Ask Hackaday: Are Conductive Inks Going To Make It?

It’s amazing how affordable PCB fabrication has become. It has long been economical (although not always simple) to fabricate your own singe and double-sided boards at home, the access to professional fabrication is becoming universal. The drive continues downward for both cost and turnaround time. But there is growing interest in the non-traditional.

Over the last year we’ve seen a huge push for conductive-ink-based PCB techniques. These target small-run prototyping and utilize metals (usually silver) suspended in fluid (think glue) to draw traces rather than etching the traces out of a single thin layer of copper. Our question: do you think conductive in will become a viable prototyping option?

Voltera V-One Circuit Board Prototyping Machine

I recorded this interview at 2015 CES but was asked not to publish it until their crowd funding campaign went live. If you haven’t been paying attention, Voltera is at almost 400% of their $70k goal with 26 days remaining. This printer definitely works. You can print circuits, solder components or reflow them, and there’s even a second non-conductive ink that can be used to insulate between traces when they cross over one another. In the video [Alroy] suggests Voltera for small production runs of 10-20 boards. Would you see yourself using this for 10-20 boards?

Personally, I think I could solder point-to-point prototypes in less time. Consider this: the V-One will print your traces but you still must solder on the components yourself. If the board design reaches a high level of complexity, that timing may change, but how does the increased resistance of the ink compared to copper traces affect the viability of a board? I assume that something too complex to solder point-to-point would be delving into high-frequency communications (think parallel bus for LCD displays, etc.). Is my assumption correct? Do you think conductive ink will get to the point that this is both viable and desirable over etching your own prototypes and how long before we get there?

Now, I certainly do see some perfect use-cases for Voltera. For instance, introduction to circuit design classes. If you had one of these printers at the middle school or high school level it would jump-start interest in electronics engineering. Without the need for keeping chemical baths like Cuperic Chloride or Ferric Chloride on hand, you could walk students through simple board design and population, with the final product to take home with them. That’s a vision I can definitely get behind and one that I think will unlock the next generation of hardware hackers.

Correction: [Arachnidster] pointed out in the comments that Voltera is still working on being able to reflow boards printed by the V-One. On their Kickstarter page they mention: “(Reflow onto Voltera printed boards is currently under development)”

55 thoughts on “Ask Hackaday: Are Conductive Inks Going To Make It?

  1. I think this concept is really cool (even more so for the school scenario mentioned above), but the cost of entry is REALLY high, IMHO. I think this might be useful if you could send designs to your local staples/ etc and have them print the board for same day prototypes, but other than that I think the cost/ wait-time ration for domestic boards is pretty reasonable.

    I do hope that conductive inks get better though, as it lends itself to ease-of-use for ‘alt’ uses, such as for art installations (where the added resistance won’t necessarily be the end of the world/ you only need one board.)

      1. Not really. Registration holes make it easy. I have done 4 layer boards at home without much effort. Now when you get to LARGE sizes home printing introduces distortions that make it a pain. but a 3″X5″ 4 layer by making 2, 2 layer boards and then gluing them together with a thin insulation board is absolutely doable without much extra effort.

      1. I’m assuming you mean “for less than ~$50K”, as I can guarantee that you can make a system that does exactly what you ask. With crazy precision (~5 mil trace/space).

        How do I know? Because there’s one about 10 feet from me right now. There are high-end PCB mills out there, they’re just expensive.

    1. I’m with you… I’ve always lusted after a small PCB mill. To play devil’s advocate, I do think some will raise concerns of throwing fine particulates of substrate into the air with a mill. Think of my example above of use in a school.

      1. If only there were a way to extract, or ‘vacuum’, particulates from the air, and a way to protect – say, ‘enclose’ – the workspace so that there’s no fiberglass dust worries…

  2. Last time I worked out the numbers here was about 1ohm per inch for a 10 mil track . To give a prospective: 10 mils or finer is needed to breakout/connect modern day packages. 6 mils is the lower limit of those $10 for 10 prototype boards deal from China. 5 mils or finer can routinely be made for production boards. So quite okay for signal tracks, but bad for handling power/decoupling. Without effective decoupling, your high speed (high edge rate) circuits might be erratic. Control impedance for signal integrity requires multilayer printing.

    If I am in a rush, I would either pay extra shipping form China or pay through my nose for local/North American board ship. Last thing I want is to debug the ink.

    1. The Voltera printer can’t do anything in a ‘modern’ package anyway – its minimum pin pitch is 0.8 mm, which means no SSOP/TSSOP or anything finer. You do get QFPs, though (which is probably the reason for hitting that target).

      But I agree, the China board option is the best for me. Extra shipping isn’t a problem if you batch the boards together. For 10 different designs the $35 shipping is a total non-issue.

      1. And therein lies the problem. QFP’s are now more common than SSOP etc and I don’t think that this product will create enough demand for manufacturers to start putting more chips in non square packages. With both layers on one side you can’t rout out from under a QFP etc.

      2. The common TQFP have 0.5 or even 0.4mm pitch, so even that is not an option.

        QFN are also becoming common. A lot of them have a thermal pad under the chip (or for mechanical reasons). So you would need a plane of some kind for thermal management. Some NXP LPC chips even have that pad as the only ground connection. :P

    2. If you can wait a couple weeks, then getting boards made is a good option. OSH Park has a 12 day turn around for quality U.S. made boards. $5 a sq. Inch isn’t the cheapest option, but considering that is for 3 copies, it’s really 1.67 (provided you need 3 copies)

      I can see where this would be cool for doing circuits on glass, or something else unconventional. I don’t see it as anything that would be able to hold up to much abuse. Though as far as resistance goes, you could probably put a surface coat of solder over the traces if that was an issue. There is also the importance of solder mask when working with snail SMD components. That does a lot to prevent bridging of solder joints, as anyone who has home etched a board will know.krux

  3. Thesy should sell the cartridge-system as strap on add on for 3D printers, would reduce the cost for an item which is for some costumers too expensive or yet another machine in a shop where there a compareable already ( also mills are in this category). Lets imagine a 2 x 1 m pcb board (you would get problems with the increased resistance) or on the fly implementing of circuits (i.e. traces) in or on 3D prints (with use of conductive resin to mount the PCB afterwards).

  4. Errata: Their own kickstarter says they can’t currently reflow boards they print, only professionally fabbed boards.

    I think the idea’s a very neat one. I’d totally prefer it to point to point soldering; you’re (probably) not going to point to point solder SMD parts, and SMD is increasingly the only way to go. This would allow you to knock out a quick SMD prototype in very little time, and iterate a lot faster. A board doesn’t have to be overly complicated to be quicker to SMD solder than to make up on protoboard.

  5. I’ve been in electronics manufacturing and design for the past fifteen years. The primary benefit of etching/milling/conductive inks is that mistakes in hardware projects are expensive in time lost debugging, time not spent moving the project forward, and materials (A fully placed, turnkey small pcb can run a few thousand. A large, complicated board can be significantly more). None of these solutions can replace actual fab. Part density, multi layer, impedence matching are all either impossible, or so different than a final board that you literally would have to run the board again just to shake these out.

    That being said, the tremendous advantage to all of these systems is that you can unit test individual circuits and test basic assumptions. This means that I can proof out new sub-designs today instead of waiting weeks. The benefit of conductive ink depositing and CNC versus etching is that it takes less human intervention. Additionally, either CNC solution can eventually be modified for a pick and place. Investing in that tech is worth it.

    Would I ever use any of these solutions as they stand to make 20 prototypes of a real end-use PCB? Probably not, because my needs go fat beyond 2 layer, slow speed throughole. Am I going to buy this new kit and give it a whirl? Yes!

    1. Exactly. For high speed circuit, the PCB itself IS a component. If you are not using the production layout, you are wasting time. We spend most of our time on the proper design and simulation so that complex boards only requires a couple of revs before they are good enough for the software people to get started. There are plenty of stuff to do between the 1 month churn cycles. Whether or not I get the board in 1 day vs 5 days done have no bearing as the board still have to line up with the CM for stuffing it with BGA.

      Raw cost is around $10k range per board so we don’t exactly do 20-30 rev to save time. Software people are very quick to blame the hardware if you do not have a good rep in making a good solid design at the first couple of attempts. When you waste their time, they waste yours by breathing down your neck.

    1. I like this idea. Conceptually it’d be like printing in flux only, then having solder reflow to create the actual circuit.

      Doing it with solder paste (extruder anyone) might be even better.
      Still no via soln though…

    2. Molten solder would try to form a ball as it has the minimum surface tension.

      You cannot print paste directly as it needs to stick to something and a way to “stretch” it to form a trace. Solder formed traces e.g. ones on a pad per hole protoboard are ugly and requires excessive amount of solder.

  6. Most of my projects are more along the lines of robotics, meaning handling relatively high currents. The impedance of the conductive inks vs. solid copper is a deal breaker for me.

    That, and the fact that it can’t do plated through-holes.

  7. Most circuits today are high speed enough so there is no point in testing it on a board where the conductivity is way worse than a copper PCB. Right now they are at least 12 times higher. Milling a board is way better than printing conductive ink, plus you can also make holes in it.
    If somebody manages to make an ink comparable with copper…then maybe yes.

  8. I think that comparing it to PCBs is a little misplaced. There appears to be little argument against the fact that PCBs are better at being PCBs.

    This seems like it is currently a solution in search of a problem (or problems) and I think it will find its niche(s). There are a number of commercial applications where screen printed inks are used — For example, a good bit of the circuitry in PC keyboards is screened on conductive material. The focus should be on what this method can achieve that PCBs can’t.

  9. At this stage in the tech. I can’t really see it being useful. But there was another company… I forget the name… that was doing somewhat the same thing except also putting down some plastic… this allow components to be suspended in a 3D printed medium…

    If this gets to a stage where you can work with extremely small pitch SMT components, BGA, 0402 resistors… I KNOW THIS IS A STRETCH… I think companies like Apple, Samsung would be the first to start pouring money into this… because with current technology even though we have 12+ layer PCBs … we can only put components on one layer… and the name of the game is to get as many small components into the smallest packages as possible so small devices can be made…

    1. Actually, at the high-end of traditional PCB fab, embedded components are already a possibility – either by placing discrete components in pockets milled in the board, or by using traditional fabrication techniques with special board materials (e.g. high resistivity) to directly etch flat passives on internal layer(s). Both methods are pretty hefty cost adders though, so you have to be in a real bind to go down that path, and money should be no object (military, medical, etc). Cost-conscious commercial applications probably will stick to 01005 passives for a while :)

  10. Compared to their cellphone/computer for non-technical market, are the DIY hardware individuals a big enough market for them? Would this fit into their overall picture? May be companies like google would be more likely, but don’t count on it.

    Also this is yet another few step above the Arduino level (i.e. even smaller market) having to route the traces vs just buying shields or poking wire onto breadboards..

  11. Why are people still wasting time with inks? The hacker movement has lowered the cost and increased the availability of 3D printers. There are several common techniques for using an FDM based printer to create a circuit mask out of things like sharpie pens or even using a cheap laser module to sinter a thin layer of black spray paint. Lasers can have a very fine focal point and it’s pretty simple to convert a black and white circuit image to gcode. For more precision, you can even convert it to raster. No need for mills. They are loud and create a lot of dust. Ferric Chloride, or even having a small electrolysis setup to remove the copper from around the mask would be cheaper than investing in special purpose pcb printers or a pcb mill. Neutralize the acid with baking soda if you want to be environmentally friendly. And the best part of all? You still have a 3D printer, and with a laser module you now have a low power laser engraver. Let’s expand the slicing tools to do more than extrusion. Skip inks; it’s far too costly and difficult to perfect, we already have a great platform just waiting for expansion.

    1. Dear Mc,

      Would be curious about your response to my (just) recent post. I also ‘totally agree’ with what you have to say, but I think at least the promise of this tech is ‘boards with many layers’… I mean these days, an no offense to anyone (as I have built many myself), if you are doing, even at least a ‘two layer board’, you are probably designing either a ‘personal toy’ or at best, a ‘gift for a friend’.

      All of which is of course ‘none of the designer’s choice’, but at minimum, all the really ‘hot’ packages are only broken out in BGA, and it is not that I really feel, with the tools/knowledge available, ‘at home’ designers could not work with these chips. Rather (and slightly oddly, though yes, ‘economies of scale’) one might find that producing a 6, 8, 10 layer board suddenly costs 50, 60, 70x the silicon.

      So I agree with you, whether through light reactive methods or lasering, we need a product we can register, sandwich together, and then ‘bake’.

      However, still, as said, I like they are at least ‘perusing’ this product as I feel there may be some interesting aspects to ‘3D routing’– Or, really, I think one could already do that even with the ‘present’ PCB technology, but I am not sure of any software package that takes advantage of it,

    2. Reselling ink and other consumables is part of the business plan, so unlikely you would actually find someone that make sensible machines for the DIY market.

      For some reasons everyone in the mass media have been brainwashing people to think that etching chemicals are icky. e.g. see multiple HaD “editor” writeup references to Ferric Chloride is toxic, but IRL it is used in drinking water and sewage treatment. wiki: Ammonium persulfate is a standard ingredient in western blot gels and hair bleach.

      It is the copper ions from the PCB that is toxic. It is a matter of converting it to a solid and disposing it responsibly. With the right masking technology, etching can get you the accuracy and resolution.

    3. Regular ovens make much better food than microwaves, but people still prefer the convenience factor of nuking last night’s pizza. Acid baths add a step and an order of magnitude more mess.

      That said, I think these technologies are duds. Beyond the fact that this can’t do vias, layers, or through holes, it also leaves you with a board with no plating or soldermask. I built a pcb mill that worked fine but stopped using it after board #2 because tabbed boards were so much nicer to work with.

      The future I think belongs to vastly cheaper and easier simulation tools. Modern computers should support robust simulation of a wide variety of arduino type circuits so that you have an excellent chance of getting a working board on version 0.1.

  12. I do like the concept/advancement of technology (there were cars before the ‘Model T’, so I would not yet give this project that designation), but from my personal experience, my concerns relate to the ‘jetting’ technology they are using… I mean just the other day, was there not an article on Hack-a-day concerning issues with the Makerbot head (which is basically the ‘commercial standard’ for the mid-to-low end) ? There are viable heads that can print various metals in solution quite reliably (see Fuji/Dimatix)… but they are quite expensive.

    On a side note, personally I have always kind of wondered, say using the ‘photo-reactive’ process (ex MG Chemicals), why it is not possible to build a multi-layer board by applying this process to a number of ‘thin sheets’, with registration marks, lining them up and the ‘baking’ the bit, sawing the excess off (and yes, I know to do the rest then you also have to be a ‘screen printing shop’, but a very basic setup is pretty cheap and using the same principles as designing the board in the first place)… I gather, really, mostly, this is the process most board houses use, but I have not yet seen a widely available product one could by for this process, or a really good tutorial on it.

    One last thing, just to think about (and why I believe companies like Autodesk are investing in it)– With a combo ‘make and trace’ tech, your ‘vias’ are now in ‘3D’ rather than just in ‘two dimensions’. Thus, where appropriate, you can now easily do angled cross overs, etc and potentially greatly lower the number of layers you have to ultimately print in the first place.

  13. I’m not sure that they will succeed in their initial market, but I could see the machine morphing into something more out-of-the-box.

    I’ve been thinking a lot lately about what would make my life easier doing prototypes and a simple pick and place machine that threw down a bit of solder paste and placed, say, 10 or 15 components that are common to all projects (1k resistors, 0.1uF caps, etc.), but got out of the way when I didn’t need it (it would have to work on my bench with no modifications, so an articulated arm; and retract into a corner) would cut the drudgery out of the process and that would be worth money to me.

    And/or a system that simply took a low intensity laser dot and pointed it at the location on the board as I stepped through the BOM on the PC would be helpful.

    They are attacking a market that’s already pretty mature, but with as much money as they raised they have a bit of an R&D fund to create a new market if they are clever enough.

  14. It looks pretty interesting, but their resolution and conductivity are pretty limited. Here’s a comparison against oshpark:
    Min trace width: 8 vs 5 mil (0.2 vs 0.1 mm)
    Resistance for a 32 mil wide trace: 0.25 vs 0.015 Ω/inch (10 vs 0.6 Ω/m)

    IMO, the killer application for this is SMT parts, but the resolution isn’t high enough for anything smaller than 0603. Most hobbyists aren’t going to be working with smaller passives than that, but ICs requiring finer pitches are quite likely.

    The other issue is the conductivity – its 20x worse than copper. If it were within a factor of 4, then you could compensate for it by doubling the trace width and height. You can probably ignore it though as long as you’re not doing anything really sensitive or high current, and don’t mind higher power consumption.

    It looks like it could be a half-decent alternative to PCB mills, if you want something cleaner and don’t mind the higher resistance.

  15. It’s an interesting application and some will find significant benefit. It’s a one step process to make the board and the materials are not corrosive, so maybe for schools. Downside is that it’s speed depends on the area to be deposited and the complexity of the circuit. Photoresist or other mask and etch is almost entirely independent of those factors, but can’t be left unattended. CNC can be unattended, but time increases with perimeter.

    I think someone will find a way to combine fine stranded copper wire with an FDM printer so that the FDM embeds the copper along the path, leaving the ends exposed for soldering. It needs a trimmer and a rotating head to change the pull direction. It would allow multi-layer effects by insulating cross-overs and leaving vias exposed to be soldered together. In addition, the FDM system could also follow 3D contours (5 or more axes). If there is a strand threader, then the section of copper could also be varied by placing different numbers of strands.

    I think the FDM would be a standout using flexible materials so the item could be deformed.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.