Shop OBEX P1 Docs P2 Docs Learn Events
P2D2 - An open hardware reference design for the P2 CPU - Page 22 — Parallax Forums

P2D2 - An open hardware reference design for the P2 CPU

1192022242538

Comments

  • jmgjmg Posts: 15,140
    Publison wrote: »
    Peter, given that P2 RevB will be here soon, is there anything you would change on the original BOM for the original P2D2 to accommodate them? I still have 4 boards that nobody in the Americas wanted, (probably the SMT fear). I my get qty 10 of all the parts and apply the toaster oven magic to distribute with the new chips. Seems a shame to let these boards sit here not creating code. My be worth getting a stencil made.
    For just 4 boards, maybe it's better to put new ES2 parts onto latest P2D2R2 boards ?
    Quite a few changes were made to the BOM, all of which make a better platform, and that avoids multiple variants.

    It maybe worth considering 4 layer boards too, as that's a PCB-FAB option for P2D2 boards ?
  • Just trying to get some mileage out of these boards that have traveled a few thousand miles thanks to Peter. Staring at 5 naked boards make me sad. :)
  • Keep them as drink coasters or hand-outs, I've still got some of those left and a whole bunch of r2s as well which I am not going to use.Now I have a whole stack of r3 pcbs that I can make up with original ES chips or wait for some new B chips. When I get more made up I will send you one :) (and one for Chip too).
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2019-10-22 12:05
    I'm looking forward to getting hold of new silicon to load onto my newer r3 version of the P2D2.

    As a side note I am quite impressed with the RV-3028 RTC chip and tiny tiny supercap that powers it along. Here's a photo from the r2 version (which had too wide a footprint for the RTC) alongside a 0603 resistor, so you can see the RTC with integrated crystal is so small, as is the capacitor. The CPH3225A 11,000uF cap charges within minutes, does not wear out, and keeps it powered for days easily since the typical standby current is around 45nA. The r3 pcb also allows for a larger supercap too but I find that once I have equipment installed it is rarely ever turned off and if it is turned off then it must be out of service anyway.

    Anyway, such a tiny integrated 1ppm RTC including UNIX time, and some EEPROM and RAM etc. I'm glad I found this after so many years of putting up with "traditional" mediocre chips.
    TAQOZ# .FDT --- 2019/10/22 TUE 22:05:28 ok
    TAQOZ# UTIME@ . --- 1571745931
    

    1573 x 1008 - 254K
  • That's a nice little RTC.
  • I'm looking forward to getting hold of new silicon to load onto my newer r3 version of the P2D2.

    As a side note I am quite impressed with the RV-3028 RTC chip and tiny tiny supercap that powers it along. Here's a photo from the r2 version (which had too wide a footprint for the RTC) alongside a 0603 resistor, so you can see the RTC with integrated crystal is so small, as is the capacitor. The CPH3225A 11,000uF cap charges within minutes, does not wear out, and keeps it powered for days easily since the typical standby current is around 45nA. The r3 pcb also allows for a larger supercap too but I find that once I have equipment installed it is rarely ever turned off and if it is turned off then it must be out of service anyway.

    Anyway, such a tiny integrated 1ppm RTC including UNIX time, and some EEPROM and RAM etc. I'm glad I found this after so many years of putting up with "traditional" mediocre chips.
    TAQOZ# .FDT --- 2019/10/22 TUE 22:05:28 ok
    TAQOZ# UTIME@ . --- 1571745931
    

    I know I'm looking far ahead, but I hope they develop a drop-in replacement that isn't vulnerable to the Year 2038 problem.
  • Hi Peter,

    I can't see the capacitor. Where is it?

    The specs of the RV-3028 are impressive. The consumption current is very small, and the wide operating voltage range it offers definitely allow for use with a supercap (no battery needed).

    Kind regards, Samuel Lourenço
  • I'm looking forward to next year, because then we will see with 2020 vision :)
    As for the 2038 problem, I still don't see why they can't just work with it as a full 32-bit unsigned integer.
    samuell wrote: »
    Hi Peter,

    I can't see the capacitor. Where is it?

    The specs of the RV-3028 are impressive. The consumption current is very small, and the wide operating voltage range it offers definitely allow for use with a supercap (no battery needed).

    Kind regards, Samuel Lourenço
    I've never really used batteries, always supercaps or a combination of a tiny battery and a supercap. It was always so laughable back in the early PC days when they would use this huge rechargeable NiCd battery for the RTC that would then deteriorate badly over a few years. Then they got a bit more sensible and used a huge Lithium coin cell instead which could power the clock on standby for a year. But that would mean the unit has been out of service and on the scrapheap already if that was the case.

    The cap is the metal component at the bottom against the 0.05" holes.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2019-10-26 03:59
    Expect to find P2D2r3 boards available in the next week or so once I can get some new P2 chips. The r3 version includes USB serial either via a CP2102N or a programmed EFM8UB3 chip although you can still override the USB serial with a Prop plug due to the resistor coupling. The reason that the P2D2r3 is not compatible with other boards is simply because it is a general-purpose module that can be used for stand-alone, for prototyping on matrix board, or as a final product in a custom PCB. There is a small thin sandwich board that will come with it that includes RPi pinouts that don't compromise the general 64 I/O of the P2.

    The r3 uses a dual switching regulator for the 1.8V and 3.3V supplies but since switching regs can be a bit noisy for A/D etc, the switcher is actually set for 3.6V and use a tiny dual LDO to supply clean rails to the the A and B side of the P2D2, that is 32 I/O each. Most LDOs can supply high current but only if power dissipation is kept low and so the 3.6V means that even at full 300ma x2 current that the power dissipation will only be in the order of 600ma x 300mV = 200mW. That's if every I/O was sourcing 9ma constantly all at the same time but that never happens. The 3.6V rail is available on the connector to make it easier to supply other devices and regulators. BTW. 3.6V is still quite safe to feed into a 3.3V input since CMOS are spec'd for +/- 300mV beyond rails.

    Some LDOs are specified as being low-noise but most of this has to do with the output caps that are used and they normally specify good 10uF caps for this low noise figure rather than the 1uF required for stability.

    As for the dev board that this can plug into I have a request to incorporate mikroBUS connections so I don't see a problem with that either. Same goes for PMOD P2 eval accessory kit boards.
    Many of the common interfaces will be incorporated on the dev board as standard without the need to plug in sometimes awkward modules.
  • Awesome, I've been waiting for one of these for a while and when it's ready I can resume my other pet project...
  • The P2D2 looks like more my cup of tea for starting out exploring the P2, rather than a P2-ES.

    The RV-3028 looks like an awesome find as well, which would be fun to play with as a potential go to RTC for other projects.

    So I guess my only question is how long is the P2D2 buyer queue :-) and cost incl shipping over the pond to NZ?
  • @GregC - I think a small satchel only costs around $10 but I could send it as a regular "letter" much cheaper. A printed circuit board is after all "printed". I'm filling small orders first so I don't think there is any problem with getting one.
  • Peter,
    Please count me in on one of the P2D2r3 assemblies. I have been waiting patiently through the whole P2 evolution with most others but the time seems right to finally have the real deal. Your board is exactly what I need to continue the journey, Thanks for your time and efforts to support the community especially with your Forth endeavors.

    Bob
  • Peter,

    I mention the r3 but I see r4 mentioned also. If r4 is ready that is what I think I would want. Please specify the differences between the r3 and r4.

    Thanks
    Bob
  • r4 will simply be r3 with an extra break-away edge with the RPi pinouts. So if you don't want those pins, just snap them off. Like all pcb revisions though it will incorporate any fixes and improvements although I can't see that there would be any other changes necessary at this stage.
  • For the life of me, I can't see the advantage of any direct connection from a P2D2 to the RPi unless I wanted an introverted P2. Wouldn't we want a hat that the P2D2 plugged into instead? Or would the P2D2 simply use RPi hats instead? In which case even then a hat adapter would make more sense too and physically support the hat and provide power connectors etc. Thoughts?
    1842 x 1185 - 420K
  • Connecting a Raspi could be a good way to cheaply get a full IP stack accessible the Prop, or Wifi or access to more USB devices etc (indirectly), plus another external screen/keyboard/mouse for software development/download to it etc, if some of the tools like loadp2 and the compilers get ported to Raspi. If a variant of the P2D2 could use HATs or go via an adapter that would be good too and open up fun things to play with in a different mode of operation, more standalone.
  • What if I did an RPi footprint compatible pcb that the P2D2 could plug into. So you would have the 40-pin RPi header, the microUSB for power (and serial USB), HDMI, audio, Ethernet either through a W5500 or just a suitable MAC+PHY, four USB host sockets, a display header, and use an ESP32 for WiFi? The P2D2 module 0.1" headers does fit in between all this except for maybe the camera connector.
  • jmgjmg Posts: 15,140
    For the life of me, I can't see the advantage of any direct connection from a P2D2 to the RPi unless I wanted an introverted P2. Wouldn't we want a hat that the P2D2 plugged into instead? Or would the P2D2 simply use RPi hats instead? In which case even then a hat adapter would make more sense too and physically support the hat and provide power connectors etc. Thoughts?

    You can do either or both.
    P2D2Pi can replace a Pi, or you can plug any Pi Hat straight into P2D2Pi, on either 40 pin header. (or both)


    Note: P2D2Pi makes both 40p connectors the same, the original P2D2 does not quite match each side's pin-map, so you cannot simply swap-plug into either.

    eg The myriad hats allow rapid and easy direct connect of LCD displays, and a 40 pin header is good enough to mount a P2D2Pi board fine, for any development and proof of concept development.

    Here is one example - a 128 x 64 OLED module.

    That drops neatly onto P2D2Pi, almost like it was designed for it... :)

    There are also 125MHz SPI LCDs, 2.8" / 3.2" and up, and this is a Motor Driver and Power module

    It gives :
    Power input range: 7V~40V
    Single motor output current: up to 5A
    Power supply current for Raspberry Pi: up to 2A, @ 5VDC

    That Motor/Power module probably works better with P2, than with a RaspPi....


    Notice these modules are all differing sizes, the connector is the only common element, & even there, it can be as small as 6 pins of the 40, (as in that OLED example) so 'physically support the hat' is open-ended on both Pi, and P2D2Pi
  • I'm looking forward to getting hold of new silicon to load onto my newer r3 version of the P2D2.

    As a side note I am quite impressed with the RV-3028 RTC chip and tiny tiny supercap that powers it along. Here's a photo from the r2 version (which had too wide a footprint for the RTC) alongside a 0603 resistor, so you can see the RTC with integrated crystal is so small, as is the capacitor. The CPH3225A 11,000uF cap charges within minutes, does not wear out, and keeps it powered for days easily since the typical standby current is around 45nA. The r3 pcb also allows for a larger supercap too but I find that once I have equipment installed it is rarely ever turned off and if it is turned off then it must be out of service anyway.

    Anyway, such a tiny integrated 1ppm RTC including UNIX time, and some EEPROM and RAM etc. I'm glad I found this after so many years of putting up with "traditional" mediocre chips.
    TAQOZ# .FDT --- 2019/10/22 TUE 22:05:28 ok
    TAQOZ# UTIME@ . --- 1571745931
    
    Thanks, Peter. And sorry for the late response.

    That cap strongly resembles a crystal, hence the confusion.

    Kind regards, Samuel Lourenço
  • jmg wrote: »
    ...
    P2D2Pi can replace a Pi, or you can plug any Pi Hat straight into P2D2Pi, on either 40 pin header. (or both)

    Note: P2D2Pi makes both 40p connectors the same, the original P2D2 does not quite match each side's pin-map, so you cannot simply swap-plug into either.

    ...

    @"Peter Jakacki"

    I've not been following the differences between theP2D2 R3 and Pi versions. I am even not interested in the RPi compatibility.

    But if it is only to swap 2 pins on one of the two 40 pin then I think that is better to do it, provided you haven't any plausible reason for not doing it (eg already made mother/dev boards.
    If this opens the doors to a higher demand for the P2D2 module I think that a higher production volume of the modules can, in future, only decrease its costs.

    My interest into P2D2 is for home/cottage/boat/caravan automation beside industrial control. This will be in din-rail enclosures with stacked PCBs anyway, so having a P2 module sweets my needs. And having it as cheap as possible of course is better.
  • I am a big fan of Nano Pi Neo (not NEO2) : http://www.nanopi.org/NanoPi-NEO_Feature.html

    If all you want is linux, ethernet and USB, you don't need anything else.
    There is also a core board : http://www.nanopi.org/NanoPi-NEO-Core_Feature.html
  • From the perspective of STEM education, having RPi hat compatibility would be a win. Getting access to that whole ecosystem of prepackaged functionality not only keeps interest up, it helps on the budget. One set of hats. Several different processors.
  • jmgjmg Posts: 15,140
    edited 2019-10-28 18:45
    dMajo wrote: »
    But if it is only to swap 2 pins on one of the two 40 pin then I think that is better to do it...
    It does only swap some pins, ( but more than 2 :) ) - both headers are 40 pins and both have 5V/3v3/GND, so they are already very close.

    The P2 package dictates the primary pin order, I've just mapped +5V, 3v3's and GND's to the same-pins as Pi first, then P2 connects to all others.
    See the attached image below (click to see all 40 pins) for the lower 32io attached, with netnames on. (Upper 32 are mapped exactly the same, just add 32, allows either/both connector choices & simpler DOCs )
    Broadly P0 is bottom left, and P31 is top-right.
    The 50mil connector has minimal change, mostly to insert GND,3v3 where Pi needs them.

    Main targets are the serial UART/SPI/i2c Pi connected boards, and there are lots of those.
    1334 x 189 - 47K
  • Cluso99Cluso99 Posts: 18,066
    The original Pi pinout was a disaster (we'd call it a dog's breakfast here in Oz).
    But they sold so many that the disaster remained for compatibility and now we're stuck with it :(

    Quite similar to the Arduino with their silly offset set of pins so no breadboard works.
  • I guess the takeaway point is don't finesse the connector header layout, just get to market!...
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2019-10-29 01:55
    One of the important points to remember is that the P2D2 is both a stand-alone board as well as a plug-in module, so it is important to bring all 64 P2 I/O out and in order just as if it were a big chip. RPi compatibility, although handy in some circumstances, is second to this requirement. Having the RPi header in addition to the standard header keeps both worlds happy and I will mostly break off this extra bit myself if it gets in the way.

    @Cluso - yes, it is a dog's breakfast. What were they thinking?
  • jmgjmg Posts: 15,140
    edited 2019-10-29 02:20
    .. so it is important to bring all 64 P2 I/O out and in order just as if it were a big chip...

    I'll admit I'm lost here ?
    In the Pi version all 64 IO do connect, and if you look at the 50 mil pinouts (image above), those are in order (ie sequentially increasing) - at least as much as P2 itself, is in any order.
    Remember, P2 has power pins interleaved, so any 'order' breaks every time a power/other pin appears.
    Certainly P1 does not appear on Pin1, and P31 does not appear on Pin31 (& there is no Pin 0... ), so users will always have to use a pin map, no matter what board they use.
    That's entirely expected, surely ?

  • @jmg I think you are putting way too much emphasis on what is essentially a cosmetic option. It's not rocket science whereas the work you did on the EFM8UB3 for serial USB is really useful.

    The RPi only has 26 I/O all together anyway and a breakaway tab just for it is a cleaner and more general purpose option. Even the 100mil headers are optional on the P2D2 since it also has the 50mil holes that can be castellated to allow the module to be surface mounted. Anyway, I will have ten of these modules made up by next week and then I will do another run of 50 (or more) immediately afterwards.
  • roglohrogloh Posts: 5,122
    edited 2019-10-30 23:32
    On my own P2 board which I have on hold I was able to get RasPi header compatibility with pins mapping over 1:1 to GPIO numbers, but this certainly made the PCB routing somewhat harder for me and would be no good for getting wide parallel busses in a nice sequence on the header or much good for very high speed applications, so I can certainly see arguments for yet another pin mapping that abstracts it away. But there also was a convenience to not needing another mapping if you are simply toggling pins to control an existing HAT. It's more of an issue if you want to design a HAT specific for the P2 with parallel buses.

    My own very simple mapping was this...
    P2  RasPi
    ---------
    P0 - ID_SD  - can be used for an internal I2C bus, and/or read HATs
    P1 - ID_SC
    P2 - GPIO2
    P3 - GPIO3
    P4 - GPIO4
    ...
    P27 - GPIO27
    
    It left me 4 pins at the top of port A which is quite good for other things on the board, such as an internal SPI bus to some peripherals/2nd SD card/USB/RGBS etc.

    One problem might be using P2 drivers that expect all pins to be in a sequential pin group such as SPI. You really want independent pin control to send your signals wherever on the header. But given there is only really a SPI bus group even defined on the RasPi, and drivers can be made to adapt it is unlikely to be that big a deal.
Sign In or Register to comment.