What I learned selling 2,500 MIDI recorders: Hardware is not so hard

我在卖出 2500 台 MIDI 录音机后学到的：硬件其实没那么难

What I learned selling 2500 MIDI recorders, part 1: Hardware is not so hard. July 19, 2026. A year and a half ago, I launched Jamcorder. A Jamcorder, in situ. It marked the completion of two life goals of mine. First, I finally had the piano recording device I’d always wanted: a fully automated device that captures everything I play, no human-involvement required. And second, after a career in software, I got to build hardware. It was super fun. 2500 units have already been sold, and counting! People genuinely love it! And it’s able to stand on its own feet as a business. In my eyes, that’s a success. For my first blog post, I want to reflect on the biggest surprise I encountered creating Jamcorder.

我在卖出 2500 台 MIDI 录音机后学到的经验（第一部分）：硬件其实没那么难。2026 年 7 月 19 日。一年半前，我发布了 Jamcorder。这是 Jamcorder 的实拍图。它的诞生标志着我完成了两个人生目标。首先，我终于拥有了梦寐以求的钢琴录音设备：一个全自动设备，无需人工干预即可捕捉我弹奏的所有内容。其次，在从事软件职业生涯后，我终于有机会制造硬件了。这过程非常有趣。目前已经售出了 2500 台，而且还在增加！人们真心喜欢它！它作为一项业务已经能够自负盈亏。在我看来，这就是成功。在我的第一篇博客文章中，我想反思一下我在创造 Jamcorder 时遇到的最大惊喜。

Hardware is not so hard

硬件其实没那么难

Coming from a career in software, I expected building hardware to be the hardest part. After all, hardware is hard. That’s the saying, right? Electronics design, plastics, manufacturing, fulfilment, component shortages, etc, etc, etc. I expected these & more to make hardware hard. But, it wasn’t. To refine the process, I hand assembled the first 500 units myself. It took 4 days. Unexpectedly, everything went completely smoothly and I made no changes. Yes, really. I kept waiting for something to get me by surprise. A scrapped production run (my worst nightmare). Or component sourcing issues, perhaps? It never happened. (Though Trump’s tariffs were a close call).

由于我之前的职业生涯都在做软件，我原以为制造硬件会是最困难的部分。毕竟，“硬件很难”，这可是行话，对吧？电子设计、塑料模具、制造、物流、零件短缺等等。我预料这些因素会让硬件开发变得困难。但事实并非如此。为了优化流程，我亲手组装了前 500 台设备，只用了 4 天。出乎意料的是，一切都非常顺利，我甚至不需要做任何改动。是的，真的。我一直在等待意外发生，比如生产线报废（我最担心的噩梦），或者零件采购问题。但这些从未发生。（尽管特朗普的关税政策曾让我虚惊一场）。

The hardest part of building Jamcorder was still, by far, the software — roughly 200K lines of code spread across the firmware, app, and manufacturing tooling. It took over 3 years and many long nights in a pre-LLM world. When compared to that, the hardware was undeniably smooth sailing. For the record, I don’t think I’m special. It’s just that hardware’s reputation for being difficult is, IMO, overstated.

制造 Jamcorder 最困难的部分依然是软件——大约 20 万行代码，涵盖了固件、应用程序和制造工具。在 LLM（大语言模型）普及之前的世界里，这花费了我 3 年多的时间，度过了无数个不眠之夜。相比之下，硬件开发简直是一帆风顺。声明一下，我不认为自己有什么特别之处。在我看来，硬件开发“很难”的名声被夸大了。

Now, granted, Jamcorder is — very much intentionally — a simple device. I get that. The Jamcorder PCB. 25 unique components. The MIDI connectors are made to order, everything else on the PCB is off-the-shelf. Assembly is just a single screw, for a single PCB. The injection mold has generous draft, no slides. I cut low battery detection, ambient light detection, the power-button, even USB-C. All these things kept Jamcorder simple. I’m also under no illusion that 2500 units is a lot by most standards. But you know what? The hardware side would have been easy regardless.

当然，我承认 Jamcorder 是一个非常简单且刻意为之的设备。我明白这一点。Jamcorder 的 PCB 板只有 25 个独立组件。MIDI 连接器是定制的，PCB 上的其他所有零件都是现成的。组装只需要拧一颗螺丝，固定一块 PCB。注塑模具采用了大脱模斜度，没有滑块。我砍掉了低电量检测、环境光检测、电源按钮，甚至 USB-C 接口。所有这些设计都让 Jamcorder 保持了简洁。我也清楚，以大多数标准来看，2500 台并不算多。但你知道吗？无论如何，硬件方面的工作依然会是轻松的。

Don’t get me wrong. If Jamcorder was 10x more complex, or 100x more scale, that would be a different story (and a very popular product!). Or if I was trying to compete in the smartwatch market, or the car market, or any number of very established industries with low margins, good luck. But for me the take away still is: “hardware is as hard as you make it”. If you’re thinking about building a hardware product & have a way to protect your margins, don’t let building hardware scare you. It’s not as hard as the saying goes.

别误会我的意思。如果 Jamcorder 的复杂度增加 10 倍，或者规模扩大 100 倍，那将是另一回事（那会是一款非常受欢迎的产品！）。或者如果我试图在智能手表市场、汽车市场或任何利润率极低的成熟行业中竞争，那祝你好运。但对我来说，结论依然是：“硬件的难度取决于你如何设计它”。如果你正在考虑制造一款硬件产品，并且有办法保护你的利润率，不要让“硬件很难”这种说法吓倒你。它并没有传说中那么难。

POSTSCRIPT

附言

I don’t want to end this article without some practical take-aways. So here are my top recommendations for successfully shipping hardware that worked for me at medium scale:

• Keep your BOM simple.
• Avoid single-manufacturer components wherever possible.
• Avoid complex assembly and calibration.
• Partner with a Chinese assembly house and suppliers. Alibaba is your friend.
• Aim for at least 70% gross margin or more.
• Keep your company lean. Scaling hardware is slower.
• Have a strong anti-counterfeit strategy. Don’t overlook this.
• Do final Q/A in-house & hold finished inventory locally.
• Request samples before every production run.
• Write a step-by-step manufacturing & assembly guide with pictures.
• Keep your packaging small. A value dense product makes everything easier.

我不想在没有提供一些实用建议的情况下结束这篇文章。以下是我在中等规模生产中成功交付硬件的几点核心建议：

• 保持物料清单（BOM）简洁。
• 尽可能避免使用单一供应商的组件。
• 避免复杂的组装和校准流程。
• 与中国的组装厂和供应商合作。阿里巴巴是你的好帮手。
• 目标毛利率至少达到 70% 或更高。
• 保持公司精简。硬件扩展的速度较慢。
• 制定强有力的防伪策略。千万不要忽视这一点。
• 在内部进行最终质量检测（Q/A），并在本地持有成品库存。
• 在每次生产前都要索取样品。
• 编写一份带有图片的详细制造和组装指南。
• 保持包装小巧。高价值密度的产品会让一切变得更容易。