Shop OBEX P1 Docs P2 Docs Learn Events
P2D2 with P2-revB - taking orders! - Page 4 — Parallax Forums

P2D2 with P2-revB - taking orders!

1246712

Comments

  • roglohrogloh Posts: 5,122
    edited 2019-11-07 01:01
    rogloh wrote: »
    I am keen on the 2A regulator and 16MB flash and hopefully 20MHz capability, though I could potentially fit a different crystal with any luck.

    I need to play with 252MHz+ for DVI/HDMI etc, so won't want voltages to sag under heavier load. I doubt the 2A would be needed on the 3.3V rail, just good to have the headroom for the 1.8V rail under extreme situations.

    I do have 2A regs available so I can make them up as well except with the clock gen fitted and you still have the option of removing the jumper resistor that couples the clock gen to the P2 and fit a crystal. What I can do is fit a custom part that ends up generating 22.222MHz at power-up but software, either from the EFM8 or the P2 can override this and set it to whatever you like.

    Sounds reasonable though I think the clocking stuff needs to be figured out. One problem I foresee is that software written to be used across different systems won't be as easily portable if you need special additional boot up code variants for the P2D2 target to setup the clock to be the same as the EVAL board. It would be ideal to default to the same 20MHz the P2-EVAL uses and then have the ability to setup the external oscillator clock source separately from the P2 only if the users want/need to. For many people they many not need to set things differently to the 20MHz. Plus if the EFM8 sets things up in the Si5351A for 20MHz you'll always need the EFM8 version, not the version with that other USB chip. It could be bit of a gotcha later if you get the wrong version now. That's how I see it anyway.
  • jmgjmg Posts: 15,140
    edited 2019-11-07 02:54
    rogloh wrote: »
    Sounds reasonable though I think the clocking stuff needs to be figured out. One problem I foresee is that software written to be used across different systems won't be as easily portable if you need special additional boot up code variants for the P2D2 target to setup the clock to be the same as the EVAL board. It would be ideal to default to the same 20MHz the P2-EVAL uses and then have the ability to setup the external oscillator clock source separately from the P2 only if the users want/need to. For many people they many not need to set things differently to the 20MHz. Plus if the EFM8 sets things up in the Si5351A for 20MHz you'll always need the EFM8 version, not the version with that other USB chip. It could be bit of a gotcha later if you get the wrong version now. That's how I see it anyway.
    Yes, I'd agree 20MHz would be nice to have, to make for easier drop-and-go user experience.
    The CP2102N does not exclude Si5351A, it just means you need to run some P2 code before you enable the XTAL source.

    EFM8 is cleaner, of course, (less for users to remember) and I just now have a EFM8+Si5351A doing a KISS 5-line init, to pop Xtal straight thru to CLK1, for P2.
    Init is done by table, & I've placed that at 0x8000 in EFM8UB3, so users can patch that memory page using the UB3 bootloader if they need a more complex INIT exit.
    All that means the Si5351A should be shipped with a 20MHz reference (Xtal or Osc), which SiLabs conspire against a little, but you can select a Si5351C CLKIN and then patch regs to flip CLKIN->XTAL.

    If a custom Si5351A was created, that could use a 26MHz low cost VCTCXO, to default 20MHz out.
  • FLASH
    Seems like the standard model, even on the special will be 16MB since nobody likes not having more :)

    SWITCHER
    The original P2D2 was equipped with a 1A switcher for 1.8V and although I overclocked it severely it was fine but it seems everyone is a bit reticent about having "only a 1A" switcher so I will make 2A standard on every board.

    EFM8
    Things however seem to be improved for the EFM8 chip since it seems to work fine at 2Mbd although loadp2 on Linux takes a long time to load whereas PNut works fine, but this needs a bit more checking.

    Si5351 Clock gen
    @jmg is integrating the clock gen loader option into the USB serial firmware too so we can use the standard Si5351 clock gen set to 20MHz. User software can write to the clock gen after power-up and select a different frequency if need be or program the EFM8 Flash via the C2 interface on the P5,6 jumper option. Just use the Silabs Clock Builder to generate a register map if need be and load in those values.

    Those who have prepaid for the special already will get the improved and standard unit no extra charge (because they're special). That way there are no units out there that need different configuration.

    STANDARD UNIT
    * 16MB Flash
    * 2A switcher
    * EFM8UB3 USB serial and ancillary functions including internal DTR reset pulses.
    * Si5351 clock gen with 20MHz P2 clock (loads the Si5351 at powerup)
    * RTC + 11mF cap
    Price $70 USD

    Express shipping is $5/unit if consolidated with a bulk order such as through @Publison in the US.

    In the meantime I will go ahead and make up one standard configuration with all the options and post the photo and results. I will have more parts on Monday so that will be a nice package going out to publison and elsewhere.

    I have included a personal PayPal.me link in my sig but make sure to pay in USD as I'm getting confused with all the different amounts and fees (probably normal)
  • All sounding pretty good now Peter!
  • Cluso99Cluso99 Posts: 18,066
    PAYPAL

    As Peter said, make sure you override to pay in USD as PayPal goes to an extraordinary extend to try and convert the currency to AUD because they make heaps on the conversion. It's about 5% each way.
  • PublisonPublison Posts: 12,366
    edited 2019-11-07 14:30
    For those of you in North America doing the bulk order, after you PayPal Peter, send me a PM so I can put you on my spreadsheet. I have my first entry!
  • EFM8
    Things however seem to be improved for the EFM8 chip since it seems to work fine at 2Mbd although loadp2 on Linux takes a long time to load whereas PNut works fine, but this needs a bit more checking.

    Have you tried with a recent loadp2 on Linux? The original loadp2 forced the load speed and the final baud rate to be the same on Linux (there's a Linux "feature" where changing baud rate on an open port toggles DTR, thus resetting the board). Thus the download speed will default to 115200 bps :(.

    This is worked around in recent loadp2s. I can't remember the exact change date, but the current one in github (https://github.com/totalspectrum/loadp2) certainly has the fix.

    BTW, congratulations on the new P2D2, sounds like a great board!
  • I finally got around to making up my first test unit and I've posted a photo back into the first post. You may notice two little solder "blobs" C & D, and these are just for loading up the EFM8 chip over the C2 interface from the P2 after which the blobs are easy to remove with the tip of a hot iron.
  • I finally got around to making up my first test unit and I've posted a photo back into the first post. You may notice two little solder "blobs" C & D, and these are just for loading up the EFM8 chip over the C2 interface from the P2 after which the blobs are easy to remove with the tip of a hot iron.

    Per photo in Post 1, Work Of Art!
  • Publison wrote: »
    For those of you in North America doing the bulk order, after you PayPal Peter, send me a PM so I can put you on my spreadsheet. I have my first entry!

    Please include your address, email, and phone if it's required for customs.
  • jmgjmg Posts: 15,140
    I finally got around to making up my first test unit and I've posted a photo back into the first post...

    Nice photo.
    How do you avoid the USB connector lip shorting the 0.1" header pins on the board edge ?
    I swapped placement on those, so the USB connector overhangs the board edge as per normal, and moved the 0.1" header pins behind the USB connector.

  • I'd be interested in a maxed out board... CO, USA.
    Please PM me with details.
  • Publison wrote: »
    I'd be interested in a maxed out board... CO, USA.
    Please PM me with details.

    All the details are in the thread. Is there something that is missing?
    After Peter's recent message about the 16MB flash chips becoming standard I'm not sure I understand the difference between the 10 "special" boards and the eventual standard boards.

  • Peter, I plan on next week. Please set the other price. I will pay that gladly and am in no rush.

    Just wanting to tool up on current rev for up and coming free time :D
  • ersmith wrote: »
    EFM8
    Things however seem to be improved for the EFM8 chip since it seems to work fine at 2Mbd although loadp2 on Linux takes a long time to load whereas PNut works fine, but this needs a bit more checking.

    Have you tried with a recent loadp2 on Linux? The original loadp2 forced the load speed and the final baud rate to be the same on Linux (there's a Linux "feature" where changing baud rate on an open port toggles DTR, thus resetting the board). Thus the download speed will default to 115200 bps :(.

    This is worked around in recent loadp2s. I can't remember the exact change date, but the current one in github (https://github.com/totalspectrum/loadp2) certainly has the fix.

    BTW, congratulations on the new P2D2, sounds like a great board!

    As a followup, and somewhat related to Peter's other work, the latest loadp2 has a -xTAQOZ flag to enter the ROM version of TAQOZ (and -xDEBUG to enter the ROM debug monitor), because I can never remember which special keys to press to do that (ESC and control-D respectively). These will be exposed in FlexGUI in the next release.

  • ...
    STANDARD UNIT (Updated Nov 8 )
    * 16MB Flash
    * 2A switcher
    * EFM8UB3 USB serial and ancillary functions including internal DTR reset pulses.
    * Si5351 clock gen with 20MHz P2 clock (loads the Si5351 at powerup)
    * RTC + 11mF cap
    Price $70 USD
    ...
    Configurations and Options:
    * Preloaded 16GB or 64GB SanDisk Ultra A1 (TAQOZ RELOADED + various media files)
    ...
    EDIT: I will release details soon of the dev board that the P2D2 can plug into as well. You can then order this with the P2D2 as a kit.
    ...

    @"Peter Jakacki" as previously stated I am interested in 2 P2D2 boards and one DevBoard with one preloaded SD. May you quote the shipping costs to Italy?

    PS1: If you have availability I would like, if possible, a 220mF supercap on one of the two boards

    PS2: If you have some group shipments requests fore Europe pls let me know. If there is no other volunteers and others are of course interested I can manage the european distribution.
  • roglohrogloh Posts: 5,122
    edited 2019-11-08 00:07
    Nice picture Peter. Hey do you happen to know or estimate with your final on-board component choices what is the minimum current possible that your board takes (eg. P2 in internal slow RC clock & IO pins tristated) where the P2 might still be able to set itself up to wake up periodically, perhaps using RTC alarm, or just with waitx intervals etc...?

    I am now wondering if I might be able to sleep the P2D2 instead of powering it down.
  • jmgjmg Posts: 15,140
    rogloh wrote: »
    ... Hey do you happen to know or estimate with your final on-board component choices what is the minimum current possible that your board takes (eg. P2 in internal slow RC clock & IO pins tristated) where the P2 might still be able to set itself up to wake up periodically, perhaps using RTC alarm, or just with waitx intervals etc...?
    I am now wondering if I might be able to sleep the P2D2 instead of powering it down.
    There is also base-currents of other parts, like EFM8UB3/CP2102N which is ~ 9.5mA and I measure Si5351A at 11~14mA. Oscillator modules are ~ 1.5mA for GPS clipped sine.
    Those are low by P2 running standards, but likely high by P2-deepest sleep standards.


  • Yeah ok I won't be able to sleep the board with those sorts of numbers. I will still have to use my other micro to control P2D2 power. I am targeting < 100uA sleep ideally, though up to 1mA is probably bearable.
  • Ordered standard version.
    Peter, thanks for all your innovation and hard work.

    Tom
  • jmgjmg Posts: 15,140
    rogloh wrote: »
    Yeah ok I won't be able to sleep the board with those sorts of numbers. I will still have to use my other micro to control P2D2 power. I am targeting < 100uA sleep ideally, though up to 1mA is probably bearable.
    Even the regulator's Iq's are also significant on < 100uA scale, but you may be able to wire to the main regulator ENables, from your sleep MCU - those have a pullup.

  • cheezuscheezus Posts: 295
    edited 2019-11-08 07:23
    Getting very excited, less than 12 hours before I can pull the trigger on my first P2D2. :smile::smile::smiley: I will probably be ordering a 2'nd and devboard… next month if I'm lucky?? but likely before the 1st of the year.

    I'm starting to look at the design of my current prototype board and how best to incorporate the P2D2. While I'd like to use those nice castellated holes to surface mount to the board I have yet to redesign, I'm thinking it might be better to use the pin headers for thermal considerations. My original idea was to place a large copper pour to carry the P2's thermal pad to the bottom of my board, then place a small heatsink on the bottom of my pour. The RTC seems to complicate my design though, with P32 -P36 being tough to get to without a cutout in my pcb to allow clearance... Need to think on that more.

    I'm wondering if @"Peter Jakacki" has done any testing of heatsinked vs non-heatsinked module's performance and could give some advice? In the small cardboard case the ES1 gets quite warm as the LCD's backlight puts off some heat but the unit isn't fully enclosed yet. I was considering building a case out of MDF, possibly wrapped with vinyl decal but I might end up using metal instead. Not sure how I would thermally connect the module to the case.

    Once again, very nice work @"Peter Jakacki" This board has solved a bunch of design considerations allowing me to think about everything BUT the P2 and support circuitry. I think I'm getting to the point of trying to wrap my head around TaqOZ again. I've been focusing on SW in my free-time but I think some HW development is due. I'm hoping some nights spent routing pcbs will help my eyes quickly finding the paths for my wire at work. :lol:


    *edit

    After closer inspection it looks like I could pull that resistor, as well as the dual 100k since I'm using all of port B. Looks like port A is wide open, other than the c2 interface. Might be able to take care of that with a couple jumper wires. Looking good indeed!
  • If anyone prepays this week I will offer it with 16MB and still at the special price of $60 except only the first nine units will ship this week while I wait for more parts and ship those next week.

    Hi Peter, I want one board from first batch if still available. If not available, then I wait for the second batch. Shipment to taiwan.

    I am confused about the SD card socket option, the picture of the board shows already a socket. There are different socket size options? full vs micro SD? I am ok with either option, just want to make sure there is one SD socket there.

    Thanks !
  • Ramon wrote: »
    If anyone prepays this week I will offer it with 16MB and still at the special price of $60 except only the first nine units will ship this week while I wait for more parts and ship those next week.

    I am confused about the SD card socket option, the picture of the board shows already a socket. There are different socket size options? full vs micro SD? I am ok with either option, just want to make sure there is one SD socket there.

    Thanks !

    There is an option to mount the microSD on the bottom side of the board. The connections are there, so it could be added later.

  • I always say that the guy that designs the circuit schematic should be the guy that selects the parts. The guy that selects the parts should be the guy that orders the parts (what a dilemma that can be). The guy that orders the parts knows about price, availability, and lead-time etc. Now the guy that does the schematic should be the guy that designs the pcb, and to cut a long long story short, the guy that designs the pcb should be the guy that builds it, at least initially. That's just for a start, and only then can you fully appreciate design choices and see ways of improving the "design" in a wholistic way.

    Here's my feedback from building the latest pcb that is featured in the first post.

    1. I need to change those silly Schottky diodes due to their having one large smd pad and one fine pad, they have a tendency to creep on a standard 0603 footprint.

    2. The reset switch is really a fraction too big so I need to find a better (but cheap) switch or make more room on the next pcb.

    3. This is the big one - I can't use 16MB SOIC Flash as it is 2.16mm high and the microSD card base is at 1.75mm.
    I could use the WSON package for the Flash even on the same footprint, it really needs a better footprint but I can also order taller sockets instead (3.35mm vs 2.65mm). However, after much to and fro and after creating a pcb footprint for WSON8 and testing it on the layout I can't see any problem with using it. The exposed pad can be left floating or grounded but in this case it will be sitting on id and mask and all vias are tented anyway. This package is only 0.8mm allowing for plenty of clearance without raising the socket height. Anyway, the SOIC8 package is fairly bulky otherwise.

    I will probably make up a little pogo pin jig to plug into the P2D2 and have it program the EFM8 chip in one hit rather than having the P2 solder jumpers. Even though my test board had a couple of tantalums, I am more comfortable with reliable ceramics.

    Time to order more parts if I want them shipped over the weekend so I can get these built early in the week.


  • jmgjmg Posts: 15,140
    edited 2019-11-08 19:29
    3. This is the big one - I can't use 16MB SOIC Flash as it is 2.16mm high and the microSD card base is at 1.75mm.
    I could use the WSON package for the Flash even on the same footprint, it really needs a better footprint but I can also order taller sockets instead (3.35mm vs 2.65mm). However, after much to and fro and after creating a pcb footprint for WSON8 and testing it on the layout I can't see any problem with using it. The exposed pad can be left floating or grounded but in this case it will be sitting on id and mask and all vias are tented anyway. This package is only 0.8mm allowing for plenty of clearance without raising the socket height. Anyway, the SOIC8 package is fairly bulky otherwise.
    Unusual problem...
    I checked the PSRAMs, and the two I found both spec MAX 1.75, or a standard SO8-150 dimension.

    Looking at lcsc, I find these in SO8-150, all 1.75 hi

    BY25Q32BSTIG BOYAMICRO FLASH FLASH 32M SPI SOP8150mil RoHS 1+$0.1970 100+ $0.1242
    P25Q16H-SSH-IT (PUYA) FLASH FLASH 16Mb (16M x 1, 8M x 2, 4M x 4) SPI - Dual/Quad I/O SOIC-8_150mil RoHS 1+$0.2121 100+$0.1348
    W25Q16JVSNIQ Winbond Elec FLASH FLASH 16Mb (2M x 8 ) SPI - Dual/Quad SOIC-8 RoHS 1+$0.2848 100+ $0.1818
    FM25Q16A-S0-T-G Shanghai Fudan Microelectronics FLASH FLASH 16Mb (2M x 8 ) SPI - Dual/Quad I/O, QPI SOP-8_150mil RoHS 1+$0.3606 100+$0.2242
    W25Q128JVPIQTR Winbond FLASH FLASH 128Mb (16M x 8 ) SPI - Dual/Quad I/O WSON-8_5x6mm RoHS1+ $1.4212 100+$0.9091

    Digikey has MX25L3233FM1I-08Q 32Mb (8M x 4) SPI - Quad I/O 133MHz 8-SOP-150 100+ 0.54530
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2019-11-08 22:19
    @jmg - It's the old bit vs Byte capacity gotcha. I'm talking 16MB = 128Mb.
    But the W25Q18JVPIQTR is what I am ordering although it is from Digikey this time. Maybe I will try lcsc next time, they were good when I ordered those PSRAMs last time.
    (lcsc website is barely crawling if at all at the moment)
  • RossHRossH Posts: 5,336
    Publison wrote: »
    RossH wrote: »
    @"Peter Jakacki"

    No hurry for a board - lots of other things to do - but if you let me know how to pay I'm happy to pay in advance. Apologies if this has already been posted somewhere and I missed it!

    Ross.
    10 special units available this week at $60 (USD) each and to keep shipping costs down this may mean that I ship this as one thanks to @publison in the US.
    Add an extra $5/unit shipping to make that $65 and PayPal peterjakacki@gmail.com and you will have them next week.

    Sent!
  • RossH wrote: »
    Publison wrote: »
    RossH wrote: »
    @"Peter Jakacki"

    No hurry for a board - lots of other things to do - but if you let me know how to pay I'm happy to pay in advance. Apologies if this has already been posted somewhere and I missed it!

    Ross.
    10 special units available this week at $60 (USD) each and to keep shipping costs down this may mean that I ship this as one thanks to @publison in the US.
    Add an extra $5/unit shipping to make that $65 and PayPal peterjakacki@gmail.com and you will have them next week.

    Sent!

    No worries Ross, I canned the special units offer and upgraded them to standard for all those who paid and you are still included too :)
    Hopefully I will get more parts on Monday including the 16MB Flash so since you are here in Oz you should have your unit during the week.
  • I'm wondering if I can get measures for the centers of the 1.27 mm and 2.54 mm pin headers. I'm working on a board layout and it's pretty much to that point. I was not sure how I was connecting the p2d2 to my board when I started but now I'm pretty sure of using the 1.27mm pin headers. Female on component side of p2d2, male on back of my board. Leaves the thermal pad exposed for heatsinking. I'm also thinking port A will be accessed using right angle 2x40 2.54mm header...

    I have a bunch of extra room but haven't decided what to add, other than a barrel jack 9-24v dc in. I've been thinking about a couple usb ports but I kind of like the idea of leaving all of port A open.
    795 x 793 - 28K
Sign In or Register to comment.