PicoCNC - a grblHAL breakout board. #3
Replies: 23 comments 66 replies
-
|
I have a question for you all. Socketing vs soldering in place. And pin headers vs castellation surface mount. There are 3 possible ways of mounting the Pico. What do people prefer? Pin header and socket. Pin headers soldered to the Pico. Sockets soldered to the base board. Insert Pico into sockets. Kind of the "Arduino" way. Relatively secure. Allows easy replacement of the Pico if it becomes damaged. Pin header and solder Pico Pin headers soldered to the Pico. Pico inserted in base board holes and soldered in place. Is very secure. Pico soldered directly The Pico is directly placed on the base board. Soldered at the castellations on the edge of the Pico. This is very secure. It is fairly easy and actually the fastest way to do it. This option requires the PCB to be designed for it and has some implications on the layout. |
Beta Was this translation helpful? Give feedback.
-
|
The 2 first solutions have the same footprint I believe, so the user is free to choose between header only or header + socket. I personally don't have a preference, the 3 solutions works for me. |
Beta Was this translation helpful? Give feedback.
-
|
There has been a request for a voltage regulator to provide 5V from 12V. Here are my thoughts but I would like to hear from others. Pros
Cons
So, in my mind, this boils down to how common is running without a USB connection? Note: I am looking at the RP2040 Nano which has Wifi and BT, a board for that will very likely need a VR. |
Beta Was this translation helpful? Give feedback.
-
|
An optional Traco TSR-1 regulator could just about fit under the Pico? Expensive but nice, I have used these before. |
Beta Was this translation helpful? Give feedback.
-
|
That is certainly an option. There are a lot of ~10mm tall modules in that SIP formfactor. They range from $2 to $5 in price. A key question - what are the 5V power needs of the board? If you add up all the possible 5V outputs - 17 (19 if including spindle) optos at 6 mA we are looking at max 114 mA plus a bit more for the 5V logic. It is likely to be a fair amount less than that due to duty cycle and other factors, though. Like I said earlier, I really don't recommend driving 5V relays from onboard power - I have contemplated removing the 5V/12V relay jumper and only supporting 12V relays. On the Teensy 4.1 board, I have never seen a non-relay situation where the draw was more the 210 mA in a 5 Axis scenario. The Teensy alone draws 100 mA so the rest of the board was pulling 110 mA. Another key question - What are the non-USB connected scenarios? I do not have a good handle on that. In the USB connected situation, there will be plenty of power available. On a somewhat related note. I am looking at what it would take to surface mount the Pico. That would preclude mounting things under it. There is room for the Traco module elsewhere on the board. But, on the other hand, there is a lot of free space under the Pico that is temping me... |
Beta Was this translation helpful? Give feedback.
-
|
As you said, the 5V voltage regulator is mainly for "USB less" usage, so when a offline controller is used (Terje Io MPG/DRO). Also if you add the SD card header on the board I guess it makes sense to have the 5V VR. I think it is worth and does not increase the price (it takes space though) to add only the footprint for the VR. I also think that maybe the 5V relay and 5V logic should be separated, 5V logic coming from USB or onboard VR and 5V relay coming from the 5V header? |
Beta Was this translation helpful? Give feedback.
-
|
After some deliberation, I decided to add a regulator. The one I chose is rated to 1A but with the copper area I was able add, is good to about 1 Watt of dissipation - 140 mA approximately. I will test it to see what it really does, if I get to 120 mA without it smoking, I'll be happy. In the process, I removed the 5V terminal but kept the relay voltage selector. The 5V relay coil option may go at some point but I will keep it for the first round of boards. |
Beta Was this translation helpful? Give feedback.
-
|
Update on picoCNC. Changes
As a proof of concept, I laid out a small microSD daughter card using a Molex socket. |
Beta Was this translation helpful? Give feedback.
-
|
I think there is a tiny mistake in the silkscreen of the stepper connector, there is missing one and the others are all X axis |
Beta Was this translation helpful? Give feedback.
-
|
Yes, I caught that. I used the wrong screen shot. sigh. Thanks for letting me know. |
Beta Was this translation helpful? Give feedback.
-
|
Boards back from the Fab. Looks pretty good. Quick photo with my phone (i.e. not the best photograph...). Hoping to get my HCT595s soon from Mouser. |
Beta Was this translation helpful? Give feedback.
-
Ok with me.
For now I think not, perhaps something to consider for later revisions based on user feedback?
Again, not for now. Code has to be added for spindle sync and I want it verified by users for the drivers that supports it, currently MSP432, iMXRT1062 and STM32F4xx, before I add it to further drivers
That depends on the target users? Some want it cheap, some are willing to pay for higher end stuff. The RP2040 on its own is very cheap so the total cost for the combo is going to be competitive anyway? Get this design out and see how users react? When I get a board for testing I want to determine the max step rate that can be safely configured. Using PIO code for step pulse generation is a plus but lack of a FPU could be a minus? I could have done it with my test setup but that is no more as it was quite a mess of boards and breadboard wires... If the max rate is good the board could become interesting for laser engraver use? |
Beta Was this translation helpful? Give feedback.
-
|
OK, that sounds good. I am running into a time crunch soon - I get my Avid Pro4848 kit tomorrow and will be spending a fair amount on the new toy. Still need a few pieces for the electronics but mechanical assembly will take some time first. I'll try to not totally starve the Pico effort. |
Beta Was this translation helpful? Give feedback.
-
|
While waiting news for the final tests so to get one PCB... Circuit looks good. If you think there is no problem I can live with that. |
Beta Was this translation helpful? Give feedback.
-
|
In regards to the cheap DC motor controllers and spindles. I have no experience with them but if both grounds are connected, it should be ok. Earth ground shouldn't be an issue and the controller shouldn't put high voltage on the interface connections. Do you have a link to one that you are concerned about? I would like to buy one and test it out. I suspect they all originate from the same factory in china... The pro T41 board that I will be releasing this week has an isolated spindle section. It adds cost and complexity that I am not sure is something the typical Pico based CNC user cares about. But am happy to revisit the issue any time. |
Beta Was this translation helpful? Give feedback.
-
|
This is the controller I use. Circuit made from a guy by reverse engineer it. |
Beta Was this translation helpful? Give feedback.
-
|
Hello, I want to design own CNC control card and software(in the future). I'm planning to use grblHAL with RP2040 but I couldn't find a schematic for rp2040 for connections like shift register, eeprom etc. I think you're designing on kicad, can you post the schematic file or a photo of it? I will use rp2040 fully embedded in my design. Sorry my english :) Thanks for interest |
Beta Was this translation helpful? Give feedback.
-
|
The PicoCNC schematic is linked to in this comment above. The latest pin assignments for the board can be found here. You may also create your own pin mapping if you want to assign pins differently. Be aware that the current PIO code for the shift registers requires the clock pins to be consecutive. There are photos and screen dumps posted above of the prototype version of the PicoCNC board, access to the design files is something you have to ask Phil for. I have made a prototype "motherboard" for my CNC BoosterPack breakout board, the pin mappings for that can be found here. It uses an I2C I/O expander for a few extra pins. |
Beta Was this translation helpful? Give feedback.
-
|
I will be putting the schematic up once the V1 design is finalized. There is one change that is being made and I am hoping to hear if the SR support is solid. I still plan on making the boards available. |
Beta Was this translation helpful? Give feedback.
-
It looks good one the scope and on LEDs, a bit of jitter (< 1us) but that is to be expected as the main stepper routine is interrupt driven. Pulse length is very stable (with less than 100ns jitter? - I have not bothered to determine that exactly) and max step rate for simple moves is likely > 250 kHz. I suspect the step signal will not break down in the same way as other drivers does as the PIO code handles resetting the pulse independely of the processor. More tests with lots of short moves (laser engraving) should be done to determine a safe upper limit? Since yesterday I have run three 10 hour tests via a Bluetooth connection (HC-05) back to back without any issues, I just started the fourth. I'll run a number of similar tests over the USB connection later. |
Beta Was this translation helpful? Give feedback.
-
|
That sounds very promising. A step rate of 200K or better is very good since most stepper drivers spec that as the max rate. V1.2 has the SR load input/keypad strobe pin swap we discussed. I'll give it a couple more days and send it off for a build of 10 or so. Also, I am looking at a design with the RP2040 (rather than the Pico). Probably won't be cheaper than the Pico version but easier to set up. |
Beta Was this translation helpful? Give feedback.
-
|
Is there anything i can do to help? |
Beta Was this translation helpful? Give feedback.
-
|
Looking very good. Thank you for putting your time into it. I agree with having a grblHAL standard. My point was, I looked to see if there was anything close to a standard and found nothing. So, I say your pinout is it. Yes, language matters so we are purging slave. In some cases it even makes more sense. Using slave for ganged motors is imprecise. One is not a slave of the other, they are ganged together. |
Beta Was this translation helpful? Give feedback.
-
|
These developments make me happy. I started to make a simple layout design, but I don't understand some places because I don't experience for CNC Boards. If you've drawn a new schematic, please post it. There are a few issues on my mind, is it suitable for box designs to use connectors in the USB part of your design? My other question is, why do we use shift register for pins that require speed, wouldn't it be better to use it in the slower relay part? Again, I am sorry for my English. Also, is there an application you use for communication? It's a little difficult to communicate from here. Thank you :) |
Beta Was this translation helpful? Give feedback.
-
|
As mention several days ago, the schematic has been posted in this thread though there a couple of small changes. I will post a final schematic when I go for a build. USB is what it is. There are different types of connectors (USB A, for example) that are a little more reliable and I will consider that for the RP2040 board version. My Teensy 4.1 breakout board uses the same microUSB connector (well, same format) and I've sold around 550 boards now with zero complaints about that. (actually, almost no complaints at all.) If you protect the breakout board in a cabinet, microUSB does not cause any problems. The problem with the RP2040 and Pico is that there just aren't enough pins. There are several ways to get around this: I2C I/O expanders and shift registers have been discussed at length in this thread. There is a plugin and a board that uses an I2C I/O expander. One of the coolest features of the RP2040 is PIO and that has allowed an amazing implementation using shift registers. Terje's post of a few days ago shows it works really well and can generate fast enough signals with minimal jitter. There is one single 8-bit SR used for step and dir. And there is a 16-bit 2 SR chain for enables, relays and several low frequency spindle signals. We did look at using RP2040 pins for step and dir but it didn't help that much. Also, adding the SPI/SDCard interface would not have been possible. The nice thing about the SR design is you can increase the chain to add features. Finally, SRs are dirt cheap. Using an I/O expander that costs more than the RP2040 seems out of balance. |
Beta Was this translation helpful? Give feedback.
-
|
For the machines I have designed I have brougth out the USB port to a USB-B connector, typically via a short cable to a small PCB (USB-A on the side facing inwards to USB_B facing outwards). USB-B connectors are IMO better suited for bringing out to a panel than any other USB connectors.
The RP2040 has the PIO peripherals ("primitive" processors in a way) that allows shifting data to out to the shift registers without any processor intervention and can reset the step pulses after the appropriate delay - again without processor intervention. Step pulse resolution is sub microsecond and pulse length very stable (close to no jitter). The PicoCNC board is currently the only one using shift registers. Step and dir pulses are generated via a 8 bit shift register, stepper enable etc. via a slower 16 bit shift register (two chained 8-bit registers). The generic board map and the CNC BoosterPack uses a simpler PIO program for outputting pulses directly to GPIO pins, but this leaves less pins for other signals. The CNC BoosterPack overcomes this by having an I2C I/O expander for the slow signals. There are therefore several alternatives to using the shift registers - but IMO they are a very cost-effective way of adding a lot of output pins both in terms of monetary cost and processor cycle cost.
I have occasionally used Skype or email, but I prefer to keep communication in the open on github. Phil is going to publish an updated schematic when completed so be patient. |
Beta Was this translation helpful? Give feedback.
-
I don't know the PIO feature on the RP2040 because I know little about microcontroller theory, I will investigate. I couldn't fully explain the problem due to my poor English, I think it would be better to explain it with a photo. If you use a terminal (power input, etc.) on the side with USB, having connectors in this side will be a problem for box designs. Thanks for reply :) |
Beta Was this translation helpful? Give feedback.
-
|
I understand your point now. But generally people do not put the board up next to the wall of their box (enclosure). I use something like this picture below. It allows freedom to place the board where ever you want and allows you to put the connector in a more convenient location for plugging your sender PC in. |
Beta Was this translation helpful? Give feedback.
-
|
Update on PicoCNC. I have been doing a bit of hand wringing over costs. The original plan was to use pluggable screw terminals but these will about double the end user cost. Given that it is targeted at a lower end user, I decided that a lower cost approach is in order. So, I have redesigned it to use the more common 3.5mm screw terminals. In addition, I switched to a different, though similar, opto isolator. It is smaller and has a somewhat higher current transfer ratio (better bin select fwiw). I am building a run of 10 boards - V 1.30. It should be done in about a week and will announce availability here. I will distribute to anyone who has contributed to the software effort here. Please reply with an email account for me to contact so I can get your address. (obfuscate your address to avoid email scrapers). These are "unkits" - includes most surface mount components (no eeprom, though) and the trimmer resistor. I will also include a screw terminal kit. If I have any left over after that I will put them on tindie for $10USD plus shipping (kits will be extra). Product price will be higher than that by the way - I am thinking of around $25USD Here's a view of the base board - I don't have models for a number of the components so I left them off. |
Beta Was this translation helpful? Give feedback.
-
@phil-barrett I am working with the KiCAD nightly build version (frozen KiCAD6 in reality) a year now. Extremely robust. Go for it. No crashes at all (I have plenty of crashes on my Ubuntu 21.10 I just upgraded and feel very sorry). KiCAD6 must be out till the end of the year as they say. But go for KiCAD nightly build version. Because it is frozen it will be the same as KICAD6 |
Beta Was this translation helpful? Give feedback.
-
|
Wow, a lot to address here. It may take several posts to fully respond. Cheap. Sure, cheap just to be cheap isn't good. But too often, solutions are far more expensive than they need to be. Here is the reason why I see the need for a low cost controller - a lot of people shop on price. For they first CNC, they buy a $5 breakout board to use with an arduino and have a really poor experience. A high quality and inexpensive drop-in replacement meets a need. That is the goal with the picoCNC. Isolation. I look at this from the CNC perspective. A PLC has a different set of priorities. For CNC input, isolation is pretty important, especially for devices (switches, sensors) that are outside the electronics case as they can be an EMI path. And, input isolation for low speed signals is easy and inexpensive. For output, you have to look at each signal to determine if isolation is needed. For the stepper/servo signals, isolation is not needed because the drivers usually have isolation built in. And, isolating high speed signals is expensive so you do it only when really needed. For relay and similar outputs EMI is not really an issue for hobby style machines. Spindle signals are worthwhile isolating because of the prevalence of noisy VFDs in the CNC environment. For communication, isolation can be very worthwhile. That's why I put ethernet on the Teensy based products - easy isolation. I've looked at isolated USB but it pushes the cost a fair amount. For async, I've left it up to the adapter - like Modbus/RS485. There are lots of easy solutions there and the cost is added only when it is actually needed - no need to burden the base board cost. I2C - there are solutions but is it really needed? This is a case where the add on feature should provide isolation. In the case of the I2C ADC board I made, isolation is in the add-on board. I could go on... Modularity. What's not to like? Flexibility is good. Reusability is good. But (there's always a but), it is not free. You need to use good quality connectors and that is more expensive than silicon. And any time you add a connector, you add a reliability point. All other factors being equal, a modular solution will always have lower reliability than a monolithic solution. So, I pick and chose carefully. Each board has an overhead cost so a modular solution will cost more than a monolithic one. And, it adds manufacturing, supply chain, fulfillment, ... complexity. More balls to juggle. |
Beta Was this translation helpful? Give feedback.
-
I will wait until it is out and in the hands of lots of people before switching. I don't want to juggle 2 different versions and am not willing to take the risk of v6 since you can't move a project back to v5. Don't get me wrong, I see a lot of great things about v6 but v5 isn't broken. Patience is a virtue. |
Beta Was this translation helpful? Give feedback.
-
@phil-barrett this link will make you happy. |
Beta Was this translation helpful? Give feedback.
-
@phil-barrett appreciate for your time spend answering in detail and all your comments are taking into account. terjeio/CNC_Breakout_Nucleo64#3 (comment) I think something really nice will come out of it and everybody will benefit (especially the community) |
Beta Was this translation helpful? Give feedback.
-
|
A quick update. I received the boards but have not had a chance to do anything with them. I am about to leave for a 4 week holiday in Africa so will be testing and sending out in September. Sorry for the long delays. Hopefully it will go quickly once I return. |
Beta Was this translation helpful? Give feedback.
-
|
@phil-barrett did you actually populate the PCB? How it goes? |
Beta Was this translation helpful? Give feedback.
-
|
I have a prototype that I use for driver testing. |
Beta Was this translation helpful? Give feedback.
-
|
@terjeio Thanks. I will look for it now. My idea is to design both on NUCLEO64 concept (one version based on your design is already on the road for printing as you know) and one using the Raspberry Pico concept based on @phil-barrett design. This guy here made a similar NUCLEO64 REPRAP design as well |
Beta Was this translation helpful? Give feedback.
-
This is about the (Trinamic) stepper drivers? They are still Step/Dir/Enable driven but configurable either by SPI or UART. I guess both interfaces/protocols will be available for quite a while into the future. The board designer has to decide which to support. The shift registers are for supplying the Step/Dir/Enable signals.
For generic use? I have added a HAL call for that, now used by the keypad plugin. A new symbol |
Beta Was this translation helpful? Give feedback.
-
|
I am studding the shift registers for Step/Dir/Enable signals. In general I am preparing a paper on what the two boards have in common and what is not common, thus to include everything on a COMMON M10CUBE interface (I/O) board. |
Beta Was this translation helpful? Give feedback.
-
|
London News talks about @phil-barrett GRBLHAL and Raspberry Pico. Congartulations. |
Beta Was this translation helpful? Give feedback.
-
|
I believe the original article is here... |
Beta Was this translation helpful? Give feedback.
-
|
Yes you are right. I am getting this newsletter but I miss it. His note on "...is that rather than having a single resistor in series with the LED of an opto-coupler for an isolated 12V or 5V input, I would split the resistor in two equal halves..." is interesting and I will give it a thought on the next design. In M10CUBE INPUT module is like the image below. LED is on a 2 resistors voltage divider. Already working with an UNO GRBL shield with no problem. |
Beta Was this translation helpful? Give feedback.
-
|
Where is this project currently? |
Beta Was this translation helpful? Give feedback.
-
|
Hi @Deneteus . At the moment I migrate the original design NUCLEO 64 design from @terjeio and here its: terjeio/CNC_Breakout_Nucleo64#3 (comment). This design is waiting in JLCPCB to finish some M10CUBE I/O modules . That is nearly finished and next week all will go for production. |
Beta Was this translation helpful? Give feedback.
-
|
One more note on the M10CUBE GRBLHAL adaptation. The idea is to free main (CPU) board from any I/Os. Next M10CUBE I/O modules will include practically anything. Moto is : "What is needed we will design it". New boards are already for fabrication. M10CUBE GRBLHAL CPU module will contain only The Step Motor Drivers and in future encoder inputs. This way we can have at least six axis on board. Finally M10CUBE concept is not for "one board Funs" but for a broad range of people whiling to "sacrifice" some space and build a really "ever expandable system". Reusable too for any automation work you can imagine. The actual idea behind M10CUBE |
Beta Was this translation helpful? Give feedback.
-
|
Wouldn't it make sense to create a new topics for the M10Cube Pico module ? |
Beta Was this translation helpful? Give feedback.
-
|
Well @Volsolive. It is a nice recommendation but I do not want to hijack the space here. I can do that if @phil-barrett permits so since this is his space. |
Beta Was this translation helpful? Give feedback.
-
|
This is only one conversation in the RP2040 repo of the GRBLHal git. It is a conversation Phil barrett started but it is not his Git Repo for the PicoCNC board, not sure he did one for it yet?! (as he did for the Teensy BOB). So I guess it would be better to start of a new conversation for the M10 cube as another board that runs GRBLHal-RP2040 or even better start a Git for the board if you want to share the design files. |
Beta Was this translation helpful? Give feedback.
-
|
Ok Volksolive. Without wasting time and following your directions I started a discussion here and see how it goes. M10CUBE PICO CNC. A Universal Controller ? I hope people put their ideas there and something nice will come out for the community. I am sure everyone can find the discussion interesting, educational and practical since new technologies will applied, constructed and try to work. |
Beta Was this translation helpful? Give feedback.
-
|
This is some work done so far. |
Beta Was this translation helpful? Give feedback.
-
|
To every one interested to "do business" with Raspberry Pico and CNC control. https://hackaday.io/project/171770-m10cube/log/202098-m10cube-pico-cnc As I state there work will be based on excellent designs found here: I am not in a hurry because it is a complicated idea for everything to work nicely inside M10CUBE platform. |
Beta Was this translation helpful? Give feedback.
-
|
A new PICO CN board discussion started here: |
Beta Was this translation helpful? Give feedback.
-
|
@phil-barrett |
Beta Was this translation helpful? Give feedback.
-
|
@timgow I do not know where Phil's design stands today but we took @phil-barrett + @terjeio idea discused it here: I do not know if that is any help to you. I am planning to do mine after summer, but the project is alive and we need some more time to finish it since it is not one board but many. |
Beta Was this translation helpful? Give feedback.
-
|
@phil-barrett Would also be curious about this! Perhaps consider just releasing the design files if your project has ground to a halt? |
Beta Was this translation helpful? Give feedback.
-
any progress? Its now 2023 and I cant find the software and/or a place to buy it... :( |
Beta Was this translation helpful? Give feedback.
-
|
Yes, wonder of wonders, I am back. PicoCNC is alive. Perhaps even better news, I have placed an initial order of boards and will get them in perhaps a week. I also have ordered parts for the assembly kit so it will be a race to see which gets delivered first. I will be selling it on Tindie to start with. What is exciting to me is that the low cost Pico provides very strong competition to Arduino+CNC Shields. Here is the first installment of the user manual. I would love to hear any feedback on the design and especially the manual. It is a bit rough and could very much use some sharp eyes. Don't spare my feelings! Phil |
Beta Was this translation helpful? Give feedback.
-
|
Hey Phil, cool stuff! User manual and schematics are not accessible though, private repo? |
Beta Was this translation helpful? Give feedback.
-
|
Sorry will fix in an hour or so. I'm away from the house. |
Beta Was this translation helpful? Give feedback.
-
|
Fixed. I keep forgetting that github defaults to private. [edit] I don't think the default IS private. 2 other repos I created were public. I'm not exactly sure how the PicoCNC one was set to private. Mystery. |
Beta Was this translation helpful? Give feedback.
-
|
Great! Just wondering if the kicad files are available, and if you're willing to share? I had a great resource in the M10-cube files, allowing independent production of a PCB, allowing to iron out kinks..etc. |
Beta Was this translation helpful? Give feedback.
-
|
I am sharing the schematic on the github site and will put the KiCAD schema up once I get my first batch back and verify them. Not sure I will share the PCB files though. FWIW, I already have a couple of changes to V1.5 - added 2 additional LED indicators. One for 5V which allows you to know if powering from USB or 12V and a spindle on indicator. On the Teensy, having an LED come on when spindle is running is a very useful diagnostic and V1.51 did not have that. |
Beta Was this translation helpful? Give feedback.
-
|
A few more photos. |
Beta Was this translation helpful? Give feedback.
-
@wakass if @phil-barrett produced his design or give us PCB files we could make it on a M10CUBE format. He can do it as well and still commercialize it but in M10CUBE format. That way he can sell more boards since many are using M10CUBE I/O (controller) boards (now preparing one wtih CM4 and an LTE chip) |
Beta Was this translation helpful? Give feedback.
-
Getting close to having some PCBs made. I will probably just get 5 bare boards to start and then do a batch of "Unkit" boards (all SMDs) once all the major issues are resolved. Features:
Here's an image of the latest board.
Note that it is a 2 layer board but most of the traces are on the top to make for a solid ground plane.
Beta Was this translation helpful? Give feedback.
All reactions