TPS51200: VO and REFOUT pulsing when hot

Hi Matt,

Do you have a schematic and layout that I might be able to review? Additionally, scope captures of the device at these temperatures and operating this way would be helpful (VO, VOSNS, VLDOIN, REFIN, REFOUT).

• These temperatures might increase any stability issues as a higher temperature can increase noise or interference within the circuit, which could be an indicator of insufficient decoupling. But thermal shutdown should happen at around 150°C. 
• Instable input supply would also cause an instability on the output here.
• Depending on where VOSNS is connected and if instability is seen here, it could cause this.
• Are you applying a load at this temperature? Changes or undesired fluctuations in the load could also have this effect.

As mentioned earlier a schematic/layout would be helpful in reviewing this. And an oscilloscope capture of VLDOIN, REFIN, VO, VOSNS, and REFOUT would be helpful as well in reviewing this as well, when these cases are seen with this device. Would these be possible to get so I can investigate further?

Thanks,
Field

Field, here is the schematic of the device. I am working to get layout. I have a screen grab of the VO and REFOUT outputs, but the technician used long wires and poor grounds, so it is noisy. I will ask if they have screen grabs of the input voltages but was told they were checked and looked good.

VO:

REFOUT:

Hi Matt,

Thanks for the response! 

• The reason I ask about the input supply is instability on VLDOIN or REFIN would be seen on VO. Additionally, instability on VOSNS could also drive this behavior. This can be seen in the functional block diagram in section 7.2 on page 11 of the datasheet linked here. VLDOIN provides VO. REFOUT (which is driven by REFIN) and VOSNS drive VO. Based on the bottom oscilloscope shot it looks like REFOUT is pulsing and correspondingly VO is. Pulsing on REFIN seen may also cause this. VOSNS may be harder to check based on the layout. But higher temperature could have an effect on these supplies as well and lead to what we are seeing.
• On the schematic VLDOIN is 1.5V, but REFIN is 0.6V from the voltage divider. If you are supplying 1.5V to VLDOIN then this might want to bring VO to a different level than REFIN/REFOUT and since REFOUT drives VO then there might be some stability issues because of this. But this may be seen on VO and maybe not on REFOUT. This may be something to look into, bringing VLDOIN = 1.2V, or this may be an error on the schematic. 
• The layout could be a factor based on capacitors and traces/planes but would like to review this.
• VIN decoupling is very high and may also have higher than normal ESR because of this. This also helps drive the internals. TI recommends a ceramic decoupling capacitor here of 1-uF to 4.7uF here. This may be something to look into, lowering C755 from 47uF to 4.7uF. 
• I can't see where PGOOD goes to but this is open-drain and needs to be pulled up to 3.3V. Just make sure this is the case.
• VO may have a high ESR since there is just one 47uF capacitor, although there is the 0.1uF in parallel. TI recommends 3x10uF to reduce ESR here. VLDOIN may also have high ESR since there is just one 47uF capacitor. TI recommends 2x10uF or up to 1/2*COUT (whichever is larger) to reduce ESR here. REFOUT also has 4.7uF but there have been instances with an increased capacitance helping with stability or overshoot issues. So these may be something to look into, but I would look into the above bullet points first as I believe the issue may lie with one of those above and that these capacitance values may be fine. 

These could all be issues that are minimal during normal operation but then exacerbated by an increase of temperature. There is also the voltage dependency and dielectric absorption of the capacitors themselves that can also vary with temperature and additionally vary between the different types. But you are right that over-temperature condition should cause these levels to fall and not rise, so I don't believe this is the case. Can this problem be seen on one device or multiple? If they have an EVM they could put this device on the EVM and run the same test and see if this was seen. This might help in determining if it has to do with that device itself or by this specific design/layout/schematic.

Please let me know if you can get an oscilloscope shot of VLDOIN and REFIN during these occurrences, as well as the layout.

Thanks,
Field

Hi Field,

I just received another screenshot from the team seeing the issue in the lab. It looks like REFIN is dropping out from 1.2V to roughly 0V when they see the pulse on REFOUT. Also, the 1.5V on VLDOIN is correct. 

vtt - blue

vref - green

Hi Matt, 

Thanks for the quick response and information!

• So this looks like it could be a sourcing/supply issue similar to what I explained above in my previous response in the first bullet point about instability on the input, and the device works at lower temperatures where I assume this is also not seen. This instability on REFIN causes the instability on REFOUT and in turn VO. Like you see in this scope capture, REFIN change causes a change in VO, which is as expected. I'm assuming that VLDOIN and REFIN have different supplies since REFIN is 1.2V and VLDOIN is 1.5V. This means that the problem may lie in the REFIN supply source. I can not speak to the device of this source/supply of REFIN, but there may be some component characteristics that are temperature dependent within it. There could be some thermal expansion, thermal feedback, or thermal shutdown of the supply too. But again, I can not speak for this other device.
• May be that C489 is having issues with the voltage dependency and dielectric absorption that varies with temperature. But I believe the issue lies with the above bullet point in regards to the source/supply of REFIN.

However as noted in the first bullet point, it looks like the device is operating as expected at this temperature, and that a change in REFIN drives a change in REFOUT which drives a change in VO. Can you please look into the source/supply of REFIN and let me know if you find anything? Additionally, you can try this device on the EVM and conduct this testing again to see if there is this issue. This might help demonstrating that/if it is a supply/source issue. 

Thanks,
Field