Reprap style SMD options

I've been batting around ideas relating to SMD. As most of you are well aware, SMD is where the majority of new technology goes for a variety of awfully good reasons. This does leave hobbyists in the dark, unfortunately, as soldering 0.5mm pitch pins is not particularly easy, especially with home etched or milled boards that have no solder mask.

This brings up the topic of how to place and solder SMD parts. The task can be broken down into three stages: Paste application, part placement and soldering

1)Solder paste can be distributed with a paste extruder head.

2)Parts can be placed using a vacuum wand. But therein lies the first problem: How do we get the parts in range of a vacuum wand? A conventional pick&place machine has a complex reel dispenser system that the wand can access to place parts. While this is a highly effective method, it poses two problems for Reprap: a) the technique is complex and expensive. b) the reel dispenser would likely take up 4-5x the space of the Reprap itself.

These points say to me that we need a small-scale, innovative solution to the problem.

3)The SMD oven technique is good, and with the aggregate experience that we have as a community, it should be no problem to build up a standard method for SMD oven work. I would recommend a modified toaster oven approach, where we control the temperature in a toaster with third-party electronics to produce a good SMD soldering profile.

There is another, more Reprapy (is that even a word?) option. We could produce a heat wand toolhead which could do point soldering. This would be good for most SMD packages. BGA would require a fairly large wand, but should still be doable. There are some issues with this technique: Too much airflow would obviously cause parts to drift, etc.

Thought I'd get the ideas out there and get some feedback while I work on my granule extruder.

With all due respect, let me disagree with several points in your presentation.

> As most of you are well aware, SMD is where the majority of new
> technology goes for a variety of awfully good reasons.

SMD was invented to facilitate smaller printed circuit boards and, more importantly, to facilitate the use of industrial robots to make pcbs. You invest in industrial robots of the sort that are made these days because you want to make a lot of something on a production line. This is, not to put too fine a point on it, the antithesis of the whole idea of Reprap.

We can undoubtly make an open source pick and place machine by leveraging what we've already done with Darwin and other reprap machines. The important question is, however, not whether we can do it (we can), but rather WHY we would want to do it?

What is the point of some guy who's building his Darwin or Mendel or whatever having to leverage his printer as a pick and place machine when at most he is going to replicate it either for export or spares a few times per year?

When he initiated the Reprap project in 2004, Adrian Bowyer rightly referred to things like stepper motors and IC's as "vitamins", that is, an irreducible minimum of outsourced material that you had to obtain to make a Reprap machine.

In that regard you might want to review the background page for the Reprap project...

[reprap.org]

It's all still there.

Here's the situation. Right now, a Darwin weighs 14-16 kg, depending on the version you put together. The printed parts comprise approximately 1.4 kg. That is to say that we are currently replicating at a rate of about 10%. That's a great start, but nowhere near where we need to be.

Another way to look at things is to talk about the percentage of replication by value. Not that long ago, printing a set of Darwin parts would cost you about $2,400. We used to be quite proud that we could basically bring that big, indigestible lump of the cost of a Reprap machine down to effectively the cost of the ABS. Here in the states you are talking about that much ABS costing in filament form about $20. Right now, since various people went for laser-cut acrylic, a nasty habit we picked up from the fab@home people {bolted, laser-cut acrylic looks really flash}. That's brought the "printed" parts baseline to about $200 or so.

One way of enhancing replicability is to simplify the design. eD has made a good start at that. His design, I'd guess, should reduce the weight of a follow on to Darwin by a third to half.

What you've got to ask yourself, though, is whether 15-20% replicability is really what we ought to be shooting for.

It is SO easy to just outsource more and more of Darwin to factory-friendly technologies like SMD. The question is, however, is that the right thing to do or ought we be, instead, racking our brains trying to get that "vitamin" percentage down?

---

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

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

I know the feeling with SMD! It seems that all the cool toys are coming out in SMT packages now - either TQFP or (shudder) BGA. I've hand soldered of SOIC (rectangular with leads on 2 sides) and TQFP (square with leads on all sides) myself with OK results, but it wasn't exactly easy.

Forest raises a good point: pick and place is hard and how many people will need it? As a community effort one could find an accomplice to hand solder the few SMT parts that sneak in.

On the other hand, if you feel that automated PCB assembly is a good idea, read on...

The boys over at Sparkfun [www.sparkfun.com] have had a go at this before, and present a variety of options [www.sparkfun.com] . A hot wand is a possibility, but I like the look of their hot plate - very simple!

A pick and place head would be very neat, but as you need to get within half the pin spacing for the chip to solder correctly, a big TPFQ with 0.5mm pin spacing would require

Edited 2 time(s). Last edit at 01/19/2009 03:49PM by jbb.

Annirak Wrote:
-------------------------------------------------------
> 1)Solder paste can be distributed with a paste
> extruder head.
Ok, possible.

> 2)Parts can be placed using a vacuum wand. But
> therein lies the first problem: How do we get the
> parts in range of a vacuum wand? A conventional
> pick&place machine has a complex reel dispenser
> system that the wand can access to place parts.
> While this is a highly effective method, it poses
> two problems for Reprap: a) the technique is
> complex and expensive. b) the reel dispenser
> would likely take up 4-5x the space of the Reprap
> itself.
Why not use a normal vacuum cleaner?

> 3)The SMD oven technique is good, and with the
> aggregate experience that we have as a community,
> it should be no problem to build up a standard
> method for SMD oven work. I would recommend a
> modified toaster oven approach, where we control
> the temperature in a toaster with third-party
> electronics to produce a good SMD soldering
> profile.
This is the part where the alarm bells ring... oven means heat, heat meats melting, melting means higher error margin...

I think you can use an oven, but you need to be careful not to get too hot for too long

Forrest Higgs Wrote:
-------------------------------------------------------
> With all due respect, let me disagree with several
> points in your presentation.
>
> > As most of you are well aware, SMD is where the
> majority of new
> > technology goes for a variety of awfully good
> reasons.
>
> SMD was invented to facilitate smaller printed
> circuit boards and, more importantly, to
> facilitate the use of industrial robots to make
> pcbs. You invest in industrial robots of the sort
> that are made these days because you want to make
> a lot of something on a production line. This is,
> not to put too fine a point on it, the antithesis
> of the whole idea of Reprap.
>

Frankly Forrest, that was unnecessary and counter to the community concept of Reprap. You didn't pause to consider why I would be wanting SMD, you simply assumed that I had the same goals as you, but was "approaching them wrong."

My goal is not to increase the replicating ability of the project. While a fine goal, it doesn't really serve me well. I am, however, decidedly interested increasing the capabilities of Reprap. Just take a look at the topics where I am most active, and you get some idea of my approach to the project: Granule extrusion or filament manufacturing, Conductive plastics, Electronics. None of these make Reprap more replicable. All of them make Reprap more useful.

When Reprap was released to the community, the spirit in which it was done was that anyone could use it for anything they wanted. Your focus seems to be that replication is the only important goal. In taking this approach YOU have gone counter to the spirit of the project. Perhaps fab@home took this more seriously considering that they have videos of chocolate fabbing. One thing that I want to do with Reprap is fab SMD electronics. Is that inherently wrong? Counter to the project? I don't think it is. Particularly because Mendel is supposed to have circuit fabbing capabilities. I am simply trying to develop that goal in a somewhat more advanced way. I never said it should be included into the base release of the project, just as the solenoid cut-off valve is not part of the base release. I do think that there would be quite a number of people who would find that SMD soldering was a very valuable addition.

I am entirely correct in stating that new technology is released almost exclusively in SMD. I am correct in stating that there are good reasons for this. They may not be our reasons, but they are still good reasons.

I went to try and create a better thermocouple reader for Reprap. There is a solution available which is more accurate, smaller, cheaper than the existing one, and provides two input ports, not just one. Guess what? It's SMD, and in a package which isn't hand solderable.

As you have said, chips are essentially vitamins of Reprap. You can't practically break the parts down any further. That is why I'm trying to do SMD soldering rather than silicon fabbing.

It just so happens that my motivation for a SMD soldering solution is that I want rapid prototyping capability for electronics. Regardless of whether this is inline with the goals of Reprap, it is a valid approach and if all it takes is a few toolheads, one of which is already being built anyway, it's a very useful addition.

jbb Wrote:
-------------------------------------------------------
> Forest raises a good point: pick and place is hard
> and how many people will need it? As a community
> effort one could find an accomplice to hand solder
> the few SMT parts that sneak in.

I can solder most SMT parts. I get stuck on BGA and anything with a powerpad. I realise that hand-pasting and a heat gun could get by this, but I don't think it's a good solution. A paste printer at the very least is necessary. We don't really need pick/place, and we don't really need a proper oven, but both would increase on the reliability and repeatability of the process.

> A pick and place head would be very neat, but as
> you need to get within half the pin spacing for
> the chip to solder correctly, a big TPFQ with
> 0.5mm pin spacing would require

Annirak Wrote:
-------------------------------------------------------
>
> Frankly Forrest, that was unnecessary and counter
> to the community concept of Reprap.
>
Nonsense.
>
> You didn't
> pause to consider why I would be wanting SMD, you
> simply assumed that I had the same goals as you,
> but was "approaching them wrong."
>
No, I didn't. I was criticising the direction that you advocate going in.
>
> My goal is not to increase the replicating ability
> of the project.
>
I gathered that.
>
> When Reprap was released to the community, the
> spirit in which it was done was that anyone could
> use it for anything they wanted.
>
Indeed.
>
> Your focus seems
> to be that replication is the only important goal.
>
From the background page...

"The self-copying rapid-prototyping machine will allow people to manufacture for themselves many of the things they want, including the machine that does the manufacturing. It is the first technology that we can have that will simultaneously make people more wealthy whilst reducing the need for industrial production."

What part of that don't you understand?
>
> In taking this approach YOU have gone counter to
> the spirit of the project.
>
Try not to invoke the "spirit" of a project the goals of which you don't appear to have much interest in. It's kind of offensive.
>
> Perhaps fab@home took
> this more seriously considering that they have
> videos of chocolate fabbing. One thing that I
> want to do with Reprap is fab SMD electronics. Is
> that inherently wrong?
>
Nope.
>
> Counter to the project? I
> don't think it is.
>
That is certainly an opinion.
>
> Particularly because Mendel is
> supposed to have circuit fabbing capabilities. I
> am simply trying to develop that goal in a
> somewhat more advanced way.
>
That assertion, to quote the late Werner Heisenberg, isn't even wrong.
>
> I never said it
> should be included into the base release of the
> project, just as the solenoid cut-off valve is not
> part of the base release. I do think that there
> would be quite a number of people who would find
> that SMD soldering was a very valuable addition.
>
I am sure of that.
>
> I am entirely correct in stating that new
> technology is released almost exclusively in SMD.
>
Yup, Zach likes it that way and Zach heads the RRRF which flogs pcbs. That doesn't mean that Zach's preferences are consonant with any goals save his own.
>
> I am correct in stating that there are good
> reasons for this. They may not be our reasons,
> but they are still good reasons.
>
That's debateable.
>
Okay, Annirak. Don't get the idea that I'm coming after you personally OR that I have anything in particular against SMD or even pick and place machines.

What I DO feel is that Reprap is starting to seriously lose sight of it's original goals. The announcement of the Rapman release...

[www.unimatic.co.uk]

last week made it impossible for me to ignore just how far astray we'd got.

I'm squawking about it and can be depended on to be nailing my thesis to some doors around here till things starting getting a bit back on track.

---

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

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

Hi Annirak,

I think it's a great idea to make it possible to use SMT.

Thanks to John Meacham's pick and place head type thingy:
[forums.reprap.org]

we know it is straightforward to add a vacuum head. Also, from JBB's excellent sparkfun links, you can start at:
[www.sparkfun.com]

and get a good idea on the steps and equipment required.

For prepositioning the components in small quantity, we could fabricate trays with custom insets for each component. Once you insert all the components into the tray, the vacuum head would blindly pick the component from the tray and deposit it on the board. Normally custom trays would be expensive, but since we have a reprap they'll be dirt, or at least plastic, cheap

For the pick and place gcode, we could start with a custom test script. Then once I have finished some skeinforge extrusion code, I'll try to integrate pick and place code into the skeinforge tool chain.

I saw a video of a CNC machine doing pick and place.

It simply picked the part up with a vacuum hose and then moved to a square hole where it bumped each wall to move the part to the exact center and squared it up before placing it.

In another video of a different machine it would move the component over a camera where the vacuum head would then rotate to square it up.

The boards we make are not that complicated maybe 10 to 20 different components?

So having sections of tape on the side might be doable. Heck we can certainly RepRap a guide for them to slide in!

Maybe we simply need one user in each major tariff/customs area that can do the feeding of a machine?

Buy the "vitamins" locally and have base PCB boards produced in bulk and shipped around the world?

How long before the Sanguino motherboard is ordered in 5000 quanity once a quarter?

I have kind of wondered why the stepper motors are not removable so the electronics and motors can migrate around the shop. Maybe during the day attached to a foam cutter and at night printing out parts?

I do agree that we should advoid SMD as long as we can, but once something has been stablised and could be consided a "vitamin" then let it, while continuing to publish alternate paths and DIY info.

While I agree replication percentage is a very desirable goal, sheer volumn of users using these machines in open source way is going to generate a lot of users around the world.

We may reach the point where we say "Salvage stepper motors from HP Laser xxxx and side mirror controller from Ford Pickups 1997 though 2004" to get most of the electronics. (Note totally pulled out of my hat, but the mirror controller does have XY movements and a heck of a lot of components to handle that on it's PC.

Good thinking to suggest a custom tray, Enriquie - that way it will have precisely controlled dimensions.

Bumping the components up against a guide to slide them into place is also a good thought thanks Freds - it might not let us get down to

//The announcement of the Rapman release... last week made it impossible for me to ignore just how far astray we'd got. //

I don't understand this.

We have a group committed to improving Reprap's replicability (say those last two words five times fast..!)

We have at least two vendors who are committed to simply distributing repstraps.

What's the problem? How is Unimatic/BfB interfering?

Edited 1 time(s). Last edit at 01/19/2009 07:29PM by hardness.

Quote

A further thought - we will be bringing a chip directly into contact with the PCB, and chips have different heights... we need some way of detecting when the device and landed and to stop pushing.

We know in advance how high the parts are so we can just program the machine to push them into the paste at the correct depth. They don't get pushed all the way to the board, some paste is left underneath.

---

[www.hydraraptor.blogspot.com]

Annirak,

I'm new here, but I'd like to encourage you to continue your research on this. Like you, I'm interested in this project because of all of the *possibilities* it could have in revolutionizing everything we all do. And for me, that includes SMT. I'd love to see DIY Laser cutters, SMTs, pick & place, CNC, etc. I may only build an extruder, but I'd love to *be able to* build other heads.

I realize that Dr. Bowyer's goal is to get as many Darwins going to achieve critical mass, but the more adaptable the machines are, the more mass appeal they may have.

I follow the blogs, relying heavily on the works of nophead and Forrest. I find it interesting that they are two of the most prolific posters, yet neither of them (to my knowledge) operate a Darwin. Forrest is making great advances in the whole movement, which Mendel will likely benefit from. He's doing this on Tommelise though, not Darwin, and his latest posts are about CNC subtractive processes for gear making, and references to Tommelise 3.0 (suggesting he is continuing on that path). That doesn't seem to be adding to the strictly-Darwin knowledge set, but to me, that's okay.

I look *UP* to Tommelise and HydraRaptor. I'm a bit disappointed that one of the people successfully using deviant technologies would criticize others who want to do the same. Although I understand Forrest's position, I don't necessarily agree with it. Even the choices of names, Darwin and Mendel, invoke thoughts of diversity, specialization, and survival of the best-mutated designs. Why else would Dr. Bowyer choose to name the devices after a geneticist and the father of evolution? IMHO, going off on these tangents is exactly what I'd like to see the "spirit" of reprap.org to be, as long as the core project thrives.

Randyy Wrote:
-------------------------------------------------------
>
> I look *UP* to Tommelise and HydraRaptor. I'm a
> bit disappointed that one of the people
> successfully using deviant technologies would
> criticize others who want to do the same.
>
Let me be clear if it is already not that way. I am not attacking Annirak and I am not saying that SMD is a defective technology.

Nor am I saying that SMD and pick and place, two areas that interest Annirak are not how he should be spending his time. The Lord knows, I've been tempted to build a pick and place machine often enough.

What I AM saying, however, is that if the goal is to create a 3D printer/milling machine/whatever that exhibits a high degree of self-replication and can be built by an ordinary person, SMD is most definitely NOT the direction we ought to be going.

I got onto the Reprap project back at the beginning of 2006. The universal Reprap design criterion of using 12v direct current was set in order to let somebody in Zambia run one off of car batteries. Think about the implications of that mindset.

The point of Reprap was, and for me still is, the creation of a fabrication technology that people-friendly and can be replicated without a lot of outsourced vitamins and minerals, NOT a technology that can much more easily be made by a factory in Guandong province, shipped all over the world and bought in Walmart.

What is upsetting me lately is that we are spending more and more effort making reprap what I will call HP-friendly and less and less time making it revolutionary in the sense of best being produced by individuals without a lot of capital behind them.

It appears, though, that a big chunk of the Reprap "community" with all it's "spirit" which I am offending has no interest in bringing about a meaningful technological revolution whatsoever, only in designing a new consumer product.

Edited 1 time(s). Last edit at 01/19/2009 10:20PM by Forrest Higgs.

---

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

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

You know, fine pitch SMD stuff is invariably a pain for hand soldering, but big, like 1208 components, are pretty easy to hand solder for most folks with a technique or two.

Using 1208 0 ohm resistors to jump tracks instead of vias, and making everything else surface mount - ie, splay out the pins on a DIP socket, and surface mount it - this could make our boards a one sided no-through hole operation. Something that could be reprapped in a single process, either with a Dremel cutting Voroni isolates or a resistive pen head and an etchant bath. No holes means no two step cutting/drilling processes, and means our absolute positional accuracy can be right out the window and we'll still make perfectly useable boards, as long as we leave some margin at the edges. Those with tighter machines and patience can try for smaller and smaller chips, but the rest of us should be able to get by on 0.1" centers.

Single sided, easily reproducible PCB's could go a long way towards easing self reproduction and rapid (really rapid!) prototyping... The stuff we're playing with right now is dead simple PCB wise; if making PCB's was really easy, it might become more popular.

I've got the same feeling regarding plastic part production, btw - it's got to get a lot easier before it's going to start happening on a wider scale. That'll happen, and it is happening right now, but it's going to take some time and development yet.

Wade

> What I AM saying, however, is that if the goal is
> to create a 3D printer/milling machine/whatever
> that exhibits a high degree of self-replication
> and can be built by an ordinary person, SMD is
> most definitely NOT the direction we ought to be
> going.

I can see where you seem to be coming from, but for the purposes of the mainline RepRap using SMT technology after having enabled the machine to use it itself would fit very well with the ability of the ordinary person to use it. It would mean that the ordinary person with a machine now has the capability to fabricate circuitry with modern chips (which often /have/ to be surface mount to work at the frequencies they use) easier than a random person using DIP, because all they do is put trays in slots on the machine and say "go". It would be much harder to make a RepRap machine directly use DIP. DIP is only terribly important for RepStrap machines and other designs focused on non-replicating production.

SMD is primarily for machines to use, we /have/ a machine we want to use, and it makes everything else simpler and easier, so SMD makes a huge amount of sense.

Wade Wrote:
-------------------------------------------------------
>
> I've got the same feeling regarding plastic part
> production, btw - it's got to get a lot easier
> before it's going to start happening on a wider
> scale.
>
And you are DEFINITELY the guy who ought to know.

Let me tell you. I can do milling on Tommelise pretty well now. It's going to be some time, however, before it becomes easy enough for a 12 year old to try. I've REALLY got to do a lot of idiotproofing of the steps that you have to go through to mill something before it's going to be ready for prime time.

Edited 1 time(s). Last edit at 01/19/2009 11:28PM by Forrest Higgs.

---

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

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

jonored Wrote:
-------------------------------------------------------
>
> SMD is primarily for machines to use, we /have/ a
> machine we want to use, and it makes everything
> else simpler and easier, so SMD makes a huge
> amount of sense.
>
From the Wikipedia article on SMT

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


MAIN ADVANTAGES

Smaller components. Smallest is currently 0.5 x 0.25 mm.

Much higher number of components and many more connections per component.

Fewer holes need to be drilled through abrasive boards.
Simpler automated assembly.

Small errors in component placement are corrected automatically (the surface tension of the molten solder pulls the component into alignment with the solder pads).

Components can be placed on both sides of the circuit board.

Lower resistance and inductance at the connection (leading to better performance for high frequency parts).

Better mechanical performance under shake and vibration conditions.
SMT parts generally cost less than through-hole parts.

Fewer unwanted RF signal effects in SMT parts when compared to leaded parts, yielding better predictability of component characteristics.

Faster assembly. Some placement machines are capable of placing more than 50,000 components per hour.

MAIN DISADVANTAGES

The manufacturing processes for SMT are much more sophisticated than through-hole boards, raising the initial cost and time of setting up for production.

Manual prototype assembly or component-level repair is more difficult (more so without a steady hand and the right tools) given the very small sizes and lead spacings of many SMDs.

SMDs can't be used with breadboards (a quick snap-and-play prototyping tool), requiring a custom PCB for every prototype. The PCB costs dozens to hundreds of dollars to fabricate and must be designed with specialized software.

------------------------------------------------------------------------------
A good read of the whole Wikipedia article on SMT rewards the effort.

[en.wikipedia.org]

SMT is designed for large scale, serial production environments. It's a pig to prototype with, which means that board design is going to happen where there is capital concentration adequate to buy the equipment to get past the awkwardness of the components for circuit designers. That limits the number of people who can participate to those with significant capital backing. Does this sound familiar? Hint, hint... corporations?

You SMT advocates may indeed be able to scale down this technology to where it is less developer and one-off production hostile. IMO, however, you've got your work cut out for you and frankly, given that it took us three years to get from the Mark II extruder design to where we are now with much smaller technology hurdles I frankly think that you SMT sorts are whistling in the wind.

Edited 1 time(s). Last edit at 01/20/2009 12:07AM by Forrest Higgs.

---

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

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

> SMT is designed for large scale, serial production
> environments. It's a pig to prototype with, which
> means that board design is going to happen where
> there is capital concentration adequate to buy the
> equipment to get past the awkwardness of the
> components for circuit designers. That limits the
> number of people who can participate to those with
> significant capital backing. Does this sound
> familiar? Hint, hint... corporations?

And until recently, playing with FDM-type machines required corporate backing. SMT is done by hobbyists already anyways; we'd just be making it easier and more accessible. It's a much easier task.

> You SMT advocates may indeed be able to scale down
> this technology to where it is less developer and
> one-off production hostile. IMO, however, you've
> got your work cut out for you and frankly, given
> that it took us three years to get from the Mark
> II extruder design to where we are now with much
> smaller technology hurdles I frankly think that
> you SMT sorts are whistling in the wind.

It really isn't nearly as terrible as you make it out to be. There are stripboard equivalents for the not-terribly-fine-pitch components, and I've done fixes on prototypes with 50 mil SMT components, and I'm a CS guy... You do need to solder them, but that doesn't make them impossible to work with.

To really effectively access it without wanting a company other than the one we buy parts from we need to work out a PCB fabrication method that can do alternating conductor and non-conductor at something like ~.5 mm wide each. To do it without requiring people to learn alternate soldering techniques would require a vacuum head, a solder paste dispenser, and perhaps a hot air head.

It's not like we're suggesting BGAs, just QFP or SOIC. Claims of the developer hostility of SMT are overblown, and based partly on needing slightly different soldering techniques to work with them well.

Quote

The point of Reprap was, and for me still is, the creation of a fabrication technology that people-friendly and can be replicated without a lot of outsourced vitamins and minerals, NOT a technology that can much more easily be made by a factory in Guandong province, shipped all over the world and bought in Walmart.

It is cheaper and easier to buy a ready made SMT PCB and ship it across the world than it is to make it from discrete through hole components, how is it more people friendly to force them to but dozens of out of date components locally (at inflated 1 off prices) and mill and assemble a PCB themselves?

For making 1000's of PCBs the same, i.e. RepRap electronics, mass production is by far the cheapest, and unlike say injection molding, the tooling is cheap. The only capital you need is enough to pay for a reasonable size batch.

If RepRap develops SMT assembly capability then it will enable hobbiests to make one off designs using up to date components, a massive enabler. The fact it could also make its own electronics is only important from a completeness point of view. It will still be easier and cheaper to but it ready made because of the discounts of bulk buying components.

This is opposite to the plastic parts. A RepRap can make its own plastic parts cheaper than they can be laser cut and it requires a lot of capital to injection mold them.

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
> The point of Reprap was, and for me still is, the
> creation of a fabrication technology that
> people-friendly and can be replicated without a
> lot of outsourced vitamins and minerals, NOT a
> technology that can much more easily be made by a
> factory in Guandong province, shipped all over the
> world and bought in Walmart.
>
>
> It is cheaper and easier to buy a ready made SMT
> PCB and ship it across the world than it is to
> make it from discrete through hole components, how
> is it more people friendly to force them to but
> dozens of out of date components locally (at
> inflated 1 off prices) and mill and assemble a PCB
> themselves?
>
> For making 1000's of PCBs the same, i.e. RepRap
> electronics, mass production is by far the
> cheapest, and unlike say injection molding, the
> tooling is cheap. The only capital you need is
> enough to pay for a reasonable size batch.
>
> If RepRap develops SMT assembly capability then it
> will enable hobbiests to make one off designs
> using up to date components, a massive enabler.
> The fact it could also make its own electronics is
> only important from a completeness point of view.
> It will still be easier and cheaper to but it
> ready made because of the discounts of bulk buying
> components.
>
> This is opposite to the plastic parts. A RepRap
> can make its own plastic parts cheaper than they
> can be laser cut and it requires a lot of capital
> to injection mold them.

I agree with you. What Forrest seems to be missing is that a lot of people do not want to have to learn to solder, and even if they do it doesn't mean they want to. The ability to prototype is all fine and good, but a lot of people don't want to prototype their boards and so why should they be forced to? Having to produce your own PCBs which would likely involve a number of nasty chemicals is even worse.

Also I fail to see how someone being able to make the board at home, if they are willing to solder, make PCBs etc; effectively spending a lot of time and money they may prefer to use for something else (like actually developing the RepRap in some way) serves the goal of the project. That isn't self-replication. That is human assisted replication. All the parts etc are ultimately coming from the same place as SMD parts. The number of parts (call them vitamins if you want) that are not made by the machine are all the same. The only difference is one case, you're relying on someone who can do it cheaply to do, in the other case your spending your own time and money to do it yourself. Neither case involve any more self replication.


In other words, the goal of the project is for a "machine that can make itself (when it can't we have to do it in some other way until we can)". The goal isn't "the machine that can make itself, and when it can't you have to do it yourself no matter how much more time or money that costs, you're not allowed to do it in any other way particularly if it involves a factory".

For true self replication, we need the machine to be making and assembling the PCBs. As it stands, from what I've seen and read, it seems to me that assembling SMD base boards is far more likely to be possible with a RepRap then assembling through-hole boards. Both have a long way to go, but the first one seems more likely to be successful. So if anything, going SMD is better for the goal of self-replication then through hole.

Don't get me wrong, I'm not saying we should ignore the ability to easily prototype and solder yourself completely. There are reasons why we should preserve that as much as possible. But there are also reasons why a lot of people don't want to get involved in that side at all, and forcing them to doesn't advance the interest of the project. Nor is it people friendly.

Ultimately, IMHO what it comes down to is this. People have interest in the project for a variety of reasons. Some simply want a 3D Printer. While that is not the RepRap's design purpose, it something which is a key part of it and we shouldn't dismiss such people completely because they may ultimately advance the project in some way, directly or indirectly. Many may be interested in advancing parts of it. They may have no interest (or perhaps simply no ability or desire to learn) in advancing others. Forcing them to do something in some way because they can then do it themselves if they want to, even if they don't want to and have no desire to is not likely to make them so include to contribute where they actually can contribute. Preventing them from getting RepRaps because they can't afford it (in time or money) means they will not be able to contribute whatever they could.

In other words producing RepRaps cheaply and easily means we get more out there. By getting more out there, it's far more likely that there are people who can and will, intentionally or not, advance the goal of self-replication. If we are elitist and say 'well sorry we're not going to do that even though it will cost a lot less and make it a lot easier because it will have to be done in a factory (or somewhere else) and we don't want things being made in factories' we're helping no one.

You mentioned in the blog that there are over 1000 RepRaps but so few that have made a complete set of parts for another RepRap. Well firstly, we don't know for sure how many have. It's likely some people have but simply never mentioned it. More importantly, even if that's true so what? Would it be in the interest in the project if the developers and anyone who made a RepRap had spent all their time making RepRap parts? Would it be better if we had 100 RepRaps in the wild but they were all made from RepRapped part? Or is it better that the developers and others can choose to devote their time however they want, improving whatever they want? Is it better that we do have 1000+ people contribution in whatever way they want? And as others have pointed out, producing the plastic parts is one area where we can do it more cheaply with RepRaps then what we are doing at the moment (lasercut). So now that we actually do have 1000+ out there, we can resonably expect that there are more people out there who will be producing RepRap parts. And yes these 1000+ lasercut RepRaps can hopefully mostly produce parts for themselves. They may not be the same parts, but they are compatible. And they can product all the parts for another RepRap if the owners want to, and if someone is interested in acquiring.

For better or worse, anything which gets more RepRaps out there is likely to be a good thing because it means there are more people contributing some of which would surely go towards advancing the RepRap's ultimate goal. Refusing to admit people to the project, i.e. refusing to do something which will help get more RepRaps out there simply because it doesn't fit your view of the project is far less likely to achieve that goal. I'm not of course saying that you personally have to like what people do, or that you have to contribute. I am saying that you should consider it, and I don't think you should try to convince those who are doing it, including the other core developers, to stop.

P.S. A far more interesting conflict, which doesn't seem to have arisen yet, is when we can do stuff more cheaply in a factory, but we can also get a RepRap to do it. PCBs, are an obvious one here. When we actually start making them, my gut feeling is they're going to cost a lot more then the mass produced stuff particularly if we use something fairly uncommon like Fields Metal.

Edited 3 time(s). Last edit at 01/20/2009 06:13AM by Nil Einne.

Wasn't planning to mention this but decided to go ahead. Remember as well that unless I'm mistaken one of the goals is to get these things out into the developing world, particularly those least developed ones. Would this goal be best served by requiring them to have soldering irons, the ability to solder, to make PCBs, etc. Remembering they may not have easy access to a reliable power sources and in many cases won't even know what a soldering iron is? And that they may not have much access to any of this. And that because of the poor standard of health care they may not have good eyesight and suffering burns from soldering iron is a risk best avoided. And do you really thing it's a good idea to mess around with the chemicals to make a PCB etc? Or that it would be cost effective to send them all the stuff? As opposed to just sending pre-made PCBs?

I haven't properly introduced myself around here. I'm an electronics designer. As such, I'm a pretty young guy, only having done design for 5 years. Never the less, I have my finger on the pulse, as it were, of the electronics industry. The only reason that parts are still produced in DIP by many manufacturers is that there are legacy customers making large volume orders. When they start losing money on DIP parts, they will stop making them. Even SOIC parts are becoming significantly more rare, being replaced by QFN parts for small pin counts and QFP variants for the larger ones.

Forrest seems to think that there is some kind of argument as to whether or not to use SMD. There is no argument. People are phasing out DIP and it will become more and more difficult to find as the years go by. This isn't a foolish pursuit, it's an attempt at survival. You see, what Forrest doesn't get is that components aren't made for us, or for individuals (Parallax excluded). Components are made for corporations. The fact that we can still get DIP parts is amazing. That's not going to last. If we don't have SMD manufacturing, there will come a time when we can't build our boards anymore.

I'm attempting to remove barriers to entry into SMD prototyping for hobbyists. I don't really care what the advantages and disadvantages of SMD are. My efforts are purely because SMD normally requires corporate backing. I want to remove that barrier to entry, and I want to remove it simply because all the fun stuff is only available in SMD.

Frankly, I'm sick of this thread. I just wanted to open up the floor to ideas on the construction of a pick & place system. Paste printing is pretty well taken care of by the syringe extruder head, and the SMD-toaster is a well known entity. Instead of a constructive environment, I got a philosophical debate. If I had the tools to lock this thread, I would. I don't think any more discussion along the line which this debate has taken will be constructive. When I achieve a reprap with toolhead swapping, I will try this topic again. If I get the same response then, I'll just take the discussion to the EE section of the piclist. People there are significantly more likely to try and problem solve rather than debate the merit of the project.

Annirak,
I don't think you need to take your bat home yet. All the posters in this thread are supportive apart from Forrest so no reason why the thread cannot continue.

I also have plans to explore SMT assembly with my machine. I have just been too busy with other aspects to get round to it, but I will someday.

---

[www.hydraraptor.blogspot.com]

Annirak Wrote:
-------------------------------------------------------
>
> Frankly, I'm sick of this thread. I just wanted to open up the floor to ideas
> on the construction of a pick & place system. Paste printing is pretty well
> taken care of by the syringe extruder head, and the SMD-toaster is a well
> known entity. Instead of a constructive environment, I got a philosophical
> debate.
>
The first thing you have to understand, Annirak, is that the debate I brought up isn't about the merits of your proposal. As I said earlier, what you propose doing is doable. It is a big job and it will take, undoubtedly, quite some time, but it is most definitely doable and ought to be done. It's been a problem that has been building around the Reprap project for some time.

Personally, I don't think it is healthy for Reprap to concentrate controls design in too few hands. That's why going over to SMT, which tends to do that naturally, gives me the shakes. You, however, have made an excellent case for the reality that the chip industry is going to finish doing that regardless of what the little Reprap project's desires might be.

Right now we have a situation where Zach at the RRRF has leveraged the foundation and considerable of his own resources to effectively get control of boards design for Darwin. Zach is diligent about presenting his board designs for comment by the core team. He is also very good about incorporating suggestions.

The problem with that is that while you can debate the details of Zach's work, you can not question Zach's basic assumptions about how the control system goes together as Nophead, another experienced electronics boffin like yourself, has discovered.

>
> When I achieve a reprap with toolhead swapping, I will try this topic again.
>

Don't wait that long. I hope you will also blog your work here regularly. I'll not interfere with your posting again. I would like to apologise for having upset you as well. That was not my intent.

Edited 4 time(s). Last edit at 01/20/2009 11:14AM by Forrest Higgs.

---

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

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

Forrest,
I got upset because I couldn't see why you were pushing old technology which is gradually disappearing instead of trying to ramp up support for new technology. The "there's no point, no one's going to use this for building more repraps" argument got me distracted from the important point that DIP is going away.

Foolishly, perhaps, I explained my motivations rather than the benefits to the project. This didn't help.

To most of us, your talk of vision didn't really make much sense. I don't understand why someone selling repraps is an issue. It gets more of them out there, which was sortof the point.

When you started talking about SMD being friendly for large companies, I should have twigged on the problem in communication that we were having. I guess I figured that from the vehemence of your arguments, you must have some kind of picture of what the semiconductor industry looks like. I'm going to blame lack of sleep for that one. I'm glad that we've managed to get onto the same page now.

By all means, interfere in my threads. But let's keep it constructive and try to assume that I've got some kind of good reason for saying what I say (not always the case, but usually a fairly good bet ).

I'll just drop a semi-related comment here for you: I'm (where "I" == an electronics enthusiast and a robotics student) using SMT parts whenever possible, for my own convinience. I mean, I can begin soldering a board right away after etching and rinsing it - no drilling! Seriously, drilling is the single most tedious and annoying thing required to make a board, trading it almost entirely for just a slightly more precision-demanding soldering method is a blessing to me. Actually, I couldn't care less about the fact that it's more friendly to high-volume manufacturing, as long as it's more friendly to my own, fully manual methods. And, no, I don't have any "professional" equipment at hand. I'm using an (approximately, those exchange rates are jumping like crazy right now...) $40 soldering station and an $60 hot air rework station, which isn't exacly an expensive setup, but nevertheless I'm just fine with QFP, SOIC, TSSOP and whatever else SMT packages. BGA is a different beast entirely, I admit, but that's still a moot point here, as it would be plain impossible to make a practical through-hole version of most BGA chips anyway. Oh, and even if it could be done, I'd really, really would like to see some of you drilling 400 0,6mm holes in FR4, just for the fun of seeing you going mental with the drill after breaking the n-th $5 bit or lifting the final pad...

I don't think BGA will be too hard to do at home.

You need a heater under the board and a thermocouple on the board, and either hot air or focused IR on the chip and an IR thermometer looking at it. A control system to ramp the heaters.

You also need some means of alignment but it doesn't have to be too accurate as the pitch is generally bigger than TSSOP and surface tension will pull it in.

You don't even need any solder because the balls are actually enough themselves.

---

[www.hydraraptor.blogspot.com]

BGA would be impressive to get to work. Hard, but impressive. In terms of getting parts from the state they come in the mail to positioned on a toolhead, I've detailed my thoughts in the thread in general discussions: [forums.reprap.org]
and I've almost got enough information to get a model of what I'm thinking put together. Cut tape specifically seems to me like it's probably the way to go to load a reprap pick-and-place with parts.

What do the people who work with this stuff more think of essentially fitting a hot air source around a vacuum wand - so that the part can be put on the board, and at least tacked down with solder in the machine, but also kept from being blown away by the air? Is that even a problem that needs consideration? It seems like placing parts dry and then having the user pick the board up and move it would be asking for a mess with no easy recovery mechanism, but I suppose you might be able to get heat to the board without pushing the parts around. Perhaps passing a fragment of nichrome up at red hot over the chip, to do radiant transfer and avoid a risk of flying parts?

I think we have to aim for BGA and QFN because that is the way the market is rapidly going.

They require the board to be heated from underneath to somewhat below reflow temperature and then heat applied to the top to bring the device up to reflow for a few seconds with defined ramp up and ramp down.

You can simply do the whole board in an oven or individual chips with focused hot air or IR.

I think a focused IR beam is nothing more than a projector bulb and some big lenses and a red filter to remove most of the visible light, just to make it easier to look at.

You can buy a manual BGA rework station from China for ~

---

[www.hydraraptor.blogspot.com]