Gathering a parts list of DIP parts for printable electronics

Special note: A printable circuit will probably not be more efficient, or smaller or easier to make or use. Making printed curcuits is more an exercise given the ojectives of self replication. Using the SM electronics from makerbot or others will probably always be an easier method. Designing circuits for printing is alot of work for not too much advantage apart from getting closer to the theoretical replication point. This design will be similar to the Gen 2 electronics(I crimped the old chips from there), only as a single large board for extruder and motherboard each as per the 3rd gen. (though actually the way I'm designing the boards is that there are actually 3 boards that are printed seperately, then stacked with common connection points between layers, looking somewhat similar to 3D chess board, this gets around the problem of double sided curcuits.)

Anyway, now that we've thrown any idea of ease of production of simplicity aside, on to the semi-pointless exercise.

I've been working on gathering together a list of parts for the motherboard and extruder board that are DIP(through hole) as well as SM(surface mount). The current boards work ok (though my extruder board is currently experiencing huge problems, for some unknown reason. Which is what put me down this path) but some of the parts are only listed as available for SM. Usually in the SOIC design. The ATMEGA chips all come in some sort of DIP version, but others do not or there is some slight variation in the name.

The reason I'm looking for DIP over SM/SOIC is because printable circuits should be the objective within the short and long term future for the project. The exact spec for solder printing hasn't been released yet by Adrian (one of his final yr students worked it out, but the needs of marking rules meant he couldn't release the report immediately), but looking at the closeup photos from the blog post I'd guess that the channels are all at least 2mm wide(guessed from the grain in the photos, seem to be about 4 layers across). It's always safe to double to needed precision on a real machine. Thus the best resolution would be about 4mm (DIP chip spaces are often around 8-9 mm, leaving us space for channel seperation). This is too fine for SMT but fine for most DIP parts. Thus DIP is the way to go for printed curcuits until we get the quality way up.

I was originally just going to design the circuits themselves as stl's, but ran into the obvious troubles with replacement chips.

Solder would be used for now as the curcuit conductor. The problem of high resistance in solder is an issue, but for now I'll ignore it.

The biggest problem (and in fact was the reason they went to SM originally a year ago) is replacing the A3949 chip. The old L293D chip has a much lower amps capacity than the A3949 chip(it would blow up constantly, and was a constant headache in my machine last year, due to it's low ampage capacity) and needed an additional logic chip. The only way out of this problem that I can see is either replace it with a LMD18200 chip(which is the same as a L293D chip, just higher capacity for amps), or to essentually expand the entire L293D chip out with transistors, and do the same with the logic chip. It is essentually a large array of diodes and mosfet(dmos) or other type of transistor chips(we would need 4 for the chip, and 4 more for logic control of one motor at least). These would take up alot of area, but if we are only using one of them, it wouldn't be especially difficult, just time consuming to build/print.

Anyway, the list thus far for the extruder board, which is what I'm working on(I'm 90% sure it's my RS485 that's not working properly, but can't figure out why not). Things like resistors and capacitors I assume can be very easily replaced and are not listed.

Note: 4 stars(****) indicates it's the exact same chip,or is the same chip, just with a different sizing spec

Current chip used--------------------replacement chip/original chip
ATMEGA168----------------------------ATMEGA168****
SN75176A-----------------------------SN75176AP****
7805---------------------------------7805****
NIF5002N (5003)----------------------TIP120
A3949--------------------------------LMD 18200 or L293D expanded chip (add logic chip as well)

Chip by chip issues:
7805- This is the voltage regulator, it tends to generate heat. Hense they may need a heatsink. However, this appears to be a component that is very common and is not made by just one producer. The reason for this is that some parts need 5V, some parts need 12V.

NIF 5002N(there is some confusion here, since NIF5002 are whats on the board and spec, and whats supplied with kits, but 5003 is what described on the wiki page) these are made to take 42V and 20 amps(though the website for the chip on onsemi.com says they can only take 2 amps, not 20. Can someone please look into this. [www.onsemi.com] , the spec seems to match the chips I got from makerbot, as well as the photo on the wiki. Are these the same chips? if they are, then they can only take 2 amps, not 20 and the wiki has a minor type). The TIP ones replacing it can only take 60 V and 5 amps. Plus there seems to be a slightly different design basis, given that the NIF's have 4 connectors, while the TIFS have only 3. Can someone see if there has been any major design change based on this. As far as I can see, the basic design is unchanged from the PWM board. The NIF's main advantage is the removal of a resistor(on the signal section) and a diode(on the power) from the board. They also have safety advantages etc to prevent overheating, though I'm not sure if this is really part of our system (I've managed to blow my heater circuit by accidentally contacting it with the thermistor, hense why i'm trying to figure out a way to produce this system while I wait and order new extruder controllers)

A3949- As stated above, this chip gives by far the biggest headache. Apart from expanding the old Gen 2 chips outwards, I don't see an easy way to get around the problem. the A3949 chip is a great chip given it's high amps and inbuilt processing etc, but is only available as Surface mount, and is too fine to be fitted in a printable circuit. The LMD18200 is pretty expensive, and expanding out the L293D chip with higher amp parts would be annoying. But hey, this is mostly for the purposes of an exercise.

Edited 1 time(s). Last edit at 08/21/2009 06:41AM by letsburn00.

Nice reseching you did so far.
I hope you don't mind if i throw some ideas around that i had while reading this. :-p

My first question is quite a rough one:
- Why bother with THT (Thru Hole Tech) when most of the electronics industry isn't really dealing with this old technology anymore for obvious reasons? One could almost say it's pretty much a waste of time and it would be better to invest this effort into improving resolution/precision of the printing and the actual fabrication of small conducting connections on a board. (This is more what the ultimate goal will be for this task)
From what i've seen so far, it would probably be better to use the RepRap as a milling and drilling machine to produce your own circuit boards for the near future. Copper coated epoxy boards do not cost that much to buy and you get them all around the world, so sourcing the raw material isn't a problem for now. Converting the RepRap into a milling machine for circuit boards should not be a big deal either. The motor doesn't have to be extremely powerful, as you will not take a lot of material away and can play with movement speed to get a clean result. Just some high rpm would be nice, and you have to have a well centered rotation point for the milling to keep track spacing narrow enough. The mill used to cut into the copper isn't really that special either, you could probably do with a regular circuit board drillbit, as you only use the tip of it to get a narrow channel milled out. Only problem here might be the precision of the current z-axis, but i'd say you could improve that by adding a pretty short but more precise z axis to the milling head instead of using the platforms z axis.
Just an idea, but probably easyer to realize, and you could use all the fancy SMD parts just as it would be an etched board.
Plus you get the advantage that you can use normal soldering tech to get the components on the board. With an extruded board you run into the problem of a melted board as soldering temp is higher than the meling point of the plastics we can extrude with current setups. (Using conductive glue probably would work, but that seems imho messy to actually realize in an "easy to do" production stage)

Okay, enough bashing of your idea, back to some constructive help

If you can't find any substitutes for SMD IC's, another easy way would be to use a conventional adapter board that "extends" the SMD component into a THT component. You'd have some old school etched board in your setup, but if you get a large batch of these small PCBs manufactured (or etch them yourself) they would not cost that much and you'd have a pretty nice temporary solution (Single layer board would be enough).

As for the 7805 linear regulator, if you're still getting your power from an ATX-Powersupply, you can cut that circuit out for good. Just use the 5V rail from your ATX-PSU, or even go down to the 3.3V voltage, as this is getting more and more the standard logic supply voltage these days anyway.

And yes, you are correct about the NIF5002. 2.0A is the rating, not 20. Actually the NIF5003 is rated to 14A, same size and all (lower Rds on!), and from what i can tell from Digikey, the price isn't that much more for the much better FET.
(1.54$ for the 5003 vs. 0.88$ for the 5002, in single unit prices)

Like I said, this isn't the most effective method, more a development of the reprap being able to make circuit boards from raw parts (ie plastic and solder) as part of the exercise of being able to print things without external additions (like a drill would be). Copper coated epoxy boards are sometimes hard to get at times or are annoying to procure if you live far out, plus if you damage them you need to wait ages to get new parts in(the components on the other hand are often more tough.

It was found that you can print solder and it doesn't really effect the plastic printed boards much. [blog.reprap.org] . I think that for some of the parts, a SM to DIP connector may be the way to go, but I avoided it at first to focus on DIP because I wanted to avoid any jerry-rigged systems from the get go.

One issue with milling is that milling heads need a considerably stronger and stabler system than a print head. Printing heads will work reasonable well,
Plus I think alot of people would rather use as few toolheads as possible. Especially with the mendel system which looks like it'll need a new design for a head swapping system.