A
1
Hardware Startup for Dummies

I didn’t know anything about hardware. The closest I came to building a hardware product was by building my PC, and even that was mostly modular building, like putting LEGO pieces together. So when I began building Halo Cars, my former rooftop digital monitor startup (still can’t figure out the right set of words to describe it), I went in blindly, but eventually came out the other side of the tunnel a bit more educated, yet still relatively woefully helpless, especially as I build my next hardware startup. Still learning! That other side of the tunnel was selling Halo Cars to Lyft, where I ended up working for 3 years doing the exact same things, except with less discretion for (still responsible) spending because I now had a corporate card, meaning I could experiment way more. After leaving Lyft, I didn’t reflect too much on the entire experience until I started working on my next project and realizing that there isn’t really a “guide” for non-hardware founders to build hardware products. Disclaimer: this isn’t that guide, BUT it is a list of things I wish someone had to told me about hardware before I made several mistakes that led to car fires, run-ins with regulators (not the FBI-raid-into-office kind unfortunately), and very, very angry customers blowing up my phone as if I had just rugged them on an NFT project. Could this have been a Twitter thread? Probably. Will I summarize it into one? If my dopamine receptors don’t get their daily fill, probably. Anyways, onto the list!

1. It probably already exists on Alibaba

Unless you're building a nuclear fusion reactor that'll bring humanity unlimited energy, your "novel" hardware idea probably exists on Alibaba (crazily enough, there are apparently components of nuclear reactors listed on Alibaba). And that's a good thing! Your hardware idea probably isn't an invention in of itself, but rather an invention of implementation -- a different usecase of hardware for a new type of problem. Usually, the largest bottleneck for any non-technical hardware founder is building an MVP to see if the fundamental function of the hardware even remotely addresses the problem they're trying to solve. There's literally no reason to reinvent the wheel if you can just buy the wheel on Alibaba and make minor improvements to it in conjunction with the manufacturer. You buy a sample, configure it to solve your problem, then put on your "Product Manager" hat and ask the manufacturer to add features, remove things, etc. in order for it to more specifically accomodate your usecase. Cheap, fast, and saves time by showing you that a hardware startup is not worth starting without spending months building the damn thing!

My mistake: there actually are no mistakes here. We made the right decision to buy some samples off Alibaba and iterated upon those up to this very current day. No significant redesigns -- just improved material and new switches here and there.

2. Know what’s coming from China before it’s at your doorstep

But let's say you somehow convinced yourself to move on and oversold the manufacturer on your "promise" to buy 1000s of units if he/she just makes a few changes to your liking. Those won't be the last changes you'll ever make... Over the lifespan of your hardware product, you’ll likely iterate upon it a few times to meet customer/cost/regulator needs. These will manifest in the form of changed screws, bolts, body material, etc., all of which can be tricky to keep track of. This is exacerbated by the fact that you’re not on the assembly line everyday in Shenzhen to become intimate enough with the changes.

My mistake: I would sometimes receive shipments of a new hardware version expecting to need a Phillips head screw to make repairs, only to discover that a PM had unilaterally decided with our OEM to replace all Phillips heads with Torx heads, which admittedly did offer better security and robustness, but at the expense of suddenly rendering all our drill bits useless and needing a trip to the supply store for new heads. The consequence here isn’t severe, but it’s likely the first one you’d experience.

My recommendation: keep track of the BOM of every hardware version, usually in collaboration with your PM. Even the slightest change in the BOM can alter your process/tool-set multi-fold.

3. Know what every single component does

As a hardware founder with not much technical experience, you might grow lazy at trying to understand how your different hardware components work — what matters to you is that it ultimately works as intended. This can be a slippery slope. Yes, on Day 1, the time spent learning how your power supply interacts with your relays might seem like a waste of time, but when your relay fails on Day 100 due to natural depreciation/wear-and-tear, you won’t know to check the power supply first when troubleshooting the issue and instead begin a frustrating goose chase dissecting every other component.

My mistake: one day, our screens started rebooting upon turning them on and entering what you call a “boot loop” (read: never ending rebooting). I genuinely had no idea where to start troubleshooting and picked the relay as the first component to check because I had previously known through a car mechanic friend of mine that they regulate power draw. If I had known that our power supplies had not only an amperage rating but also a built-in safeguard mechanism that shut them down upon exceeding that amperage rating, I would’ve known that was my exact issue. Instead, I wasted about a month down multiple rabbit holes until I landed upon a random manual online for our power supply that detailed how a “boot loop” was not a bug, but a feature.

My recommendation: get smart on electronics, material science, etc. by doing online research and asking “stupid” questions to mech/elec/reliability engineers (if they’re at your disposal). Also, ask your supplier what different components do and why they decided to use them, along with a diagram of the components that displays what connects to what.

4. A multimeter is your best friend

A multimeter is an instrument designed to measure electric current, voltage, and usually resistance, typically over several ranges of value (I pulled this directly from Google). In other words, it measures important “electrical values” that represent power draw, rate limits, and more.

5. DO NOT THREAD YOUR SCREWS

This one is simple: get the right screw bit to remove a screw otherwise you risk threading it. Threading a screw means eroding the head’s design that matches a specific screw bit (e.g. using a Phillips-head screw bit to screw a Philips-head screw).

6. Test the edge cases

Anxiety is a cruel thing and it just might cripple you to the point of thinking every possible worst case scenario has an equal likelihood of happening and that all is lost in the world. Fret not, unless you're building a hardware product... This might not apply later on, but there is no such thing as "Oh, I'll just fix it later, let's just get this thing produced." Once you've delivered a hardware product, it's extremely difficult to recall. Every component in hardware is a critical component, otherwise it wouldn't be there in the first place because who's spending already hard to come by money trying to make something prettier than it needs to be (teenage engineering get a pass, they deserve it). If you think something will happen to your hardware product, it'll happen at some point and it's hard to size just how frequently.

7. Regulators are human too (but not in a sympathetic way)

The average person tends to think regulators and law enforcement are infallible beings with unilateral power over their jurisdictions. This is not true -- they're human and that's why lobbyists exist to influence them and their decisions. This can work in your favor or against it. Even if written regulation exists, a regulator may interpret what's written differently than you would, and the only way to find out who's right in a disagreement is to go to court. Bad idea! We do not have money to pay lawyers, and unless it's regulation that will literally make or break your company (see Waymo/Cruise/Zoox with California's DOT), it's not worth fighting them on it. Instead, it's best to be as open as possible to their interpretation of the law, take that back to the lab, and try and satisfy it. If they see a goodwill effort on your part to comply, chances are they'll be lenient and look the other way at minor transgressions.

My mistake: We wanted to expand our screens into NYC, and we knew there'd be regulatory hurdles. The TLC was notoriously anti-rideshare. They'd pass laws and policies that weren't consistent with how taxis were enforced, and no matter how much you complained, you were left shaking your fist at the sky. One of those laws was that any exterior "device" mounted onto a vehicle registered with the TLC had to pass an environmental test. This was to ensure that any externally facing device to the public was structurally sound and not susceptible to damage that could cause it to malfunction and cause disproportionate harm (e.g. explode). Things like water ingress, dust ingress, and inability to withstand vibration were examples of concerns, and valid ones too! However, you'd expect the regulator to know about the details of the tests you need to run and not just point you to some military-grade environmental tests that were massively excessive in scope, right? Right? Welcome to government incompetence. The regulator told me "You need to pass the MIL-STD-810G test", and upon further questioning of what this test entailed, he replied with "I'm not sure, just look it up." Fun! The test involves completing 30 sub-tests ranging from fungus exposure to gunfire shock. Makes sense for NYC, but I guarantee you all this wasn't necessary nor was it financially feasible. It would've taken minimum 6 MONTHS just to get these tests going and who was to guarantee we'd pass any of them. And the worst part is there weren't even parameters set for "Pass" or "Fail". I had to find a testing lab to give me a recommendation for those parameters, and I shared them with the TLC and they okayed it. They were arbitrary! You can tell that I'm still a bit fueled by frustration from 5 years ago, but it doesn't do justice to how difficult it was to deal with this. My mistake was thinking I could squeeze out more details from this guy beyond just "This is the test, get it done." I was able to get the number of tests required down to 12, and even though our screens did not pass with flying colors, with the right framing, we eventually passed and were granted access to NYC.

My recommendation: Treat regulators like they're babies. What they tell you is probably what they know. Try and find answers for them for questions you may have and they'll likely just adopt your answer as the word of God.

8. McMaster-Carr and Digi-Key are your Amazons

Perhaps one of the greatest struggles when dealing with hardware is not having all the nuts and bolts and screws you need immediately available to you all the time. It's hard to stay organized and I don't blame you. Most people default to thinking they can buy and receive things quickly via Amazon, but the quality of the products tends to suffer and are therefore not reliable when it comes to testing and drawing conclusions about the durability of your product. That's where McMaster-Carr and Digi-Key save the day. They both offer very fast delivery AND pickup in metropolitan areas, and the quality of their products at their level of affordability are second to none. Highly recommend ordering anything from them!

Slightly abrupt ending, but that's all the trauma dumping I've got for now. Hopefully new topics soon. Peace.