TPS65131: The negative output cannot keep regulated

Hello John,

Can you post some scope plots that demonstrate this behavior? what would be of interest is what happens with the power stage (OUTN, inductor current, output voltage) while you are increasing the load? We need to figure out whether this is related to the control circuitry or something else (measurement setup, layout, BOM..) The schematic looks ok from this point.
Are -2.5 V and 18 mA fix values or do they change? Is it possible to check if this happens while the positive rail is disabled?

Thank you.
Best Regards.
Ilona

Thank you. Here is the PCB layout (pls note, the vias that grounded the thermal pad beneath the IC is not shown in this picture):

Here is the waveform on switch node (INP, OUTN) with +/-18mA load:

here is the waveform on FBN with load:

Obviously, the inverter power switch only working at very low percentage of time though the feedback voltage is consistently kept on positive level.

Regards,

John 

Hi Team,

Any suggestion about this problem?

Thank you,
John

Hi, 

I have done some further tests to help to find out root cause:

1. Inductor current (yellow) and switching node (OUTN pin, blue) waveform. inductor current is derived through a 0.27 ohm serial resistor between inductor and GND. You can see the switching activity stopped due to some reason though the peak inductor current is only arround 0.5A. What is the cause that stops the switching operation though the negative output is far from regulated?

2. Zoomed in: (seems working properly?)

3. Even more strangely, The DC/DC, either positive or negative output seems working properly once I reduced the input voltage down to 3.3V. Below is the waveform (inductor current and switching node voltage). The negative rail may keep regulated -7V output with load from zero up to more than 150mA and 3.3V input. Whereas the -7Vout can only keep regulation with load < 9mA and 5V input voltage.

Thank you for help!

John

Hi,

Thank you.

1. With 5V input and -7V is loaded with 10mA (actually negative out dropped to 0.6V).

Here is the VIN scope capture with DC coupling: (there is not any significant drop. The scope kept un-trip mode if switched to single trip mode)

switched to AC Coupling and zoomed in:

(the periodical glitch on VIN is due to the booster periodically stops switching)

INN & VIN:

2. With 3.3V input:

Hello John,

To be honest, this behavior is really strange. Let me try to summarize your observations:

1. The positive converter works properly.

2. The negative converters seem to work properly when the input voltage is reduced to 3.3V.

3. The output voltage of the negative converter is correct, when it is not loaded.

4. The output voltage of the negative converter drops when loaded with a few mA and the input voltage is at 5V.

At the moment I do not have a good idea about the root cause, other than layout issue. I think that there is a noise issue caused by the current in the buck-boost converter influencing the grounding or AVIN of the IC.

For getting closer to the root cause, you could check how the IC behaves when you disable to boost converter and just operate the buck-boost converter and in addition check if the behavior changes with PSN connected to GND or to VIN. Please observe as well the CN pin and connect the GND side of the probe to the AGND pin of the IC.

Best regards,
Brigitte

Ok, I will try. However, as a QFN device and customer PCB is very tight, it is very difficult to rework the connection.

Thank you,

John

Hello John,

Could you please change the resistor-network of the negative converter? Use 10x higher values for the resistors, means R4= 100kOhm and R3 = 570 kOhm. This will ensure that the current drawn by the REF pin is not bigger than 5µA.

Let me know how it behaves.
Best Regards.
Ilona

The datasheet requires the current >=5uA. I had checked the reference voltage, found the reference is always correct no matter the output voltage is regulated or not. So I do not think the divider resistance is the problem.
Thank you,
John

Hi,

Here is the waveform on CN and OUTN with 5VIN and 10mA load. Seems the abnormal behavior is correlated to CN voltage. After several cycles of switching, The CN pin voltage drops to zero suddenly and switching stopped while the output voltage is far from regulation (see below CN vs FBN waveform). 

CN (yellow) vs. FBN (blue):

Thanks,

John

Hello John11513,

Please measure as well the voltage on VREF pin. Ilona is correct that the output buffer of the VREF pin is not designed for driving too high current. The current should be somewhere between 5uA and 10uA, otherwise the voltage reference might drop.

The observation on CN is the other way around, when the IC stops switching, CN is discharged.
Please measure FBN with no load on the output.
Please measure FBN with 3.3V input voltage and load on the output.

Up to now, I still do not have another idea but layout issue with grounding. To make sure that the external components are ok, you could solder them onto the TPS65131 EVM and check if this still works properly.

Best regards,
Brigitte

Hi,

I have changed the negative out feedback resistor to 130k/750k, such the VREF load is 9.3uA, while no any change with the problem.

There is a small difference with this higher feedback resistance: the -7Vout is regulated if the scop is not hooked on and no load applied. While the output wandered around -7V once the scope gauge was hooked on (only at 5V input, no such issue with 3.3V input).

Here is the FBN waveform at 5VIN, no load:

FBN, 5VIN, no load, zoomed in:

FBN, 5VIN, 10mA load:

FBN, 3.3VIN, 10mA load:

ALso, I measured the VREF with 5V input and 10mA load:

Thank you,

John

Hello John11513,

Please exchange the IC with a new one and check if the failure still is present. The only other explanation I have is that the layout is not good and therefore I recommend using the EVM to check that the problem is no longer present.

Best regards,
Brigitte

Hello John11513,

Did the swap test solve the issue?

Best regards,
Brigitte

Hello John,

Thank you for the update. I will close this case then.

Best regards,
Brigitte