People are missing the point here. Carls entire channel is filled with really cool experiments in making PCBs do things people wouldn’t normally think of. There doesn’t have to be a practical purpose really, he just loves experimenting.
I really love his flexible PCB experiments too, he’s made a lot of awesome things from them.
A self-heating board is more efficient than heating an entire enclosure. A sensitive circuit or part could use this trick to efficiently keep key componants warm on demand without an entire external heating system. And dont let Louis Rossmann about this. Inbuilt desoldering circuits underneath key componants might make at-home board repair too easy even for his tastes.
We had this use-case back when I worked for a pro-audio company building analog synthesizers.
One of the components of a voltage controlled oscillator (VCO) is a linear to exponential converter, that converts the 1V per octave control voltage to a per-hertz reset pulse for the sawtooth waveform.
Such components can be done in various ways, but one of the most traditional is to use the exponential part of a transistor's voltage-to-current transfer function (not sure what the actual term is here). The issue is that this function is very sensitive to temperature, and so fluctuation in temperature on this part can easily detune the VCO by a very audible amount. Fluctuation is less of an issue at higher temperatures (around 70-80°C), so the solution was: heat!
The trick we implemented in the synth was to use a transistor array in a chip: use two transistors for the Lin/Exp converter, another as a heat source (we nicknamed it "the oven"), and another as a heat measurement device, to get some sort of PID feedback loop. I believe there were five transistors in the chip, so there was a leftover that got used for another purpose.
Surprisingly efficient, though we could still hear some pitch drop when blowing on the transistor array chip. Once in the enclosure, we figured only applying some ice on the front panel would have an effect.
This "heating trace" in the PCB would have been a good alternative, as some of those transistors arrays were particularly hard to source (we needed a particular transistor type, can't remember which).
Oven-stabilized crystal oscillators are a time-honored technique for precision electronics.
It turns out that no matter how carefully you stabilize a circuit for temperature changes, it gets much more stable when held at a controlled temperature.
Arrangements for this go from simple (gluing the component to a PTC thermistor driven at constant voltage) to elaborate (thermocouple-controlled PID heater in multiple styrofoam insulating enclosures). There are some pictures at https://en.wikipedia.org/wiki/Crystal_oven
I've never seen PCB traces used for this, but I've always wanted to try making a clock from a piece of LCD color-change material laminated to a PCB with traces in seven-segment patterns.
If it's for an art project (where power consumption, noise, product lifespan and general sanity don't matter as much) just put it in an enclosure with a fan blowing across the digits programmed to turn on once a minute.
You can install toaster coils too. A pcb heater layer directly under a temperature-sensitive chip is much more direct, and thinner. I see a cold-weather cellphone with a touchscreen that can get up to workong temperature quickly.
exactly this. The creativity process is hampered by forcing every single possibility along the path to a new invention to something that is completely practical and useful.
On top of that he’s probably figured out a bunch of interesting things that he can apply to his next project and other things and so on. this is good stuff.
Only if the components actually remove themselves from the board. Bonus points if they make it to somewhere else in the room. Consider it like a faulty component quitting before it gets fired. I once had a bus transceiver that unsoldered itself from a card on an extender and ended up on the floor.
It's an entertaining hack. I think it should be considered in this context.
If we start thinking about it in practical terms it stops making any sense. Specifically, a standard 2 layer board costs few $ to make. Here we're paying a huge premium for extra internal layers. You may as well take that money and put it towards a hob with an scr and a cheap controller to solder them on it.
That actually depends on the specific costs. There is a clear business desire to replace capex with opex even when the resulting expenditure is higher. This achieves that at the pcb soldering level.
It does not have to be a good idea to get traction.
This is actually one of the designs we use at work for building connector-only PCBs (like what you'd find in track lighting raceways) because our FR4 board used for that application is so thin that it can't go into the reflow oven, it will flex and slip off the rails when it heats up. We just make one big PCB with an array of connector-only parts and a track heater built-in, run the board through the solder stencil, pull it out, place the connectors by hand, slap the board on a lapped-flat granite block with a heater under it, hook up power, and watch it until it reflows. We use an extremely-low temp bismuth solder that reflows just a bit above water's boiling point for these boards.
Nice to see hobbyists in the field coming up with the same concepts.
It's a great hack and a cool demo that may have (very) limited real-life niche applications, but it's not really worth it unless you only make a couple of PCB per year and can't afford a small oven.
Where this is great is that you can finely control the temperature of the board since you're in control of the current you pump into it.
Where it's not great is that it uses a full layer and could cause issues if any of your PCB needs special considerations, when using high frequency clocks or RF (say implementing a switch mode power supply can be quite sensitive to how your layers are composed). You also end-up with either a floating layer or a ground layer that has a long path instead of a real ground plane. That could also create noise issues.
If you assemble a few PCB a year, a PCB oven is fairly cheap. You can make one with a regular US$50 oven and a PID control kit or just buy a dedicated one like the T-962 (that can be further hacked and improved) for about US$250 from Aliexpress.
But that hack is still cool and Carl Bugeja's channel[1] has lots of fun projects that push PCB to new limits for hobbyists, showing what you can achieve with flex PCB in particular and using them as mechanical interfaces for creating motors, actuators, etc. It's a great channel!
The trouble with the T-962 is that the stock units heat so unevenly that you can scorch the center of the board while getting cold solder joints near the edges.
I have one of those with the mods for improved firmware and better thermocouple references. It's OK. The firmware runs the fan all the time, even during heating, which improves the heat distribution. There's a better mod which adds a second fan.
A number of companies in China made clones of those things, but none of them seem to have done a redesign to make a better one at roughly the same price. Now there seems to be a new generation of these things, with more fan power. That's a step forward. No idea if they're any good. (There are reflow oven review sites, as bogus as mattress review sites.) There ought to be a good $200 reflow oven by now.
You can buy good small reflow ovens from the US and Germany, but prices are upwards of US$3000.
There are relatively affordable vapor phase ovens- at $WORK I arranged the purchase of Vapor Phase One- it works really well. However, some people are more sensitive to the resulting smell.
My issue with vapor-phase is that it tends to cause a lot of tombstoned or sleeping components. It's pretty good for multi-leg components like power ICs and processors and BGA components, but not so good on two-conductor resistors, capacitors, jumpers, etc. This is mostly because with vapor-phase you can't really control the thermal profile of the reflow.
My understanding is that Galden vapor is dense, so you get a layer of hot vapor near the fluid, with lower temperature further from fluid (where there is less vapor)- the required thermal profile is maintained by controlling position of the boards..
From what I gathered- just dislike, mixed with some concerns. I didn't smell anything apart from the usual flux smell, but I have very little sense of smell.
You don't even need that; you can solder PCBs in a frying pan. It works very well. I never even tried an oven to be honest because a frying pan worked so well and Adafruit or someone like that said it was better than a toaster oven.
Although I do have induction hobs now. Not sure if a PCB would be too happy about that. But if not you can buy a portable mini hob for like £15.
Perhaps the particular T962a was just a bad product, but at $WORK we just could not get it working reliably across many types of PCBs- it really required tuning the profile to each type of PCB (small, large, black solder mask vs green, etc)- and yes, a colleague spent a day or two installing the cold junction compensation mod. From what I saw- there was just too large variance in IR emissivity/absorption of the PCBA and their thermal properties, which resulted in scorched connectors on one batch and raw/non-molten paste on another batch- with exactly the same profile.
I achieved much better results with just a (convection) toaster oven and multimeter thermocouple- I only added PID controller because I started using the oven for drying something as well.
I have been following this Youtuber for a while now, and I have been amazed by his creativity and unorthodox use of flexible PCBs. Strongly recommend watching a few more of his videos
Excuse me, but did he just make a cheeseless pizza!? I guess baked tomato sauce will bind toppings a little bit, but you'll lose all the texture of the sauce this way :(
Historical pizza is basically just any round bread with some sort of topping(s). Max Miller's Tasting History showed a 16'th century pizza recipe from the Vatican with powdered sugar and rose water for toppings[1].
This could be done with a normal ground plane by extending it out both sides of a board. A microwave transformer with the high voltage secondary cut out to make a 2 turn transformer could supply more than enough current... use a thermocouple the same way, and a Triac to control the primary, and Bob's your Uncle.
Doing a row of vias on both sides then paralleling the top/bottom traces to make some big T shaped terminals might make it easier to get uniform current distribution.
This misses the entire purpose of a ground plane. The point of a ground plane is that return currents can follow roughly the same path as the forward going currents at higher frequencies, so the inductive loop the combined path forms is as small as possible. Also, a real ground plane can provide shielding by allowing currents that oppose electromagnetic fields, but these currents have to go in loops also.
Yea, perhaps you could be clever with that as well later on...? Probably not the most precise design, but it seems honestly a lot less hassle than heating an oven.
It isn't really clear in the video, but that is a 4 layer board and there is another ground plane layer elsewhere. The heater loop doesn't connect to any components.
To what else then ? Leaving it unconnected would be worse, leaving it at ground level makes it at worst act like small capacitance to ground for nearby stuff
It's a cute trick but I can't imagine a worse ground plane. People say ok -- this is fine for low frequency circuits. There is no such thing as a low frequency circuit when there's a microcontroller present since the clock edges have frequency content well into the GHz. The only way I could imagine getting away with this is if you sacrifice a layer to the heater and put a ground plane between it and the rest of the PCB stack as a shield. Still I think it would be better to scatter resistors around the board and heat sink them to a solid ground plane. I don't know. I love basically everything Carl does but if anyone is considering this seriously they need to learn a bit about Maxwell's equations.
Yes. The magnitude will be lower than an STM32 running at 100 MHz but they’re there. They’re usually benign and tiny but when your return path is a loop antenna then there’s a chance there will be issues.
That was fantastic. As far as I can tell it's an original idea which in this area is rather cool by itself.
I guess I'm curious how much worse the heater-shorted-into-grounding-plane performs as compared to a classic full grounding plane.
If anyone is curious about the cool-looking pcb holder used in the early shots, that's the Omnifixo [1]. It's been on my wanted list for a while but it's not cheap and often sold out (I think it's basically a one-man company).
Thanks for mentioning Omnifixo- when you said it's not cheap, I mentally prepared myself for a price tag of several hundred EUR. I find the ~60EUR price very reasonable (IIRC a single stickvise with PTFE jaws is around the same price point).
Well, I did mill the PTFE yaws for stickvise myself, and was seriously contemplating building a stickvise style vise (as I pretty much have the needed tools)- until I needed some Digikey part ASAP and had to fill the cart to free shipping threshold :)
YouTube has been recommending this channel to me for a while. It was a long journey to get there, with a lot of prototype hotplates made out of PCBs along the way. Worth watching if you're bored some night.
The "screen" he's using in the video is called a stencil. They're made out out of thin stainless steel. These days they're pretty easy to come by, even for the hobbyist.[0]
Yes, I thought this was pretty common now. I do it every time as it hardly adds any cost.
It’s also worth looking at the costs for SMT assembly. Companies like JLC are amazingly cheap for fully assembled boards (provided you stick to their standard parts catalog).
I was almost going to do that for one of my projects, I got my design right on time for chip crisis to start....
But yeah, even for something like 10 boards the prices for assembly has become pretty reasonable, especially if you design for the parts they have in stock.
I really love his flexible PCB experiments too, he’s made a lot of awesome things from them.