Catch up on interesting hacks from the past week with Hackaday Editors Mike Szczys and Elliot Williams. This week we discuss the story behind falling lifetime ratings for LED bulbs, look at finite element analysis to strengthen 3D printed parts, admire the beauty of blacksmithing, and marvel at open source Lidar development. We delve into great reader suggestions for Blue Pill projects sparked by last week’s podcast, discuss some history of the V2 rocket, and cover Chromecast control hardware, glowing home chemistry, K40 laser cutter add-ons, and more.
Links for all discussed on the show are found below. As always, join in the comments below as we’ll be watching those as we work on next week’s episode!
Take a look at the links below if you want to follow along, and as always, tell us what you think about this episode in the comments!
Direct download (60 MB or so.)
Episode 5 Show Notes:
New This Week:
- Great comment thread on Blue Pill (STM32 dev board) and what you can do with it can be found on last week’s podcast article. (The comments are new since that episode was released)
- uTaster mass storage bootloader for STM32 (video demo)
- Elliot mentions the STM32 Nucleo has a breakoff STlink programmer
- Swapping out an aerospace-grade STM32L443 onto this $2 board.
- FOSDEM talk on the Blue Pill: Microcontroller_Firmware_From_Scratch
- Revisit this Hackaday article about building RC from salavged parts: Cheap Toy Airboat Gets A Cheap RC Upgrade
Interesting Hacks of the Week:
- Back To Video Basics With An ESP32 VGA Display
- Finite Element Analysis Results In Smart Infill
- Related reading by Moritz Walter: 3D Printering: Non Planar Layer Fdm
- Blacksmith Elevates The Craft With This Fabulous Strongbox
- Open Source Lidar Lets You Get Down To The Nitty Gritty
- This Chromecast Volume Knob Has A Certain ’70s Chic
- Make Your Own Phosphorescent Material
- K40 Gets A Leg Up With Open Source Z Table
- Simple Automata Extravaganza
- This Blinken Grid Is All Analog
- A 3D-Printed Robotic Chariot For Your Phone
- Elliot wrote an article looking for exactly this kind of hack: Ask Hackaday: Why Aren’t We Hacking Cellphones
- What Happened To The 100,000 Hour Led Bulbs
- Operation Backfire: Witness To The Rocket Age
- Paper on Laser diode lighting (PDF) (Elliot’s note: the higher efficiency from lasers only really comes into play when used as RGB, rather than reflecting off phosphors, hence the need for direct-diode green lasers. I misspoke in the podcast.)
11 thoughts on “Hackaday Podcast 005: Undead Lightbulbs, Home Chemistry, And The Strength Of 3D Printing”
When the standard Edison socket LED bulbs started to come on the market, there were soooo… many bad products. They weren’t even heatsinked properly; many manufacturers simply took the LED dies and drove them at the nominal current without any sort of de-rating and simply printed the datasheet figures on the product description. Problem being that the datasheet figures are measured at 25 C junction temperature, which is valid for about 1 milliseconds after you turn the light on.
Based on the nominal figures, they said they were 100,000 h bulbs and X Watts with Y lm/W efficacy, color temperature of such and so, but none of that was actually true. It was simply not standardized, and accurate labeling wasn’t mandated, so for a number of years the LED products on the market were just pants on fire frauds.
That is btw. one of the neglected aspects of the Phoebus cartel story as well: back in the day, many manufacturers who claimed 2,500 or 5,000 bulbs were simply lying. Consumer protection wasn’t so strong at the beginning of the century.
The goals of the cartel were to ensure a certain *minimum* quality but also to stop any company from making bulbs *too good* to ensure they all had a steady income from replacement sales even though before WW2 the market was growing rapidly due to rural electrification projects around the country.
Yes, though for the long term for survival, the cartel could not mandate bulbs that were basically rip-off – otherwise they would have been found out simply by consumers complaining about bulbs that pop constantly and cost too much to replace. In the end, they made the correct compromise as far as the physics of tungsten filament bulbs go, and the subsequent efficiency evaluations and standards set up by government research groups came to the same conclusions about lifespan vs. price vs. cost of energy.
So what really is the moral of the story?
Also, for the beginning of the century in mind, electric lighting was still competing with gas light. People (who could afford electric light) still had gas lines in their homes ready to provide the light, at comparable prices and other benefits. For the bulbs that could last for 2,500 hours or more, the light was no better than gas light – it was dim and yellow-red, and it cost a pretty penny for the low efficiency. This was at a time when people were sold gas radios operating on the seebeck/peltier effect because they were still cost-effective against buying electric service or recharging batteries at the local shop.
Gas radios? Hold my beer, I’m heading to Wikipedia…
Even worse: some Chinese manufacturers/vendors printed the datasheet nominal/max. values on the box and ran the LEDs with much lower current. The result was much less light than expected.
That open-source LIDAR would go well with this paper about our “internal GPS” (grid cells) since in essence that’s what we’re trying to do with navigating robots.
“The brain, she says, is an “inference generating organ.” She describes an increasingly well-supported working hypothesis called predictive coding, according to which perceptions are driven by your own brain and corrected by input from the world.”
Right now you can buy the Blue Pill on AliExpress for less than US$1.50/each, but the STM32F103C8 in the same package is US$3.47 with a 2,400 unit minimum on Digikey. What is the actual mechanism by which that happens? Are these surplus or scavenged chips?
Also, I’ve always found it curious that the Blue Pill, otherwise apparently designed with cost as a primary driver, includes a 32.768 oscillator for the RTC. It’s not an expensive part, but it’s also not really necessary even to run the RTC. Why not leave it out and save a few pennies?
Just been listening to the podcast and looked up exactly the same thing as I wondered about getting some for prototyping with at work (to ultimately use in final products).
While devboards are often sold as a bit of a loss-leader to get users hooked on the devices, I can’t get over the disparity between the chips (even in large quantities which is what I was looking for) compared to the cost of the devboards. Might have to prod our supplier of STM chips to see what pricing they have access to and see if they’re just as expensive from them as I’ve been using a cheap, small 8051 as my ‘throw-away’ embedded micro of choice so far (EFM8UB10F16).
Okay, having asked around – the consensus is because a lot of them are not actually genuine STM32s.
They’re legitimately designed asics (i.e. correctly licensed ARM cores, etc) but, as ST can’t stop them from doing it, they’ve just replicated the same register map tied to identically functioning peripherals. Some people may then be re-badging them as the real thing but that’s a different issue.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)