STM32H725 Vcap pins

[fing]

Hello,

I would like to know whether I understood the application note about the Vcap pins correctly. If there are 3 Vcap pins on the package, then 2 capacitors should be placed and one has to be connected to another. I'm not sure about this connection.

 

[Stassen.C · Accepted solution]

Hello fing,

Yes that's the setup you should be using according to the AN5419. If the LDO is disabled use 100nF capacitors.
Please see below:

Regards,
Stassen

[waclawek.jan]

The documentation is a mess. While AN5419 talks about VCAP1, VCAP2 and VCAP3, the actual VCAP pins in DS are not numbered.

Also, the DS is not clear about which VCAP pin can or must be connected to which other one; footnote 3 talks about "third VCAP". Also, if I see it correctly, there are three VCAP pins in all available packages.

Footnote 1 from the DS, which is also not quite clear in what "When bypassing the voltage regulator" exactly means - "LDO disabled, Vcore supply provided externally" or similar wording should be used explicitly to remove ambiguity (as "bypassing" can also simply mean the fact that there is a bypass capacitor connected).

To add insult to injury, this is the relevant portion of the relevant Disco's schematics:

which appears to indicate that all VCAP can? or must? be connected together.

It would be nice if ST could properly clean up the documentation in this regard.

JW

[Stassen.C]

Hello Jan,

You are right it's a bit messy I will report this and request an update to the Ans and documents. The rule is ,depending on the MCU, you put a 2.2nF per Vcap in parallel, if you have a 3rd Vcap pin you just connect it to the other Vcap pins without adding more capacitors.

Regards,
Stassen

[waclawek.jan]

Hi @Stassen.C ,

> The rule is ,depending on the MCU, you put a 2.2nuF per Vcap in parallel,

So, in 'H725, which - again, according to the DS, in all packages has three VCAPs - *all* VCAPs have to be connected together?

And in other 'H72x/'H3x, if there are two VCAP pins, they also have to be connected together?

Or is such connection optional? If so, to which other VCAP should be the "third" VCAP connected, and which one is the "third" VCAP?

Are there other STM32 models/families, where VCAP pins *must* be kept separated, or *must* be connected together?

JW

[LCE]

I almost had the same question:

https://community.st.com/t5/stm32-mcus-products/stm32h7-without-smps-vcap-ldo/m-p/628432#M232538

@AScha.3 said he has a few PCBs where he did not connect the VCAP pins on the PCB - but placed the caps on each.

And from little bit of info from the datasheet and the block diagram, it does not make sense to connect these, as it seems to be routed completely internally. 
@Stassen.C can you confirm that?

Datasheet: "VCAP: VCORE supply voltage", and either comes from the internal LDO, or with the internal LDO off, VCORE must be supplied externally via these VCAP pins. In case of external supply, it might make sense to connect VCAP pins on the PCB.

Anyway, all too unclear, as @waclawek.jan already pointed out.

PS: it's 2.2 µF per pin

PS2: the H723 / H733 have 2 VCAP pins in the LQFP-144 package

[AScha.3]

@LCE ,

The "view" boards - are about 600 or so, running around the world 24/7 ,  most of them.

But a mix of the CPU there, H743, 733, H7A3 , because we could not get enough CPUs from one type.

[TDK]

FWIW, I have never connected VCAP pins together and I have never had an issue (at first due to misunderstanding and now due to inertia I suppose). Many many designs using STM32F4 and STM32H7 chips. I'm not advocating this, just giving information.

[fing]

Thank you all for the answers, I think I will stick to the original connection.