New Raspberry Pi model 3B+ 1.4 GHz, 330Mbit Ethernet, 802.11ac, PoE

[W. H. Heydt]

Why not just use a 3B+ and turn off the WIFI? Instead of getting a soldering iron out? Not meaning to start anything, just honestly wondering why go to the trouble of de soldering stuff you can just turn off? Granted, a Pi Foundation 2B+ would likely cost less than a 3B+ but would it be worth the effort? Me personally, I'd be more interested in a 3A or 3A+ than a 2B+. I have several headless setups that don't need Ethernet or 4 USB ports. I currently us an A+ or a Pi Zero W if I need WIFI.

Perhaps because you're going to have people using the boards that could turn the wireless back on, and you don't want that to be able to be done?

I will accept the statement that has been given that Pi2B sales are too low to cover any new engineering costs to design a Pi2B+. So my question becomes...What--if any--would be the implications of just manufacturing a Pi2Bv1.2 with a '2837B0 SoC instead of the earlier '2837 (unspecified stepping) one? In other words, are the two chips "pin compatible" and only require setting the pick-and-place robot to use one package instead of the other?

Now...I will grant that that would lose a big chunk of the reason for the '2837B0 package, becuase it wouldn't spread the heat nearly as well by coupling it to the board for that purpose (and, indeed, that alone might be a "show stopper" for this approach), but there should still be *some* benefit, even if it only allows a Pi2B to gain--say--100MHz to be a 1GHz board...and look shiny.

I do agree with you that a Pi3A based on the Pi3B+ is a higher priority than a Pi2B+. Indeed, I think a CM3+ and CM3L+ is a higher priority than a Pi2B+. However, if I'm going to do anything other than just let them chug along, if I were to upgrade the Pi2Bv1.1 boards I'm using, I would want to go to a Pi2B+, thank you very much. Failing that, I will wait to see what happens, oh--a year or two--after the Pi4B comes out. Perhaps *that* will cause there to be a Pi2B+ or Pi2Bv1.3. Or maybe in a year or so, a Pi2Bv1.3 will quietly appear that uses the '2837B0 so that manufacturing of the '2837 can be discontinued.

[ejolson]

What--if any--would be the implications of just manufacturing a Pi2Bv1.2 with a '2837B0 SoC instead of the earlier '2837 (unspecified stepping) one?

Since the 2B runs at 900MHz the thermal advantages of the new packaging are not needed. In other words, the 2B didn't need fixing, so there is no need to fix it.

[W. H. Heydt]

What--if any--would be the implications of just manufacturing a Pi2Bv1.2 with a '2837B0 SoC instead of the earlier '2837 (unspecified stepping) one?

Since the 2B runs at 900MHz the thermal advantages of the new packaging are not needed. In other words, the 2B didn't need fixing, so there is no need to fix it.

Ah! But having read through the entire (current) set of comments on the launch blog entry thread, I think you're wrong. The B0 stepping includes fixes in the boot ROM to the USB boot process. Since the Pi2Bv1.2 can also have the OTP bits set for USB MSD and PxE booting, it *would* benefit from getting the B0 step of the '2837 SoC.

So now that the boot ROM support issue is exposed as a benefit, is there any reason why a '2837B0 couldn't be put on an otherwise unmodified Pi2Bv1.2 board? (And Eben did say in one reply that they're "looking at" the Pi2B, so an upgrade is not off the table...at least not yet.)

[fruitoftheloom]

Why not just use a 3B+ and turn off the WIFI? Instead of getting a soldering iron out? Not meaning to start anything, just honestly wondering why go to the trouble of de soldering stuff you can just turn off? Granted, a Pi Foundation 2B+ would likely cost less than a 3B+ but would it be worth the effort? Me personally, I'd be more interested in a 3A or 3A+ than a 2B+. I have several headless setups that don't need Ethernet or 4 USB ports. I currently us an A+ or a Pi Zero W if I need WIFI.

Perhaps because you're going to have people using the boards that could turn the wireless back on, and you don't want that to be able to be done?

I will accept the statement that has been given that Pi2B sales are too low to cover any new engineering costs to design a Pi2B+. So my question becomes...What--if any--would be the implications of just manufacturing a Pi2Bv1.2 with a '2837B0 SoC instead of the earlier '2837 (unspecified stepping) one? In other words, are the two chips "pin compatible" and only require setting the pick-and-place robot to use one package instead of the other?

Now...I will grant that that would lose a big chunk of the reason for the '2837B0 package, becuase it wouldn't spread the heat nearly as well by coupling it to the board for that purpose (and, indeed, that alone might be a "show stopper" for this approach), but there should still be *some* benefit, even if it only allows a Pi2B to gain--say--100MHz to be a 1GHz board...and look shiny.

I do agree with you that a Pi3A based on the Pi3B+ is a higher priority than a Pi2B+. Indeed, I think a CM3+ and CM3L+ is a higher priority than a Pi2B+. However, if I'm going to do anything other than just let them chug along, if I were to upgrade the Pi2Bv1.1 boards I'm using, I would want to go to a Pi2B+, thank you very much. Failing that, I will wait to see what happens, oh--a year or two--after the Pi4B comes out. Perhaps *that* will cause there to be a Pi2B+ or Pi2Bv1.3. Or maybe in a year or so, a Pi2Bv1.3 will quietly appear that uses the '2837B0 so that manufacturing of the '2837 can be discontinued.

The SoC on a RPi 3B+ is upside down compared to a 2B / 3B.............

[user879]

Hey all!

Plugged in and booted up my pi 3B+ and am using the ethernet port for internet. Everything is working fine except the green light on the ethernet port is not lighting up. I do have internet (I am sending this from the 3B+), the wifi is turned off, and the yellow/orange light works.

Any ideas on how to figure out why the green light isn't on?

Thanks

[bensimmo]

What--if any--would be the implications of just manufacturing a Pi2Bv1.2 with a '2837B0 SoC instead of the earlier '2837 (unspecified stepping) one?

Since the 2B runs at 900MHz the thermal advantages of the new packaging are not needed. In other words, the 2B didn't need fixing, so there is no need to fix it.

Ah! But having read through the entire (current) set of comments on the launch blog entry thread, I think you're wrong. The B0 stepping includes fixes in the boot ROM to the USB boot process. Since the Pi2Bv1.2 can also have the OTP bits set for USB MSD and PxE booting, it *would* benefit from getting the B0 step of the '2837 SoC.

So now that the boot ROM support issue is exposed as a benefit, is there any reason why a '2837B0 couldn't be put on an otherwise unmodified Pi2Bv1.2 board? (And Eben did say in one reply that's they're "looking at" the Pi2B, so an upgrade is not off the table...at least not yet.)

Given PoE, the PxE and cooler and better power managment even at the slower 900/1GHz the new revision say it introduces.
Then dumping the WiFi part would be a cost saving for anyone using it, who would need the big WiFi sat there.

Saying that though a 3A+/2A+ with/without WiFi, but with Ethernet would be a better target and remove the audio/video socket.

A redesign but hey, stick a Wireless keyboard and mouse in the one usb slot and you have a small powerful end terminal.
Remove the SD card slot while you are at it.


Market, that's a different question...

But you could order your "2B+" from Farnell as a custom part if you wanted I would assume. (As the trimmed 3B+)
Quantity and cost may be different.
https://www.element14.com/community/doc ... C-RPi-page

There was a nice pictures of a Japanese stall with various itterations (Probably 3B) a few weeks ago.

So it is possible, but you may have to pay for it.

[Gavinmc42]

The BT chip has Host wakeup output and 32khz input.
Could it be used in a Zero WH+ for low power mode, BT/RTC wakeup and shutdown?
A Zero WH+ with the 43455 becomes a Cypress RF module partner development kit?

[hippy]

I am simply suggesting not fitting the WiFi/BT electrical parts on a Pi3B+ and calling that a Pi 2B+. As W. H. Heydt had suggested earlier ... Maybe someone will unsolder the WiFi/BT parts and report back how well that worked.

Why not just use a 3B+ and turn off the WIFI? Instead of getting a soldering iron out?

That would be the best way to go. I was suggesting unsoldering to assess if there would be any electrical or other issues if things were removed, to see if a Pi 2B+ could use the same board as a Pi 3B+.

I think the Pi 2B is, and Pi 2B+ would be, a niche product and probably not worthwhile producing when WiFi/BT can be turned off on the Pi 3B+.

I would expect both Pi 2B and Pi 3B production to ramp down as focus shifts to using the BCM2387B0, little need for a Pi 2B+, unless some commercial users wanted those in quantity.

It was mostly speculative musing that a Pi 2B+ could potentially use the Pi 3B+ board to avoid development costs if having a Pi 2B+ were considered desirable to have. Nothing more.

[drgeoff]

The SoC on a RPi 3B+ is upside down compared to a 2B / 3B.............

?

[dom]

Plugged in and booted up my pi 3B+ and am using the ethernet port for internet. Everything is working fine except the green light on the ethernet port is not lighting up. I do have internet (I am sending this from the 3B+), the wifi is turned off, and the yellow/orange light works.

Any ideas on how to figure out why the green light isn't on?

Orange LED means gigabit ethernet is active.
Green LED means 10/100M ethernet is active.

So flashing orange and no green LED is good.

[hippy]

The SoC on a RPi 3B+ is upside down compared to a 2B / 3B.............

?

What I believe is meant is that the silicon placed within the BCM2837B0 chip package has been inverted to the earlier version so heat sources are thermally closer to the pins and board so heat can be better dissipated through the board as well as through the heat spreader further from the board.

[RaTTuS]

The SoC on a RPi 3B+ is upside down compared to a 2B / 3B.............

?

viewtopic.php?f=63&t=208057#p1286841 may help

[drgeoff]

The SoC on a RPi 3B+ is upside down compared to a 2B / 3B.............

?

What I believe is meant is that the silicon placed within the BCM2837B0 chip package has been inverted to the earlier version so heat sources are thermally closer to the pins and board so heat can be better dissipated through the board as well as through the heat spreader further from the board.

What relevance has that to the discussion on producing a B2+ by not fitting the RF block in an otherwise normal B3+? (That was where fotl posted his "upside down" comment.)

[W. H. Heydt]

?

What I believe is meant is that the silicon placed within the BCM2837B0 chip package has been inverted to the earlier version so heat sources are thermally closer to the pins and board so heat can be better dissipated through the board as well as through the heat spreader further from the board.

What relevance has that to the discussion on producing a B2+ by not fitting the RF block in an otherwise normal B3+? (That was where fotl posted his "upside down" comment.)

The potential relevance is a question of how the chip is wired to the ball grid array that mounts the chip to the PCB. That still doesn't address the concept of using a Pi3B+ PCB without installing the wireless circuits.

[hippy]

?

What I believe is meant is that the silicon placed within the BCM2837B0 chip package has been inverted to the earlier version so heat sources are thermally closer to the pins and board so heat can be better dissipated through the board as well as through the heat spreader further from the board.

What relevance has that to the discussion on producing a B2+ by not fitting the RF block in an otherwise normal B3+? (That was where fotl posted his "upside down" comment.)

I simply answered the query you seemed to have on what fotl posted.

I imagined that others might also be wondering what was meant by what fotl had written, how anyone could fit an entire SoC upside down, and sought to clarify that for you and others.

If I misinterpreted your single "?" as questioning what fotl was meaning when you were not then I am sorry. But I interpreted it as best I could.

If you were questioning the relevancy of what fotl had written, rather than what was meant by what fotl had written, you could perhaps have made that clearer and I would have interpreted your post differently.

[fruitoftheloom]

What I believe is meant is that the silicon placed within the BCM2837B0 chip package has been inverted to the earlier version so heat sources are thermally closer to the pins and board so heat can be better dissipated through the board as well as through the heat spreader further from the board.

What relevance has that to the discussion on producing a B2+ by not fitting the RF block in an otherwise normal B3+? (That was where fotl posted his "upside down" comment.)

I simply answered the query you seemed to have on what fotl posted.

I imagined that others might also be wondering what was meant by what fotl had written, how anyone could fit an entire SoC upside down, and sought to clarify that for you and others.

If I misinterpreted your single "?" as questioning what fotl was meaning when you were not then I am sorry. But I interpreted it as best I could.

If you were questioning the relevancy of what fotl had written, rather than what was meant by what fotl had written, you could perhaps have made that clearer and I would have interpreted your post differently.

Actually I should of said "flipped" which was mentioned somewhere......my bad!


The relevance was comparing making a 2B out of a 3B to feasibly making a 2B+ out of a 3B+........

[user879]

Plugged in and booted up my pi 3B+ and am using the ethernet port for internet. Everything is working fine except the green light on the ethernet port is not lighting up. I do have internet (I am sending this from the 3B+), the wifi is turned off, and the yellow/orange light works.

Any ideas on how to figure out why the green light isn't on?

Orange LED means gigabit ethernet is active.
Green LED means 10/100M ethernet is active.

So flashing orange and no green LED is good.

Cool, it makes sense now. Thanks!

[drgeoff]

What relevance has that to the discussion on producing a B2+ by not fitting the RF block in an otherwise normal B3+? (That was where fotl posted his "upside down" comment.)

I simply answered the query you seemed to have on what fotl posted.

I imagined that others might also be wondering what was meant by what fotl had written, how anyone could fit an entire SoC upside down, and sought to clarify that for you and others.

If I misinterpreted your single "?" as questioning what fotl was meaning when you were not then I am sorry. But I interpreted it as best I could.

If you were questioning the relevancy of what fotl had written, rather than what was meant by what fotl had written, you could perhaps have made that clearer and I would have interpreted your post differently.

Actually I should of said "flipped" which was mentioned somewhere......my bad!


The relevance was comparing making a 2B out of a 3B to feasibly making a 2B+ out of a 3B+........

@fotl
I see no relevance. Can you convince me otherwise?

[fruitoftheloom]

I simply answered the query you seemed to have on what fotl posted.

I imagined that others might also be wondering what was meant by what fotl had written, how anyone could fit an entire SoC upside down, and sought to clarify that for you and others.

If I misinterpreted your single "?" as questioning what fotl was meaning when you were not then I am sorry. But I interpreted it as best I could.

If you were questioning the relevancy of what fotl had written, rather than what was meant by what fotl had written, you could perhaps have made that clearer and I would have interpreted your post differently.

Actually I should of said "flipped" which was mentioned somewhere......my bad!


The relevance was comparing making a 2B out of a 3B to feasibly making a 2B+ out of a 3B+........

@fotl
I see no relevance. Can you convince me otherwise?

No one knows what changes were made to the actual mainboard, well no one outside RPT / RPF or has signed a NDA.

So until we do we will have no way of actually stating it is no-brainer to create a 2B+ from a 3B+.....


Yes Gregeric I talk a load of .................

Falls into the category of Pink Ponies...

[drgeoff]

Yes Gregeric I talk a load of .................

That's something I don't dispute.

[fruitoftheloom]

Yes Gregeric I talk a load of .................

That's something I don't dispute.

Yes but at least I admit it, which is more than most do..........

[W. H. Heydt]

Actually I should of said "flipped" which was mentioned somewhere......my bad!


The relevance was comparing making a 2B out of a 3B to feasibly making a 2B+ out of a 3B+........

@fotl
I see no relevance. Can you convince me otherwise?

Okay... The change from '2837A0 to '2837B0 is a "step change". It's a "minor" upgrade to the chip. so if the B0 stepping were to put into the old package, nobody would have noticed, absent an announcement, even though the changes to the boot ROM may have some fairly profound effects. The bigger issue here is that the actual Silicon chip has been "flipped". That is, it is in the package upside down compared to the old package. The chip has a lot of connections to the package--that ball grid array that is used to solder it to the PCB. By flipping the Silicon, the chip connections are no longer in the same places they were...at least, they aren't unless a mirror image operation was also carried out, and we have heard nothing about *that* being done.

Therefore, the assumption pretty much has to be made that the package to PCB connections are reversed in some way between the the two *package* versions (otherwise, the internal package wiring would get really, really...ugly). And that means that stuffing a '2837B0 onto a Pi2Bv1.2 PCB will not work. The connections are in the wrong places. That leads to two possible solutions, both of which are *probably* viable if the RPT wants to use them. The first is either a modified Pi3B+ PCB that doesn't connect or terminates the WiFi/BT connections or just uses a Pi3B+ PCB and doesn't even populate the WiFi/BT circuit area. The second is to have '2837B0 Silicon packed in the "old style" (that is, the one being used for the '2837A0 chip) and use that on the current Pi2Bv1.2 PCB.

The former approach would gain all the benefits of the Pi3B+ that pertain to the Pi2B (faster wired Ethernet, improved boot ROM, improved thermal management, PoE capability). The later approach would gain the improved boot ROM and perhaps a bit of a speed boost, but not much..

As I see it...is the improved boot ROM alone worth the change? I would argue that it is. Others will--obviously--differ.

[drgeoff]

And that means that stuffing a '2837B0 onto a Pi2Bv1.2 PCB will not work. The connections are in the wrong places. That leads to two possible solutions, both of which are *probably* viable if the RPT wants to use them. The first is either a modified Pi3B+ PCB that doesn't connect or terminates the WiFi/BT connections or just uses a Pi3B+ PCB and doesn't even populate the WiFi/BT circuit area. The second is to have '2837B0 Silicon packed in the "old style" (that is, the one being used for the '2837A0 chip) and use that on the current Pi2Bv1.2 PCB.

The former approach would gain all the benefits of the Pi3B+ that pertain to the Pi2B (faster wired Ethernet, improved boot ROM, improved thermal management, PoE capability). The later approach would gain the improved boot ROM and perhaps a bit of a speed boost, but not much..

The second version of the first solution (ie a 3B+ with the RF bits not fitted) is what I understand was being postulated. The die inside the 2837B0 being flipped has no bearing on that.

I'm not campaigning for or against a B2+ (however it might be fabricated) to be made available. Just would like any discussion to be based on relevant scientific and engineering facts.

[Imperf3kt]

why not stop this silly discussion and make your own Pi2b+ from a Pi3b+?

Buy a Pi3b+, and make with the stabbing of the wifi chip or whatever you are suggesting to remove.

If you remove it without damaging the Pi, and it works, you have a "Pi2b+" at no extra research cost to the Pi foundation.
If it doesn't work, the Pi foundation didn't spend money that could be better spent on other things, on a failed niche product that will probably soon be deprecated anyway.

[davidcoton]

And that means that stuffing a '2837B0 onto a Pi2Bv1.2 PCB will not work. The connections are in the wrong places. That leads to two possible solutions, both of which are *probably* viable if the RPT wants to use them. The first is either a modified Pi3B+ PCB that doesn't connect or terminates the WiFi/BT connections or just uses a Pi3B+ PCB and doesn't even populate the WiFi/BT circuit area. The second is to have '2837B0 Silicon packed in the "old style" (that is, the one being used for the '2837A0 chip) and use that on the current Pi2Bv1.2 PCB.

The former approach would gain all the benefits of the Pi3B+ that pertain to the Pi2B (faster wired Ethernet, improved boot ROM, improved thermal management, PoE capability). The later approach would gain the improved boot ROM and perhaps a bit of a speed boost, but not much..

The second version of the first solution (ie a 3B+ with the RF bits not fitted) is what I understand was being postulated. The die inside the 2837B0 being flipped has no bearing on that.

I'm not campaigning for or against a B2+ (however it might be fabricated) to be made available. Just would like any discussion to be based on relevant scientific and engineering facts.

@drgeoff: I appreciate your efforts to keep discussion focussed, however you seem to be overlooking the contribution quoted below, which led to the comments about flipping. Now can we stop this, and instead discuss how many Pink Ponies fit on a 40nm (or 28nm, or ...) trace....

So my question becomes...What--if any--would be the implications of just manufacturing a Pi2Bv1.2 with a '2837B0 SoC instead of the earlier '2837 (unspecified stepping) one? In other words, are the two chips "pin compatible" and only require setting the pick-and-place robot to use one package instead of the other?