Shop Learn
what happened to the P2D2 and Peter Jakacki? - Page 2 — Parallax Forums

what happened to the P2D2 and Peter Jakacki?

2»

Comments

  • @Cluso99 said:
    I have spoken to Peter a couple of times over the past week.
    He's trying to get the P2D2 done but struggling with a couple of issues.
    Hopefully he will comment here.
    BTW he is not 100% which doesn't help either.

    Now you got me worried again. What do you mean by saying "..he is not 100%..." ?

  • @Maciek said:

    Now you got me worried again. What do you mean by saying "..he is not 100%..." ?

    It's been implied in earlier posts in this thread that he is somewhat, somehow, physically unwell. I don't specifically know, and I don't care to speculate. Whenever he feels better, he'll be welcomed here. Until then...

  • Cluso99Cluso99 Posts: 17,839
    edited 2021-07-13 22:59

    I speak to Peter every week or two on zoom. He’s been converting TAQOZ/Tachyon to run on RISC V and is on the forth group.

    He still says he’s working on P2D2 but there are a few issues he’s trying to work through. Perhaps putting all that stuff on the P2D2 was not the best idea? So he’s spending less time on P2 for now while still building P2D2 one at a time. Some times you just need to take a break for clarity.

    Peter’s health hasn’t been 100% for some time now, but it’s not for me to say, so please don’t comment.

  • Sounds like good news again.
    Having Tahyon/TAQOZ ported to Risc-v opens up a whole lot of possibilities for both the Risc-v based chips and for the P1/P2.
    I hope for the best and ask no more.

  • ErNaErNa Posts: 1,620

    What will happen to the P2D2? I still see a lot of chances to apply such a board and indeed, Peter's absence gave me a punch as I was just starting a project that will make use of many P2's and will run a network of TAQOZ processes. I hope he will find his way back soon and we will establish something like a special interest group to promote the most promising set of hardware/software I can imagine! Whoever is interested in such an open group may wistle!

  • Cluso99Cluso99 Posts: 17,839
    edited 2021-07-14 20:18

    Peter is still interested in TAQOZ and the P2. In fact he’s given a zoom presentation on TAQOZ and P2 on the Forth group recently.

    As for the P2D2, I’m willing to do a P2D2 semi-equivalent but I need to know what parts are required as I see some isssues in producing a direct copy. Perhaps you may care to comment on these points
    1. Switcher power supply
    2. Selectable clock input frequency vs 20MHz xtal
    3. Multifunction USB 2102 onboard
    4. PCB size and connectors
    5. Any other features

    These are my thoughts on these points
    1. The switcher chip used is creating an assembly problem as the footprint is tiny and the chip doesn’t always self align during reflow. I prefer the old LDO method as it results in a quieter board.
    2. I prefer to use a 20MHz crystal. It just works reliably when used correctly.
    3. The soft USB implementation has some glitches that I know little about. I would prefer to either use a cheap external USB to serial dongle or place a cheap CP2102 or CH340 chip on the pcb. I have some CH340 chips that I’m trying out.
    4. PCB size and connectors. What is important? The 0.050” pitch connector and/or 0.1” pitch?
    5. On the underside P2D2 allows for directly attaching expansion boards. I’m inclined to remove this.

  • jmgjmg Posts: 14,756

    @Cluso99 said:
    Peter is still interested in TAQOZ and the P2. In fact he’s given a zoom presentation on TAQOZ and P2 on the Forth group recently.

    As for the P2D2, I’m willing to do a P2D2 semi-equivalent but I need to know what parts are required as I see some isssues in producing a direct copy. Perhaps you may care to comment on these points
    1. Switcher power supply
    2. Selectable clock input frequency vs 20MHz xtal
    3. Multifunction USB 2102 onboard
    4. PCB size and connectors
    5. Any other features

    Maybe Layers ? 4 layers is better cooling and lower noise.

    I did a design fork of the P2D2 as P2D2Pi, my comments below

    These are my thoughts on these points
    1. The switcher chip used is creating an assembly problem as the footprint is tiny and the chip doesn’t always self align during reflow. I prefer the old LDO method as it results in a quieter board.

    LDO's generate too much heat, and thus limit the usable-P2

    Wrt original P2D2, I removed that tiny switcher, as it made me wary and it was giving issues, and changed to a TLV62585PDRL, and a 'oversize' inductor. (basically chosen to fill the spare space)
    I also split the single-package LDO into 2 x NCP187AMT330TAG, for better thermal spread, PowerGood, and lower 3v3 noise, and that part has peak 1.2A rating.
    I added a second USB connector for 1.8A 5V power, this makes PCB 57.7mm, up from P2D2 55.5mm

    1. I prefer to use a 20MHz crystal. It just works reliably when used correctly.

    Xtals are 'cheap and cheerful' but note the new Edge has changed to a TCXO as the P2 really has precision well above a vanilla crystal, and crystals have cross-talk issues with adjacent pins.
    The P2D2 has Si5351A, which self loads on UB3 boot, so is invisible to users, but there are wiring options for other choices.
    I added a NTC Xtal footprint, and also added a 74AUP1Z125GW as a optional buffer to bypass Si5351a, but still allow TCXO to be used.
    Edge uses 74AUP2GU04 + R + C, which is another alternative.
    P2D2pi footprint choices are 2016/2520 on top-side (CMOS or Clipped Sine) and 3225 and 5032 on bottom, those larger ones allow access to 0.1ppm level TCXO.

    1. The soft USB implementation has some glitches that I know little about.

    What are the issues with EFM8UB3 ? I thought that was stable last I heard ?

    I would prefer to either use a cheap external USB to serial dongle or place a cheap CP2102 or CH340 chip on the pcb. I have some CH340 chips that I’m trying out.

    The EFM8UB3 gives higher performance than CP2102N, and loads Si5351a and adds i2c-access measurements, and careful reset handling.

    Other UART parts I've looked at are :

    • CH343 is new version of CH340, and they have a CH9102 which maps to the higher pinout CP2102 choices.

    • PL2303GC is a new, faster version of PL2303, it gives a virtual baud clock of 96MHz/N and can set up to 12MBd with sustained speeds above 7MBd . DSD-TECH do a module using this new part.
      On minus side, it exhibits a strange ~ 420ns stretched stop bit. That does not affect classic UART, but it means testing results are never quite full speed, and it makes checking baud settings harder.
      On most UARTS, I can send a large file of 0x55 and connect a counter to determine actual baud setting.

    • XR21B1420 - this has very good fractional baud support, with jitter free Baud of 48MHz/N and fractional values of 480MHz/N. The appeal here is P2 (or P1) SysCLK ca be more freely chosen.
      It also sustains the highest rates of any FS-USB Bridge, it can give over 10MBd sustained average - that's way above CP2102N / FT231X
      On minus side, their tech support is not stellar, and it is not mainstream. CLKOUT works but I've found no means to access < 24MHz settings.
      I've done a P2D2Pi variant layout with XR21B1420IL28 + EFM8 QFN20 part while I mull this over.

    • FT232H is HS-USB, but only comes in large packages, too big for P2D2, and is costly.

    1. PCB size and connectors. What is important? The 0.050” pitch connector and/or 0.1” pitch?

    I changed the P2 pinout to make the 2 x 40 pin connectors swapable/identical pins for both sides, (avoids wrong-side oops), and made the pinouts match RaspPi.
    Solder jumpers allow Pi or full P2 pin access.
    I kept 0.050” and 0.1” connectors. Removing the 0.050" is possible, but board would only shrink a small %, and they are useful probe points for development and debug.

    1. On the underside P2D2 allows for directly attaching expansion boards. I’m inclined to remove this.

    That direction I never followed.

    Other choices : If you like simpler, there is also this P2ME2 (nice name :) ) by rogloh
    https://forums.parallax.com/discussion/comment/1525488/#Comment_1525488

    That drops the 0.050" connectors, changes to external USB, and shifts PSU focus to wider supply than P2D2, so does SMPS to 3v3 and then compact integrated inductor SMPS to 1v8 from 3v3 and uses double sided parts placements.
    That has an Oscillator or XTAL layout option (not sure what package ? ) but may allow upgrade to moderate TCXO like ECS-TXO-2016-33-200 ?

  • Cluso99Cluso99 Posts: 17,839
    edited 2021-07-15 09:01

    @jmg Did you build and test your board?

    My RetroBlade2 doesn’t suffer noise issues. The crystal runs fine. I use a 2 layer design. It’s one of the better/best overclocking designs. I just don’t get what all the fuss is about not using crystals!!!
    My RetroBlade2 is only marginally different in size to the P2D2 at 2.5mm 0.1” wider and 3.3mm 0.15” longer. Of course I don’t bring out the top 16 I/O on headers as all bar 6 are on other connectors and I don’t have USB 2102 on board. However my board plugs into breadboards as I use single row 0.1” pitch for P0-15 and P32-47 and you can also use the 0.1” P16-31 across the bottom.
    I have some tiny heat sinks used for RPi coming and expect this will remove the heat for my LDO regulators and P2.
    For these reasons, I still very much favour LDOs rather than switchers. BTW the newer P2D2 uses a different switcher and it’s this footprint I understand is causing problems. However I am happy to use a switcher if necessary as long as the footprint is easy to place and reflow.

  • VonSzarvasVonSzarvas Posts: 2,424
    edited 2021-07-15 10:03

    @Cluso99 said:
    @jmg Did you build and test your board?

    My RetroBlade2 doesn’t suffer noise issues. The crystal runs fine. I use a 2 layer design. It’s one of the better/best overclocking designs. I just don’t get what all the fuss is about not using crystals!!!

    If you can hook your board to a high freq. Scope you will see that your board does also suffer from jitter, just less than other boards due to the location and trace path to the xtal. The line is thin, and your board is just on the better side, but will still be unreliable/marginal in some applications. But maybe not the applications your design targets.

    Chip might be able to explain better tonight.

    Product-wise we took the view that for a major customer group the Edge targets, then having a super accurate and reliable clock will allow for interesting experiments and projects in the classroom; which would not be practical with the simpler crystal design.

    That's one reason why we went the whole hog and fixed the issue fully with the TXCO once rogloh spotted the issue on p28-p30. Sure though, the simple crystal is totally fine for many applications, and I'm sure some future Plx boards will use them too.

  • We're talking custom boards here. The one that is just perfect for the intended use case might not be even close to adequate for the different purpose for which it was not considered. The only notable exception I see to that statement is the eval/development type of boards, These should indeed be designed with the broad variety of applications in mind. If the board meets the requirements for the particular purpose it was built for, then it is THE RIGHT BOARD for that purpose. It might be good or even excellent for other purposes but that's just a bonus.

  • Hi,
    in my opinion a low cost (!!) but complete P2 board, that can directly be used with bread boards and experimental pcbs is still needed very much. It should be easily available too.

    Why low cost?
    1. I have not yet seen any maker-project at all with P2, that cannot be done with a different controller or linux-board. With 8 cores and the smart pins it might just be less difficult, if you are willing to learn special software first. So the cost has to be attractive even if you don't use all resources of this chip in every project.
    2. While somebody might be willing to buy a single edge board + pin adapters + usb-adapter just to learn this chip, the price might perhaps be OK for this learning set. But if you want to build it into a fixed project, you will come to think, why don't I use a xyz?

    Why and what is minimum of complete?
    Just to make it easy to be used and to start. USB-Serial, SD-card and power supply from an usb-connector is needed.

    Why DIL?
    Because this can be used directly with cheap bread boards or wires and it can be used very easily with experimental PCBs too. The "edge" connector is of no help.

    So in my opinion to get it done, use linear regulators and a crystal and some usb-serial chip, that you know and that has drivers available. If it is helpful to lower the price, than put some gpios onto a special connector, to have a smaller board. Design it, that it can be produced in high volumes for low cost.

    Just my 2 cents. :-) Christof

  • RaymanRayman Posts: 12,099

    @"Christof Eb." Isn't that what P2D2 is?

    Looks DIL, has linear regulators, not too much extra besides the Flash and two uSD (not sure why two...).

  • @"Christof Eb." As much as I would like to see a P2 board that is what you described - I hate to say it but I will - low cost and complete are relative terms.
    Low cost means: low quality components or assembly or both, incomplete or a high volume product. On the other hand complete means additional components, more complicated design, higher assembly cost. Do not forget that you also need to take into account the distribution and, quite possibly, documentation and marketing costs (forget the warranty or replacement) that are associated with the product and that doesn't help either. Fortunately we have a choice of boards right on this forum and most probably more to come. Money wise, I don't see how could I purchase a P2 board for much less than what we have available here even if I totally ignore the cost of my time and tools needed to make such a board.

    But, we have derailed this thread, sorry.

  • jmgjmg Posts: 14,756

    @Rayman said:
    @"Christof Eb." Isn't that what P2D2 is?

    Looks DIL, has linear regulators, not too much extra besides the Flash and two uSD (not sure why two...).

    The P2D2 has 2 SD's facing different directions, so the top one sits internally, over the flash, and the bottom one allows card access from over the board edge (eg a slot in a case)

  • msrobotsmsrobots Posts: 3,419

    the basic idea is to use either or...

  • @"Christof Eb." said:
    Hi,
    in my opinion a low cost (!!) but complete P2 board, that can directly be used with bread boards and experimental pcbs is still needed very much. It should be easily available too.

    Hi Christof,

    the KISS board is available with open source schematic and layout. It shouldn't be too hard to add an SD card slot and a USB connector. It's relatively expensive because I make it in small batches and I'm not good at shipping single boards around the world. If somebody would buy 200+ at once I could make it for $25.

  • PublisonPublison Posts: 11,966

    Looks like @Christof Eb is in Germany. Local shipping for you.

  • Cluso99Cluso99 Posts: 17,839
    edited 2021-07-16 22:14

    And there is my RetroBlade2 board that includes a uSD card slot and single 0.1" pins in each side that plugs into a breadboard.
    https://forums.parallax.com/discussion/172262/clusos-retroblade2-single-chip-computer-small-stock-available-for-new-orders-payment/p1

    Like @ManAtWork I assemble these in low quantities. I hand place but I use a professional reflow oven and then hand test. I spend well over 30 minutes per board assembling/testing. And this is without the extra cost of buying components in bulk and keeping them in the hope I will use all of them. Otherwise that unused inventory has to be amortised over the number of shipped boards. The shipping of these parts to me is significant too. Just because a component costs 1c doesn't mean that is the landed cost. If you buy $200 of parts and it costs $50 to ship it, that means those parts cost +25% assuming you use them all. There are minimum buy quantities and they never balance out, so you always have wasted inventory. Then you pay GST or whatever it's called in your country. Here in OZ it's 10%. Because it's a hobby, you don't get to claim it back when you sell the final product. And so you can see the impact of shipping, now (as a covid excuse) Australia Post discontinued the $3 letter rate so now it costs ~$25 to ship the same packet around the world (ie to you). Thankfully Publison ships around America cheaper, but it still costs a large shipping cost to him, and then we hold the cost of him holding inventory waiting for orders which may never come.

    For a company like Parallax, they have certain cost reductions because of pick-n-place machines, but they cost hundreds of thousands up front, they have staff overheads, they have large premises, etc, etc. Ken once said a product cost $30K just to design the product and I don't doubt this - I ran my own companies since 1976 designing and programming hardware and software so I'm fully aware of these overheads too. And if you go thru a retail chain like DigiKey or Mouser, they want big margins too.

    Unless you can get to critical mass, like the Raspberry Pi, the products are always going to cost way more. And remember, Raspberry Pi pay no taxes and do not run for profit - they are a charity! IMHO it's a scam now because it's unfair business practices against the legitimate companies. But the users don't complain because they get cheap boards, and now expect everyone else to compete.

  • Basically it comes to this (and this is my view only and possibly not the universal truth): if you need something for a professional development of a commercial product then you need to consider all the costs and benefits to make a profit on your final product in the end. In that case, even an unreasonably expensive development board or tool doesn't matter much because you need it to achieve your final goal.
    On the other hand, if it is just for your hobby use then, well, you do not expect to make a profit on your hobby but you just burn money on it. Weather you have that money and the desire to burn it is entirely up to you. Some hobbies are just too expensive for an average person and some people save money for years just to get in to that dream hobby they so desperately want to pursue.

    I think @msrobots summed it up nicely, you either use what is available and a closest match to your needs or...you chose otherwise.

    Maybe you end up making your perfect board yourself eventually, like some of the users here did, from which fact I myself benefited hugely. And believe it or not, I'm not a big money burner just because I do not have that money to burn. I save money for most things I need and the best (my) way of doing so is to NOT BUY things that I don't REALLY NEED or that I can't AFFORD. Works for me.

  • roglohrogloh Posts: 3,392

    Yeah. The problem with targeting a P2 board for hobbyists is that they are typically a very price sensitive group.

    Commercial organizations have no real qualms spending hundreds of dollars on development boards and their test setups while for a hobbyist that can often be rather hard to justify (depending on how keen/wealthy you are).

    It will be interesting to see how low a price a P2 board could ever get to. Tubular and I were talking about this recently and with a BOM cost of say ~$20 for the simpler P2 boards it seems hard to get the price down much lower than a $50 ballpark for low volumes when you factor in all the extra costs and choose to not operate at a loss.

    That same $50 is starting to look pricey for some hobbyists/students etc, and unfortunately these days the mass produced RasPi and other devices on AliExpress, Ebay etc have dropped many people's price expectations of simple microcontroller boards where they already get a lot for their $5-20 (though obviously not a P2).

    This is a real pity because these same hobbyists may one day be involved in purchasing/design decisions at their workplace, or could help promote the P2 via interesting Hackaday projects, or also donate their time for providing useful P2 software.

  • Actually, there is some benefit to the higher entry point price too. It may seem stupid at first but think of it as a safety valve.

    What happens, in most cases I see when I look around, is that if something is dirt cheap you might just snatch it, even in multiple quantities, not because you need it but because it's cheap and you think you'll use it at some point. While it sometimes ends this way, most of the times you end up spending money on the items that fill your drawers and never see an actual use. A waste of time, money and the items you bought and never used.

    If you're smart enough and can see what's coming if you follow that hoarding route, you wake up and start learning to refuse yourself such buying impulses. You end up with some free space in the drawers, no buyer's remorse and some pocket change that you can put towards the purchase of an item that isn't so cheap, or maybe it even is overpriced, but the one you actually need, want and will use.

    When buying something, doesn't really matter what it is exactly, I ask myself these three simple questions:
    1. do I need it ?
    2. do I want it ?
    3. can I afford it ?
    in that order.
    If the answer to any of these questions is a NO, I postpone the purchase. Things might change in the future and I may get all of them answered a YES and THEN I will go ahead with that purchase. It takes discipline, yes, but it works for me.

  • ErNaErNa Posts: 1,620

    Reportedly there are some issues with the P2D2. And it looks like the boards, Tachyon, TAQOZ and Peter are type of amalgam. We together have to join efforts to create success. Peter went to another forum as there is just a bigger echo chamber. He gets more resonance and he knows that the members of the forum will not discuss about the best language but just select the optimal forth vocabulary. That's my perception. As I'm interested in the Propeller and TAQOZ but reluctant to create a facebook account, I just don't know if it would be advantegous to join that group. On the other hand there is a chance to have Peter back here.
    So I ask those interested in TAQOZ to join efforts to create an application the very likely is of interest for everyone. As an initial step I asked ManAtWork to produce this next batch of the KISS boards and as he knows how to select the right components and can pick and place them with his automated machine we should be able to create a base of hardware and people dedicated to the P2 and TAQOZ.
    The proposal I have is like follows: build smart home components and build a smart motor controler. The first comprises a window shutter controller and a smoke detector that doesn't create false alarm every second day. So it needs an new technology. The second will create a sophisticated motor control solution. Of importance is identical hardware. That will be reached as Parallax will offer a simple board that can be used for different motors. I will contribute some code examples based on the P1 and demonstrate some ideas on why it is possible to control a motor sensorlessly.
    So a short "me2!" from those that are willing to contribute.

  • ManAtWorkManAtWork Posts: 1,079
    edited 2021-07-17 13:34

    @Cluso99 said:
    ...And remember, Raspberry Pi pay no taxes and do not run for profit - they are a charity! IMHO it's a scam now because it's unfair business practices against the legitimate companies. But the users don't complain because they get cheap boards, and now expect everyone else to compete.

    Yes, it's the same with printers and ink cartridges. Printer manufacturers sell the printers for less than the production costs and expect to get their expenses back by selling overpriced ink. I also remember that when brand new Altera FPGAs came out the eval boards were "subsidized" to promote the chips and were actually much cheaper than bare ICs even at medium quantities ~100. So we ordered aval boards and soldered out the FPGAs :D

    Parallax could probably sell cheap boards in the hope to make more profit later on chips. But that would mean the chips would cost a little more to all others.

    @ErNa said:
    As an initial step I asked ManAtWork to produce this next batch of the KISS boards and as he knows how to select the right components and can pick and place them with his automated machine we should be able to create a base of hardware and people dedicated to the P2 and TAQOZ.

    Have I missed something? (never mind, forgot to check my mail)

  • @ManAtWork - congratulations for hitting the 1k mark with that post. I know this 1k mark means nothing to you but I value your post highly. But again, not for their quantity :).

    @ErNa - let me share my views here as well.

    Firstly, I do not see any real issues, technically speaking, with the P2D2 board itself. It is an excellent and feature reach board, well thought out and nicely done. The issues I see are manufacturing and distribution of these boards but even these can be overcome if addressed properly. I am sure of that. Bear in mind, I am not referencing any price point because the final price will depend on too mamy factors to be discussed here on the forum. I recall (and that can probably be found here in the posts or maybe elsewhere as well) Peter has stated he doesn't fancy manufacturing these P2D2 bards and that somebody else should do it but nonetheless he committed to do a batch or two.

    Secondly, I think Peter is a very practical and pragmatic man and likes to do his stuff in his own way, asking for advice or comments but making his decisions based on his personal judgement. I do that too and am fine with that. We are all free men.

    And thirdly, I think your project ideas deserve two separate threads (also separated from tis one) where we could discuss the matters, in which threads I will participate with pleasure.

  • Christof Eb.Christof Eb. Posts: 429
    edited 2021-07-20 06:38

    @Cluso99 said:

    .... And remember, Raspberry Pi pay no taxes and do not run for profit - they are a charity! IMHO it's a scam now because it's unfair business practices against the legitimate companies. But the users don't complain because they get cheap boards, and now expect everyone else to compete.

    Hi Cluso, that with low volume it is not possible to directly compete with Raspberry, I totally understand.
    Rpi pico is less powerful and is therefore not a direct competitor for P2. Nevertheless pico shows, what production and selling cost can be for a high volume, small board, clever design with usb-serial/upload and power regulator on the chip. Clever to design it, that it can run directly from flash with a cache, so together with the thumb instruction set less ram is more. Small chip area with these 8 super simple PIO-Processors + "only" 2 full blown cores. I assume, that using only M0+ but giving it a hardware divider as a peripheral helped also to give good power/price relation. Using an existing kernel with existing compiler for existing languages is a free design decision and keeps cost down for software development and for documentation. They really just copied the instruction set of M0+ from ARM documentation into their datasheet.
    So rp2040 and pico was designed from the beginning with power/price relationship for the users with very high priority in mind. Paying no taxes and even running for no profit seems to be only a minor contribution to the low price!

    I think, that the competition by rpi and by others (!) simply is reality and has to be faced, if more volume is wished.

Sign In or Register to comment.