 5Vin after correcting mosfet U5 drain to 6V.

We just found that AGND is connected to thermal pad. We are planning to cut the trace and connect to gnd of nearby decaps. Can this be the issue for PMIC instability? 

Also please let me know how to share the layout file only to you instead of this e2e forum.

Prema,

I sent you a friend request on e2e, where you can share the files? What format is the layout file in? 

At some point AGND needs to be connected to power GND using Net Tie. But, how the cap GNDs are reference is important. All the power traces need to be connected to power GND.

Sathish

Prema,

With 300mA load at Buck1 output, when it is dropping to 2.5V, how does the 6V looks like? Is there any current limit on the supply?

Sathish

Prema,

I think, i found the problem.

For Buck5, external components are added but is FB5 still connected to GND?

When FB pin doesn't detect the proper voltage, then it causes emergency shutdown of PMIC and it because CTLx pin high, it tries to power up again and the cycle repeats ~every 10ms.

Sathish

Hi Satish

We cut the trace for AGND connected to thermal pad after removing PMIC IC. Later we connected the agnd to vref cap gnd. But still PMIC is unsatable. The feedback signal of buck 1 is below.

Hi Satish

FB5 is disconnected from GND and connected to the external components(capacitor)

Hi Prema,

10ms re-try indicates that one of default power rail is having issues with not able to come up.

Please refer to the below FAQ and towards the end there is list of outputs to check in the order. Please check all the default rails are coming up wrt to CTL pins and check which one is causing the shutdown. There is a link to another E2E post within that.

https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/992402/faq-tps650864-regulator-is-shutting-down-or-measuring-x-lower-than-expected---what-is-the-cause/3665210?tisearch=e2e-sitesearch&keymatch=TPS650864%252520faq#3665210

Sathish

Hi Satish,

We have set 3A on 6V

Hi Satish

We have left LDOA3 also NC. we will try connecting 4.7uf and check if this helps

yes, every default rail needs to come up.

Sathish

Hi Satish

we tried connecting cap to LDOA3 and it didn't help. we are still seeing the same kind of waveforms. 

Were you able to review the layout? Any feedback on that

Prema,

10ms re-try indicates that a power fault on a rail causing the PMIC to initiate shutdown.

I see C54 cap on SWA1 as DNP. Is this populated? Can you please try this?

If it didn't work out, please send me the scope plot of all the rails including CTLx pins as shown in the power up sequence in the datasheet, including VSYS.

Sathish

Hi Satish

We already have an existing board with the same PMIC part number. SW1 capacitor is not mounted. It works fine. So I don't think C54 would help. SWA1 is not part of power sequencing also. Anyways we will try it. Also I'll send all the signal waveform in the power sequence tomorrow.

Today we tried to correct all the feedback traces to connect near load capacitor at the load point. We disconnected the trace connecting near PMIC and did this rework. Now we are observing the below waveform. the freq has changed. Does this provide any hint what could be the issue? 

Prema,

correct, SWA1 is not needed. Where did the FBx pins connected originally? Can you please check all of these are corrected?

Sathish

Hi Sathish

Yesterday we did feedback correction and our PMIC is up now and stable with 6Vin for both PMIC and Mosfet drain