Sunday, April 19, 2009
First reprapped circuit
Ed and I have a final-year student - Rhys Jones - who's working on RepRap for his MEng research project. He's been taking the old idea of depositing metal in channels and an observation of Forrest's and Nophead's (that you don't need a low-melting-point alloy because the specific heat of metals is so low that they shouldn't melt the plastic anyway).
He's adapted the pinch-wheel extruder to drive ordinary solder (without core flux - that just makes a mess) down through a thin tube with a nichrome heater wrapped round it.
Here's his first circuit. It was made entirely automatically by the RepRap machine. It's a bit blobby, but...
It's easy to put the components in from the other side. It's the RepRap opto endstop circuit.
Finally, here it is fitted to the RepRap machine that made it and working:
Labels: reprapped electronics
I've been wanted to try this for a while, I'm so happy you all have gotten it working already.
I think that this is a serious cross over point for the project. And I can't be happier
If surface mount components are placed bottom up first on a base layer, then a layer of plastics with channels in it added, and then solder extruded in the channels, we could make a whole circuit in one pass!
Although, the question is how accurate the resolution is / can be made of the solder deposition approach.
I'm getting close to starting my senior year Mechanical Engineering project and I look forward to reading Rhys' paper on this. Again, congrats on the progress!
>You just make the circuit board with holes for the components, stick the components through, bend the leads or something to keep them in place, and then flip it over and add the solder?
That is the next step. For this test I set the extruder to add the solder into the channels, then once complete used a soldering iron to melt the solder, whilst inserting the components.
If you're interested, there is a video of the extruder in action here
Just to clarify that video is running at 3x normal speed. The actual circuit took about 3 minutes. I reckon the extruder in its current form can produce around 100mm of track in a minute. Still its by no means slow.
Now it's going to be interesting to see if it's possible to reliable print multiple layer boards by printing plastic with channels followed by eutectic followed by more plastic with channels and wells, and so on. :-)
Very well spotted. The fire cement surrounding the tube just came into contact one of the "blobs" in the next channel and nudged the entire board. I intervened and rotated the board so that it would fill the last remaining channel.
On a side note I think the issue of the blobs are the biggest problem with the design at present. Firstly because of the problem above, and secondly I don't think it would be possible to have the multilayer boards Forrest mentions without the solder having a consistent (and predictable!!) cross section. I've have a few ideas that may solve the blobs but unfortunately I now have the small matter of a dissertation to write in the next ten days.
Also I cannot take all the credit for this work, Adrian and Ed have been extremely supportive and I wouldn't have gotten anywhere near this far without their help.
As an aside, coincidentally that circuit was produced exactly two years to the day after Adrian started building the very first Darwin machine
And good luck on your paper Rhys. I did my final year research project about 2 years ago (in chem eng though, not mech) and I know that if you're done with the actual work a few weeks ahead of when it's due then you're way ahead of most people.
I agree that the circuit shown isn't going to win any prizes on the quality front, but I'd like to point out that I originally designed this extruder for channels double the size used in the circuit. The quality of these were much better. (Pics are here. Bad news is, the channels were too big to produce the optoswitch circuit, the pins of the H21LOI chip are simply too close together.
I managed to get the extruder to extrude into these smaller channels by upping the axis speeds. But the nozzle is now larger than the channels, and I don't think this is helping on the accuracy front. I did try to implement a smaller diameter solder/needle, it failed immediately due to buckling of the filament. This could of course be improved by reducing the distance between the pinch wheel and the extruder head, but alas I ran out of time. Another benefit of the smaller nozzle is that the effects of surface tension will be reduced, and should result in accuracy being much better
Ultimately, it would be nice to have a wire bonding head. Unless that's extremely difficult, it may be wortwhile to skip surface-mount entirely and go straight to chip-on-board (or, more precisely, chip-in-bulk).
Thats exactly what the plan is. We should probably work a bit more on the consistancy 2D wise first. But once that is done the goal is to go a bit further than 3D circuit boards. The goal is to have the circuits contained within the structural components of the machine. Thats a long way off though:D
Were the channels strictly necessary? It seems that if you blow cool air at the solder it might just hold together.
I must say, I assumed they were needed from the beginning (based on prior student's work with field's metal). However, I started off with channels with a rectangular cross section, and I didn't feel that the track was as secure as it could be. So the tracks are actually shaped like our logo ... a tear drop, tapering at the top so the solder cannot be removed once the channel its filled.
That being said, if we cranked the temperature up, it might be possible to allow the solder to melt into the ABS forming its own channel. God knows what this would do the shape of the ABS (it does warp a tiny bit after extrusion now)
Also, for some circuit boards, you may be able to draw circuits by drawing solder bridges between a closely-spaced grid of copper pads. They sell through-hole boards in Radio Shack that have copper around each hole. I don't know if the copper is close enough to make reliable solder bridges, but if so, you might be able to use this as a template to keep the lines more consistent (and also use less solder).
Just my $0.02, but would it be possible to use silver solder/brazing material instead (with a higher temperature heater, of course). I would think this might alleviate some of the problems with the limited electrical conductivity of regular solder as the solder wire is made from a mixture of silver and copper with some zinc. I was able to find some small cartidge heaters at McMaster-Carr (http://www.mcmaster.com/#cartridge-heaters/=1nnhq0) that can go up to 1400 degrees Fahrenheit starting at $27.35. They also carry the solder/brazing material (http://www.mcmaster.com/#7761a13/=1nni54). In fact, one of the solders they have is a 56/22/17 mix of silver/copper/Zinc with a melting point of 1145 - 1205 degrees Fahrenheit which might be ideal. I'm sure that both the heater and the solder can, probably, be found for a little less elsewhere as McMaster tends to be a little expensive.
My only warning for you, or anyone else that may decide to experiment with this, is that some of the silver solders out there have Cadmium in the mix, which is very poisonous. I recommend looking for an alloy that doesn't have any.
If you would have a nozzle which is not pointy but flat at the bottom, but does contain a small orifice. The bottom should have some film over it that the solder doesn't stick to. I'm immediately thinking of some nichrome film, since we all know solder and nichrome heating wire do not mix...
Perhaps the nichrome could even be used to guide the solder and heat it at the same time.
Sorry if this is a dopey question but why do we need to involve melting? Why not use fine metal power is a sort of plasstic coating so it acts as a wire carrying the signal?
For 3D circuits, if you can't make entirely get rid of the blobs, wouldn't it be possible to build plastic bridges over the traces without actually contacting the trace (I know you can do some degree of overhang), and thus hide the unevenness?
I'm not entirely sure on this one to be honest, I did think about it myself some months ago. I think it may work, but there would be alot of problems that need to be solved. Having a really fine powder will mean its very hard to control. Equally at some point were going to have some holes in the board to attach components to the circuit,we could run into problems with the powder escaping. Also, how would you solder to it??
>wouldn't it be possible to build plastic bridges over the traces without actually contacting the trace?
Thats the plan, in fact what would it matter if the plastic came in contact with the trace, if anything it would ensure the trace is held securely...I must admit I think a bit of a redesign is required to solve the problems rather than tweaking settings. In short, I think the diameter of the orifice needs reducing. The problem is, we can't machine anything out of brass (brass appears to be slightly soluble in solder - ok for normal use but if your extruding meters of filament the nozzle slowly disappears ), This extruder was based on a stainless steel needle. The problem with that is the filament needs to be a smaller diameter than the needle, I did try reducing the orifice and the filament a bit more but ran into loads of buckling problems....I'm hoping to come back to this at some point once we have a head changer.
Doesn't teflon stand up to molten solder? Could you use a teflon-lined brass orifice?
Thats true, but I have a suspicion that as the infill isn't 100% solid it will self itself out after a few layers
>Doesn't teflon stand up to molten solder? Could you use a teflon-lined brass orifice?
To be honest I don't know. If I was going to have another go I think I'd try to hack something together based around a vented cap screw...
make two repraps :
1 - one with a laser head capable of vaporising copper.
2 - one that deposits reflow solder
then use the number 1 to "etch" the virgin copper board with your circuit design.
the user gets the board and sitcks all components into it (using glue)
them places it into the second reprap that drops smalls droplets of reflow solder into each component
the plate goes back to the machine number one, that uses the laser (at a lower power) to melt all solder on the board.
You'd probably do better putting a Dremel-like tool on the head and physically cutting the traces. There are PCB-prototyping machines that do this.
Or just deposit a mask on the copper (either a Sharpie or melted plastic, if it'll stick well enough) and chemical etch the copper.
Do you think you can pull it off?
And with components - this is simplest problem. Is need to have a holes, with height equal to height of component, with pin separators. And after component installation - simple add an solder in 'filling' mode. After this, we add another plastic layer, and can place a next circut and next component level above installed.