The 57 °C problem
A week with a thermocouple and a pile of competitor products, and the argument it started for taking the electronics out entirely.
I've spent this week with a thermocouple and a small pile of competitor products, and I want to write down what I found while I'm still annoyed about it.
The numbers
The clinical trials on heat therapy — the ones showing it works about as well as ibuprofen for menstrual pain — use what they call continuous low-level heat. In practice that's around 40 °C, held for hours.
Several of the cheap heated plush toys I tested settle at around 57 °C.
That gap matters more than it looks, because skin injury isn't linear in temperature. At 44 °C you have about six hours. At 51 °C you have about two minutes. At 55 °C you have under thirty seconds. The curve doesn't slope, it falls off a table.
Nobody is getting a burn from these in normal use — you'd move first. But "you'd move first" stops being true the moment you fall asleep on it, and falling asleep on it is precisely the use case.
Why they do it
I don't think it's malice. I think it's the demo.
A product that reaches 57 °C in ninety seconds feels powerful when you first pick it up in a shop, or in the first thirty seconds of an unboxing video. A product that holds 41 °C feels like nothing much for the first minute and then quietly does the thing you wanted for two hours. One of those is easier to sell.
What I keep coming back to
Every one of these problems is downstream of the same thing: there's a heating element in there, and an element will hold whatever temperature it's told to for as long as there's power. Which means the safety of the product is entirely a matter of the electronics someone specified in a hurry. Get the sensor wrong, or skip the shut-off to save eleven cents, and there is nothing underneath to catch it.
The alternative I keep circling is the boring one. Heat it once, in a microwave, and let physics take it from there. No element, no cell, no board, no firmware. It comes out at its hottest and gets cooler every minute after, which means the failure mode I've spent this week measuring simply cannot happen. You can still overheat it — that's on the label and I take it seriously — but you can't leave it on.
It's less impressive on a spec sheet and I've made peace with that. I'd rather ship something whose worst case I can describe in one sentence.