TPS62745: TPS62745

Igor,

It is probably difficult at this time.  We can only recommend that teh abs max rating be observed.  I will contact our design team about internal details.  For sure it will require an NDA to share any details.

Thanks for your complete and so deep reply John.

It seems that our board don´t have to face many problems working below 7V (maximum 1 per billion).

After your answer, we have tested today 200 boards, just connecting 6,3 V Lithium battery, and 28 units are damaged, so current consumption increases 10-100 times. After changing TPS62745 regulator it works fine returning to the normal current consumption.

So, do you think there is another possible problem, maybe on our schematic? Or could be possible to be a batch problem on these regulator, I don´t think so, but do you have any information about it? 

Thanks in advance.

Hello John,

Thanks for your answer.

So, if we are working under any probed specifications, there is no explanation for having to replace your regulators on some of our devices in order to work properly.

If you are able we can share our schematic with you, maybe there is something wrong that we cannot detect.

BRs,

Igor,

if you post your schematic, I will review it.

Hello John,

Attached you can find schematic referred to TPS62745. Where D1, D2 and T4 are not mounted. So J2-battery power go directly to R2 0 ohm. 6V lithium battery is connected to J2 in normal activity (initially 6,5 V).

In parallel 3V3 are applied to Vout from TC1 connector when device is programming. Sometimes could be 3,4 - 3,5 V.

Can you see any issue on that schematic?

Thanks in advance.

BRs,

Igor,

If you apply an external 3.3 V to the output of the TPS62745, and the battery is not connected the high side body diode can conduct and you may possibly have enough voltage at the input to allow the device to start switching.  Can you check it in that condition?

Hello John,

I have checked what you said and there are 0,25 V on the high side body diode along the range of 2-4V applied externally on Vout / SW pins.

So, in normal situation, when we apply external 3,3 V at the V output we have 3,05 V at Vin pin.

If we apply external 3,4 --> 3,15 V at Vin

If we apply external 3,5 --> 3,25 V at Vin

So, it could be a problem to damage the chip?

Thanks

BRs,

Igor

Igor,

Do you see any switching activity on the SW pin?  Is EN tied low?

Hello John,

Attached you can find both signals when connecting 3,5 V to Vout, on an already damaged chip.

Vout & SW --> GREEN

Vin & EN --> YELLOW

BRs, 

John,

Here you have Vin/Vout signals when connecting 2V5 to Vout on a non damaged chip.

Vin is 0,25V less than Vout, also decresing Vout to 2V.

With Vin&EN pulled low, connected to GND, there is just 0,25V on this pin after connecting 2,5V on Vout. 

BRs,

Igor.

John,

In our typical procedure, while programming the board by external 3V3 on Vout, EN is not low and couldn´t be low because Vout would be connected to GND through the high side body diode.

It seems that applying some voltage to Vout-SW could damage the chip, if EN is high,.

Am I right or is there any other conclusion?

Thanks.

Igor

Hello John,

Undervoltage lockout threshold is 3,1 V. 

Diode drop is 0,25 V

We can apply 3,1 + 0,25 = 3,35 V minimum at the output for not damaging the chip. Is it correct?

In this case, we would have 3,1 V at Vin & EN. It could be still a problem or there is no risk on this case?

BRs,

Igor.

Hello John,

We have tried to program the chip at 2,8 V and it works. So, we are going to program a bigger batch of boards, check the battery consumption and then I will let you know the results. I hope it should resolve this issue.

In parallel i wanted to tell you that we followed the instructions of TI reference design and smart lock, using same regulator, as you can see in the following links:

http://www.ti.com/lit/df/tidrmq3a/tidrmq3a.pdf

http://www.ti.com/lit/ug/tidubv3c/tidubv3c.pdf

It uses the same regulator and asks to connect the board with CC2640R2 directly from the programmer at 3V3. Same than we did.

Having in account UVLO threshold, it seems that there is a mistake on this TI design reference. Please confirm that in order to avoid future similar issues on new designs.

Thanks.

BRs,

Igor 

Hello John,

Thanks.

What do you mean with VFB?

We are below 2,9 in Vout to avoid problems, as you asked.

BRs,

Igor

Hello John,

Thanks. VFB is V_TH_UVLO?

V_TH_UVLO minimum value is 2,9 V. This is why we have powered 2,8 V at Vout for chip programming, so Vin should be maximum 2,55 V. 

After programming a batch of 200 units at 2,8 V we have less regulators damaged: 6%, but the problem is still there.

Is there any other reason to explain that problem?

BRs,

Igor.  

Thanks for the explanation John.

Both signals are stable during programming:

SW=2,8 V

Vin=2,55 V

After that, we make a system test, where SW voltage decreases below Vin during 100 msecs aprox. as you can see in the following picture:

Is it a problem?

BRs,

Igor

Igor,

I would have expected just DC voltages.  It is difficult to comment without being able to see the amplitude and time scales as well as the ground reference point.

John,

Green signal is Vout=SW. 2,8V

Yellow signal is Vin. 2,55V

During the test, Vout decreases to 2,4 V, and stays below Vin, during 100 msecs, until Vin decreases also 0,4 V (0,25V below Vout).

Scales are 500 mV / 500 msecs.

BRs,

Igor

Igor,

Why do the voltages change?

Hi John.

Voltages change because some LEDs are blinking during system test.

I have been doing same procedure: to power the reader through regulator-Vout to 2,8V and do system test, but after testing 200 readers, Vout doesn´t decrease below Vin on most of the cases.

So, regulator Vout & Vin are stable during programming and also system testing.

There is another reason for damaging the chip...

Are you doing any internal test to reproduce this failure?

BRs,

Igor

Igor,

I am transitioning out od the support role for these devices.  Our applications manager will assign someone to pick this up.  I don't think this can be easily replicated here.  It seems it may be unique to your programming sequence.