During the prototype and pilot phases, you can get by with using hardware that’s been hacked together from off-the-shelf parts. But when you go to deploy a full IoT system, you’re likely going to need to work with a manufacturer.
Manufacturing could be an entire ebook in and of itself (and maybe it will be, let us know if you’re interested!), but here we’ll give you some of the important considerations for manufacturing production-grade sensors/devices. Some of the points below won’t apply when you’re using sensors/devices that have already been built and manufactured, and whether you need to build entirely new sensors/devices will (as always) depend heavily on your specific use case.
The most important consideration, is that it’s going to take a while. Do not expect to go from a few prototypes to multiple thousands of production units within a couple months. We’ve done it in four or five months, but that’s like trying to sprint a marathon. It’s possible, but it’s more than likely you’ll burn out and fail.
Expect several months to a year to go through the entire process. And scale also factors in here; a manufacturer will be much more willing to move quickly if they know they’ll be producing hundreds of thousands of units than they will to produce just a few thousand.
This process takes a considerable amount of time for a number of reasons.
With software, if there are bugs or areas of improvement, you can update the system after it’s live. And with the connectivity layer, you might be using proven standards and existing infrastructure that carry relatively low risk (like WiFi or Cellular). Even if you’re using a relatively new standard, making updates to the network can also be performed after-the-fact.
With hardware, you are not making changes to your sensors/devices once they’ve been produced. This is why having a small scale pilot is absolutely critical, it allows you to rigorously test your prototypes to identify any bugs or weaknesses in the hardware so that the final units can as purpose-built for the specific use case as possible. And manufacturing usually involves building injection molds, which help reduce per-unit costs but have massive upfront fixed-costs, meaning that it’s extremely costly and time-intensive to make changes.
Also, for any sensor/device that’s communicating wirelessly, you’ll need to get FCC certifications (if you’re in the US that is; for any other country you’d get certifications from the relevant governing body). Wireless means that the device is communicating using electromagnetic waves over a certain spectrum, so these certifications are to ensure that your device isn’t harmful to people nor infringing on licensed bands.
Another reason that the manufacturing process can take a while, is that you need to source all the materials. For a given sensor/device, there may be dozens or into the hundreds of individual components necessary to build the full device.
The manufacturer will need to set up production lines too. This involves setting up all the equipment and assembly lines to actually manufacture the sensors/devices, as well as setting up the testing processes for each important stage. You want your sensors/devices to be tested throughout the entire process to ensure that you don’t produce an entire batch of units with defects.
Finally, once the units have been produced, they’ll need to be shipped to wherever they’re needed (perhaps directly to you or to the location where they will be deployed). If you’re manufacturing in China, which is likely, this means that the devices will first need to pass customs because they’re being imported from abroad. This process can be several days or weeks and doesn’t have a set time, so make sure there’s a bit of cushion in your delivery schedule.
And shipping itself is an important consideration. If units have lithium batteries, there are regulations that may prevent them from being transported on airplanes (due to possibility of combustion).
Also, it’s critical to consider device behavior (which ties into our earlier chapter on hardware capabilities). Do your devices “know” they’re being shipped? Do they stay in sleep mode? We once made the mistake of shipping units that, because of movement, “woke up” and began trying to find a network. Since they were in transit, they couldn’t find a network and ended up completely draining their batteries trying to connect. Most were dead when they arrived on location.
It’s no accident that hardware has “hard” in its name. Here at Leverege, we develop software solutions and act as the overarching systems integrator on end-to-end IoT solutions; we don’t manufacture hardware. However, we’ve been fortunate to have hardware partners with extremely deep experience and high aptitude and recommend that you listen closely to the words of any hardware partners you work with.