LM76003: Issue with varying output voltage

Hi Will,

Was it fully no load condition when testing power on/off? can you test it with 10mA--50mA load?

The Vout will increase about 1.5% at no load condition 

Pls try below:

1. Short R149 as the Cout should be low ESR capacitor.

2. Remove C109

3. If problem still exist, try to connect Pin 17 SYNC/MODE  to Pin 8 VCC

B R

Andy

Hi, the load is about 200mA.  The overall power input is about 1W.  This is the lowest current the system can draw - I cannot remove any more load without taking off CPUs and FGPAs.

There is 100uF of low ESR caps nearby on the board (see the note in the schematic).  Is this proximity not good enough?

I can try removing C109 and possibly the Pin 17 strapping, depending on whether the artwork will allow getting to that trace.

OK, pls try above suggestion 1&2, you can bias additional 100mA load by resistor or E-load on output, then no need to try suggestion 3#.

B R

Andy

.Step 1) Removed C109.  No change in behavior.

Step 2) Shorted R149. No change in behavior

Step 3) Added 100mA (via 50 ohms resistor).  No change in behavior.

Step 4) Added external load up to 2A total load - output remained stable.  I did not power-cycle the unit at full load.

I took some measurements at the the feedback node (pin 12) and when this unit outputs 5.2V, Vfb = 1.03V, and when it comes up at 5.4V, Vfb=1.07V.

There are no weird oscillations that I can observe under either operating condition.

Can you upload the PCB layout of LM76003 section?

Hopefully this format is usable for you.

LM76003_Layout.pdf

It is an 8-layer board.  There are contiguous ground planes on the 2nd and 7th layers, uninterrupted under the entire PS section.

Thanks Will!

The VCC cap layout was not so good.

On bottom layer (Blue),  the VCC pin first connect to a via, can you try to solder a 0402 or 0603 0.1uF 16V cap between this via and Pin 7 (connected to GND) ? and check if any improvement

B R

Andy

Sorry - we cannot do the detail rework required for that mod in-house.

This problem never occurred on the first units we built a few years ago, and the parts were sourced from either DigiKey or Mouser.

The last two runs, where we have had problems, the parts have been sourced from a reseller in Hong Kong due to the global parts shortage.

Is TI aware of any grey-market parts being sold in this fashion?  They are genuine TI parts as far as we can tell, but we have experienced this issue on 30% of units in our last two builds.  The issue of Vfb running at 1.07V means the part is running at 7% error.  This is well beyond the datasheet values.

Hi Will,

In that case, pls check it with your vendor for product guarantee. 

B R

Andy

Well they of course claim everything is legit.  Which is why I'm looking to TI for assistance.

Hi Will

We had long shipping history of LM76003 and never saw such product issue and failure rate, you can do some IC Failure analysis compared with old good units.

B R

Andy