Tiny Hotplate Isn’t Overkill

When working on a new project, it’s common to let feature creep set in and bloat the project. Or to over-design a project well beyond what it would need to accomplish its task. Over at Black Mesa Labs, their problem wasn’t with one of their projects, it was with one of their tools: their hot plate. For smaller projects, an 800W hot plate was wasteful in many ways: energy, space, and safety. Since a lot of their reflow solder jobs are on boards that are one square inch, they set out to solve this problem with a tiny hot plate.

The new hot plate is perfectly sized for the job. Including control circuitry, it’s around the size of a credit card. The hot plate is powered from a small surplus 20V 5A laptop power supply and does a nice 4 minute reflow profile and cools off completely in under a minute. Compared to their full-sized hot plate, this is approximately 29 minutes faster, not to mention the smaller workspace footprint that this provides. The entire setup cost about $20 from the heating element to the transistors and small circuit board, and assuming that you have an Arduino Pro sitting in your junk bin.

It’s a good idea to have a reflow oven or a hot plate at your disposal, especially if you plan to do any surface mount work. There are lots of options available, from re-purposed toaster ovens to other custom hot plates of a more standard size. Overkill isn’t always a bad thing!

31 thoughts on “Tiny Hotplate Isn’t Overkill

  1. Hey now… I don’t normally mention this kind of error, but this instance is pretty extreme, so maybe drawing attention to it will help people become more self aware of themselves when making the same mistake.

    I’m pretty sure just about everyone hated studying “significant figures” in school, but that doesn’t justify the subtraction there. If the hot plate took about 30 minutes to cool down, and the miniature one cools down in “under a minute”, you really can’t say:
    “about 30 minutes” – “under a minute” = “approximately 29 minutes”

    Using direct quotes from the article makes it worse. There’s no reason to expect that the difference between “30+ minutes” and “just seconds” is “approximately 29 minutes”. In fact, it seems much more likely to me that “30+ minutes” – “just seconds” = “30+ minutes”.

    It’s analogous to saying there are “about 40000” people at a concert. If ten more people show up, that doesn’t make it “about 40010”. It’s still “about 40000”. What’s the difference? Saying “about 40000” implies that there’s between 35000 and 45000 people there. Saying “about 40010” implies that there’s between 40005 and 40015 people there, which conveys a false sense of precision.

    Similarly, “approximately 29 minutes” implies that it’s between 28.5 and 29.5 minutes. Meanwhile, “30+ minutes” doesn’t carry that level of precision, even though one might say “under a minute” does. Regardless, in the scope of the uncertainty of “30+ minutes”, a change of “just seconds” gets lost in the margin of error (as it should).

    As an alternative, saying that it cools down “approximately 30x faster” would be OK.

    Don’t even get me started on when a politician wins an election by a number of votes that falls well within the margin of error.

    With regard to the project itself… I want one! It just feels wrong to waste energy by using huge equipment for small jobs. This is definitely going on my “eventual weekend projects” list.

      1. I didn’t celebrate the new millennium, I celebrated the new year. But that’s not your point.

        I suppose you’re drawing attention to the fact that the year following “1 BCE” is “1 CE”, rather than “0”. This convention is awful. Something created in 8 BCE is aged 7 years in 1 BCE, 8 years in 1 CE , and 20 years in 13 CE. Quick subtraction (13 minus negative 8) yields 21 years of age, which is wrong. Everyone is expected to remember to adjust their result by subtracting 1 whenever crossing the BCE/CE boundary.

        By contrast, using ISO 8601, that same thing would be said to have been created in -0007, which makes the subtraction much more clear (+0013 minus -0007).

        Furthermore, as represented in accordance with ISO 8601, the year preceding +0001 is +0000, which does, in fact, make 2000-01-01 a logical date for the celebration of the new millennium.

        But if it were up to me, the year would start in February (so that the names of the X-ember months reflect their numbers), there would be 13 months, each of exactly 28 days (makes weekdays consistent from month to month and year to year), and days and months would be zero-indexed, with one or two (leap year) bonus days at the end of the year which fall outside of any month or week. Zero-indexing the days and months might seem absurd, but we already do it with years, hours (24-hour clock), minutes, and seconds, so why not make it all match?

        …but that’ll never happen.

          1. According to wikipedia, it was used as far back as the 15th century– first known use by Johannes Kepler in 1615 who used the latin form, i.e. vulgar era. Jewish academics started using the modern CE (common era) notation in the mid 19th century to avoid implicitly invoking “year of the lord” in all their dates.

          1. Good point, but I blame the seven-day-week for that. (365=2*2*7*13, so there’s no way to avoid that factor of 13 without ditching the 7 too by switching to a product other than 365.)

            I’d be happy with 12 months of 30 days each, where each month consists of six weeks of five days each. Then each year gets a bonus week (five days), possibly plus a leap day. Heck, there’s a lot of ways to break down 360. Keeping things in the right ballpark and maintaining the five-day-week (so there’s something convenient to append to make 365), we could also have 18 months of 20 days or 9 months of 40 days.

          2. [Pat]: regarding the Egyptian calendar, I wonder if there’s any relation to the number of degrees in a circle, which I believe also dates back to a mathematically-inclined ancient civilization, though I forget which one…

            [Ren]: If it were that simple, then the full moon (and the new moon) would always be on a Monday, for example. It would work if the lunar cycle was 28 days, but it’s actually 29 and change. We’d need a bonus day every month (#29) and a “leap” day roughly every other month (#30). Come to think of it, that’s remarkably close to what we have now, except we have one too many days in each month, and they’re still counted as parts of the week (which disrupts the “full moon on Monday” schedule). I’ve often heard that the months of the year are supposed to correspond to the lunar cycle, but again, we’d need 29 and 30 days per month to make that work, not 30 and 31.

            A side thought: it just occurred to me that a 28 day month would also make “that time of the month” very literal… for the average woman, it would fall on the same days each month. A quick search for the distribution suggests that the average is also the mode. (Not a woman, so I’ll refrain from commenting on whether this would be a good thing or a bad thing.)

          3. It all happened around the same era, a few thousand years BCE. We got 360 degrees in a circle because there are roughly 360 days in a year, and that’s the nearest multiple of 60, a highly-divisible number and the basis of the Sumerian numerical system. Ancient civilizations loved numbers like 12, 24, 60, 360, etc. because they can be split up so easily, and to a culture without calculators, that’s a big advantage.

    1. There is no margin of error in a vote because 100% of voters vote. Note, I am not saying that everyone that can vote, did vote, but that 100% of the people whose vote counts, because they actually voted, have voted.

      There is a margin of error in polling, which takes a sample of those voters, say 100 of them and then you can look up the margin of error for a sample of that size from the chart.

      1. If that were true, then recounts would yield identical results to the original tallies (in practice, they do not). You make the assumption that every vote will be tallied correctly.

        Occasionally, the number of votes cast exceeds the number of eligible voters in a given area. I’m not sure if the reported margins of error are intended to account for such discrepancies. I also don’t know if they account for the possibility of voter fraud or electronic machine malfunction.

  2. I think that cooling that fast is not a good idea. If you look at reflow charts you need a controlled cooling as to not cool too fast. No reason why the hot plate can’t do that. But in this project the heating process probably doesn’t follow the curve either. Other than that looks great. I mostly do small boards. I lucked out and my dad found a $300 hot plate at the thrift store for $10 which I am going to use just for rework or recovering parts.

    1. No, this is backwards. From the solder’s point of view, you want as fast a cooling as you can get, since the more time you spend in the transition zone, the larger the grains are that form, and fine grains provide a better mechanical bond.

      So it’s really just the components that limit how fast you can go down, to avoid thermal shock (no dunking in an ice bath). The rampdown rate that people use is typically “4 deg C/second”, but realistically that’s because that’s what free air cooling basically gets you. So this is probably around that value as well.

  3. The 1″ Hot Plate is indeed real. Black Mesa Labs designs multi chip modules using castellated via technology instead of bulky connectors ( the MCMs look like radio modules ). The modules are typically 1″x1″ or smaller – just enough to break out a chip’s 0.5mm pitch IO to either 0.100″ or 0.050″ pitch castellated vias around the perimeter. Larger and much simpler interposer PCBs then hold and interconnect multiple MCMs together.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.