jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 25408
Joined: Sat Jul 30, 2011 7:41 pm

Re: New Pi 4B v1.2?

Sat Feb 01, 2020 6:48 pm

W. H. Heydt wrote:
Sat Feb 01, 2020 6:18 pm
spooker wrote:
Sat Feb 01, 2020 8:46 am
hippy wrote:
Fri Jan 31, 2020 11:49 pm
I'm more enjoying the thread as a master class demonstration of discussion dynamics, how avoiding answering the question doesn't make it go away, rather leads to a never-ending thread.
it is really sad computer science is full of egocentric people that will do everything to
This is different from other fields of endeavor...how? (Actually, politics is *far* worse about this than CS.)
1)not admit their F-UPs
The RPT *did* admit this is their error.
2)when they can't avoid (1) they try to downgrade the severity of their mistake.
This is a minor error. Unless one is going to particularly expensive power cables or supplies, one would never notice the "problem".
What he said. The number of egocentric CS people who come on here saying we are hiding this issues, when its utterly obviously clear that we have freely admitted it, is really odd.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I own the world’s worst thesaurus. Not only is it awful, it’s awful."

ejolson
Posts: 4460
Joined: Tue Mar 18, 2014 11:47 am

Re: New Pi 4B v1.2?

Sat Feb 01, 2020 7:08 pm

Burngate wrote:
Sat Feb 01, 2020 6:13 pm
hippy wrote:
Fri Jan 31, 2020 11:49 pm
Back in the day the Foundation were pre-announcing revision changes -

https://www.raspberrypi.org/blog/upcomi ... d-revision
But back then, the Foundation was different, and so was the whole Raspberry Pi scene.

The Pi started as a punt in the dark - they had hoped to sell 10,000 - and the idea of having there baby copied was thought to be proof that education was needed and might really work.

Now, real money is involved - money that goes via the Foundation into education. Any pirated hardware that takes money away from the Foundation is seen as a Bad Idea, so the hardware is kept as far as possible hidden away.
It's unfortunate how too much stress can cause both people and companies to make poor decisions. Competition can lead to the creation of adversarial relationships, not only with competing companies, but with existing customers.

In the past I made an analogy between the Pi and how the IBM PC became the most popular and copied hardware design of its time. After having thought about this a bit more, another fitting analogy might be to Sun, the manufacturer of the first Unix workstations.

Sun is a good comparison because of the size, youthfulness and the reliance on hardware sales for profits. While Sun seldom let financial concerns get in the way of innovation, they ended up nearly bankrupt more than once and were finally destroyed by an acquisition which appeared to monetize the existing customer base rather than expand it.

Some of the innovations, however, like the Sun Ray thin clients

Image
https://en.m.wikipedia.org/wiki/Sun_Ray

are still unsurpassed, not to mention Java, Virtual Box, ZFS and the use of encrypted RAM to mitigate Spectre-like side-channel attacks long before they were even called that. In my opinion, given the lead starting in 2015 that SPARC had with encrypted memory, Sun could have seen major deployments in the cloud had the whole architecture not already been cashed out.

I'm not sure what lessons can be applied to manufacturing the Raspberry Pi except how important it is to guard against acquisition by another company that has different goals and ideals.

Has anyone seen a 1GB or 2GB Pi 4B with support for USB Power Delivery versions 2 and 3? Which memory size do you think will appear with the 1.2 revision next?

Edit: Updated to replace the incorrect term power distribution with the correct power delivery for clarity.
Last edited by ejolson on Sat Feb 01, 2020 10:13 pm, edited 1 time in total.

W. H. Heydt
Posts: 11663
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: New Pi 4B v1.2?

Sat Feb 01, 2020 8:49 pm

ejolson wrote:
Sat Feb 01, 2020 7:08 pm
Has anyone seen a 1GB or 2GB Pi 4B with support for power distribution? Which memory size do you think will appear with the 1.2 revision next?
Fair point... I'd actually expect the Pi4B1 before the Pi4B2. My reasoning is that the major demand appears to be the Pi4B4. Only those comfortable with what they're doing are likely to opt for the 1GB. Far too many people will opt for the 4GB (Bigger! "Better"!). This leaves the 2GB model as the "red-headed step-child" of the Pi4B world. (Notwithstanding all that, while I have one project for which I want 3 Pi4B4 boards, I anticipate using more Pi4B2s than Pi4B4s. And so far I have one Pi4B1, one Pi4B4, and two Pi4B2 boards.)

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 25408
Joined: Sat Jul 30, 2011 7:41 pm

Re: New Pi 4B v1.2?

Sat Feb 01, 2020 8:55 pm

ejolson wrote:
Sat Feb 01, 2020 7:08 pm
I'm not sure what lessons can be applied to manufacturing the Raspberry Pi except how important it is to guard against acquisition by another company that has different goals and ideals.

Has anyone seen a 1GB or 2GB Pi 4B with support for power distribution? Which memory size do you think will appear with the 1.2 revision next?
I'm not sure what point you were trying to make, but with regard to someone buying Raspberry Pi - and not blowing one's own trumpet - the main USP is the engineers. WIthout them, and Eben's direction, it would not be where it is now (and note, this is my personal opinion). If another company took over and their goals did not make the engineers happy and match their ideals, then they would soon find themselves short of engineers. It's not like the job market isn't pretty good for the sort of people who work here.

What do you mean by power distribution? And don't forget rev 1.3, and rev 1.4 and rev 1.n where n is a number > 2 and less than 100.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I own the world’s worst thesaurus. Not only is it awful, it’s awful."

W. H. Heydt
Posts: 11663
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: New Pi 4B v1.2?

Sat Feb 01, 2020 9:50 pm

jamesh wrote:
Sat Feb 01, 2020 8:55 pm
What do you mean by power distribution? And don't forget rev 1.3, and rev 1.4 and rev 1.n where n is a number > 2 and less than 100.
I took it as being an assumption that the v1.2 applies to the PCB, regardless of RAM size, and therefore, which of the other two models (1GB or 2GB) was likely to be first to flush through the distribution channel and show up in end users hands. It is, of course, all handwaving, as none of us has a decent handle on the breakdown of sales volume by variant, save that--so far as anyone on the outside can tell--the 4GB are by far the highest volume of sales of the three.

Effectively, my guess is that the 1GB variant will be the next to be found in the wild through a combination of demand and production. Personally, I seem to be a bit of an outlier for preferring the Pi4B2 for general use (and recommendation to beginners) over the Pi4B4.

ejolson
Posts: 4460
Joined: Tue Mar 18, 2014 11:47 am

Re: New Pi 4B v1.2?

Sat Feb 01, 2020 10:09 pm

jamesh wrote:
Sat Feb 01, 2020 8:55 pm
What do you mean by power distribution?
Whoops! I meant USB Power Delivery versions 2 and 3--the technology that allows you to use the USB-C port on a new notebook as an expensive power supply for the Pi.

I've updated the original post.

My understanding is more 2GB models were originally produced than 1GB. This would also suggest a revised 1GB Pi 4B will be found before a 2GB model.

hippy
Posts: 6842
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: New Pi 4B v1.2?

Sat Feb 01, 2020 10:59 pm

jamesh wrote:
Sat Feb 01, 2020 8:55 pm
And don't forget rev 1.3, and rev 1.4 and rev 1.n where n is a number > 2 and less than 100.
Less than 16 given the 4-bits reserved for revision version, though I suppose unused bits elsewhere or a whole new scheme for revision encoding could be used -

https://www.raspberrypi.org/documentati ... /README.md
ejolson wrote:
Sat Feb 01, 2020 7:08 pm
Which memory size do you think will appear with the 1.2 revision next?
My prediction is that all 4GB will be 1.2 from now on. All 2GB and 1GB will remain at 1.1 until old boards are used up, when both will then become 1.2

So I don't think anyone can make a prediction. I expect it will simply come down to the luck of the draw as to when 1.1 boards are used up and which happens to go through production first after that.

I don't know what batch sizes of PCB are purchased, nor build quantities of each variant. Knowing that may give some indication of how many were bought, how many have been used, how many remain, and how long they may take to get used. Could be days, could be years.

But equally, the Foundation could simply bin unused 1.1 boards, take the hit on doing that.

And maybe they'll all jump to 1.3 if and when the Boot Eeprom changes in size. Depends if that requires a new revision code or not. Could be 1.3 variants of 2GB and 1GB appear while 4GB is still 1.2

ejolson
Posts: 4460
Joined: Tue Mar 18, 2014 11:47 am

Re: New Pi 4B v1.2?

Sat Feb 01, 2020 11:12 pm

hippy wrote:
Sat Feb 01, 2020 10:59 pm
I expect it will simply come down to the luck of the draw as to when 1.1 boards are used up and which happens to go through production first after that.
Here on the liberal frontier, where high tech is second only to high stakes at the casinos, the luck of the draw is an event of intense focus. Then again, when there are so many problems already in the world, why invent more. It's better to plug a Pi 4B into the official power supply and commence solving problems with code, engineering and fun work.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 25408
Joined: Sat Jul 30, 2011 7:41 pm

Re: New Pi 4B v1.2?

Sun Feb 02, 2020 7:45 am

ejolson wrote:
Sat Feb 01, 2020 11:12 pm
hippy wrote:
Sat Feb 01, 2020 10:59 pm
I expect it will simply come down to the luck of the draw as to when 1.1 boards are used up and which happens to go through production first after that.
Here on the liberal frontier, where high tech is second only to high stakes at the casinos, the luck of the draw is an event of intense focus. Then again, when there are so many problems already in the world, why invent more. It's better to plug a Pi 4B into the official power supply and commence solving problems with code, engineering and fun work.
Quite.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I own the world’s worst thesaurus. Not only is it awful, it’s awful."

takyon
Posts: 57
Joined: Wed Jul 24, 2019 6:05 am

Re: New Pi 4B v1.2?

Fri Feb 21, 2020 2:27 pm

Get in the C: Raspberry Pi 4 can handle a wider range of USB adapters thanks to revised design's silent arrival
We asked Upton this week if the fix was out and he confirmed the update had indeed been rolled into a PCB Design for Manufacturing (DFM) process. He added that he would have expected the update "to have reached end users by now".

Possibly keen to avoid what some term "The Osborne Effect" and clogging up the channel with older models as customers awaited the new device, the revision has emerged with very little fanfare.

The update, Upton told us, had also moved "the WLCSP SD card voltage switch to the top side" to protect it from damage, and also "silk screen tweaks to reduce solder bridging in manufacture".

Upton said the 4GB version of the Pi 4 remained "the high-volume runner". Back at launch, he reckoned it would be the 2GB version that would fly off the shelves, but by September he acknowledged that the 4GB was most popular.

User avatar
Burngate
Posts: 6222
Joined: Thu Sep 29, 2011 4:34 pm
Location: Berkshire UK Tralfamadore
Contact: Website

Re: New Pi 4B v1.2?

Fri Feb 21, 2020 6:56 pm

So the schematic at https://www.raspberrypi.org/documentati ... educed.pdf is now possibly out of date?

Would a quick fix be to add a small chunk to the right and down a bit of the "old" USB-C POWER IN chunk, where there's a nice big blank space? (possibly left deliberately blank in anticipation of this change, somewhat like Fylingdales was left off the OS maps, to fool the Russians)

I'm also assuming that any other changes haven't altered any of the rest of the schematic.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 25408
Joined: Sat Jul 30, 2011 7:41 pm

Re: New Pi 4B v1.2?

Fri Feb 21, 2020 7:40 pm

Burngate wrote:
Fri Feb 21, 2020 6:56 pm
So the schematic at https://www.raspberrypi.org/documentati ... educed.pdf is now possibly out of date?

Would a quick fix be to add a small chunk to the right and down a bit of the "old" USB-C POWER IN chunk, where there's a nice big blank space? (possibly left deliberately blank in anticipation of this change, somewhat like Fylingdales was left off the OS maps, to fool the Russians)

I'm also assuming that any other changes haven't altered any of the rest of the schematic.
I will try and find out if any new schematics are in the pipeline.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I own the world’s worst thesaurus. Not only is it awful, it’s awful."

scorpia
Posts: 21
Joined: Mon Oct 08, 2018 10:54 am

Re: New Pi 4B v1.2?

Sat Feb 22, 2020 12:37 am

Good to know a fix has been released, Is anyone able to confirm if the backpowering issue over the USB-C connection was fixed as well? Will then update my blog posts on the issues (both the CC resistor and backpowring ones) with the fact that a new one has been released and how to tell which version people have (still get a lot of trafic from people looking for info about the problem).
Last edited by scorpia on Sun Feb 23, 2020 10:37 am, edited 1 time in total.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 25408
Joined: Sat Jul 30, 2011 7:41 pm

Re: New Pi 4B v1.2?

Sat Feb 22, 2020 8:23 am

scorpia wrote:
Sat Feb 22, 2020 12:37 am
Good to know a fix has been released, Is anyone able to confirm if the backpowering issue over the USB-C connection was fixed as well? Will then update my blog posts on the issue with the fact that a new one has been released and how to tell which version people have (still get a lot of trafic from people looking for info about te problem).
What back powering issue?
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I own the world’s worst thesaurus. Not only is it awful, it’s awful."

Technocolour
Posts: 104
Joined: Thu Jul 04, 2019 6:23 pm

Re: New Pi 4B v1.2?

Sat Feb 22, 2020 10:10 am

I haven't analyzed the USB C connection and have no original thoughts on the matter right now, but after creeping though scorpia's post history I'm guessing this is what he's (?) worried about:

scorpia:
Also the 5v on the connector is connected directly to the 5v rails. If the pi is powered over the GPIO then the USB-C connector is vbus hot which means it could backfeed devices when using the type C port on the pi as a USB client.

hippy
Posts: 6842
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: New Pi 4B v1.2?

Sat Feb 22, 2020 10:53 am

Technocolour wrote:
Sat Feb 22, 2020 10:10 am
I haven't analyzed the USB C connection and have no original thoughts on the matter right now, but after creeping though scorpia's post history I'm guessing this is what he's (?) worried about:

scorpia:
Also the 5v on the connector is connected directly to the 5v rails. If the pi is powered over the GPIO then the USB-C connector is vbus hot which means it could backfeed devices when using the type C port on the pi as a USB client.
Not so much "could backfeed devices when using the type C port on the pi as a USB client" but clearly will.

But, while that is something to be aware of, it is not exactly an "issue" in the sense of something which is a flaw or needs fixing.

If it didn't back-feed one couldn't power via GPIO and use the USB-C port as a host connection without a separate power supply to whatever was connected to the host.

If one wants to power the board via GPIO and connect it to a host for use in gadget mode one would need to cut the power wire to the host. In that respect it's exactly as it is for a Zero and Zero W.

ejolson
Posts: 4460
Joined: Tue Mar 18, 2014 11:47 am

Re: New Pi 4B v1.2?

Sat Feb 22, 2020 4:04 pm

hippy wrote:
Sat Feb 22, 2020 10:53 am
Technocolour wrote:
Sat Feb 22, 2020 10:10 am
I haven't analyzed the USB C connection and have no original thoughts on the matter right now, but after creeping though scorpia's post history I'm guessing this is what he's (?) worried about:

scorpia:
Also the 5v on the connector is connected directly to the 5v rails. If the pi is powered over the GPIO then the USB-C connector is vbus hot which means it could backfeed devices when using the type C port on the pi as a USB client.
Not so much "could backfeed devices when using the type C port on the pi as a USB client" but clearly will.
Now that the resistor needed to key the right device type for USB-C power delivery is fixed, there should be no need for an external supply with the 4B. Just use a suitable cable and it will receive enough power directly from the host computer. Once PC's that support USB 3 power delivery are widely available, I think this will revolutionise how the 4B is used in school computer labs.

dickon
Posts: 800
Joined: Sun Dec 09, 2012 3:54 pm
Location: Home, just outside Reading

Re: New Pi 4B v1.2?

Sat Feb 22, 2020 6:11 pm

I doubt it. AIUI, the Pi 4 doesn't support USB-PD, it just requires 5V at 3A to be delivered. If that requirement needs to be negotiated, or the port can't supply it -- and the Wikipedia entry is a little hazy on this -- the Pi 4 won't boot. All that's been fixed is the odd resistor setup which flags the device as either something reasonably normal, or a pair of headphones. The Pi 4 won't negotiate anything, and won't handle anything outside the usual 4.75-5.25V on the USB power pins -- certainly not the higher voltages that may be requested by the USB-PD spec.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 25408
Joined: Sat Jul 30, 2011 7:41 pm

Re: New Pi 4B v1.2?

Sat Feb 22, 2020 7:14 pm

dickon wrote:
Sat Feb 22, 2020 6:11 pm
I doubt it. AIUI, the Pi 4 doesn't support USB-PD, it just requires 5V at 3A to be delivered. If that requirement needs to be negotiated, or the port can't supply it -- and the Wikipedia entry is a little hazy on this -- the Pi 4 won't boot. All that's been fixed is the odd resistor setup which flags the device as either something reasonably normal, or a pair of headphones. The Pi 4 won't negotiate anything, and won't handle anything outside the usual 4.75-5.25V on the USB power pins -- certainly not the higher voltages that may be requested by the USB-PD spec.
Correct - we don't negotiate, just identify, so the power source will simply supply 5v.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I own the world’s worst thesaurus. Not only is it awful, it’s awful."

ejolson
Posts: 4460
Joined: Tue Mar 18, 2014 11:47 am

Re: New Pi 4B v1.2?

Sat Feb 22, 2020 7:37 pm

jamesh wrote:
Sat Feb 22, 2020 7:14 pm
dickon wrote:
Sat Feb 22, 2020 6:11 pm
I doubt it. AIUI, the Pi 4 doesn't support USB-PD, it just requires 5V at 3A to be delivered. If that requirement needs to be negotiated, or the port can't supply it -- and the Wikipedia entry is a little hazy on this -- the Pi 4 won't boot. All that's been fixed is the odd resistor setup which flags the device as either something reasonably normal, or a pair of headphones. The Pi 4 won't negotiate anything, and won't handle anything outside the usual 4.75-5.25V on the USB power pins -- certainly not the higher voltages that may be requested by the USB-PD spec.
Correct - we don't negotiate, just identify, so the power source will simply supply 5v.
Agreed. My understanding is that this identification allows USB-C with power delivery to provide 3A at 5V, which is enough to run the Raspberry Pi 4B.

dickon
Posts: 800
Joined: Sun Dec 09, 2012 3:54 pm
Location: Home, just outside Reading

Re: New Pi 4B v1.2?

Sat Feb 22, 2020 7:53 pm

*If* -- and this is the important point -- the hardware is capable of it. My reading of the Wikipedia entry is that the port doesn't have to supply more than 5V at 0.1A without negotiation, which isn't enough by a long shot.

Now, how many ports supply more than that is a good question. I'd expect an awful lot to default to 5V at 1A -- the older base USB spec. -- which won't be enough for a stable Pi 4. We'll have to see what the next-gen hardware actually supports, but don't be too surprised if most motherboards don't ship more than the base <=USB 2 without negotiation, for PSU and over-current reasons of their own: motherboards typically have quite a few USB ports, and overloading the 5V rail on the inbuilt PSU must be a concern at some point.

scorpia
Posts: 21
Joined: Mon Oct 08, 2018 10:54 am

Re: New Pi 4B v1.2?

Sun Feb 23, 2020 1:16 am

jamesh wrote: What back powering issue?
Technocolour is correct. It is the direct linking of the USB Type-C VBUS lines and the main 5v rail which can cause issues when used as a usb gadget and powered through the GPIO. Some more details and workarounds are on my blog post at https://www.scorpia.co.uk/2019/07/16/mo ... over-gpio/.
hippy wrote: If it didn't back-feed one couldn't power via GPIO and use the USB-C port as a host connection without a separate power supply to whatever was connected to the host.
Just to note that this will work with some devices but not with all (most simple things like keyboards will probably work), as the pi still identifies as a sink on its cc pins, non clent only default current peripherals are meant to check the correct CC connections before starting a connection
dickon wrote: *If* -- and this is the important point -- the hardware is capable of it. My reading of the Wikipedia entry is that the port doesn't have to supply more than 5V at 0.1A without negotiation, which isn't enough by a long shot.
.

Type-C current allows for charging currents to be specified without USB negotiation. As well as the default USB current Type-C src ports can also optionaly advertise 1.5A or 3A power capability to connected devices by using different pullup resistors on their end of the CC line. This can actualy be detected by the pi (at least in the first pi4 version) as the vgencmd has a function to read the CC resistors. For info Power delivery functionality takes precidence over the Type-C current advertisment and allows for higher currents and voltages to be negotiated.

scaramonga
Posts: 27
Joined: Sat Jul 13, 2019 4:08 am

Re: New Pi 4B v1.2?

Sun Feb 23, 2020 4:19 am

Devices are cheap enough anyway.

I'll just buy each revision regardless, I see no problem :)

W. H. Heydt
Posts: 11663
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: New Pi 4B v1.2?

Sun Feb 23, 2020 5:45 am

scaramonga wrote:
Sun Feb 23, 2020 4:19 am
Devices are cheap enough anyway.

I'll just buy each revision regardless, I see no problem :)
+1. It's been a tempest in a teapot.

hippy
Posts: 6842
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: New Pi 4B v1.2?

Sun Feb 23, 2020 12:34 pm

scorpia wrote:
Sun Feb 23, 2020 1:16 am
hippy wrote: If it didn't back-feed one couldn't power via GPIO and use the USB-C port as a host connection without a separate power supply to whatever was connected to the host.
Just to note that this will work with some devices but not with all (most simple things like keyboards will probably work), as the pi still identifies as a sink on its cc pins, non clent only default current peripherals are meant to check the correct CC connections before starting a connection
Powering via GPIO and then using the USB-C socket to connect USB devices is a rather niche situation I would have thought. Most people would just grab an off-the-shelf USB-A to USB-C cable and connect their USB-C connector devices that way.

Given the USB-C socket was only intended for power input, having the SoC's USB 2.0 bus connected to that is mostly a bonus, I can't fault how it is. If one wants to use the USB-C power socket with USB-C devices, the solution would probably be to implement an external interface which bridges the Pi and the device. I wouldn't expect the Foundation to include all that on the Pi by default.

If it were insisted they had to I would guess they would simply disconnect the SoC USB 2.0 bus from the USB-C socket, make it purely power input, and more people would lose out than could have gained.

Return to “General discussion”