Raspberry Pi RP2350A SMPS - Inductor Polarity

[EEVblog]

One of my Patrons pointed me to this Raspberry Pi design document that talk about them having issues with SMD inductor polarity on their new design.
How sensitive must the controller (built into the new RPi chip) be to be impacted by this? 
https://datasheets.raspberrypi.com/rp2350/hardware-design-with-rp2350.pdf

The layout of the regulator on the minimal boards closely mirrors that of the Raspberry Pi Pico 2. A great deal of work has gone into the design of this circuit, with many iterations of the PCB required in order to make it as good as we possibly can. While you could place these components in a variety of different ways and still get the regulator to 'work' (ie, produce an output voltage at roughly the right level, good enough to get it running code), we’ve found that our regulater needs to be treated in exactly the right way to keep it happy, and by happy, I mean producing the correct output voltage under a range of load current conditions. While performing our experiments on this, we were somewhat disappointed to be reminded that the inconvenient world of physics cannot always be ignored. We, as engineers, largely try and do exactly this; simplifying components, ignoring (often) insignificant physical properties, and instead focusing on the property that we’re interested in.

For example, a simple resistor does not just have a resistance, but also inductance, etc. In our case, we (re)discovered that inductors have a magnetic field associated with them, and importantly, radiates in a direction depending on which way the coil is wound, and the direction of the flow of the current. We were also reminded that a 'fully' shielded inductor doesn’t mean what you think it might. The magnetic field is attenuated to a large extent, but some does still escape. We found that the regulator performance could be massively improved if the inductor is 'the right way round'. It turns out that the magnetic field emitting from a 'wrong way round' inductor interferes with the regulator output capacitor (C7), which in turn upsets the control circuitry within RP2350. With the inductor in the proper orientation, and the precise layout and component selections used here, then this problem goes away. There will undoubtedly be other layouts, components, etc, which could work with an inductor in any orientation, but they will most likely use a lot more PCB space in order to do so. We have provided this recommended layout to save people the many engineering hours we have spent developing and refining this compact and well-behaved solution.

More to the point, we’re going so far as saying that if you choose not to use our example, then you do so at your own risk. Much like we already do with RP2040 and the crystal circuit, where we insist (well, strongly suggest) you use a particular part (we will do so again in the crystal section of this document).

2.1. New on-chip voltage regulator
The directionality of these small inductors is pretty much universally ignored, with the orientation of the coil winding impossible to deduce, and also randomly distributed along a reel of components. Larger inductor case sizes can often be found to have polarity markings on them, however we could find no suitable ones in the 0806 (2016 metric) case size we have chosen. To this end, we have worked with Abracon to produce a 3.3μH part with a dot to indicate polarity, and importantly, come on a reel with them all aligned the same way. The TBD are (or will very shortly) be made available to the general public from distributers. As mentioned earlier, the VREG_AVDD supply is very sensitive to noise, and therefore needs to be filtered.

We found that as the VREG_AVDD only draws around 200μA, an RC filter of 33Ω and 4.7μF is adequate. So, to recap, the components used will be…
• C6, C7 & C9 - 4.7μF (0402, 1005 metric)
• L1 - Abracon TBD (0806, 2016 metric)
• R3 - 33Ω (0402, 1005 metric)

The RP2350 datasheet has a more detailed discussion on the regulator layout recommendations, please see External Components and PCB layout requirements.

[moffy]

That is a very nasty issue, maybe a case for a small shielded enclosure around this psu and its components.

[floobydust]

You don't want any stray B-field from the inductor coupling to sensitive nodes. I haven't seen any Abracon part numbers yet, the freq., the current.
Having a polarity dot, flipping that - solves nothing IMHO. I think it's another problem that exists. Try applying the Right-hand Rule for more confusion lol.
Many "shielded" inductors use glue with ferrite particles in it.
I would guess having that AVDD resistor R5 nearby is bad practice, and a stability problem with the Vcore reg might be manifesting.
More in 6.3.8.3

edit:
"Even with nominally fully shielded inductors, leakage magnetic field coupling into the loop formed by the output VREG_LX node through the inductor and output capacitor (COUT) seems to affect the regulator control loop and output voltage. Field orientation (and hence inductor orientation) matters - the inductor has to be the right way around to make sure the regulator operates properly especially at higher output currents and for higher load transients. This necessitates an inductor with marked polarity."

[idpromnut]

From the document the RPi people released, the switcher is integrated into the RP2350 IC itself, and the problem stems from something in that circuit (or coupled near it?) going screwy if the inductor is installed "backwards". The only thing I've been able to find was that it involved inductor kickback. As was said above, I'm only "familiar" with the magnetic field collapsing inducing a current/voltage in a circuit.

[Xena E]

This is interesting.

I have come across issues where the outer winding layers of inductors have had to be connected into the circuit in the most 'electrically quiet' orientation to stop a problem, self screening if you will, but I'd say that shows a design to be very marginal.

If needs must I suppose, but it seems a bit wank.

[EEVblog]

You don't want any stray B-field from the inductor coupling to sensitive nodes. I haven't seen any Abracon part numbers yet, the freq., the current.
Having a polarity dot, flipping that - solves nothing IMHO. I think it's another problem that exists.

That was my first reaction.
Problem for them is they are stuck with the switcher being inside the RP2350A silicon.
Could very well be some instability or some other issue at the die design/layout level.
Can't say I've ever heard of this issue with dedicated switcher chips.

[EEVblog]

I would guess having that AVDD resistor R5 nearby is bad practice, and a stability problem with the Vcore reg might be manifesting.

R5?

[PCB.Wiz]

This is interesting.

I have come across issues where the outer winding layers of inductors have had to be connected into the circuit in the most 'electrically quiet' orientation to stop a problem, self screening if you will, but I'd say that shows a design to be very marginal.

Yes, I've recall building two identical power supplies and one popped the uA723 sometimes and the other never did.
A lot of digging and un-twisting of the transformer primary 2 blue wires, uncovered the mains winding was switched-pin nearest secondary layer in one case and switched-pin further away in the OK case.
Easily fixed.

I can see it would make sense to connect the outside winding layer to the CAP rather than the SWITCH, but that's really a capacitive effect.
Any magnetic effect must be second order as a rod-core will be close to rotationally symmetric in field, unless the shielding ferrite itself is asymmetric ?

If needs must I suppose, but it seems a bit wank.

An easy check will be to swap one around on a board, and see what actually happens 

[ataradov]

While there are theoretical concerns, if it really matters, then the designs is just broken. Or they are over cautious.

Switching converters inside the MCUs are tricky in general. Atmel's SAM D51 has a noisy buck converter and the errata basically says to not use it if you also want to use PLL. No amount of strange requirements is going to fix things, so you have to errata out the broken feature.

Swapping it around and seeing what happens is likely not going to show anything. If it does, then the converter is broken for sure.

[Xena E]

While there are theoretical concerns, if it really matters, then the designs is just broken. Or they are over cautious.

If the inductors mag field is causing the problems when orientated one way, this is a parasitic degenerative/regenerative issue.

The design is just broken.

[langwadt]

While there are theoretical concerns, if it really matters, then the designs is just broken. Or they are over cautious.

Switching converters inside the MCUs are tricky in general. Atmel's SAM D51 has a noisy buck converter and the errata basically says to not use it if you also want to use PLL. No amount of strange requirements is going to fix things, so you have to errata out the broken feature.

Swapping it around and seeing what happens is likely not going to show anything. If it does, then the converter is broken for sure.

from the quoted section of the datasheet: "We found that the regulator performance could be massively improved if the inductor is 'the right way round'."

that leaves broken ...

[radiolistener]

I think this is their big mistake to put switching converter inside MCU. Someone wrote that they also had ADC nonlinearity issue. Most of all, this issue with converter is also the root of cause for ADC issue...

And even if they fix this issue, it would still be a problem. Because such MCU cannot be used in solutions which is critical to noise interference and power supply purity and since they embedded that regulator inside MCU, there is no way to replace it with linear one...

[ataradov]

Oh, I did not see that it was basically mandatory to use it in a switching mode.  Usually switching mode is included as an option if you really want to shave some power consumption. 

In that case the whole device is junk. I don't see that being too popular if it requires components from a single supplier that need to be mounted the right way.

[ArdWar]

I'm still not exactly sure what the problem is. If it's B field coupling then surely it's the orientation that is critical instead of the polarity. Like if you assemble the inductor with the right polarity but laid the PCB out the other way then the B field problem is exactly the same??

Edit: now reading a bit more :duh: it's indeed the orientation that matters. Now even with the correct polarity and orientation, some indicator can be placed sideways, that will be interesting huh.

Oh welp... at least you can bypass it right?

[ejeffrey]

Yeah.  It's hard enough to make sure that diodes and LEDs get assembled in the right orientation and people know to look for that.   Making sure inductors get assembled the right way is just too much trouble.  Not to mention if the manufacturer starts winding inductors the other direction, or some manufacturing engineer tries to substitute an alternate part. 

Of course all those problems are something you can deal with with assembly notes and proper manufacturing checks and supply chain management.  But for what should be a simple low power regulator it's not worth the trouble.  And it means the whole design or at least the board layout is marginal, so I wouldn't really trust it long term anyway even if made "correctly".

[radiolistener]

Yeah.  It's hard enough to make sure that diodes and LEDs get assembled in the right orientation and people know to look for that.   Making sure inductors get assembled the right way is just too much trouble.

The main problem is not the wrong inductor orientation in space, but the fact that changing inductor orientation affects MCU functions... That is an indication about serious issue in MCU design. And even if you rotate inductor in such way that MCU start to work, it don't fix the issue. It's just hide it...

But for what should be a simple low power regulator it's not worth the trouble.  And it means the whole design or at least the board layout is marginal, so I wouldn't really trust it long term anyway even if made "correctly".

Yes, exactly

[floobydust]

Raspberry Pi Foundation isn't doing IC design, IMHO. It would be high-level architecture only.
I have seen semiconductor IP licensed, the Lego blocks such as FLASH memory are a chunk you buy (license) from a specialty design firm. Developing that in-house is too costly and time consuming.
This Vcore buck-converter, I'm assuming it's a Broadcom piece of IP. 1.1V at several 100's mA for the two cores?
Inductor is a custom Abracon 3.3uH 2016 [metric] guessing like the ASMPM-0806 series but with a polarity dot.
Wonder what happens if you place it on its side, and what is picking up the B-field in the MCU. Very strange and bad news.
"The regulator’s output voltage can be varied between 0.55 V and 3.3 V, but RP2350 may not operate reliably with its digital core supply (DVDD) at a voltage other than 1.1 V"

I would guess having that AVDD resistor R5 nearby is bad practice, and a stability problem with the Vcore reg might be manifesting.

R5?

It's the AVDD R_FILT or R5 on the Pico schematic that I thought was originally picking up noise - not Vcore reg. I'm an optimist lol.

[SiliconWizard]

In this case, regarding Flash, I doubt it's on the same die. It's probably a second die in the same package.

Regarding the buck converter, who knows if they have designed it or bought an IP. (Similarly for the ADC.)
Even if they bought some IP, integrating it properly on the die requires skills. Even just the pinout may influence the results.

I haven't followed closely the problem here, but I've already seen this kind of problems: switching converters with EMI coupling to the feedback network and throwing regulation completely out of whack.

[profdc9]

I would guess there's probably a synchronous buck converter consisting of a P-channel high-side and a N-channel low side on the chip.

Think that perhaps adding a Schottky diode (SS14) or a small snubber capacitor (1000 pF) at the VREG_LX terminal might help?  Maybe this would smooth any transients when the high-side switch turns off.

[EEVblog]

Oh, I did not see that it was basically mandatory to use it in a switching mode.  Usually switching mode is included as an option if you really want to shave some power consumption. 
In that case the whole device is junk. I don't see that being too popular if it requires components from a single supplier that need to be mounted the right way.

That's a huge red flag. Potential show-stopper. If I was the design engineer deciding whether or not to implement this micro in a design, and I knew this, I'd do anything to avoid using it.

[oPossum]

An external linear regulator can be used.

An example:

https://x.com/dangerousproto/status/1821903736446550197

https://x.com/dangerousproto/status/1822232583104495992

[ataradov]

An external linear regulator can be used.

I'm surprised that this is not an official solution. You don't even need to think hard to figure out that an LDO is cheaper than the inductor and all the supporting components.

This makes me think there may be issues with this. May be sequencing issues or something like that. Things may appear to work until they start to break.

Hopefully Dangerous prototypes cleared that use with the RP guys.

[gamalot]

In the schematic on page 5 it is R3, while in the PCB silkscreen on page 6 it is R5.

[Geoff-AU]

With this kind of hocus-pocus the RP2350A would be perfect for audiophile projects.

 

Having a manufacturer custom-design an indexable inductor is an end-run around understanding and fixing the real cause.  It also ignores all the lessons hardware design companies learned about supply chains during the pandemic. 

Meh.

[Hydron]

Datasheet states "Alternatively, DVDD can be supplied directly from an off-chip power source."

I've got a few RP2350s to play with but none of the special inductors, happy to hear a LDO has been used successfully, and datasheet seems to back it up (though I haven't read enough to know if there are sequencing considerations). Shame I didn't see this thread earlier, I would have asked the RPi guys directly.