Beyond the Sanguino

Quote

I guess you could also construct a hot air blow gun and nozzle Tool Head to reflow in situ on the Darwin....

That wasn't something I had considered. I vaguely assumed that it would be easiest to toast the finished board. It would be very nice to have an all-in-one solution, certainly.

Quote

Why wait for dispensable metal tracks ??

I regard those with a certain amount of suspicion, to be honest. I have visions of motor drivers falling out because they had insufficient heat dissipation, and melted the field's metal underneath them, or whatever. Uninformed paranoia, perhaps

Quote

A big question is do you actually really need milling capability or only hole drilling (via's) ??

I'm of the opinion that milling is a very useful technique to have to hand, even if reprap can 'only' manage wood and plastic. Given that I have little faith in the extruded curcuit approach right now, I envisage being able to carve out your own circuits by more conventional means almost essential!

Quote

the question, however, is how we get from here to there.

I reckon a good start is for a few people with the more advanced projects to do a little experimenting with milling

Skillet reflow.

[www.sparkfun.com]

[www.nearfuturelaboratory.com]

[www.reconnsworld.com]

Toaster Oven

[www.seattlerobotics.org]

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Isolation milling?

Here I though the plan was for the reprap to embed all the components in plastic, and then wire them together with molten solder?

That way we don't need a milling head (few heads overall) and we a lesser variety of build materials.

It'd be nice to need only solder, plastic, and IC's. The question would be if the needed acuraccy is achievable - laying down a bit of plastic to isolate two adjacent pins on a fine-pitch surface mount chip isn't within what the current extruder head can do...

---

--
I'm building it with Baling Wire

Quote

Here I though the plan was for the reprap to embed all the components in plastic, and then wire them together with molten solder?

This sounds like a far greater technical challenge than doing isolation routing, which is why I'm very skeptical of it. If a complex working circuit is made this way I'll be pleasantly surprised, to say the least.

It seems like a very contrived way of making circuits, when access to blank circuit boards isn't hard to come by, nor painfully expensive as is the case with rapid prototyping, for example. I'm going to gamble on copper foil being cheaper and easier to work with than field's metal, I think

Quote

That way we don't need a milling head (few heads overall)

But instead you need a low melting point metal extruding head. What else would that be useful for?

Now what else might a lightweight milling head be useful for?

I know which one I'd rather have, given a limited palette of tools to work with.

A thought I had about a hot air reflow head vs reflow oven;

With a reflow head it would be possible to remove/scavenge parts, to paraphrase Haynes' manuals 'disassembly is the reverse sequence to assembly'. If I had a method of using SMDs then ideally I'd like to be able to harvest parts from old circuit boards, as I currently do with through hole stuff.

In rapid circuit prototyping it would make even more sense; Components being transplanted from the previous version's board to the new design.

In thinking further about it while there is obvious overlap there are also quite a lot of differences. I've listed the differences I came up with below, Despite these differences a reflow head would be an important step towards being able to 'recycle' SMDs.

Circuit creation:
- the type and location of the parts are known
- a solder paste dispenser is needed
- A method of forming circuit boards is needed
- circuits can be designed around limitations of the various tool heads
- the pick and place only has to overcome gravity

Scavenging:
- nothing is initially known about the components
- a camera, either on the frame or on a tool head is needed
- software is needed to identify and select components (see below)
- Some components may not be retrievable owing to limitations of the various heads
- the pick and place will have to overcome the surface tension of the molten solder as well as gravity.

Although the machine vision and parts selection is far from trivial surface mount parts make it much simper by mostly being rectangles with identifying markings on them. The machine vision doesn't need to be perfect and so could initially only identify easily found parts; retrieving some components is better than binning them all.

This is kind of off topic, but oh well (I think the discussion of the original topic went as far as it could go, and it is indeed interesting how the thread evolved into a discussion of RP methods for creating SMT boards).

As far as desoldering SMDs, I think it might be worth considering putting the heated skillet on the platform (that is, the platform normally used as a build platform), putting the board on the skillet, and using a pick and place head to remove each device and place it on the side (where it goes after removal is a different discussion), instead of having to operate two toolheads at once (or a single more specialized toolhead with a pick and place and a hot air dispenser). I imagine it could be a bit less efficient, but the difference in efficiency could be minimized by proper insulation, quick operation, and a few other techniques which may or may not be effective.

It may be interesting to also use a flux dispensing head before using the pick and place to clean the board (perhaps allowing the components to be removed more easily) or perhaps a solder paste dispensing head to apply a bit extra solder before desoldering (I'm not sure if this would help at all, but if you have such a head for SMD soldering anyway, it's worth a try).

As to finding the components on the board, perhaps a simple rangefinding toolhead could be used to create a heightmap of the board, showing the locations of devices, leads, etc. Personally, I would prefer a sound or light rangefinder (building a sound rangefinder shouldn't be too expensive, probably $10-$50), but AFAICT we already have a mechanical rangefinder, thanks to Andy Kirby (the touch probe, [reprap.org] [though I don't know precisely what stage it is at in relation to creating a heightmap]). I think the biggest problem with this (any rangefinder toolhead used for mapping, not just a mechanical one) would be the limited accuracy of the positioning system, but, depending on the size of the tip of the pick and place, it may be irrelevant (as you would simply place it in the middle of the component, and if the positioning is off by a bit, it might miss the center but still grab the component).
It is a sort of camera, but I don't think a traditional [digital] camera (like a CCD) would be a cost-effective solution.


Of course, I don't know whether the skillet method would work for desoldering through-hole devices. I think the heat should be enough to allow a different tip on the pick and place to grab the device (or perhaps the same, depending on the device, really) to grab and remove the component, but there would be friction between the leads and the inner sides of the through-hole, which might complicate things. Perhaps applying heat directly to the lead (with a heated tip, perhaps similar to a soldering iron toolhead) would be more effective and efficient (perhaps this would also be a better idea for SMDs, as well, perhaps using Chip Quik [dispensed by a toolhead] or some other method).

In any case, it would probably be a good idea to remove the through-hole devices before removing SMTs and also to remove non-IC devices (like capacitors, etc.) before flat IC-style devices. I think that the software/firmware (probably software) will be able to differentiate between the two sets by analyzing the heightmap. Depending on the accuracy and precision of the positioning system, the system may or may not be able to produce a heightmap with high enough resolution to differentiate between SMDs and through-hole devices.



As to soldering, I'm not sure how it's usually done (in mass production and such), but the paste solder and skillet reflow method seems fine. I do wonder if screen printing is necessary, or if we can simply dispense it with a toolhead as others have suggested (I worry about the fine pitches of SMDs). Then again, if direct dispensing doesn't work, I suppose we could create the screens (perhaps using another toolhead) and then use the dispenser head to apply the solder paste through the screens, or some variation on that idea.

In any case, I must cut this post short (too late) and say that the most important part (well, the most relevant part that needs to be developed) of either process is the pick and place toolhead. The rest of the stuff is either not specific to to the RepRap or requires a small amount of work.

I think there is room for both deposition and milling.

Maybe they are complementary techniques, maybe they are competitive.

Either way the end result is we can fab electronic circuitry. Down to Surface Mount sizes.

Personally the objective is electronic prototyping/construction on home built kit (or home built by something that I home built)

Right at this exact second as we have neither I will take the first along and if something that makes the first redundant comes along I will move to that and so on.

I guess it's bootstrapping capabilities.

In so far as we cant currently readily meet the need to surface mount the next device beyond sanguino, yes it is off topic but is also a stumbling block along the topics route.

If we can work out solutions for that then I guess Surface Mount devices with high pin counts (but how high given our current level of resolution ??) are available to us for post sanguino implementation.

Which way next then ?? more Atmega or Custom Logic keeping backwards software compatability with the Atmega or port/rewrite for something else ??

---

Necessity hopefully becomes the absentee parent of successfully invented children.

I like the idea of putting a hotplate/skillet under the PCB on the RepRap deck.

It fixes a something that always worries me which is how do you move a board full of components without slopping them either off the board or out of place.

I know the solder paste has some grip, paranoia say not enough though.

cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Quote

(complex machine vision circuit board disassembly plan)

If you're dismantling a board which you can already fabricate, this is easy. If you're dismantling a board which you have not fabricated, but migt need to build or dismantle more in the future, it might be worth trying to automate the process.

For one off boards? Have a jog controller, and mark the corners of each IC and each discrete device.

Dull? yes. For larger circuits, mind numblingly so.

But unless you're making an industrial scavenging machine, you don't really need it to be any other way. It would be responsible if reprap coudl recycle its own products, but merely a bonus if it could recycle others too.

Quote

It fixes a something that always worries me which is how do you move a board full of components without slopping them either off the board or out of place.

I recall some SMD soldering guides suggesting a small blob of glue to hold the larger components (ie, ICs) in place. Once you've sorted the precision paste dispensing problem, this shouldn't be a challenge. Small devices like resistors and the like should have so little inertia that they won't slop off.

But don't take my word for it.

Some systems use a mechanical glue dispenser as well as a solder paste. One thing to note is that the board will only be moved up and down on a RepRap, not vigorously side to side - hopefully this will reduce the odds of components detaching.

Joshua: our SMT assembly gear at uni uses dinky little syringes of solder past that apply a controlled blob to the board from a gantry arm. This sounds eminently RepRap friendly.

If you have a look at SparkFun (www.sparkfun.com) they have a tutorial on soldering fine pitch SMT stuff using a skillet or electric frypan as we call it down here. They apply solder paste with a mask, but there's no reason it wouldn't work with syringe applied stuff.

I expect that in terms of Pick and Place it would be easier to use as many SMT components as possible, because they can be easily (hah) placed, then do the through hole stuff (which may be harder to align) by hand if necessary.

I have heard from a few friends that've tried it without that the solder mask is NECESSARY for fine pitch SMT work, or all of your pads bridge over in the oven.

jbb

I agree with the techniques for component holding, I have done them assembling Surface Mount by hand. small component counts, large component counts you don't want to go there.

If pick and place is one operation per device, reflow time is everything at the same time. so you can just about discount, but stick every component doubles your production cycle time. Plus impedes scavenging and rework.

ie a solution that makes more problems.

Better to reflow in situ. Saves that careful carrying of the board followed by lots of swearing as you trip over the cat/ducks and drop it on the floor..........

BTW

Scavenging I knew a guy years ago who used to scavenge double sided Through Hole boards in a determined way.

He used a blow torch.

It is basically a hot air gun when you think about it just a bit fiercer.

His technique (which works cos I blagged quite a few components of him over time) was to wave the flame over the underside of the board far enough away and for long enough to make all the solder in an area molten (both sides, the through holes conduct heat). He used to then hit the board on a lip of a wooden tray and all the components used to fall out of the board and into the tray to be sorted when they cooled.

It takes a little practice (but as he blagged boards for nowt, he could afford to waste a few) but is very quick and effective when you get the technique right. You dont hang a bout doing this or the components are toast, quick is the key to melting the solder and extracting the parts before the components become heat damaged.

If you have ever tried professional reworking/repairs on multilayer boards you will appreciate how difficult extracting components is with out screwing the tracks and PTH's is.


There is no reason why using the skillet reflow you can't do the same thing ie

Reflow the solder then very quickly pick up the board with tongs and use the same extraction technique.

Sorry guys, no Darwin neccesary.

cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

... sometimes i have to repopulate SMD-PCB's manually - it's no problem for passive components with two to four pins with a fine soldering bit.

For desoldering SMD-IC's i use a 50-microns-sheet of steel as blade for cutting through the melted solder, so i release pin after pin.

By the way - here: [forums.reprap.org] - a photo of two pneumatic dispensers - left the small one is for soldering-paste, the big one is a heated glue-dispenser.

*** Edit: and atached in the same post a photo of some vacuum-grippers suitable for SMD-holding and positioning ...

Could be easily atached as second (or third) tool-head in a sturdier repstrap or reprap ...

Viktor

Edited 2 time(s). Last edit at 09/05/2008 06:05AM by Viktor.

I like the little one.

I wonder it it's at all possible to make a simple small RP version of the valve/top bit.

Acquiring compressed air for those sort of flow rates is trivial and very cheap (using recovered domestic components).

I have been messing about with a handfull of syringes I acquired for messing about with.

The 10mL one I have (OD 16mm) fits realy well in the Universal Pen Holder and Touch Probe.

Just need a simple and cheap way to run it or an equivalent dispenser syringe plus needle combo. (remember our previous discussion re LuerLock etc)

cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

I velly surplised that no one hav sugestered a WepWap RireRap mashine valiant yet.

The WepWap rap the rire awound the foo hole leggies of eech componerent and groin the leggies with an elechical costechion to eech other. A spitter mechanism take a the insultation offa the copper wir to bondi to the plateleting of the leggies.

Noting: no murcury or gallium are drip out of prastic when hot melting.

grael Wrote:
-------------------------------------------------------
> I velly surplised that no one hav sugestered a
> WepWap RireRap mashine valiant yet.
>
> The WepWap rap the rire awound the foo hole
> leggies of eech componerent and groin the leggies
> with an elechical costechion to eech other. A
> spitter mechanism take a the insultation offa the
> copper wir to bondi to the plateleting of the
> leggies.
>
> Noting: no murcury or gallium are drip out of
> prastic when hot melting.

Got a lot of time on your hands, do you?

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

Never enough Forest

These are my intended pin assignments for the single PCB Arm Cortex-M3 based CNC system I'm designing.
(see attached gnumeric spreadsheet)

Even after losing quite a few pins to parallel power duplication and higher functionality state control, a 100 pin package has plenty left to handle what the Repraps currently do, and a lot more as well.

Major features:
4 symmetric dual full bridge drivers, driven directly from the microcontroller, each enable controlled from a dedicated PWM output, and 2 analogue inputs assigned per dual bridge for detection of missed steps and stalled motors.
Each driver is the L298, rated at around 2 amps continous, when correctly heatsinked.

7 general purpose outputs to go to an on board 7 channel ULN2003 darlington driver, including heater control for the extruder.

3 PWM outputs for servo control (like radio control servos)

USB

JTAG for programming

3 x axis limit state monitoring

2 general purpose inputs

Extruder cold zone temperature monitoring analogue input

Platform temperature monitoring input

SD memory card interface

Status Piezo output

2 status LEDs

One analogue input for control button detection.
3 x ctrl + 8 x data reserved for connector, for common standard or graphics LCDs

RS485 (2 wires+ ground, industrial communications) bus

I'm using the free version of eagle, so space is rather constrained on the PCB for double sided, non stacked. There has been some interest expressed, and also the point made that it is not an efficient development in terms of existing planning for the official reprap development design path.
The situation, is that I've come in, as an outsider, after the project has some established history, and as a former professional electronics and software developer, I've formed my own opinions, based on my own experience of design and production.
I can understand how the existing development has occured, and I am impressed with the team leaders efforts to make Reprap available and functional, but I don't agree that the functionality of the existing modular design outweighs the cost savings and logistical clarity of a single board approach.
I am expecting to take around 3 months before I have my machine beginning to process Gcode, The processor is new to me, and I'm used to 8 bit assembler on various micros, but not C, nor 32/16 bit micros.

I welcome any criticisms and suggestions for improvements. I'm working on this only part time, and so any ideas that would help to make this a generally versatile PCB are appreciated. I don't want it to be a total mass of general purpose connectors, I want clean functionality, with common expansion options. RS232 I'm dumping, although it's what I've been used to, because it's getting too out of date, and also, expansion hubs for USB, and the lower parts count of USB are advantages.

One tweak I'm considering, is to have the LCD option addressed, so that it can be multiplexed with CPLD expansion/s. There is a lot of functionality available already though..


Graham Daniel

grael Wrote:
-------------------------------------------------------
> I don't agree that the functionality of the
> existing modular design outweighs the cost savings
> and logistical clarity of a single board
> approach.

You're far from alone in that opinion.

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

grael, I think we come from similar backgrounds. I completely agree with your move to ARM. I'm sticking with ATMega for now out of respect for those who don't have PCB manufacturing capabilities (including me!)

I'm designing a modular control system which, I understand, you don't like. I agree with you on cost, but I do have a few technical reasons for going the route I am.

My system has one primary controller which runs the GCODE interpreter but in addition to that, it dedicates one controller to each toolhead, one controller for the toolhead swapping system, and potentially one controller per axis, but let's leave that last one out for now. It's just an idea I'm tossing around.

This sounds excessive, but there are a lot of good reasons behind it.

The toolhead at a minimum needs to have an independent controller because of swappable toolheads. In order to control each toolhead correctly, it makes far more sense to have a dedicated controller that knows how to manage the toolhead. It also cuts down on cabling and related problems: I don't want an analog line running next to the motor drive PWM. I think this makes perfect sense. In my system, the motor drive PWM is generated on the toolhead, and the only connections to the toolhead are full duplex RS-485 and power.

For the toolhead swap system, the dedicated controller isn't as necessary, but it makes sense because that is a complex task to manage. In addition, it will take a fair number of signal lines to perform the toolswap management, and cabling to it will become problematic.

I think the mother board makes a lot of sense, but integrating the motor drivers on it may become problematic.

As I see it, there is a need for a minimum of two PCBs: The motherboard and the toolhead controller. If we add multiple toolheads, the toolswap controller becomes necessary too.

I'm still not sure what to do with the stepper motor drivers.

Guys

I have been musing for a little while on similar.

My thoughts are that the systems should indeed have a master controler board with a better processor core for motion planning and Gcode conversion etc featuring.

Host Comms, Keypad, LCD, Media Card, Environmental control/monitoring and Z axis control.

A controler dedicated to X and Y axis motion control featuring limit detection for both X & Y axes as well as the head swap management (if any as it is largely mechanical anyway) and the RS485 Bus termination.

The Currently active tool head/s ie the ones that are mounted on the carriage to have a dedicated controler on each and connected using stub RS485 connected back to the main bus. The tool head controlers should have hot plug capability built in and the ability to uniquely identify themselves to the controler.

This is pretty much as far as my thoughts have got so far, for what they are worth.

I went with X & Y under one controler as it is easier then to synchronise movement on these two critical axes from one controler.

The slave controlers could well idealy be AVR/Arduino based for ease of transition.

The main controler could well be Sanguino or perhaps something more powerful......

The Z axis could well be split off onto a dedicated slave controler again RS485 bus connected.

Using slave controlers in this way also means that alternative actuator technologies can be readily accommodated within the same principle design as only the slave controlers and not the master need know what is being driven and how.

The master issues higher level go-to/how-fast-to commands to the controlers.

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

aka47,
I think there's a growing number of us on the same page.

I'm not sure that headswap goes well with X/Y monitoring.

I'm beginning to wonder if CAN might be more suitable than RS485 given the complexity we're starting to create.

Annirak,
I've had no experience with CAN bus, I know it's popular in the automotive field though, which would put it in a similiar environment, i.e. no more than a few meters maximum size, electrically noisy environment.

I haven't researched CAN properly, but just spending a couple of minutes now to look at it, I have these points to note:
~It looks like CAN still needs external tranciever chips (as with regular serial and RS485)
~CAN bus on the STM32F103VBT chip is an alternate function of the USB pins (pins 70, 71 on the 100 pin chip.)
~CAN is still rare on the smaller microcontrollers last I looked. I'm sure they are around, but you may be much more limited in your package, manufacturer and feature options compared with standard asynchronous serial.

RS485 is very robust, and doesn't need many wires.

The analogue line running next to the toolhead that you mention,
I'm thinking it's for monitoring heat at the extruder ?
If so, the changes in temperature are probably not going to exceed 50 deg Celcius per second, less again in practice, due to thermal inertia of the sensor.
Is this case, you can capacitively or digitally filter the received voltage to obtain the correct temperature. It's probably not even worth taking reading more than say... 4 times per second ?

I know what you mean about separating out tasks for each axis or toolhead to a different controller, but assuming I have enough toolheads and axis as standard, then the separation can be done at a software level quite easily instead. I'm thinking I will have to run a tasking OS, but I don't think they are too hard to set up from scratch anyway. There is a performance penalty with register swapping, but if you reserve RAM for all the required functions for each hardware device, then the penalty for slightly more sophisticated register use significantly outweighs the performance lag involved with serial communications to slave devices.

Correspondingly, in a modular hardware design, you have to propagate software changes through all affected slave microcontrollers too, and the serial conversations become somewhat complex in themselves, even if you only have regularly polled register updates over the serial links.

I don't quite understand what all the toolheads would do though, it seem to me that you need well controlled drive speed for extrusion, valve if the extrusion is via a seperate pressure reservoir, heat if melting something. I've noticed on the forums though, that stopping mid print and restarting seems to create imperfections. If the toolheads work in the same 3-D point when active, then there must necessarily be a stop start action, even it's only an end of colour layer process. That's why I was thinking of a single extruder heater, but with multiple inputs to the melting/mixing cavity. There would be some diffusion and bleed out during changes, but on quite a fine scale, especially if the mixing chamber and extrude nozzle is non-stick.

Even solder could run through the same mixer potentially, but normal solder properties aren't quite right. The viscosity to surface tension ratio would need to be increased, and the melt temp matched to the plastics.

Which leads to:
Could my variant handle the tasks that you and Andy are envisaging ?

If you can read the gnumeric file I attached to my post above, I have 3 PWM outputs available for servo motor control and another 6 spare outputs going to ULN2003 darlington driver pins (the seventh is the extruder heater).
If a fast interupt clocks counters for each general purpose channel, and runs a compare against channel "position" values, then you can have a total of 9 servo motors driven, quite easily. These could be running triggers of budget electric screwdrivers, dremmels, scissors, pen up/down etc. Alternatively, you could run them to a PLC board, or drive relays, fans etc. Once the basic machine functionality is done, you could have some upgrade parts to produce a ratcheting tool head swap device, with an inbuilt rotary switch.

I'm seeing two main themes though:
Additive (extrusion- variations on rate, thickness, temp, material/colour)
Subtractive (milling- variations on milling bit, bit speed.

Past this, you are looking at pick & place or other automated factory type functions. By it's nature though, a REPRAP CNC machine only works at one point at a time, and at high resolution, so it's slow. I don't think it's very productive to have the same machine doing assembly operations, because once you make one part, it's faster to make the next 5 pieces via mouldings. There are exceptions of course, shapes that are otherwise difficult to make in one piece. Once the decision is made to produce in quantity though, it makes sense to produce the moulds instead.

I do have what could be a 4'th axis on my design so far, but I'm thinking that could be for a stepper motor extruder drive. If you WANTED a rotating toolhead, then you could run the 4'th axis stepper motor for that, and lock it with a servo motor once in place.

I have a friend with a CNC machine, and he has a 4'th axis that is rotary, so he can do lathe like operations.

Graham

grael,
If you read through my thread on toolhead controllers, you'll notice a few important points that you may have missed.

Just because two toolheads are both subtractive doesn't mean they need the same control system: Take milling vs. EDM, for example.

For milling, you spin a tool quickly and it cuts. You may also spray lubricant onto the piece you're milling.

For EDM, you charge up a capacitor to a very high voltage, then discharge it through a wire feed into the piece you're machining. You might also spray an electrolyte onto the work surface.

To mill you need a high speed motor drive and a pump drive. To run EDM, you need a low speed motor drive, a capacitor charger and a pump drive.

EDM actually has more in common with a thermoplast extruder than with a mill.

No two toolheads will take exactly the same command set either. They'll all be similar, but not the same. As always, the devil is in the details. Handling this in software will get really complex really quickly. This is part of the reason that I want to run modular toolhead controllers.

Consider this: any kind of extruder will report, when asked, max & min extrusion size. When the main controller requests an extrusion size, the toolhead will report back a required toolhead speed, or range of speeds. The controller will return a requested speed, then the toolhead will set any operational parameters that are necessary to achiive that speed (extrusion temperature, etc.) and when it's ready, it will indicate that to the controller. There are all manner of other useful parameters that could passed, like time to cool per unit area, which allows for hardening between layers.

Why is this important? It completely decouples the controller from what kind of extruder it's using. That means that if you want to build an extruder for some strange new substance, you don't need to redevelop the code for the controller, just the toolhead. Furthermore, this makes testing a lot easier. You don't need to re-qualify the whole platform, just the toolhead. It also forces the interface to the toolhead to be well defined. This is of critical importance if we're talking about a multi-toolhead setup, because it makes the project maintainable.

The other part of the reason is cabling. The toolhead is a dense I/O region. It takes a minimum of 10 I/O lines to do a complete thermoplast extruder.
(3xSPI for thermocouple sensing, motor PWM, motor Direction, 2xquadrature feedback, 1xheater pwm, 1xfan pwm, 1x fan tach). Running a toolhead controller requires 2 power + 4 data for full duplex RS-485. What I have in mind is using SATA cables for the RS-485, so it's actually only 3 cables.

Perhaps my goals have swollen past what is useful for reprap. That may be. But I believe that yours have as well. My primary goal now is extensibility. I believe this is the key area where reprap needs to improve. I believe that toolhead controllers are an integral step to making that happen without having the codebase for the controller swell to unmanageable proportions.

I can't say I disagree with anything there Annirak,
But, I do want a base machine that can handle the common tasks with ease, and have room for expansion. With what I'm planning, I think that most common extra tasks can be implemented with existing I/O, and tasks of your nature can be done by extension.

The SATA cables sound like a good idea, however you may find that they don't have enough strands for the required flexibility anyway. I have to admit, I've been more focussed on electronics lately, and put the whole cable routing issues aside for the moment. I had contemplated IR for communications, but power has to be routed anyway.

I haven't checked your design for this.. but if your ground breaks, hopefully your signals will be protected from damage.

Part #2
I've just looked up "EDM" to see what the term meant, also known as "spark erosion" Yep, I'll have one of those too Annirak.
[en.wikipedia.org]

I like making radio controlled Jetboats, and I want to have betterfastercooler jet pumps, so far I'm making them by hand, and better quality, better performance is a big driver for me to make my own CNC/reprap machine.
I'm also interested in creating moulds for injection moulding, so that seems like a fair addition for my machine too.
This is the before CNC jetboat running:
[nz.youtube.com]

Graham

Sorry guys....

EDM is Electro Deposition Modeling

(Imagine having a tool head which is a mig welder torch and building up models in molten weld..... This is EDM)

As opposed to the plastic thing we do which is

FDM, Fused Deposition Modeling

ie fused by heat alone.

I am waiting for someone to name the syringe and UV cure resin thingys

CDM, Chemical Deposition Modeling <|

Oh and Additional tool heads should include, Touch Probe/Pen Holder and perhaps other methods of digitizing objects. Perhaps even when the technology for Solid State Lasers catches up a laser cutter/engraver toolhead.

Cheers

aka47

PS Actuator technologies we could embrace other than stepper drives are to name a few Linear, DC Brushed Servo, Brushless Servo, Servo Pneumatic and Servo Hydraulic.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Have you got any bookmarks to UV cured extrusions Andy ?

I can see that if you had the viscosity right, then the tool head could UV cure as it goes.. very cool.

I'm thinking expensive extrude material though...

Graham

Hey Andy,
Hand-wound linear induction motors are on my list of experiments once my cartesian frame is built. This is part of the reason that I've started work on rebuilding the stepper motor controller. I could see it being used to drive a 4-phase LIM provided that it can deliver higher currents.

Brendan

Hey, neat to see some crazy ideas bubbling away over here.

Two things to say:

1) The choice between a master controller and a nodal architecture is really interesting. A central controller chip would be good for the motion control work (XYZ, R, theta...) because to get good motion all of these must be carefully synchronised, and comms lag could bite us badly. However, we don't want too many lines going up to the extruder from a reliability point of view, especially if there are plugs involved. Therefore I suggest a high power central control chip which handles motion control etc. and secondary processors for the extruder heads. This also means that any tinkering with the extruder won't stuff up the main controller programming.

2) People seem to be getting a little off topic - Beyond the Sanguino - maybe we should fork this?

jbb

JBB,
Point #1,
I agree with, with the proviso that I think the basic board should actually do all the standard reprap tasks by itself.

Annirak is working on the slave devices side, and I haven't put as much thought or effort into it as he has.

I have, however changed a few pin assignments on my list of designated outputs for the STM32F103VBT I'm planning to use.

Where we're not completely in agreement, is in the demarkation line between main board, and non essential extras.

I think Annirak was eyeing up a cheaper processor, I got some of the STM32 for a good price already, that suit me just fine. That was months back, and now STM have released versions with 64k ram, 512k Flash, in the same packages, + new ones also.

Point #2,
I agree, so I think it's best if I start a build thread for my STM32F103 version, for those who are interested. Working on it now...

Guys

A quick word on model/development and path.

There is no one size fits all.......

It is quite rare to get a disparate group like this to agree on a single course of synchronized action. It will always fragment into areas of special interest.

This is actually very good and incredibly useful, no input is ever wasted and all development is ultimately playing a part in "Breeding" the most fitting solutions. (Darwinism, chortle, pun definitely intended)

Diversity is necessary to create the hybrid vigor essential for the survival and optimization of the organism/s or in this case Electronics, Software & Mechanisms that characterize the project.

Whether a solution (of the many possible) is "the" or "an" optimum solution will be played out into the future (time will tell, my balls are definitely not crystal), you should not let this discourage you from trying out your ideas whether they appear to fit an apparent consensus or not.

All in all go for it and see how it turns out.......

I guess the same can be said of those with a perennial preference for CAN.... Go for it and see how it turns out......


A quick word on UV resins.

There is a well explored thread on the forum's that went into UV resins in great depth. I think frequent contributors to this were (amongst others) Enrique and Viktor. This is really not an area I can contribute much if anything to. Please have a rummage for this thread it is very much well worth the read as it covers everything you could ever want to know on the subject, and more. If you have difficulty finding it (I can't remember now which topic it kicked off under) drop Viktor a quick PM.

Hope this all helps

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.