Physical product businesses need to rethink supply chains. Microfactories are one of the options and in this post, we introduce you to our CNC Lab where we develop microfactory management tools.

Why the change?

For a while now product management was straightforward:

  1. Design & prototype locally;
  2. Manufacture in China;
  3. Ship from China to local;
  4. Market & sell locally.

This model heavily relies on cheap labour and high availability of production capacity of manufacturing in China. For a while now this process was so profitable that global scale manufacturing in the USA and Europe is nearly extinct.

However, rising labour cost in China, “America first” and similar policies are putting economical and political pressure to change the established system. Moreover, coronavirus ( COVID-19 ) is testing the robustness of supply chains and it seems like the virus is winning.

The system seems to be shifting towards more local manufacturing. Even the giants like Tesla are spreading their factories across the world ( eg. the USA, China and Germany ) instead of concentrating their manufacturing in a single country like China.

Few businesses are large enough to be able to establish entire factories in multiple countries. Thus for many, the solution can be found in microfactories — small self-contained highly automated production units with a capacity just enough to service the local market.

Microfactory system saves a lot of costs and time by reducing the shipping distance but is much more dependent on the local labour costs. To save the latter high degree of automation or remote operation is necessary. In order to develop efficient microfactory processes, we established CNC Lab.

Why the CNC lab?

CNC lab’s mission is:

Develop processes and tools to automate manufacturing with minimal local skilled labour and maximal remote operation.

Remote operation of local machinery benefits from reduced shipping costs while employing more cost-efficient remote labour. However, as easy as the concept is the transition is more complex than installing Teamviewer on the CNC computer, because:

  1. Lost visual contact needs to be compensated by a camera setup and monitoring software;
  2. Failsafe mechanisms for connection loss are required for safe operation;
  3. Remote operation increases operator reaction time which must be dealt with to ensure safety;
  4. The work that requires physical contact must be standardised and structured to be handled by unskilled employees;
  5. Need to identify scenarios where local operators are required and define how to solve them.

With every challenge, new opportunities arise. Having the setup monitored by cameras enables the use of artificial intelligence ( AI ) software to not only address the issues above but also supplement the process with more automation. Eg. slowing down or stopping of machinery operation upon detection of a person in the machine vicinity or detecting workpiece defects before starting to machine.

Thus, in search to solve the problems arising due to remote operation and identify opportunities of AI application in manufacturing we’ve set up the CNC lab. The core of it — is a CNC router.

How we got the CNC router

During the planning phase, we considered multiple manufacturers from Europe and China. Although European deals had a feeling of quality, better and faster support, we picked to go with the more flexible, more hardware feature-rich and much less expensive Chinese machine for our specs.

Still, buying expensive equipment from China is scary because of all the horror stories one hears about business not receiving expensive machines. To make sure we’re not joining those stories, we verified the CNC suppliers extensively by checking all the information available about the supplier online and asking the supplier to give contacts of people near us who bought the machine from them.

Out of our list, Omni-CNC had the best feedback online, had good communication and provided contacts of two companies who gave positive feedback. Negotiating the machine specs over email and WhatsApp took us a few months and then we placed the order.

Having paid for the machine, instead of silently waiting we insisted on as many pictures as we could as often as we could. Luckily, our contact person Allen helped us in this stage and provided pictures during the whole building and shipping process of the machine. Good communication is hard to come by even directly and he communicated well even over the very long distance between Europe and China.

Machine specification choices

The CNC router has a working area of 1300x2500x200mm which is the smallest machine capable to process 1250x2500mm ( or 4' by 8' in imperial ) sheets. Most of the materials are available in that format and thus no pre-cutting is necessary in order to put the sheet on the table.

It is necessary for us that the machine is driven by servo motors instead of steppers. This is because steppers can fail silently by losing steps and unless the fail is substantial those lost steps will be noticed only when the final piece is measured. Servos have an integrated feedback loop which allows them to correct for excessive loads. In the worst-case scenario, they overload and stop the machine which is easily noticed by the operator. Our router has four servo motors — two for the Y axis, one for X and one for Z.

The spindle is 3.2kW strong, is cooled using water and has automatic tool changes ( ATC ). Our choice is to go with the most affordable Chinese spindle as long as it has ATC. The latter is crucial for automation and operator-less reliability. Our tool rack contains 8 tools which have been sufficient for our cases so far.

Workpieces are fixed using a T-slot table with an MDF spoil-board on top. Figuring out how to mount workpieces faster and with less manual labour is one of the very important subjects that we are working on.

In addition, the 4th axis is available which we use rarely because of lack of support of it in Fusion 360. However, recently Fusion 360 added operations for the 4th axis and it is in the works to try it. For these operations, the setup is automation-friendly because it employs a pneumatic chuck.

Last but not least, is our CNC router control interface. Very early on we realised that having the interface as a program on a PC makes it much more remote and automation-friendly. In the simplest scenario, the operator can just use TeamViewer to log in the machine which is much harder in the case the machine is controlled using some embedded proprietary hardware. Our choice was a Weihong PM95A controller with NCStudio software. So far this combination has been working well for us and we’ve even done remote sessions using TeamViewer with the operator being 20km away!


To sum up:

  1. Global supply chains are under pressure to change and local micro-factories are an option for many manufacturers;
  2. We established a CNC lab to develop processes and tools for automation and remote operation of manufacturing plants;
  3. The choice of a particular machine setup heavily depends on the context: economic setting, acceptable tradeoffs, budget, how the machine will be used ( high volume vs high flexibility );
  4. Buying from China is a worthwhile option, but requires extensive due diligence. Omni-CNC ( ) worked well in our case and we recommend reaching out to contact Allen ( );
  5. Considering remote operation possibility during the shopping stage can save a lot of costs in the long run. Making a machine remote operation capable which was not designed for it can cost much more time and money.

So, what do you think? Considering automating your workshop, see issues with our setup or have ideas you’d like to test in our lab? Feel free to reach out with comments below or on our site!


At we’re excited about the potential of AI to improve businesses and people’s lives. CAD and CAM are two of the largely unexplored territories we’re invested in. If you think you can benefit from a decade-long experience of applying machine learning to business processes, get in touch!