Reprap Electronics Devolpment

Just though I would start a post on the electronics and the possible future I feel that the electronics now are not 100% Reprap and better fit Makerbot.

In doing this I would hope that we might look into adapting them for Reprap (Mendel).

the Motherboard I think needs to be changed to except 12 volts and go away from the PC power supply header and maybe add a USB connection?

The Stepper driver is good (even though hard to get at times) but the electronics can use almost any stepper driver as long as it is compatible with the steppers them selfs the only other issue is the end stops.

The extruder controller works I don't know if we want to change it as it works with both a motor extruder and a stepper extruder.

an other thought might be availability of parts do we go back to more % of thru hole components...

Bruce Wattendorf


I just thought that this would be a start anybody with any ideas please post.

FOR EXAMPLE **the extruder controller does not work well with Nema 17 steppers..**

At it's core, the makerbot electronics are just Arduino with daughterboards.

I think RepRap electronics should move forward with an Arduino based platform because (1) others actively develop the electronics, (2) we should see additional CPU and hardware variety compatible with Arduino both in software and in hardware (differing only in capability, memory, etc.), and (3) the two projects using Open Source allows transfering and leveraging ideas to/from other non-RepRap applications.

So, the 'official' RepRap, I believe, should simply document the number of arduino compatible boards required, (I.E, Makerbot uses two), and the electronics schematics and PCB designs for the daughterboards (possibly having configurable jumpers to change IO port numbers for specific functions.) It would also seem good to create prints for specific PCB designs for the daughterboards required that would just plug into an existing standard arduino pin compatible controller.

This approach would allow leveraging some of the other arduino compatible devices (for example, the arduino compatible 72MHz ARM Cortex-M3, if and when it becomes a reality..) without redesigning the entire motherboard and extruder board.

What problems or disadvantage do you see with this kind of approach?

Good point it is to bad that the Sanguino was not made to have the "Arduino" foot print like the Megadruino did which is compatable with older daughterboards..

the only issue i see is with mounting the stack of boards on the printer..

Personally I think there is actually a need to move away from Arduino or at least the AVR 8bit core.

I respect its an easy stepping stone for most to get into electronics/software, but the capabilities of the AVR 8-bit core are limited and the larger devices which are being approached such as the mega (ATmega1280) in order to attempt to compensate for these inadequacies are starting to show.

The ATMega1280 is ridiculously expensive, about £20 in single unit quantities!!!!

I also respect the fact that the core VHDL is open source and having this available at some distant point in the future may give us the capability for some future generation of the machine to create it's own microcontrollers. However I believe this to be a long way off and choosing a particular architecture based on this is naive.

There are a large number of other options for much less cash, the most widely used microcontroller in the world are the ARM variants.

There is a GNU GCC compiler available and a wide selection of support tools and chip manufacturers.

The best and most cost effective core available for this task is the Cortex M3.

With my personal preference being the new NXP LPC1768, 100Mhz Cortex M3,512k Flash, 64k SRAM, Ethernet MAC, USB Phy, 2 SPI Busses, I2C/S bus, 2 CANBus channels, 4UART's, 12bit ADC's and loads of General I/O in a 100pin device.

They can be obtained for around £7 in single unit quantities.

They support ISP (In System Programming) from the factory and In Application Programming, when integrated into the application software.

The ISP means there is no messing around with ICSP programmers to get the bootloader in so you can use the development environment. Which regularly sees posts on the forum for instructions on how to do this.

As a professional software engineer I find the Arduino environment clumsy and lacking, although its simplicity has gained supporters on here, porting this to the Cortex M3 is entirely feasable and not that difficult.

I do have more to say but I will leave this for discussion now, I look forward to your comments!

I am all for that since it is 32 bit. I just do not know about anybody else?


It would be cool to go with a processor that has room to grow and have enough power to do every thing and then some.
I don't know how much resistance we would get.

here is a good start to prototype with but i can see it would be cheaper to make a board with the chip on it.

[mbed.org]

Not a bad thought....

I made my own environment for programming my arduino- I've literally never seen what the arduino software looks like. I just installed avr-gcc, wrote a makefile and have been punching out C in my favourite text editor (kate) since then.

I agree that porting the firmware to something like the cortex M3 is in our future, but I'm gonna try and make it work on my atmega168-based arduino first

I have a ts-7250 waiting for something to do if the arduino isn't up to the task...


As for stepper motor drivers, almost all drivers will accept step/dir inputs, so I dodged the hard-to-get reprap ones and obtained some totally different ones from pololu. Currently I need to take a photo of them with video card cooler strapped on top.

Bruce Wattendorf Wrote:
-------------------------------------------------------
> I am all for that since it is 32 bit. I just do
> not know about anybody else?
>
>
> It would be cool to go with a processor that has
> room to grow and have enough power to do every
> thing and then some.
> I don't know how much resistance we would get.
>
> here is a good start to prototype with but i can
> see it would be cheaper to make a board with the
> chip on it.
>
> [mbed.org]
>
> Not a bad thought....

I have an mbed, I had been looking at using the LPC1768 or one of its variants for reprap for a while, it looked promising at the initial release price level, but the production units are £100, which is too expensive imo. (I paid £35 delivered for mine!)

The cloud dev environment is very useful if limited, but it does give you access to a full Keil compiler and developed libraries.

Secondly the community is quite active.

I think it would be useful for the developers until someone was able to produce a board.

Hitex do a USB plug device for about £55.

Keil/Embedded Artists and Code Red do conventional dev boards. but their in the £200 region.

>> Personally I think there is actually a need to move away from Arduino or at least the AVR 8bit core.

I disagree. I still feel the officially documented electronics should include "simple to build", "easy to find a place to purchase one", and "inexpensive".

If you want to use a 32 bit Cortex M3, just make sure you design the board with an interface compatible with the Arduino or two (To open up more IO ports if they are available). Since Arduino development targets 'C', there shouldn't be too many problems converting, consisting mostly of word size assumptions and/or IO library -- creating an Arduino compatible library seems perfect for this kind of porting task.

If development continues down the Arduino path, I believe there will exist a higher probability that other projects external to RepRap will contribute, either directly, or possibly unknowingly -- someone making a range finder for their own arduino project opens up ways to add extra features without repeating all their work.

BeagleFury Wrote:
-------------------------------------------------------
> >> Personally I think there is actually a need to
> move away from Arduino or at least the AVR 8bit
> core.
>
> I disagree. I still feel the officially
> documented electronics should include "simple to
> build", "easy to find a place to purchase one",
> and "inexpensive".
>
> If you want to use a 32 bit Cortex M3, just make
> sure you design the board with an interface
> compatible with the Arduino or two (To open up
> more IO ports if they are available). Since
> Arduino development targets 'C', there shouldn't
> be too many problems converting, consisting mostly
> of word size assumptions and/or IO library --
> creating an Arduino compatible library seems
> perfect for this kind of porting task.
>
> If development continues down the Arduino path, I
> believe there will exist a higher probability that
> other projects external to RepRap will contribute,
> either directly, or possibly unknowingly --
> someone making a range finder for their own
> arduino project opens up ways to add extra
> features without repeating all their work.

I have 2 aims for this concept:

1. Provide a cost effective, easy to setup and use, control system for the Reprap machine providing the capability to support most of the features desired by most users here and attempt to eliminate a number of common failings of the existing design.

2. Provide some elements of flexibility in the design of the architecture, with the aim to allow the reuse of the design for implementation in other projects,

My aim is not to create an Arduino clone board using the LPC1768, my previous comments in porting Arduino, related to the software to used in Arduino not the hardware, this could have been misinterpreted.

I'm guessing your aim is to try and keep the electronics as close to the standard arduino profile as possible, with the theory of being able to get hold of parts easier given the current supply problems at Makerbot.

My aim would provide some of this flexibility in that the architecture would be designed to be more flexible in terms of usage.

You do have a vaild argument in terms of supply, but not all the components required for Reprap are available for arduino without some manufacture. So currently this would not solve the problem.

Finally the cost of the shield type systems is higher than a dedicated system and can offer technical problems in the operation of the system.

good points I think we should also focus on cost and daughter boards might not be the way.

we should also bring up the idea of a separate extruder controller yes or no.

I have seen a lot of use of our code to drive cnc machines so i would think a separate controller would be cool then we could slowly design a new extruder controller but on the other side a separate extruder controller could cost more.
on a plus side a separate extruder controller can be made for each config type like thermistor vs. thermo coupler.

on a last note staying with arduino might allow our electronics to be used in other projects.

annodomini2 Wrote:
> Finally the cost of the shield type systems is
> higher than a dedicated system and can offer
> technical problems in the operation of the system.

Cost seems like a lesser concern than availability and flexibility (electronics-wise).

It's sort of the whole argument against the combo DVD/VCR/TV all-in-one devices : one part breaks, and now you have to replace the whole kit and kaboodle. There also exists the advantage of being able to upgrade / update one part at a time. I hoping for the time where RepRap evolution takes off and I can replace modules with superior modules -- replace the stepper driver with a servo driver, after printing out the servo mounts, and then print out a new clay/ceramics piston, and hooking up the daughterboard /or slave board to drive that so I can make some nice artistic custom mugs to throw in a community kiln.

BeagleFury Wrote:
-------------------------------------------------------
> annodomini2 Wrote:
> > Finally the cost of the shield type systems is
> > higher than a dedicated system and can offer
> > technical problems in the operation of the
> system.
>
> Cost seems like a lesser concern than availability
> and flexibility (electronics-wise).
>
> It's sort of the whole argument against the combo
> DVD/VCR/TV all-in-one devices : one part breaks,
> and now you have to replace the whole kit and
> kaboodle. There also exists the advantage of
> being able to upgrade / update one part at a time.
> I hoping for the time where RepRap evolution
> takes off and I can replace modules with superior
> modules -- replace the stepper driver with a servo
> driver, after printing out the servo mounts, and
> then print out a new clay/ceramics piston, and
> hooking up the daughterboard /or slave board to
> drive that so I can make some nice artistic custom
> mugs to throw in a community kiln.

If my concept behaves as I am envisioning it, then this interchangability will be a reality, but obviously this is only on the drawing board so have no supply to speak of.

The concept is also observing the potential for Hot swap, so when new heads become available, instead of having each wired up the robot can actively change heads.

Ok so we should be compatible with existing hardware.

I think 32 bit is the way to go. There are some atmel chips we could use I have a few Dev boards for them and they might be easier to talk to the Arduino Env.

Should it be open source??

I am also a very big fan of Diagnostic util. Either leds or software to test with and think that is a big thing missing in the present hardware.

My plan would be open source, stupid not to in the spirit of the reprap, the AVR chips are more expensive than the LPC, but if you wish.

Environment compatibility is irrelevant, its just adapting the core operations to provide a similar interface, this can be done for most micro's that support ISP.

I agree with Bruce we need to always maintain an upward compatible path so that every one in the Repap community. Can easily upgrade to the high tech versions as they become more involved in the RepRap project.

This will mean that older versions will gradually become available for new RepRap builders. Thus keeping the New rep-rapers starting cost down.
Its also a good way of off setting the additional cost to upgrade a RepRap system.

I am also a big fan of diagnostics that might well include self testing systems, automated calibration,

All parts of future development must maintain the original idea of everything being open source/creative commons, so its freely available to one and all.

---

Bodge It [reprap.org]
=======================================

---

BIQ Sanguinololu SD LCD board	BIQ Stepcon	BIQ Opto Endstop
BIQ Heater Block PCB	BIQ Extruder Peek clamp replacement	BIQ Huxley Seedling
BIQ Sanguinololu mounting	BIQ standalone Sanguinololu or Ramps mounting	Print It Stick It Cut it

My rep strap: [repstrapbertha.blogspot.com]

Buy the bits from B&Q pipestrap [diyrepstrap.blogspot.com]
How to Build a Darwin without any Rep Rap Parts [repstrapdarwin.blogspot.com]
Web Site [www.takeaway3dtech.com]

So here is what I have found so far

NXP LPC1768 $8.70 looks like a very good possibility.

The omap3503 seems cool (you can run Linux on it )but it does not have that many I/O pins and it is 35.00 for the chip it would make a cool main controller.

the atmel AT32UC3A3 is about 12.00 per chip and i dont think it is as world wide avail. Even though it is used in the make magazine controller.

Bruce Wattendorf Wrote:
-------------------------------------------------------
> So here is what I have found so far
>
> NXP LPC1768 $8.70 looks like a very good
> possibility.
>
> The omap3503 seems cool (you can run Linux on it
> )but it does not have that many I/O pins and it is
> 35.00 for the chip it would make a cool main
> controller.
>
> the atmel AT32UC3A3 is about 12.00 per chip and i
> dont think it is as world wide avail. Even though
> it is used in the make magazine controller.

How cheap do you want to go for limited MIPS, RAM, or FLASH?

LM3S101-IRN20-C2 $2.74
LPC1342FHN33,551 $2.98
LPC1751FBD80,551 $4.21
LPC1765FBD100,551 $7.60

Guys

I think there is merit in all that has been sugested.

My personal favourite is staying with the Arduino, The mega can be sourced in china for aorund 2UKP, and is available in the UK for around 20UKP.

The reasons this is my current favourite is that we have'nt run out of steam/headroom yet (although we might soon so other things are definitely worth considering) we can take advantage of the mass market and production to get access to a ready to go controler board that is very cost efective.

I agree that there is a lack of on machine display and control (that shoulds also do local printing and diagnostics)

Overall what I would like is to move away from lots of ribbon cables etc and to work towards is a passive modular motherboard arangement using low cost Pin headers and sockets. Where the processor board can be whatever you want and the IO Driver cards plug onto the motherboard.

This method (is realy not new and has been a favourite for instrumentation for as long as I can remmber) allows you to mix and match the drivers and controlers etc as you want. And to upgrade them as you see fit without binning everything.

You can also have simple single or double sided prototyping ards that slot into the same system so you can develop what ever else you want.

The case or rack for this should be RepRapable if you want it.

Just for conversations sake I would sugest a start point of a standard board size just larger than an Arduino Mega (to allow Megas and other variants to be mounted on to the board). With the passive backplane taking all the IO from the Mega Sheild sockets. This gives you an initial spec to work to.

Into this back plane you should be able to plug in a PSU card, It may be and empty card with an ATX connector on it, or it may be a fully fledged Switchmode PSU taking a 12v car battery style supply. The choice is yours.

Overall yes this will be more wasteful of PCB and pins/sockets but gives the flexibility that you all want to be able to do your own thing and benefit from other contributors work.

The backplane, Arduino Mega Adapter, ATX PSU Adapter and Prototyping board can all be boards that are doable on the kitchen table at home using toner transfer techniques. Or alternatively sourced cost efectivley from a local board maker.

If you want to epxperiment with CPU-U-Like (AVR32, Arduino whatever, ARM, XMOS, etc etc) do it and design a controler board that only has to have compatible IO/Power SOurce via the back plane. Plug and play (Not in the PC sense).

Thought s for what they are worth......

---

Necessity hopefully becomes the absentee parent of successfully invented children.

I really like the backplane approach aka47 proposes. The Arduino Mega exposes around 70 pins, so it probably makes sense to do a dual-row connector. We might consider card-edge connectors. Not having to solder those 70 pins would be nice.

I've been frustrated with the shield format for a while. I think this would be a big improvement.

-Tim

bothacker Wrote:
> I've been frustrated with the shield format for a
> while. I think this would be a big improvement.

Just curious.. what frustrations have you had? I would assume that a header pin connector would be about as convenient as a card edge connector, if not even more so since you can stack multiple boards on top of each other in cases where not all pins need to be used by a particular device.

the main reason i like aka47's idea is that is seems nice and clean, and i think it is an easy way to keep things in there place. over all i rather like the idea. but you are right for the most part beaglefury i do think they are the same for the most part.

one bad thing about stacking though is it can make it allot harder to add heat syncs if you need too. that is one area where the back plain is bettter i think

---

[mike-mack.blogspot.com]

BeagleFury Wrote:
-------------------------------------------------------
> bothacker Wrote:
> > I've been frustrated with the shield format for
> a
> > while. I think this would be a big improvement.
>
> Just curious.. what frustrations have you had?

I find that the with shield format, with signal wires around the perimeter of the board, it is awkward to segregate noisy circuits. And if your circuit layout is larger than the basic shield size, having to work around and in-between the "circular" connector layout is a hassle.

-Tim

bothacker Wrote:

> I find that the with shield format, with signal
> wires around the perimeter of the board, it is
> awkward to segregate noisy circuits. And if your
> circuit layout is larger than the basic shield
> size, having to work around and in-between the
> "circular" connector layout is a hassle.

Fair enough. So, you would propose several card slots? Would they all share a common bus, or would you forsee each card having a dedicated set of I/O pins?

Either could work. I suppose the exact interface may matter much less in terms of modularity; worst case, one can create adaptor cards that would create any interface needed, E.G, a card plug that provided header pins, or vice versa, a daughterboard that plugged into the headers, and provided the card slots you desired instead.

So How about the best of both worlds use this

[leaflabs.com]

We might want to look into making a board that uses the same pin layout as the Arduino mega? and then make shields??

or the Xduino [www.xduino.com]

so there are a few 32 bit arm that have ported over to the Arduino interface.

Guys

Interesting points, it is probably worth kicking this one around for a little while and then trying some ideas in reality. The backplane is in effect a passive IO back plane.

The main three obstacles to using the shield format in a multi board stack are :-

1. If you want to work on a board whose components are within the stack you need to continually assemble and disassemble the stack to get at the board.

2. The current shield model does'nt readily support board stacks of more than two boards. Without bodging extra connectors onto bottom of a mid stack board by hand. Doable but very messy and prone to breakage due to 1.

3. Heatsinks are enclosed on too many sides to work effectively, The connectors around the edges limits the number and placing of connectors you might want to place for your boards function. With a board end connector you potentially have upto three sides to work around. (Allowing for board mounting of course.

Other than the above I actually like the shield format most excellent in an educational setting. I just feel it is limiting for what we are trying to do.

Backplanes could be as many slots long as you wanted, a sensible minimum may be 4.

Like wise you probably would'nt want to export all the Mega IO pins as-is onto the back plane. It might be worth considering :-

A. Leaving off the serial IO that is used by the USB/Host Interface.
B. Leaving off the serial/ISP IO that is used by the SD Card Interface.
B. having at least one serial/UART presented onto the backplane as the RS485 Buss that is now common on the current electronics. It means you can have multiple cards that you can talk to on the buss if you wanted to experiment with this, As well as supporting the current RS485 extruder boards from makerbot.
C. Having a common reset line that all sub boards can respond to to ensure that they all come up in s sensible state at power up or reset.

We may also want to consider doubling up on some power rail pins, ie two per voltage.

On the sugestion of heading towards a single 12v input supply, I personaly like this idea very much. It makes sense to minimise the number of voltage regulators etc in the system, this both minimises cooling requirements and makes it all run a little more power efficiently. 12v supply systems are easy to create from automotive junk and using mass market automotive components.

What other power rails would you really need though. Clearly 5v but would you really need 3.x Volts ??

On connectors I proposed the standard 2.54mm (0.1") header pins and sockets because :-

1. They are pretty much dimensionally standard internationally.
2. Widely available internationally.
3. Inexpensive, board interconnects don't come much cheaper for their level of reliability (certainly cheaper than DIN 4162 connectors)


On an aside re the pin and socket connectors. My thoughts are to make the backplane all sockets. The plug-in cards have pins, but only the pins that they need to pull off the IO they want from the back plane. Unless it is a prototyping card. So most plug-in cards would have power and reset, plus only whatever pins they needed to connect to to the IO they wanted.

This reduces the electrical loading on the backplane and reduces the amount of unnecessary pin soldering to only that which is needed.

Making the backplane sockets also allows for making up pin to cable assemblies and plugging them into the IO backplane if you want to.

Sockets on the backplane protects the backplane IO connections from accidental shorting. As a plugin card is arguably not powered when removed from the back plane having exposed pins does'nt matter.

Thoughts for what they are worth.

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Since I'm not an electronics guy, I've been thinking whether to chip in to this debate or not, but here goes...

I know that for PCB wizards, getting the electronics up and running is a small thing, but by looking at the other forums, I get the impression that the success of MakerBot is holding RepRap back. RepRap shares the same custom boards as MakerBot, and there is a lot more demand than supply.

At least short term, ensuring easy availability of electronics really should have priority to new features. So anything which could help wrangle the "monopoly" (note: not meant maliciously)out of the hands of MakerBot would be an improvement.

Next, before jumping in and creating a high powered main board, I think it would be prudent to consider, why a high MIPS board is required? Will the RepRap be a dumb device driven by a computer, or should all the logic be handled on board?

Definitely sockets on the backplane, and I like the idea of moving to single 12V supply.

I did a quick part estimate on digikey. There are probably cheaper parts, but for a first cut, here's some numbers:

Connector: 0.1" Dual Row, assuming 72 Pin

header, right angle, dual row S2112E-36-ND 3.42
header, female, dual row S7104-ND 3.59

part cost/connection ~ $7

Something else to consider: I see that PCI card edge connectors are very cheap. Here's a 64 position connector for $0.87 in single quantities.

[search.digikey.com]

Some of my goals with this Discussion was to

1. make the change the electronics to be more Reprap friendly.

Mainly the power to go back to just 12 volts and do away with the computer power supply by using a separate power supply board, (and there maybe be 2 different versions one for PC power and a different for 12 volt dc power. I am not against the pc power I just dont like having to wire it up to the mother board directly.
or having to hack the present motherboard to work off of 12 volts.



2. Reprap is a development platform and I feel that we need to continue with electronics development and not stay with a platform because it might be open source.

3. I think we need to make Reprap to run with no pc attached (I am also very much for one that requires a pc and would love for a system to able to be upgraded to a larger "motherboard".)

4. as far as the discussion of do we need more robust electronics not that the present is not great but take a look at the Repman vs. Reprap Electronics and the quality of print I just think that there is more that can be done with a faster more memory processor.

I wish we would get others in this discussion like Nophead, Forrest, Vic..
It seems like we have lost Reprap support from Zach, (not to point at Zach you have been doing a great job!!). I think we need to take this to the next level..
And I also think we have some great talent and wisdom and can come up with some great options.


Also Anton anybody can open a store and sell electronics (I cant say that it does not go through my mind every day).


Bruce

I found this a while ago it is a chip that is used for I2C stepper controller

[www.avrcard.com]

not that we want to limit the system to I2C

This also looks like it could easily be made to be single sided board.

and at 30 eur.

Bruce