About KickStarter

Hi All,

I recently did a soft-launch on KickStarter(KS). This launch was done semi-silently. I wanted to see how much demand there was for EasyMaker in its current configuration directly from KickStarter over the launch week — this way, I’d know a lot more than I do about the market for the robot. I’ve seen at least  2 other 3d printer projects launch within the last few weeks on KS. At this point, I  know enough about KickStarter trends to suspect that the current robot won’t sell enough units in that channel to meet the goal.

I can’t launch EasyMaker without enough paid pre-orders. The only way I can get the various parts I want is to have them custom designed. I probably have enough skill to do any/all of the components on my own, but I don’t have enough time. In order to make easymaker, “Easy” for end consumers, the electronics and software need a lot of extra work.  This is true in all 3D printers and CNC mills.

It’s hard for me to explain the problems in those without co-mingling the concepts a little.  The electronics and the software are so tightly coupled.  So, I’ll try and explain the problems in electronics first by comparing/contrasting to what already exists or will soon exist in other projects.

Heat management/power.

Right now, whether you use RAMPS, sanguilolo, PrntrBoard, etc, you end up with an Alegro A4983 or A4988 stepper driver.  These drivers are rated to 2 amps of current and have a lot of nice features.  They also have a terrible limitation.  They thermally shut-down very easily( aka overheat ).  This means that all of the electronics that use this driver, especially at high power, need to manage heat.  This means heatinks and fans in the enclosures, which means larger enclosures and more limited operating conditions.  Since this generation of electronics using this driver are open-loop systems, thermal shutdowns manifest as skipped steps — they affect the output quality.  New electronics around the TIDRV8818 would probably solve this problem — they’re known to be more thermally stable.  They’re also more expensive, and have a different pin layout.  It would take quite some time to design a new board around this, and without high quantity, would be  expensive.

Cabling/Headers.

Right now, there are two types of cabling/headers available.  Some people use pin headers, and some people use screw terminal.  Both of these can work just fine if you don’t move the robot around much.  Both of these are a real pain to work with if you do move it around a lot.  I drive my printer around from time to time, so I have to disconnect/reconnect it often.  I get to see the problems of this up close.  First of all, it takes too much time.  Both to initially assemble the cables, and second, to disassemble and move.  Pin headers wear on every connect/disconnect cycle.  After a few dozen cycles, they make poor contact with their pins, and new pins/housings need to be crimped on the cable.  Definitely no fun.  Screw terminals stress the wires, and over connect/disconnect cycles, cause metal fatigue that breaks the exposed sections of wiring.  More over, while they take less time than pin header in initial construction, they still take a lot of time, and make the cost of disconnecting/reconnecting much higher.  The right solution is probably a design around the IEEE 1394 cabling standard.  These cables are designed for high power DC applications, are cheap, and come fully assembled.  They can be connected/disconnected many times without the wear causing electrical contact problems.  They also would mean new connectors on the boards, and possible moving the stepper drivers to the motor.  Again, without volume, this is essentially impossible to do.  I’d really have to sell 50-100 EasyMakers just to begin getting this work spec’d out.  I’d need the volume for a chip run to be affordable.

Enough drivers/features.

Right now, the cheaper electronics are going with 4 stepper drivers.  Some go for 5, and I only know of one board at six.  I think the future of 3D printing and CNC milling will be around 6 Axis systems.  For the 3D printer, 3 axis are for X, Y, and Z; while another 3 are for finishing extruder( which will be much smaller than today’s .35 — I guess a .1 will be possible/normal once skinning and infill are separated ), an infill extruder( while will probably be .5 or .75 ), and a support extruder( probably around .35 ).  In CNC milling, the same X/Y/Z axis will remain, while pitch. yaw, and table rotate will probably become more common.  While I see that the RA board will have this driver layout, it is also as expensive as RAMPS.  I actually think more will be needed.  In order for EasyMaker to fill its vision( lathe/powder/whatever you want, ) I think 8 drivers might be optimal, as well as at least a 6 pin GPIO.  All of this has to be easy to use.  Also — existing boards don’t support variable voltage.  This means that you can’t use voltage regulated motors.  In milling, there’s a nice set of relatively cheap ER spindles made in China right now.  These use a 1/4 hp variable voltage motor.  Being able to change the voltage on the fly would mean speed/power control in an affordable, precise, and powerful spindle.  It would revolutionize home milling — bringing into affordability the level of precision and quality that people are currently getting with something like a Fordham carving system( which is just a motor and spindle for $300 ).

Just the design time of the electronics would take an EE at least 3-6 full time months of work, if not more.  It would take me years of tinkering in my spare time.

Next, we come the software problems:

Firmware

Right now, people are mostly using Marlin.  Marlin is pretty nice — I use it too.  But there’s things it doesn’t do very well.  For example, it doesn’t pause when you tell it to, or it doesn’t turn on/off the heater elements when you would expect.  This is because Marlin uses a buffer and a timer to handle GCode processing.  The timer isn’t used to handle the pause GCode, but to scan the buffer for temperature changes and prioritize them.  Great for 3D printing.  Terrible for milling.  Many CAM tools will generate start/stop signals inline to the motion code.  But since those are now executed out of order, well, you end up not being able to use full control over your spindle.  Acceleration profiles are great for printing, terrible for milling.  You do not want to speed up when boring a hole.  Features aren’t there — for example, variable speed control of the spindle.

CAM

While I’m loathe to write CAM software, given that SkeinForge and Pycam both already exist — the current crop of open-source CAM tools leave a lot to be desired.  The biggest problems with them are around multi-axis machines and workpiece supports.  It is looking increasingly likely that tapes and other adhesive systems can be used to hold a workpiece down while milling it.  But, the CAM tools currently all treat the process as layers, and cut the entire profile out at each layer.  This means that you cut through your tape too early, and your workpiece isn’t as stable as I’d like when using something as easy as tape.  Other mounting systems exist — parallel bars, bolts, vices.  But nothing is home-user easier than tape.  Also — if additional axis are to be used, it’s not enough for the machine to have them and the firmware to support them — the CAM tool must use them as well.

UI

Currently, the user interface of the tools is all different.  It’s not all intuitive( or even complete ).  I think Slic3r might be the best in terms of UI — but it doesn’t do any milling.  Even Slic3r has its UX issues.  Renaming things over time.  Changing absolute numbers to percents.  Etc…  In order to make 3D printers and CNC mills easy, it’s going to take a new UI — one made by a designer, and not an engineer.

Again — I know enough about software to do it myself.  But this is closer to a 2 man-year project( aka, it would take a single person working full time about 2 years ).  Again, this needs to move from work done once in a while to a full-time person.

And this is just the technical side.  There’s a lot of work to be done on the business side.  Sales,accounting, HR, etc…  So, unless I already had a functioning side business, or funding from a VC, there’s simply not enough manpower to take EasyMaker to complete the full vision.  There’s just not enough interested buyers to justify doing much more with this project.  EasyMaker needs to sell without advertising.  EasyMaker needs to sell enough units.  I need to know that the channel supports the robot intrinsically — These are just  requirements to run a project from a garage.

The future

In order to move EasyMaker to the next level, there has to be enough unit demand.  Right now, I’m not sure that will happen.  Most 3D printers or mills hit their goal within the launch week.  It looks like the channel has spoken, and wants additional changes.  Now, I’ll have to figure out what they are.   I need to re-engineer the machine in a way that drives unit volume, while at the same time, keeping the constraints of the electronics/software/parts that are available to me at my scale.  I’d hate to do that — these constraints can be solved, and create a better overall experience for everyone.

This will likely mean a change in name.  Without the electronics and software changes I mentioned, this new robot won’t be as easy to use, and I feel the name would be inappropriate.  It’d be no harder than existing robots( aka RepRaps, Sherline mills, etc ) — but I think existing robots aren’t easy enough to use.  Rolling element bearings will have to be removed — they’re too expensive ( and they’ve been on the chopping block for a while.  Sourcing them reliably, cheaply, and at high quality is a real challenge).  I’ll likely remove the changeable drive system, and instead just focus on a single drive type.  Printer functionality will probably be a bolt-on upgrade to the basic robot, while the Z motion system will likely be changed to a more traditional system than MakerSlide.  I do have a vision of these changes — but they’ll take some time to make.  In the mean-time, the market will move forward.  Some of the problems will solve themselves, due to market changes– but some won’t.  Backing the KickStarter now is a better way to get to the end goal of an easy to use, fully functioned robot that can make almost anything.

6 thoughts on “About KickStarter

  1. “EasyMaker needs to sell without advertising.”
    I hope you are talking about paid advertising, because nothing sells without being noticed, and to be noticed everything needs to be spread as much as possible. If you were expecting that the kickstarter campaign would be noticed without talking about it in many websites and forums, you could have the best machine in the world and it wouldn’t make a difference. I follow this blog since the beginning, and I didn’t even notice the kickstarter until you mentioned on reddit, but didn’t even provide a link. So I came here, and nothing. And now this post, without even a link! And in the kickstarter, the negatives and risks seem more detailed and emphasized than the strengths. You even admitted some time ago that marketing was not your strength, so you could find a friend with those skills. I hate misleading marketing, but there is healthy marketing. I believe there would be much more interested people if you made a convincing argument and tried to sell your ideas and execution better. You are still on time to do that if you are not too rigid.

    • Kickstarter is an intrinsic market at this point, as is Amazon, and Ebay. This is very different than an extrinsic market — KickStarter, Amazon, and Ebay have a user base that actively searches. For example, a common search string might be, “3D printer”. They then find, internal to the site, what they’re seeking, examine, and buy. No advertising is needed — some % of people doing these internal searches will convert automatically. This conversion number is all important. This number is the, “How cool is it” number. This is the number that tells you how many units the product can move, just on sight. This number isn’t 100% perfect — some of it is based on how good the video is, or how much information the person supplies — so it’s not a clean number. But it is information. So, you’re right in that my video and my information is part of the problem — but if the robot was, “Cool enough”, it wouldn’t matter. I’ve seen plenty of horrid KS videos and pages that have had the conversion numbers needed to launch a product.

      The problem is this — intrinsic sales are a baseline. You use them for planning. If I get 5 unit sales from KS, and I also have a proxy model for how many people found it from KS, then I have a way to estimate cost of sales for KS and every other channel. In order for EasyMaker to go from a garage project to a professional project, I need to know there’s going to be a monthly order flow. That order flow can be small — say 10-20 robots a month — but it must be there. That order flow can’t be gotten by hitting the blog sites every month — they’ll be tired of it by then.

      Things that have intrinsic demand sell without any advertising at all. People find it via search, say, “Wow, that’s cool. I’ll post about it on my favorite site!” It ends up spreading on its own, and people talk about it. How much of that happens determines how much will sell. It also determines how much time I have to put in to the various activities. This intrinsic demand has to be high for a garage project.

      As to why I didn’t hit all the blog sites:
      I get very few bites at the apple. I can’t go to channels and test too often. There’s a fatigue point. This means I have to know, really know, how many units people want in any given time period, and how much they want it. If I go and hit all the sites with the project, then I “use a bullet”. If that bullet misses, then there’s a reload time before I can hit those sites with the next iteration. I was reaching a decision point, and I needed to know what the market thought. So, I did a soft launch. If it had spread at the speed I wanted right off KS, then I would know that the market likes this variation, and viola — I know the direction I’m going is correct, and I keep going down that path.

      But the Market doesn’t seem to like this variation. I think the reasons are:
      1. The market wants a cheaper robot in a shorter delivery time frame.
      2. The market wants lower overall risk instead of spread risk( low risk, expensive model, high risk and cheap model isn’t what it wants — it wants a single low risk, less expensive model.).

      I suspect there are a few other things the market wants — but I think these are the two big ones.

      So, this is where I have to go. This means some radical changes in design. I have to remove parts that I can’t source easily, cheaply, and quickly in the US — which throws out the rolling element bearings, as these are just too hard to get here in the price, quality, and delivery window needed to reduce risk. I have to reduce parts count and I have to integrate functionality of some parts together — that reduces time to deliver( less waiting for different parts to arrive ). I have to focus on doing only one part of the design, and rely on the market for the other elements. This I don’t want to do — I’d rather control the whole stack to my vision.

      This isn’t the end of EasyMaker. At least, I don’t think it is. I know the idea works, and I know how to deliver the technology. It’s now a matter of merging the technology with the market desires. Once I’m confident that I’ve done that, then it’s off to the blog sites and push advertising. That will hopefully get me to the monthly volume needed to professionalize the project.

  2. I read in some forums that the drivers overheat because the pcbs are too small and not well designed. Integrated boards are said not to overheat. It is one of the improvements that are claimed by 4pi and smoothieboard.
    https://groups.google.com/forum/#!msg/openpnp/89mgrM8DQos/08eUxQmqGOkJ
    “Smoothieboard can drive 1.9A NEMA23 steppers at 30v ( like what pololus theorically can, except it doesn’t burn when you try ).”
    http://smoothieware.org/smoothieboard
    https://0xfb.com/shop.html

    They only allow for 5 drivers though…and they use arm processors so the firmwares are being rewritten to modern architectures (with a modular design for smoothie). Smoothie is also more focused on cnc milling. There is also the R2C2, it was the first and has working firmware but it’s very expensive.

  3. You can also wait for smoothieboard bare that allows for external drivers. But the best for easymaker seems to be the smoothie firmware modularity:
    “I think the main advantage of Smoothie over other choices you can make, is that it’s built from the ground up to make adding new stuff to it as simple as possible.
    This means that you can add any feature to it, without having to touch/undertand any of it’s inner workings : simply create a new module, plug into events, do stuff, that’s it. http://smoothieware.org/moduleexample.”
    Atm, Smoothie knows how to laser, 3D print, cnc mill.

Comments are closed.