TPS7A8300 input capacitance affecting power up

Hi Michelle,

It sounds like this problem is not only due to the input capacitors. If you are starting up the system with the part enabled then the supply will have to charge both the output and input capacitors which can cause the startup time to be much slower. You may want to take a look at your output capacitors and make sure there is not excessive capacitance on the output. We cannot recommend removing the input capacitor as the part is designed and characterized with a 10uF input capacitor. However, there are some things you can try if a soft-start capacitor is added it will delay the start up time of the system and should reduce the amount of inrush current you are experiencing allowing power up. If this is not an option for you, your current solution of enabling them at different times is probably the best solution because it will reduce the amount of current that is required for the part to start up.

Best Regards,
Mark Sellers

Hi Mark,

Appreciate the feedback. We can power our board without turning all these on. We can then use SW to switch on all LDOs. Therefore, the output caps shouldn't be the problem. We are also using the recommended 22uF cap on the output as well.

So, even with 36 LDOs, we still need a 10uF cap on the input pin for every LDO?

How much inrush current is expected per LDO? I would like to again confirm the amount of current on our 4.38V power plane.

Thanks,
Michelle

Hi Michelle,

"We can power our board without turning all these on."
Does this mean that during power up all the LDOs are disabled?

If this is the case and the 4.38V rail still powers up, then having the 36 10uF capacitors should not be affecting the startup.

When using SW to switch on the LDOs how many are being turned on at once? The more LDOs turning on simultaneously will draw more current from the 4.38V rail.

Each LDO should still have a 10uF capacitor on the input to ensure the datasheet performance, as the parts were designed and tested using a 10uF capacitor.

The inrush current for each LDO can be on the order of Amps. If the input capacitors are already charged when the LDO is enabled the input capacitor should assist in supplying the inrush current.

Best Regards,
Mark

Yes, we control the LDOs with software. We want our SW to be able to turn all these on at once. However, when we do this there is too much of a surge and it shuts off the board. 

I read your recommendation that we need the input caps on every LDO (all 36) and that we cannot decrease the number of these or the value, is that correct? 

Thanks,

MIchelle 

Hi Michelle,

I apologize for the delay with the US holiday and weekend. TPS7A83 was designed, characterized and tested with a minimum of 10uF on the input. As a result, in order to achieve the specifications and characteristics as presented in the datasheet, a localized 10uF or greater capacitor should be placed at the input.

What Mark is pointing out is that if simply sequencing the enable pins of the LDO solves your current draw, the input capacitors are not the issue as the input capacitors will be charged as the Vin rail comes up. When you say that you want to turn on all of the LDOs at once, my interpretation is that you want to turn on all of the LDOs at some point after the input rail is already established. Is my interpretation correct?

Very Respectfully,
Ryan

Hi Ryan, 

Your interpretation is correct. We can boot the board with the LDOs off. Then we can use software to enable all the LDOs at once, but this causes a current surge. So we have to add a SW delay, use less caps or use a different power rail.  

Thanks,

Michelle

Yes. We can successfully power on our board and all LDOs are (intentionally) off. We can then use SW to switch on LDOs. If we use a cmd that turns on all LDOs at the same time, then there is a surge and the boards hang until we do a hard reset. If we use a cmd that adds a small time delay between turning on each LDO we can turn on all 36 and the board is stable.

We are also able to remove input capacitors on the board and once we remove enough, we can use the cmd that enables all LDOs at once (no time delay) and the board is also stable. Our question is if it's a viable solution to use less input capacitors or instead use all the required input caps but at a lower value? We hoped that since there were so many input caps on the same power rail (36) that all wouldn't been required when using 36 of these.

If neither of these are supported we will likely need to use the SW delay to power on these LDOs or explore using another power rail so we don't risk using these parts out of spec.

Thanks,
Michelle

Hi Michelle,

TPS7A83 was designed, characterized and tested with a minimum of 10uF on the input. Thus, a local 10uF capacitor is needed at the input to achieve the datasheet performance. Just for clarification when testing with the 36 10uF input caps was the input rail all the way to 4.38V before the LDOs were powered on?

If you choose to change the voltage rail to a rail with a higher voltage be sure to check that the thermals of the device. There will be more power dissipated in the LDO when using a higher voltage rail.

Best Regards,
Mark

Mark Sellers said:

Hi Michelle,

Just for clarification when testing with the 36 10uF input caps was the input rail all the way to 4.38V before the LDOs were powered on?

.. yes. The board is fully powered and we can use SW at any time after a full boot to power these on. 

So, is the answer that all caps must be used, we cannot use a lower value, and we need to determine how to address this?

Thanks,

Michelle

Hi Michelle,

To achieve the operation as stated in the datasheet, you must use the LDO as directed in the datasheet. In this case, yes you must use the input capacitors.

Please note that since your input rail does startup (per your last post) with all of the input capacitors, the input capacitors are not causing the issue. Since capacitors store energy, they will actually help your 4.38V supply source the current required to charge the output capacitors after the LDO is enabled.

From the information provided on this thread it seems that your issue occurs when attempting to startup all of the LDOs at the same time, after all of the input capacitors are already fully charged. This implies that your 4.38V supply (or possibly a supply further upstream) cannot source enough current to satisfy the inrush and load current of the LDOs (and anything else you may have on the rail). Since you can startup all of the LDOs at individually, it is the combination of the inrush current of all of the LDOs starting up that is causing your issue. This indicates a couple of courses of action that you can do:

1) Increase the current of your upstream supplies in order to supply the inrush current
2) Increase the input capacitance in order to satisfy the inrush current
3) Decrease the output capacitance to reduce inrush current (output capacitance cannot be reduced below the minimum capacitance stated in the datasheet)
4) Sequence the LDOs so that they do not startup at the same time. This will reduce the peak current that your 4.38V supply must source at a given time.

Very Respectfully,
Ryan