We need a different sort of 'RAMPS' boards !

This forum seemed somewhat deserted. But I'm posting this here as it seemed relevant still and I'm hoping to get some mindshare about this:

These days there are large number of 32 bit 'development boards' e.g. mostly ARM based STM32 and various NXP and some from Atmel. TI, and not to miss out Raspberry Pico etc.
You could find examples of stm32 boards here
[stm32-base.org]

You can find various of these boards from say marketplaces like Aliexpress, Amazon, Ebay etc. Then there are others like those (micro)python boards running STM32, ST itself offers various 'Nucleos' and 'Discovery' boards. Offerings from Olimex [www.olimex.com], [www.olimex.com], maybe others from Adafruit, Sparkfun etc

The good thing about them is that most of them are Arm Cortex-M based microcontrollers and there are already a large base of existing support for those mcus e.g. in Marlin etc.

One common thing about all these 'development boards' is most (about all) of them *do not share a same pinouts nor physical layouts* (even for the same MCU).
This pretty much invalidates the concept of 'shields', and the replacing concept is *connecting cables* (e.g. 'dupont' cables)

Are there any efforts to make a 'RAMPS' board, which pretty much sports the same motor control and other peripherals e.g. mosfets, temperature sensor connectors, endstops connectors, maybe these days things that could be added could include:
1) on board power supply - preferably a relatively well powered bucks converter such as LM2596 [www.ti.com]
this on board power supply would need to provide power to the 'controller board' as well as drive other power hungry pheriperials such as a servo motor for bed tramming.
possibly needing more than 1 voltage levels as well, e.g. both 3.3v and 5v. e.g place a 'ldo' between 5v and 3.3v rail.
3) interfacing for 3.3v to 5v (or some other (higher) voltages) - this is a pain with 3.3v microcontrollers really. Nearly all of them requires a level shifter (e.g. those 74HCTxxx as level shifters) to drive 5v logic level mosfets. this prossibly needs to be better thought out as well as we'd need to be able to better adjust the Vgs operating points for the mosfets. the resistances of pcb 'heated beds' are pretty much unpredictable these days.
4) possibly add other peripherals on board that has become popular with the 'standard model' of 3d printers. e.g. servo drivers and sensors for bed tramming (auto levelling), probes such as filament run-out sensors, filament change support, fans, leds, etc.
5) one common thinking is that these days 3d printers should after all be *enclosed*, not so much for the prints, but that 3d printing has *emissions*
[www.ncbi.nlm.nih.gov]
so it is good to think in terms of support for an enclosed environment, such as those 'tent' style enclosures. that is where led (it is often dark in the enclosure) and possibly additional temperature sensors may become relevant as well.

6) and most important, it is no longer shields but that most likely the interfacing connectors are 'dupont' pins, and connectivity between the 'RAMPS' board and 'controller' boards would be cables such as 'dupont' wires. it seemed it has become rather common that on boards it is 'male' pins, and all connecting cables using 'female' pins and shells.

RAMPS has pretty much become a 'standard model' of a 3d printer at least a 'virtual' one.

the idea is to have a *generic* board of these most commonly used peripherals on its own board so that this 'RAMPS' board can be connected to *any* development boards.

Edited 15 time(s). Last edit at 11/26/2021 09:23AM by ag123.

Your information is not accurate

Ramps is a dead form factor. Most are using modern 32bit all in one boards designed to be controller boards from the ground up.

DuPont is also not the standard any more, dropped in favor of latching connectors that don't fall off
See [www.youtube.com]

Edited 1 time(s). Last edit at 11/26/2021 08:31AM by Dust.

It isn't too much about 'form factor' of RAMPS, the mega pins are pretty 'useless' to most other boards as they won't fit.
In terms of size it is 'about right', but I'd think putting more stuff e.g. onboard dc-dc buck converter would result in a larger board.

in terms of 'dupont' connectors, I'd think there are many better ones out there. I'd guess even JST connectors are likely slightly better than 'dupont' connectors.
But that today, it is still a cheapest way to loose connect things, and they are about the only pins found on low cost development boards, but that latching or some other connectors could be a welcome change.

all-in-one boards are ok, in fact many are good. Just that there are still many 'loose' development boards out there that is easy to purchase but won't fit the notion of a '3d printer' board.
But the idea of placing the various peripheral electronics i.e. the motor controllers, mosfets, the additional resistors and caps for the thermistors as well as any other possible servo drivers etc on its own board still has merits as those things won't typically be found on 'development boards'.

the idea is to have a *generic* board of these most commonly used peripherals on its own board so that this 'RAMPS' board can be connected to *any* development boards.

Edited 4 time(s). Last edit at 11/26/2021 09:11AM by ag123.

... I've used RADDS boards, where the 32Bit ArduinoDue were used as controller ... and later my company developed an ArduinoDue-"shield" with level-changers for 5V to the I/O's and +5V to +10V ADC-levels, with a 40pin-cable to fit in our own "industrial boards", which werd driven with 24V and all I/O-pins opto-isolated.

For this type of boards we developed a "standard" for the 40pin-cable, so the 24V-boards will run with any other µC-shield, providing same 40pin-layout and 5V I/O-levels.

This was "proprietary" only for our own brands then (now I've changed my dayjob) ... but this sort of "norm" could be made for any other layout --- actually there is the "GPIO-pinout" of the Raspberry, giving something similar -- I have the same RasPi sockets/layouts on Jetson Nano and Xavier NX boards from NVIDIA too

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Hi guys.
I see your concern about the need of moving forward to a new form factor. Reading @ag123 requirements I grub the opportunity to present M10CUBE concept
Having speedup by the many form factors specially from China I decided to put my 40 years of experience in computers , industrial automation, and making into something that will make sense.
What is it?
The short story:
- A form factor PCB 90x90mm (but can extend up to 100x100 in some places on the three corners to accommodate connectors in face with the container in needed)
- 40 pin Raspberry connector on board on the "bottom" of the board
That is all. You use as many Raspberry pins you like according to your design

The result is modular construction. Every new design can be shared with the community. For instance on the following link
[github.com]
The maker @terjeio (involved on GRBLHAL) have constructed a CNC controller using NUCLEO64 board (a very flexible idea) . Now what I did after asking him I copied his design into M10CUBE platform . This 4 layers board costs 8 euros for 5 pieces!
The benefit is that one can use all I/O boards already designed. New maker boards will be available to the community as well. I can have Raspberry on board and many more. Not to mention the large real estate area to build more axis (after removing some I/Os not needed because that is already done by m10cube I/Os)
ST NUCLEO64 boards are very powerful, lots to choose from and most important dirty cheep (10 Euros! in Mouser). That Half the price than some "blackpill" ST F446 designs.

A Raspberry Pico design is also under consideration. First experimental is here:
[hackaday.io]

These designs can be easily converted for REPRAP. Can be six axes but with clever design can be 12 axis by connecting to boards (only one with the CPU populated). Can be done

The long story is here:
[gitlab.com]
[gitlab.com]
[hackaday.io]

The good new are that Dr Dalmaris (the writer of best seller "Learn KiCAD like a pro" ) found the M10CUBE idea very interesting and wants to make a podcast about it. That is planned for next year when many new designs will be added.
It is an idea and I like you to give me a feedback on that. Any help will be appreciated. I you are a PCB designer it will be a grate help to make beautiful designs immediately available to trhe community
And how know? we may doing something really valuable for the community.
Thanks

---

M10CUBE
[gitlab.com]
[hackaday.io]

thanks m10 cube is a nice concept!
A thing here is that 'ramps' style electronics is a little more specialized.
It creates a stereotype / model of the 'driver' stages of a 'cartesian' (actually delta and maybe more printers runs on it too ) 3d printer driver stages.
the common components these days are:
- stepper drivers X, Y, Z, extruder (maybe at least 2), that 'spare' extruder controller can be used for other purposes if it isn't used that way.
- mosfet drivers mainly the hotend and heated bed
(maybe add smaller mosfets to run the fan and leds in an enclosure)
this mosfet driver stages need to be rethought out. present hotend and heated bed mosfets are based around 5v logic level mosfets.
it is a stumbling block for 3.3v mcus e.g. stm32, nxp and ti 3.3v arm variants etc to drive the mosfets.
there are 3v logic level mosfets, but not ideal because - some of the non-logic level mosfets offer lower resistances and better costs-performance ratio.
the other thing is we need to be able to better control the vgs on voltages of the mosfets, some mosfet drivers are capable of say using 12v drive levels.
but if that mosfet driver is indeed used from 12v, we may be turning on that 5v logic level mosfets for current levels like 100 amps rather than say 10 amps.
In practice other parts of the circuit will be limiting the power available, but still being able to tune this vgs is important.
(note, 5v logic level mosfets can still be used, but a 'mosfet driver' is needed for 3.3v devices to drive it.
that 'cheap mosfet driver' is actually any of the 74HCTxxx series ics 74HC*T* as the HCTs convert from CMOS levels to TTL levels.
a convenient chip is the 74HCT245
this chip seemed to be widely stocked and deemed a common component and it seemed easy to buy them on the likes of ebay, aliexpress etc.
However, it may not be necessarily cost effective for those buying in bulk.)

- on board dc-dc power supply - sorely missing, these days both 5v and 3.3v is needed, 3.3v can probably be derived from an LDO from 5v.

note: this on board dc-dc 5v power and 3.3v power powers the mcu controller board, so pins for power rails are needed the power would be routed
outbound to the mcu board (which could have additional stuff connected like lcds and sd cards). some 'spare' power rail pins are nevertheless useful,
as this board now adds 'power supply' functionality.

5v is most likely from a buck convertor converting down from 12v. 'cheap' 3d printers these days provide only a single 12v power supply.
a buck convertor that say provides 3 amps is necessary as this 5v dc will be used to run servo motors for bed tramming etc, and possibly run
the case fans to cool down all that motor drivers and mosfets, there are 12v fans though, just that the 'cheaper' and smaller ones tend to be run off 5v or sort of,
this 5v can also be used to run rather high power leds that these days can draw like 1 amps (but for lighting up the enclosure, hopefully less is needed).
- thermistors interfaces (resistor (dividers) and caps) - well it is a FDM 3d printer unfortunately
- endstops, (limit switches) interfaces this is rather common so much that the 3-pin layout is nearly a 'standard'
- other things that seemed to creep in as 'common' these days seemed to be
- additional pins / ports to support bed tramming (auto levelling) - i'd think it'd at least involve power to drive a servo motor and its controller,
then pwm pins to the mcu, then add a sensor a simplest being a limit switch which'd also go back to the mcu.
(for bed tramming, i'm thinking of a hack for my own printer. simply have a 3d printed contraption that hold a limit switch near the nozzle.
this is needed only during tramming/auto levelling. it is (manually) removed after levelling. this does away with the servo.
But i'd guess these days things like bl-touch [www.antclabs.com] are more popular)
- enclosure support: leds and additional thermistors (temperature sensors) interfaces, and maybe additional fans
This is because an enclosure is becoming more common for 3d printers, not so much to make prints better (it helps), but to control emissions
[www.tandfonline.com]
[www.ncbi.nlm.nih.gov]
leds would likely go to a mosfet driven say from 5v or 12v, then a pin to the mcu (needs level conversion for mcu pin io to vgs)
'tent' style enclosures are becoming more popular, i'd think we'd see more of such being mass purchased.
[www.youtube.com]
(i found an accidental benefit of using (tent-style) enclosures, lower bed temperatures is possible, this saves an *enormous* amount of power.
the heated bed is a power hungry monster. i'm just short of printing ABS *in the cold*, thought to be 'impossible')
- then maybe add things like filament run-out sensor and maybe extra stuff to support a filament change procedure.

the m10 cube modular concept is relevant, but it is actually a 'standardization' of this '3d printer motor and other driver ('ramps') board', most likely some form of 1-2 rows of pin header connectors (a 'bus'). the electrical interface probably need to consider both 3.3v and 5v designs. i.e. level shifting circuits especially for the mosfets is likely needed.

Edited 16 time(s). Last edit at 12/05/2021 02:13PM by ag123.

Thanks @ag123 for responding that fast. Reading your reply we agree that both a REPRAP and CNC controller have a lot of common ground
- Step motor controllers
- I/O control

M10CUBE has exactly what we need. At the moment already working as far as I/O concern:
- INPUT OUTPUT PCB: That is 8 X Optically isolated Inputs + 8 X 1 BSP75NTA FET transistor outputs. Can be higher on a special design
- OUTPUT 8X 16 A relay. New design will have 220V 16A Triac SSR (S216SO2) as well.
- As you see there is also a PSU PCB so modules can powerd from 24V.

There is no problem to design M10CUBE analog CARD as well as one board the can be "MULTIFUNCTION" PCB with Analog in, Output FET 1A and power FET 20A, IINPUTS isolated, encoder input etc...

The idea is One PCB for the CPU and drivers and I/O on a separate card, M10CUBE PCBs can be shared on a RASPI 4 or ZERO as well.
I will look closely tomorrow your notes on what a REPRAP controller needs and contact again.
My idea off making a common controller for CNC and REPRAD makes sense and I am sure can be done!
A few minutes ago @terjeio on this discussion gave me some directions as well :
[github.com]
Hope some people will involve on making this a common controller for 3D printing and CNC but with "modular" and re usability in mind
You see I am trying to "downscale" my experience in 40 years in industrial and automation control. In the end I learned that any controller needs various types of I/O interface. One way or another to make automation everything is a PLC. Programmable Logic Controller. So soon or later you will need it in one form or another.

If more people will involve in M10CUBE we will end up with a better and clever design for everybody to enjoy

Edited 1 time(s). Last edit at 12/05/2021 01:04PM by m10cube.

hi m10cube, thanks for the quick response
well, not necessarily 'all sorts of io', mainly the 'most common' 3d printing drivers parts. the core driving chain mainly.
things like (TFT/OLED) lcd goes direct and independently to the mcu board, so do sd card etc.
Those don't make sense to be on this '3d printing (motor/mosfet) driver sub-system' board, as it'd just add more wires
other 'accessories' can also be independently directly connected to the mcu board.

note that another difference vs cnc router is :
- the mosfets on 3d printer board driving the hotend and especially the *heated bed* generates *a lot of heat*, especially the cheap ones.
i think the heated bed mostly consume at least 100 watts, e.g. 12v x 10 amps. could be higher.
cooling for the mosfets and (stepper) motor drivers has been a lingering concern for a very long time. hence having things on one board possibly helps with the
enclosure cooling design as well, a fan can blow directly at this board, and possibly heatsinks on the mosfets and motor drivers as well.

Edited 7 time(s). Last edit at 12/05/2021 01:57PM by ag123.

Thanks @ag123.
I am starting a collection of valuable info now and a helping hand from any available maker.
Design is in KiCAD6 of course.
Feedback from the community is number one factor for the success or future of any idea before it goes to drawing board.
Failures and bad design produce more CO2 all the time...

Edited 1 time(s). Last edit at 12/06/2021 11:05AM by m10cube.