TPS65131: Abnormal overshoot on positive 15V rail

See below for scope plots noted above..

TI Customer Support reference is: CS0041728

Fig1 (good board): +15V output rail ripple +/- 30mV.

Fig 2 (bad board #1): C1 = +15V output rail (overshooting by ~1V).

 Fig 3 (bad board #1): CH1 = +15V output rail; C2 = switch node; C3 = inductor current

Fig 4 (bad board #1): CH1 = +15V output rail; CH3 = inductor current - note output current being dumped back into input.

Fig 5 (bad board #2): +15V output rail with 250mV overshoot.

Fig 6: 'INP' (+15V) switch node at 300mA load

Hi,

I have notified our expert regarding this topic. Please expect a response by 6/26/19.

Thanks,

Aaron

Hello Stu,

Thank you for your detailed problem explanation and scope plots. Can you please also share schematics of the TPS65131 section of your design?

Also, if you verify that the this issue follows the TPS65131 device in an A-B swap i.e. good board becomes bad and bad board becomes good, we may need to get one or two of the problematic device back to debug this further in our labs.

I look forward to you response.

Kind Regards,

Liaqat

Hello Liaqat,

Thank you for looking into our issue - please see below our TPS65131 schematic and top layout (layer 2 is an uninterrupted ground plane).

Kind regards

Stu

Hello Stu,

Capacitance at boost and inverting buck-boost converter output is lower than than recommended int he application schematics in the datasheet. Your schematics shows less than 10uF and datasheet recommendation is for 22uF so you can try to add capacitors in parallel to  C221 and C212.

More importantly though, I requested some one in product engineering to look up the lot code information you provided and his response is "I did a search for the Lot Trace Code and didn’t come up with any results.  That means they could be counterfeit parts".

So it is a real concern. Where did you purchase these parts?

Kind Regards,

Liaqat

Hi Liaqat,

Thank you for your feedback.

We've already increased the value of the capacitor on the 15V rail but it didn't fix or improve the issue - but will check again.

I've also experimented with changing most of the other component values (inductor included) to try and find out if there is any sensitivity to component variation but with no improvement.

Our PCBA assembly house procured the parts from RS Components in the UK. Our supplier has already asked RS for traceability information, but it has not been received yet. Our board supplier is also expecting a delivery of TPS65131s from another IC distributor (Digi-Key, USA), for comparison tests.

We are still in the process of getting a faulty part fitted to a working board.

Are you aware of any counterfeit parts being in circulation?

Thank you

Stu

Liaqat - please see two X-ray images below of faulty part before being removed from PCB..

Hi Liaqat,

Please see photo of the package below - should it have the TI logo?

Thanks

Stewart

TI

Please note that this issue is not resolved, contrary to being tagged 'TI Thinks Resolved'

Please update the page

Thanks

Stu

Hi Stu,

I am not aware of the counterfeit parts in circulation but it is possible. Our QA engineer could not find any information for the lot code you provided and every lot produced by TI is traceable.

TI logo is not supposed to be on the package of this device.

I think we need to wait for the devices from Digikey to see if any of those exhibit this issue. These devices seem suspect to me. Also, let's see what the results of AB swap are.

Kind Regards,

Liaqat

Liaqat,

Further to your earlier request, I added a 10uF ceramic in parallel with the two 4.7uF already fitted to the 15V rail and this exacerbated the problem (plot below). Again, even with the extra cap, cooling the TPS65131 fixed the problem.

For control loop stability, is there a maximum value of output cap that the TPS65131 can cope with?

RS Components still have the reel that these devices came from and are trying to get a batch code and check that they are genuine.

Thanks

Stu

Faulty board with additional cap on 15V rail. 1.4V pp ripple

Hello Liaqat

RS Components still have the reel we got our parts from and have given me the batch code, 6240118t46. Would you be able to trace the parts back to TI from this code?

Thanks

Stewart

Hello Stewart,

I have sent this information to quality engineer responsible for this product to see if the batch code information you provided can help in tracing. Will let you know as soon as I hear back from him.

Kind Regards,

Liaqat

Thank you, Liaqat,

I purchased several TPS65131 from RS Components which, according to RS, should be from the batch I sent you the code number for. Please see photos below of one decapped IC - maybe this will help with die verification.

Regards

Stewart

Thanks Stu for additional information.

To debug this issue further, I suggest to first conclude:

1. Results of AB swap to see if good board becomes bad and bad becomes good.

2. How devices purchased from Digikey are working.

If devices from Digikey work well, I would suggest to work directly with distributor to replace the inventory you have.

Kind Regards,

Liaqat

Hi Liaqat,

1) We are expecting the board with the AB swap to arrive today for our evaluation.

2) The board build with the ICs from a different supplier will start soon.

3) It would be beneficial to get feedback from TI's quality engineer on the authenticity of the RS parts. RS are also very keen to find out where they stand in regard to these parts.

Thank you

Stu

Hi Liaqat,

We just received and tested a good board which received a TPS65131 from a failing board and found that the fault follows the part - scope plot below.

We have also confirmed that two boards with apparently faulty parts can be fixed by replacing the part.

We estimate that our next build of 10 boards will arrive in one week, with parts fitted from a different distributor.

Please try and find out as much as you can about the history of the parts supplied by RS Components. 

Thank you

Stewart

Hi Stu,

Our QA engineer got back to me and mentioned that based on the latest tracing information you provided, it is a TI manufactured device and was manufactured in 2009. We have no indication that there is anything wrong with this batch of devices but it is interesting that a device manufactured 10 years ago is still in distribution channel.

Based on the AB swap information that shows that problem follows the device, it appears that this is a quality issue. Quality returns are handled through distribution channels so you will need to return faulty devices to the distributor who in turn will initiate a quality return to TI.

Kind Regards,

Liaqat

Hi Liaqat,

Please thank the QA engineer for the information.

I'm very surprised that these devices were manufactured in 2009. Can you find out if there have been any die revisions since 2009?

I am now very concerned that going forwarded we could be designing in old parts, with potential issues, irrespective of the distribution channel. How can we ensure that we are getting the latest silicon revision in the future? Should we be buying directly from TI ?

We will not close this issue until evaluating TPS65131 parts from another distributor.

We will also try and get some faulty TPS65131 back to TI via the distribution channel, if we are not allowed to send them directly.

Thank you.

Stu

Hi Liaqat,

Our board builder has received alternative TPS65131 parts from Mouser and we are starting a build with these parts.

Mouser provided the following information for the parts. Would you be able to check that these parts are authentic? I'm assuming the date code is YYWW, but not sure about the +5.

Date Code           1817+5

Lot Code              8665024T48

Thank you

Stu

Stu,

Please let me know how the new build turns out. In the mean time, I will try to find out if there have been any revisions to the die since 2009. I believe you can buy this device on TI store at ti.com.

Kind Regards,

Liaqat

Hi Liaqat,

We've done some further bench investigation and it appears that TPS65131 power saving mode (which we are using) is causing the TPS65131 15V rail to overshoot excessively (1,3V), as seen on some RS sourced parts. We disabled this mode by cutting and strapping a PCB trace and this removed the 1,3V overshoot completely (output regulation now <20mV pp – see below plots).

With power saving mode disabled, the TPS65131 output is stable over temperature. With power saving mode enabled, the TPS65131 is extremely sensitive to temperature variations, getting better at cold. We have also seen this sensitivity to temperature change with TPS65131 parts which do not have excessive output overshoot when being used in power saving mode.

We received two boards fitted with TPS65131 parts from Mouser and these looked better in power saving mode. Heating the parts still showed some output regulation temperature dependency but it was limited and appeared controlled (max 250mV PP ripple at hot – scope plot below). We are now waiting on another 30 boards with the Mouser parts to check for consistency.

I have also come across another TPS65131 user who has seen a similar issue with this part. 

With above said, I’m trying to understand the root cause of this issue:

1)   Is it possible that the problem is just a characteristic of the TPS65131, which has not been fully characterized by TI? Maybe the IC design is just extremely sensitive to process/voltage/temperature/PCB layout/component tolerance variations etc, in power saving mode. I expect that the ‘failing’ parts would pass TI’s ATE tests if returned to TI for investigation.

2)   Could this problem be so subtle that it is very difficult to detect at TI’s final test – these parts are likely not tri-temp tested?

3)   Could the shelf life of the parts we got from RS be a contributing factor?

Please let me know your thoughts and if you are able to help us further on this matter.

I have managed to get the contact details of TI’s Global Key Account Manager for our parent company and will keep him informed of our intentions in regards to our future plans for the TPS65131.

Kind regards

Stu

Fig: TPS65131 from RS, power saving disabed, <20mV pp ripple on 15V rail

Fig: TPS65131 from Mouser, powers saving enabled, hot test, 250mV pp ripple on 15V rail

Hi Stu,

Power save mode in general is supposed to have higher ripple since it operates the device in PFM mode where switching pulses are generated in bursts followed by a period of no pulses. But your application with some specific devices seems to have quite a but higher ripple than what is normally observed. To me it seems that the these devices have some issue and at this point I suggest that you return the devices in question to the distributor who in turn can initiate a quality return with TI. I am not sure how else can I help further on this issue.

Kind Regards,

Liaqat

Hi Liaqat,

Thank you for summing up the issue.

As you are aware, it will take some time to get parts back to TI and for TI to perform a failue analysis on them to determine the root cause of the overshoot isue - in my experience this could take many months. For this reason, we are now considering changing our design to disable power saving mode as we are nearing production. If you could find out if TI have put containment in place to screen for this overshoot issue, we would have more confidence using power saving mode now and in the future. As you will appreciate, it is a real pain for us to screen for this issue at our end.

Kind regards

Stu

Hello Stu,

Since the issue has not been confirmed yet in our labs, considering a screen at this time is a bit premature. First step is to replicate it in our lab so I would still suggest to return the faulty devices to the distributor and ask them to start a quality return process with TI. Since we have discussed this issue at length and have concluded it to be quality related, I am going to close this thread. If you have any other problem or question, please feel free to open another thread.

Kind Regards,

Liaqat

Hello Liaqat,

From your reply, I have to assume that TI has not seen this problem before and not done any FA on devices with this issue, which is good to know.

Did you manage to find out if there has been any die revisions since 2009?

I'm assuming my question about the maximum tolerable capacitance on TPS65131 outputs could not be answered.

Thank you for your help in this matter

Stu

Hello Stu,

Recommended capacitance at the boost and  inverting converter outputs in the datasheet is 22uF and we do not expect any problem with that much capacitance. I will not be able to answer about maximum tolerable capacitance since that would require a through characterization at higher capacitance to ensure that there are no issues.

As I suggested earlier, we will be happy to look into the issue you are seeing if you can arrange to have some devices returned to TI through the distributor.

Kind Regards,

Liaqat

Thanks Liaqat,

The reason I ask is that we actually have 28uF of capacitance on our 15V rail. Two 4.7uF caps at the TPS65131 and another four close to the switching loads on the PCB.

We are getting 30x  boards next week with Mouser devices fitted - if we see any sleep mode issues with the Mouser TPS65131, I'll return to TI for FA.

Please close this post for now.

Best regards

Stu