RG1 1.8v regulator

As shown in Troy Mackay's post on Jul 28 it seems to me that we could mod our PI's to work more as the chips were designed.

I think that this could make the PI more stable, from the looks of it.

I am not a designer Just an old tech, but I think we need to find a way to FIX what we now know is an error in the board.

So could someone do some practical testing to see if there is something that can be done to easly fix the current board, rather than wait for RPI foundation to fix it by waiting for a board redesign?

i will love to try out the hack Troy mackay has done also and then test again with that fix on the board, but i have looked into this and i most say it is very well done by Troy as i think it is to small for me todo and i done have an microscope as need for this.

so yes if some one can findout where to make an cut to split the LAN9512 1v8 from the lod 1v8 then i will try this also.

Tooms

certainly there have been plenty of opportunities on the

RPi forum and twitter for someone to say yes we have a

design error and it is causing chip temperatures to get hotter

than they are supposed to.

On the recent "case with fan" thread:

http://www.raspberrypi.org/phpBB3/viewtopic.php?f=63&t=14323

bredman replies:

"1. The chips in the RPi are supposed to run hot, they are designed to operate safely to 120 degrees C."

On twitter, a similar question about fans is asked:

https://twitter.com/scottfrye/status/234232917583343616

Liz's reponse is:

"we're making good progress on the heat issue that some of

you are experiencing, with expert help from the other forum."

Oh wait, that's not quite what she said.  Instead she said:

"Why do you need a fan?"

Ron K Jeffries further clarifies: "It runs very cool, even when

overclocked to 900MHz.  Unless you live in the Mojave desert,

you'll not need a fan."

I think you said it best some months ago, that facts and honesty "interfere with her right to a new kitchen".

coder27 wrote:

 

There's a report today of a guy who thinks there is a correlation

between hot chips and ethernet cutting out and keyboard repeats.

 

http://www.raspberrypi.org/phpBB3/viewtopic.php?f=24&t=14478

 

"Mine gets pretty hot. Heat I don't mind, but when it does get hot,

the ethernet cuts out and the keyboard repeats happen more often."

Here's my latest hypothesis.  We're seeing three outcomes of the regulatory battle between RG1 (1.8V regulator) and IC3 (LAN9512).

1.  RG1 has substantially higher Vreg than IC3 and thus provides essentially all the current for 1V8.  IC3's regulator switches off and IC3 runs nice and cool.

2.  IC3 has substantially higher Vreg than RG1 and thus provides essentially all the current for 1V8.  IC3 get very hot, but still functions OK, at least for now.

3.  RG1 and IC3 have almost identical Vreg and the regulation is unstable, with the current alternately being sourced by RG1, IC3, or both, depending on RG1 and IC3 temperatures.

(3) is probably quite uncommon since it requires almost idential Vregs, but if it does occur might add enough ripple to 1V8 to cause USB and Ethernet failures.  It would be interesting to put a 'scope on 1V8 near IC3 to see what it looks like on boards with USB/Ethernet problems not otherwise explained.  It shouldn't affect IC3's PLL since the PLL has its own regulator and its own 1.8V filtering, but if IC3 is rapidly warming and cooling you might get some PLL instability.

It's also possible that (2) is causing local heating inside IC3 that's slowing a critical path enough to cause USB/Enet failure.  If path delay is right at the edge, process variation or case design could make the difference.

Good summary.  It's worth pointing out also that voltage is only one electrical parameter here.  If any instability is present, the added inductance from the incorrect connection could also be relevant given that the LAN9512 expects nothing but decoupling caps on these tracks.

The 1V8 LDO powers PLL circuitry in the LAN9512, so stability is important.  Those 3 caps won't be decoupling the LAN9512 as effectively if they're simultaneously part of a larger on-board 1.8V mesh, even if that mesh is not misbehaving electronically.

Morgaine Dinova wrote:

 

Good summary.  It's worth pointing out also that voltage is only one electrical parameter here.  If any instability is present, the added inductance from the incorrect connection could also be relevant given that the LAN9512 expects nothing but decoupling caps on these tracks.  The 1V8 LDO powers PLL circuitry in the LAN9512, so stability is important.

According to the LAN9512 data sheet (Figure 2.2 -- Power Connections), the PLL has its own +3.3V to +1.8V regulator.  It appears to be wired up correctly in the RasPi schematics.

I wasn't so sure about their complete independence inside the LAN9512 when looking at the datasheet.

The VDD18ETHPLL and VDD18USBPLL lines are also connected to L3 which is specc'd at 100MHz so it seems strongly related to dynamic PLL operation (L3 is isolating ETH and USB sections from each other for RF, similar to L4 on the 3V3 side), and neither of those PLL lines has a tank cap like the 4.7uF on the 1V8CORE lines, so it doesn't seem to be caring about low-frequency stability.  That suggested to me that the PLL's 1.8V may actually be affected by the core 1V8 LDO, if not directly as a source (which seems unlikely since it's fed from 3V3) then at least indirectly through core logic.

When SMSC go to all the trouble of inductively isolating the USB and ETH sections of both their 3V3 and 1V8 LDOs, it's easy to imagine that they're not too impressed when their 1V8CORE pins are coupled to something foreign.

There are also some additional bits of information to consider based on reports of problems with Ethernet and the X1 crystal. If there are indeed power issues on the SMSC core that may produce unstable operation of the internal clock signals and ethernet is particularly picky about clock jitter and shape which may get the part to operate out of the IEEE specs. I've seen this problem with some Ethernet controllers/PHYs where temperature produces too much drift on the crystal/oscillator and the ethernet link is lost.

And who said that the parts on the R-pi are rated to operate at 120C ?

The PoP RAM chip on top of the SoC is rated for Commercial temp ranges, it varies from mobile to normal applications but it does not exceed 95C.

There is a lot of clueless folks at the R-Pi forum contributing wrong information to the already existing confusion.

-J

There's a dedicated thread now in the other forum.

http://www.raspberrypi.org/phpBB3/viewtopic.php?f=29&t=14489

Love the classical cryptyc response from RPF minions ...

-J

coder27 wrote:

 

There's a dedicated thread now in the other forum.

 

http://www.raspberrypi.org/phpBB3/viewtopic.php?f=29&t=14489

From that thread:

JamesH wrote, on the RPF forum:

 

Cannot say much about this, except that the Foundation are aware

of the problem and are working to rectify it.

And why exactly can the Foundation not say much about it?

This is a board intended for technical education, not a secret product hiding trade secrets for commercial advantage.  Why are essential items of information like the gerbers hidden?  Why are problems hidden?  Why are solutions hidden?  Why are future plans hidden?

I think I can summarize the questions more succinctly, if less politely:

Why is the Foundation full of it?

Morgaine.

coder27 wrote:

 

There's a dedicated thread now in the other forum.

 

http://www.raspberrypi.org/phpBB3/viewtopic.php?f=29&t=14489

From that thread:

JamesH wrote, on the RPF forum:

 

Cannot say much about this, except that the Foundation are aware

of the problem and are working to rectify it.

And why exactly can the Foundation not say much about it?

This is a board intended for technical education, not a secret product hiding trade secrets for commercial advantage.  Why are essential items of information like the gerbers hidden?  Why are problems hidden?  Why are solutions hidden?  Why are future plans hidden?

I think I can summarize the questions more succinctly, if less politely:

Why is the Foundation full of it?

Morgaine.