folays
Posts: 6
Joined: Tue Dec 17, 2019 7:17 pm

RPi 4 USB-C Power Delivery Seems Fixed

Tue Dec 17, 2019 7:56 pm

Hi,

I bought a RPi 4GB at kubii.fr on the 24th of June 2019, delivered on the 21th of September.

My Pi had a solder problem on the SD card slot, which was loosening.

They offered me a replacement.

The previous (orignal) Raspi had power :
- NOT working with USB-C PD power supplies : Macbook 12" 2015, Macbook Pro 15" 2018, Anker PowerPort Speed+ Duo, Nintendo Switch, LG 32ul950 screen
- OK working with stupid non-USB-C-PD power supplies : As I only buy USB-C PD, I only succeeded booting my June RPi4 with an old iPad mini 3 "10W USB" power adapter (not USB-C nor -PD at all) + a stupid USB<->USB-C cable (of course I know that 10W could be insufficient for RPi4 so I stopped using it since I had no other 3amp power supply... and I did not want to start buying non-USB-C-PD just for RPi...)

Currently, the new (replaced) RPi 4GB I received yesterday :
- works with USB-C PD power supplies : Macbook 12" 2015, Anker PowerPort Speed+ Duo + "USB-C charger cable (apple)", LG 32ul950 screen (with a USB-C non-TB3 cable, plugged on the USB-C-TB3-70W-able port of this screen)

Are new batches of RPi4 USB-C have their PD specs fixed?
Or I am just lucky to have a RPi which could have some of its electrical resistance components damaged and unwillingly fixing the problem?

I issued a, which took 62s to complete (matching a random RPi4 benchmark found on the web) :

Code: Select all

$ sysbench --test=cpu --cpu-max-prime=20000 --num-threads=4 run

I have a funny USB-C multimeter (those that have two USB-C ports, one at each side, and which are able to mesure amperage)

With 32ul950 USB-C-TB3 70W port I got :
- 5V 0,8 Amp at boot
- 4,83V 0,74 Amp once logged, idle
- 4,78V 1,13 Amp during sysbench
- "stress-ng --cpu 0 --cpu-method fft" yielded 4,70V 1,47 Amp

(yes, I know that I should expect a CPU bug or damage with under 4,75 V, which are undervolted)

The Anker USB-C PD adaptateur yielded (this Anker is labelled at 2,4 Amp at 5V, and is one of the best-rated PD adapter by the Google guy which made the spreadsheet of USB-C cables&adaptaters) :
- 4,92 V 0,74 A once logged, idle
- "stress-ng --cpu 0 --cpu-method fft" yielded 4,86-to-4,90V at 1,40-to-1,47 Amp

So, RPi 4 USB-C PD are fixed ?

trejan
Posts: 1114
Joined: Tue Jul 02, 2019 2:28 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Tue Dec 17, 2019 9:34 pm

What is the revision code in /proc/cpuinfo ? There is a v1.2 which has some minor changes to the PCB but no confirmation as to what has actually changed.

trejan
Posts: 1114
Joined: Tue Jul 02, 2019 2:28 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Wed Dec 18, 2019 2:01 pm

spooker wrote:
Wed Dec 18, 2019 1:09 pm
where did you buy the rpi from?
They said where on the first line.

folays
Posts: 6
Joined: Tue Dec 17, 2019 7:17 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 10:10 am

trejan wrote:
Tue Dec 17, 2019 9:34 pm
What is the revision code in /proc/cpuinfo ? There is a v1.2 which has some minor changes to the PCB but no confirmation as to what has actually changed.

Code: Select all

[email protected]:/home/pi# cat /proc/cpuinfo 
processor	: 0
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 108.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae evtstrm crc32 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x0
CPU part	: 0xd08
CPU revision	: 3

processor	: 1
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 108.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae evtstrm crc32 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x0
CPU part	: 0xd08
CPU revision	: 3

processor	: 2
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 108.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae evtstrm crc32 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x0
CPU part	: 0xd08
CPU revision	: 3

processor	: 3
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 108.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae evtstrm crc32 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x0
CPU part	: 0xd08
CPU revision	: 3

Hardware	: BCM2835
Revision	: c03112
Serial		: 10000000c3ed3713

GlowInTheDark
Posts: 373
Joined: Sat Nov 09, 2019 12:14 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 10:14 am

trejan wrote:
Wed Dec 18, 2019 2:01 pm
spooker wrote:
Wed Dec 18, 2019 1:09 pm
where did you buy the rpi from?
They said where on the first line.
Trejan is a big fan of Miriam Webster's word-of-the-year-2019.
GitD's list of things that are not ready for prime time:
1) IPv6
2) 64 bit OSes
3) USB 3
4) Bluetooth

trejan
Posts: 1114
Joined: Tue Jul 02, 2019 2:28 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 1:24 pm

folays wrote:
Fri Dec 20, 2019 10:10 am
Revision : c03112
Interesting. It is a v1.2 then. The photos posted in the big v1.2 thread showed the circuitry around the PMIC to look superficially the same as the original v1.1.

User avatar
davidcoton
Posts: 4541
Joined: Mon Sep 01, 2014 2:37 pm
Location: Cambridge, UK

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 1:33 pm

@trejan
See my post here.
The change could be tracking only, repurposing the existing (v1.1) components.
Signature retired

trejan
Posts: 1114
Joined: Tue Jul 02, 2019 2:28 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 1:53 pm

davidcoton wrote:
Fri Dec 20, 2019 1:33 pm
The change could be tracking only, repurposing the existing (v1.1) components.
Yeah. If this post is accurate then that looks like what has happened. The lack of a silkscreened v1.2 isn't great though as there isn't a quick & easy way of identifying it. You need to look at that small component location on the bottom or have the hassle of booting an OS to look at the revision.

folays
Posts: 6
Joined: Tue Dec 17, 2019 7:17 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 2:04 pm

I will try to take 2 close pictures of front and back area near the power input.

Do you want another picture?

Also, I could take my ampere-meter if you want me to test the Ohmic resistance of of one of the resistance.

I do not have the time to research again what the problem were before with my initial RPi4, but I think I understood that one of the VALUE of the resistance was not good ? Or maybe a completely missing resistance?

User avatar
davidcoton
Posts: 4541
Joined: Mon Sep 01, 2014 2:37 pm
Location: Cambridge, UK

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 2:41 pm

The problem was that the two CC pins were linked, and had a common resistor to ground (and another to the A2D input).
There should have been separate resistors to ground from CC1 and CC2.

If the Pi needs to determine what is connected to the CC pins (cable or far end), it would also need two resistors going to two A2D inputs. My suspicion is that the Pi cannot react to that knowledge (when powered by another input, it cannot control the output voltage or current through the USB-C), so there is no point in the A2D connections. So the two existing resistors (which IIRC are the correct value, but probably not what was intended for linking to the A2D input) can be repurposed by a track change. This probably won't show in photographs.

Of course my understanding may be partially or completely wrong :o
Signature retired

trejan
Posts: 1114
Joined: Tue Jul 02, 2019 2:28 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 2:46 pm

folays wrote:
Fri Dec 20, 2019 2:04 pm
I will try to take 2 close pictures of front and back area near the power input.
That would be great if you could.

Can you run "sudo vcgencmd measure_volts usb_pd" and tell me the voltage reading you get?
folays wrote:
Fri Dec 20, 2019 2:04 pm
I do not have the time to research again what the problem were before with my initial RPi4, but I think I understood that one of the VALUE of the resistance was not good ? Or maybe a completely missing resistance?
The short answer is that there are two configuration lines in the USB-C connector. They're supposed to have a separate resistor on each line but the Pi 4 connected them together to a single resistor. That causes the resistance to be wrong and the Pi 4 to be detected as an audio accessory i.e. headphones and smarter chargers or e-marked cables will refuse to power it. It tends to be chargers like your Macbook charger or the expensive e-marked USB-C cables that have a problem with it.

Scorpia's post at https://www.scorpia.co.uk/2019/06/28/pi ... resistors/ explains it fully.

emma1997
Posts: 511
Joined: Sun Nov 08, 2015 7:00 pm
Location: New England (not that old one)

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 4:27 pm

As mentioned several time here it would not be smart from a profit standpoint to reveal new 'improved' versions until old ones have been flushed from the shelves. Just ask Osbourne company... oops... wait... there is no...

Heater
Posts: 14458
Joined: Tue Jul 17, 2012 3:02 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 4:43 pm

emma1997,
"flushed from the shelves"
You make it sound like they have been having a hard time selling Pi. Stocks piling up on buckling shelves covered in dust all around the world.

As far as I can tell Pi 4 have been flying off the shelves as fast as they can make them!

The Osbourne story is a) Not even true. b) Not relevant to the Pi 4 situation, even if it were true.
Memory in C++ is a leaky abstraction .

trejan
Posts: 1114
Joined: Tue Jul 02, 2019 2:28 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 4:46 pm

jcyr wrote:
Fri Dec 20, 2019 4:14 pm
Why have the RPI engineers refrained from providing a definite answer?
There is official confirmation but it was very quiet. Dougie make a documentation pull request which adds v1.2 to the list on the website and it was committed 3 days ago.

We know that the main Pi 4 manufacturers are Farnell, RS and Raspberry Pi Trading. The v1.2 boards we've heard about are all 4GB models AFAIK so it isn't Raspberry Pi Trading as they are only making the 1GB models. Has one of the others completely drained their inventory of v1.1 so they're now manufacturing and shipping the revised v1.2? The lack of an official announcement even though v1.2 boards are being delivered is because presumably the other manufacturing partner had a large inventory of the old v1.1 boards still. As Heater mentions above, they were having trouble keeping the Pi 4 in stock so large inventory just lying around in the warehouse seems a bit unlikely.

User avatar
rpdom
Posts: 16121
Joined: Sun May 06, 2012 5:17 am
Location: Chelmsford, Essex, UK

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 4:53 pm

trejan wrote:
Fri Dec 20, 2019 4:46 pm
We know that the main Pi 4 manufacturers are Farnell, RS and Raspberry Pi Trading.
Well, to be strictly accurate those are the main Pi 4 distributors. Sony is the main Pi 4 manufacturer.

trejan
Posts: 1114
Joined: Tue Jul 02, 2019 2:28 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 5:04 pm

rpdom wrote:
Fri Dec 20, 2019 4:53 pm
Well, to be strictly accurate those are the main Pi 4 distributors. Sony is the main Pi 4 manufacturer.
Yeah. The lines are blurred here with the odd system used though. RS and Farnell are the main distributors but they also make all the 2GB and 4GB boards by contracting it out to the Sony factory in Wales.

emma1997
Posts: 511
Joined: Sun Nov 08, 2015 7:00 pm
Location: New England (not that old one)

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 5:04 pm

Heater wrote:
Fri Dec 20, 2019 4:43 pm
You make it sound like they have been having a hard time selling Pi. Stocks piling up on buckling shelves covered in dust all around the world.

The Osbourne story is a) Not even true. b) Not relevant to the Pi 4 situation, even if it were true.
The 'story' may or may not be true. Like maybe meteor alone didn't actually kill the dinosaurs. But Osbourne Effect is definitely real. I find it hard to believe they would intentionally damage sales with such a poor move.

Considering the phenomenal success IMO masters of RPF are genius level sales craft. Maybe not billionaire market but up there with Google/Amazon/Tesla and the other hotshots for that. These tantalizing romance and intrigue tactics certainly worked on this poor sucker with six Pi4 cluttering up the place now.

Regarding buckling shelves, some units I purchased were sitting around for months before bid/bo. Not all were one-off vendors either. So there are units sitting on shelves, specially ones with 200%-300% markup. They don't turnover like low margin dealers but sometimes patience=profit. I think RPF/RPT take good care of their resellers, official or not.

User avatar
Burngate
Posts: 6193
Joined: Thu Sep 29, 2011 4:34 pm
Location: Berkshire UK Tralfamadore
Contact: Website

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 7:11 pm

emma1997 wrote:
Fri Dec 20, 2019 4:27 pm
... it would not be smart from a profit standpoint to reveal new 'improved' versions until old ones have been flushed from the shelves.
There seem to be two alternative scenarios:

Either some thousands of Pis have been sitting on shelves, with one end of the first shelf at Pencoed being filled up as the other end is depleted by moving stock to the next shelf, and so on to the last shelf of a DHL truck delivering to someone's home

Or an order comes into a dealer, who passes it up the chain, eventually reaching Pencoed, and the latest Pi is despatched to the end-user, never touching a shelf in its travels.

Reality is probably somewhere in between, because they're likely despatched from Pencoed by truck. You can only fit a few trucks in a loading bay, and you can't reasonably carry less than the odd thousand Pis in each truck.
And the same goes for each link in the chain.

It rains in the highlands then it floods in the delta, but there's a lot of water locked up in the Nile

hippy
Posts: 6766
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 7:50 pm

trejan wrote:
Fri Dec 20, 2019 1:53 pm
The lack of a silkscreened v1.2 isn't great though as there isn't a quick & easy way of identifying it. You need to look at that small component location on the bottom or have the hassle of booting an OS to look at the revision.
I would agree with "isn't great" but, whether one's buying the old or the new, it won't matter to the overwhelming majority of purchasers - Taking it as that being the only change made to the board.

Just explicitly specify that it must be a version 1.2 when ordering one if that's what's needed or wanted.

folays
Posts: 6
Joined: Tue Dec 17, 2019 7:17 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 7:58 pm

With the « Anker PowerPort Speed+ Duo + USB-C charger cable (apple) », both idle or during a stress-ng (same result) :

Code: Select all

[email protected]:/home/pi# vcgencmd measure_volts usb_pd
volt=1.6700V
Video of the « original » (september 2019) RPi4 4GB, which did not work with USB-C PD charger, so that you can watch various things (layout, white ink) :
https://imgur.com/a/vr0ONdK

4 pictures of the « new » (december 2019) RPi4 4GB, which works with USB-C PD :
https://imgur.com/a/txckDpr

I don’t see any différences in the number of resistances in the power area.
But there seems to no longer be a three-legged black thing, as shown on https://imgur.com/a/FH3gmYA which is a picture of mu « new » RPi 4GB december 2019.
(In front of the pen, there is no longer a three-legged black component that you can see on the previous video of my RPi4 4GB september 2019)


I took my multimeter. I measured, by using https://www.scorpia.co.uk/2019/06/28/pi ... resistors/ as a « reference to which R* are where » (assuming that they are on the same place in all 4GB v1.0-v1.2) :
- R1 as shown on the ref page : 5,09 to 5,1 kOhm
- R79 : 5,1 kOhm

I initially though that I could maybe have the « chance » that my R79 could have been somewhat damaged (a higher resistance) which would have made the USB-PC charger think that the RPi is a sink power.
But this theory seems to refute itself since R79 has the intended 5,1 kOhm.

Maybe CC1 and CC2 are not anymore both connected to R79, but I do not have any USB-C line splitter to test if CC1 and CC2 are on the same line.

Maybe the 5x5 (25 pins "holes") area near and below R1 gives access to the USB-C pins.

trejan
Posts: 1114
Joined: Tue Jul 02, 2019 2:28 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 8:10 pm

folays wrote:
Fri Dec 20, 2019 7:58 pm
With the « Anker PowerPort Speed+ Duo + USB-C charger cable (apple) », both idle or during a stress-ng (same result) :

Code: Select all

[email protected]:/home/pi# vcgencmd measure_volts usb_pd
volt=1.6700V
At least 1 of the CC lines is still connected to the analog input of the PMIC then. 1.67V would give an Rp of 10K which would signify a 3A capable charger.

How about "vcgencmd measure_volts ain1"? That measures the other analog input on the PMIC. The v1.1 board just tied it to ground.
folays wrote:
Fri Dec 20, 2019 7:58 pm
I don’t see any différences in the number of resistances in the power area.
But there seems to no longer be a three-legged black thing, as shown on https://imgur.com/a/FH3gmYA which is a picture of mu « new » RPi 4GB december 2019.
(In front of the pen, there is no longer a three-legged black component that you can see on the previous video of my RPi4 4GB september 2019)
A similar looking SOT component has now appeared next to the I in MICRO SD.
folays wrote:
Fri Dec 20, 2019 7:58 pm
Maybe the 5x5 (25 pins "holes") area near and below R1 gives access to the USB-C pins.
They're thermal vias to help conduct heat away from the PMIC.

folays
Posts: 6
Joined: Tue Dec 17, 2019 7:17 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 8:30 pm

Okay, found it !

I saw on https://www.scorpia.co.uk/2019/06/28/pi ... resistors/ that CC1 and CC2 being connected together is the screw-up, but would best rephrased to "the problem is that they share the same set of resistances (R79 + R1)".

Especially since this URL suggest that with non-e-marker USB-C "intelligent" cable, there could be no problem.

The schematic on the URL (https://www.scorpia.co.uk/wp-content/up ... /image.png) shows us that R79 and R1 have one of their legs directly connected together.

Compare : There is a major difference :

When you orient the RPi with "the SD card slot open side (where you insert the sdcard) to your right" :
- On the september RPi : The right leg of R1 was going immediately up to it, seemingly to the north of R79
- On my december RPi : The right leg of R1 goes around a corner, take a small turn, goes elsewhere, maybee to the south side of R79, which if I understood correctly would probably be the ground.

I confirmed this : I measured GPIO-ground with :
- right of R1 : 0 Ohm
- south of R79 : 0 Ohm
- south of "the resistance at the right of R79" : 1,525 kOhm (which I don't know the purpose and probably has nothing to do with the subject)

Resistance measured between "left leg of R1 and north leg of R79" is 10,23 k0hm. [edit : fixed "right left of R1" -> "left leg"]
A random webpage on the web confirmed me that "two resistance in serial disposition of both 5100 Ohm" should yield 10,20 kOhm.

That would confirm that R1 and R79 are no longer incorrectly directly connected
(well, the USB-C side of them. obviously, they have both their ground-side connected to... ground)
Now CC1 and CC2 both have each their own, isolated, 5,1 kOhm transistor.

BTW for @trejan, immediately after boot, presumably while idle :

Code: Select all

[email protected]:/home/pi# vcgencmd measure_volts ain1
volt=0.0000V
Last edited by folays on Sun Dec 22, 2019 1:56 am, edited 1 time in total.

trejan
Posts: 1114
Joined: Tue Jul 02, 2019 2:28 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 8:36 pm

folays wrote:
Fri Dec 20, 2019 8:30 pm
When you orient the RPi with "the SD card slot open side (where you insert the sdcard) to your right" :
- On the september RPi : The right leg of R1 was going immediately up to it, seemingly to the north of R79
- On my december RPi : The right leg of R1 goes around a corner, take a small turn, goes elsewhere, maybee to the south side of R79, which if I understood correctly would probably be the ground.
Great! Thanks for investigating this.
folays wrote:
Fri Dec 20, 2019 8:30 pm
BTW for @trejan, immediately after boot, presumably while idle :

Code: Select all

[email protected]:/home/pi# vcgencmd measure_volts ain1
volt=0.0000V
Ah. They've not connected the other analog input then. I was wondering if they'd wire it up to the other CC line but it looks like it is still tied to ground.

Heater
Posts: 14458
Joined: Tue Jul 17, 2012 3:02 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 9:31 pm

This gets out of hand.

When you buy pretty much any model of anything it is likely not exactly the same as the one your friend bought last week.

They work as advertised the same, even if the guts have been totally rearranged. So what's the problem?
Memory in C++ is a leaky abstraction .

Heater
Posts: 14458
Joined: Tue Jul 17, 2012 3:02 pm

Re: RPi 4 USB-C Power Delivery Seems Fixed

Fri Dec 20, 2019 9:50 pm

I can appreciate that nerdy attention to detail.

It's all the other guff I don't get. "Osborn effect", expectation of some announcement, and all that other speculation and nonsense.
Memory in C++ is a leaky abstraction .

Return to “General discussion”