Control boards for 3D printers are a dime a dozen on the usual online marketplaces, and you usually get what you pay for. These boards can burn down your house thanks to a few terrible design choices. [Scott Rider] aka [Crow] took a look at the popular Melzi board, and what he found was horrifying. These boards overheat right at the connector for the heated bed, but the good news is these problems are easily fixed.
The Melzi board has a few problems with its PCB design. The first and most glaring issue is the use of thermals on the pads for the heated bed connector. In low-power applications, thermals — the method of not connecting the entire top or bottom layer to a hole or pad — are a great idea. It makes it easier to solder, because heat isn’t transmitted as easily to the entire copper layer. Unfortunately, this means heat isn’t transmitted as easily to the entire copper layer. In high-power applications, like a connection to a heated bed, these thermals can heat up enough to melt a plastic connector. Once that happens, it’s game over.
Other problems were found in the Melzi board, although you wouldn’t know it just by looking at the Eagle file of the PCB. [Scott]’s Chinesium Melzi board used 1-ounce copper, where 2-ounce copper would be more appropriate. The connector, too, should be rated above the design power loading.
[Scott] made a few tweaks to the board and also added a tiny DS1822Z temperature sensor to the high-current area of his version of a Melzi. Imagine that, 3D printer electronics with a temperature sensor. Slowly but surely, the state of 3D printer electronics is clawing its way to the present.
31 thoughts on “Bad Thermal Design And Burning Down The House”
Love the different eyebrows on the silk for the screw terminals on the new PCB ^__^
I believe they are meant to be the Pac Man ghosts (?)
if the maker wants to save money instead of omitting copper on the board why not do a walmart?
work the employees off the clock (oh we need you to go do this (once they punch out))
rig the electric meter to get free electric (theft of service from china electric)
instead of naming it Melzi name it of one of the more respected names like makerbot or davinci
I’m betting your comment was partially sarcastic, but I wanted to share this for others:
Planned obsolescence is really not as much of a thing as people think it is…
Engineers get a material requirements document full of the objectives the thing they’re making has to meet.
Cost point is always on there, because the product always has to be sold, and cost drives the minimum price you can successfully turn a profit with.
So is weight, battery life, size, minimum operational lifetime, etc… All of those things are interrelated and some are mutually exclusive:
* A handheld soldering iron isn’t going to meet 1,000,000 hours of battery life.
* A $20 coffee pot isn’t going to meet a 15 year operational lifetime.
“Planned obsolescence” is just conspiracy talk for “you get what you pay for” 9 times out of 10.
Granted… there are documented cases out there of entire development groups sticking it to the consumer, but for most products, it’s just trying to meet the market demand and turn a reasonable (or unreasonable) profit.
New washing machines and dishwashers shouldn’t self-destruct after 4 years. They just shouldn’t.
That’s why every 5 years I do the $20 repair on my old ones… Absolutely horrible cheap electronic controls on everything new. It’s not that I’m a luddite, it’s WHY THE HELL CAN’T YOU DO THIS PROPERLY!!!!
I don’t know about products killing people, but I’d largely guess that it the cause of the length of the warranty is a dependent variable on the quality of the product. Nobody is going to warranty a product for double the lifespan they expect it to live for. What you’d do is get an engineer that’s dedicated to determining failure causes and have them asses the product and unless the lifetime of the product is going to be too low you have them estimate what say 95% of the products will live to, and set the warranty accordingly. This doesn’t mean the product was designed to fail just after the warranty ends, but in fact the other way around. The warranty was chosen so that it covers the product during it’s expected life only. This means that the product will have a warranty against premature failure or manufacturing problems but not beyond that.
Actually the $20 coffee pots are more likely to have an extended operational lifetime than the “fancy” ones, Very little to go wrong. I won’t buy one with a timer or “digital control” or any of that crap, because they fritz out a month after warranty up, whereas I get solid service from the cheaper pots. I do have an idiot problem though, where they leave the pot off the plate for an extended time and tend to cause failure that way. If I was using them myself 15 years would be a given. Record at the moment is about 7 I think between idiot cafetiericides.
Warranty on cars ends pretty quickly in relation to a cars lifetime. Still manufacturers go way out of their way to contact vehicle owners (via vehicle registration authorities) if there is a defect that may cause harm to people. Even if it’s way out of the warranty period. Getting a letter for a recall after 10 years of vehicle life is nothing special.
I think that is more a function of the cost of lawsuits vs the cost of the recall. Remember that scene in Fight Club? I’m sure it’s not so black and white as the movie makes it out to be. Consumer confidence and goodwill (in terms of money) I’m sure are also factors.
Other consumer products enjoy the same kind of “protection”. Lots of products are recalled after their warranty expired, even after long after they stopped being manufactured at all. I imagine it’s all a function of the monetary impact on the company.
The issue at hand is not the restricted heat transfer but the heat generated by high current flowing through the narrow thermal reliefs. TFA got it right.
Don’t just accept the default if it is critical. Someone that have time to draw eyebrows should spend more time looking at the gerber instead. :P
– You can play with width inside polygon fill and Supply under DRC. When you have different width requirements for fill, you can actually use 2 separate fills and overlap them.
– One can also add thick tracks to the pad and/or add additional at 45/135 degrees. This gives a bit more control.
I doubt this is an intentional mistake. Most likely an ignorant board layout engineer. If you aren’t experienced with high current designs, then it’s an easy mistake to make. But it is also under-tested. Good product validation would catch this mistake.
Yes. Never attribute to malice something that is easily explained by stupidity.
So Melzi melts, eh?
No, only Chinese cheapo Melzi does.
I have an original one and no problems whatsoever – but mine came with screw terminals and better designed PCB. Someone in China didn’t bother to copy the original design properly.
Look as the original layout of a genuine Melzi and compare it with the melting clone:
Well, my first Melzi from England a few years ago went harakiri after a few prints, when I was still printing calibration prints. Just a burnt area, a few parts left and the magic smoke was gone. I got a replacement board but I built a stand alone external direct regulator for the heated bed with propper cooling and overkill tracks instead of using the internal. Worked for years since without problems. But it was a bad design that might have been improved since then.
“Makers” + “Chinese Clones” = Epic Fail!
Unfortunately most makers don’t have the funding to buy domestic. I sure didn’t. $200 for a Chinese clone, verses $2,000 for the exact same thing sold by a local retailer.
No, no, no. There is nothing inherently wrong with using thermal reliefs on through-hole pads. There is a whitepaper by SynQor (manufacturer of high-current DC-DC modules) that has some excellent math on the subject. http://www.synqor.com/documents/appnotes/appnt_Thermal_Relief_Study.pdf – see “Example 2” on page 4. They claim only a 5C rise with four 2:1 length:width ratio, 1oz spokes carrying 60A (15A per spoke). Anecdotally, I have run many tens of amps through thermally relieved pins with negligible temperature rise.
The real problem here is the cheap connector and/or the subpar soldering connecting it. In my own experience, both by applying the calculations in the above whitepaper and observed in real products, the temperature rise in the connector itself is always a stronger driver of heat than the spokes, unless the spoke geometry is purposefully made unreasonable. In fact, removing thermal reliefs altogether is actually a detriment to joint reliability and conductivity, because the planes tend to suck heat away from the joint, preventing solder from fully filling the barrel.
This smells like truthiness, I’ve had problems with ATX PSUs where they put 100W worth of shitty connector on a theoretically 400W supply and it’s burning at 200W draw…. Higher quality PSU/Connector and you can pull 250, 300, and it’s cool to the touch still. Sometimes it’s crimping, sometimes it’s a narrowing of the terminal between crimp and clip, sometimes it’s very small point of contact between clip and pin.
Sometimes, you’re just trying to perform the impossible (Because you’re a hacker dammit) and you have to go to a solution like Cool Amp conducto lube to improve pin to clip conductivity and drop heat.
This is all wrong again.
The problem here is that the heat is being generated in the first place and adding thermal conductivity is *NOT* the solution.
High current connection need to be crimped. The absolutely *only* time you should add solder to a high current connection is after it is crimped for metal to metal compression connection and in that case the solder is *only* to add some mechanical strength or environmental exclusion and not in any way being added for solders electrical properties.
In the pictures I see two different types of connector blocks, one has larger square pins and the other has smaller flat pins. If you ever ever expect a through hole connection to carry current to a connector the the hole in the PCB *HAS TO* very closely match the pin size to minimize the amount of solder the current has to pass through. Putting small pins into larger holes is a catastrophe waiting to happen.
Here are some figures to support my claims.
The resistivity of copper is 1.7 micro-ohm centimeters
The resistivity of good old 60/40 Tin Lead solder is 3 micro-ohm cm
The resistivity of lead free solder is 12.3 to 14.5 micro-ohm cm
My advice is to use the on-board connector to control a Solid State Relay (SSR).
The SSR has screw terminals so that you can attache wires with crimped connector lugs because this is the *correct* and only safe way to connect high current connectors.
“…60/40 Tin Lead solder…”
D: LOL, just making a joke.
Some examples for those wanting to design these boards – these aren’t the best but that are better.
These come in vertical or horizontal. The predominant connection to the PCB is *NOT* solder. You bend the pins out after inserting them so the the metal of the connector makes physical (and electrical) contact with the copper of the PCB through hole. The current connection to the connector needs to be on the bottom to avoid issues with though hole plating quality. You then add solder for *mechanical stability* even though the solder can and does help electrically.
You need to look for pins that are more towards square and further from rectangular. More rectangular pins increase the force required to bend them which increases the risk of damage to the PCB also more rectangular pins forces you to use larger through holes and that substantially reduced the electrical benefit of the solder.
But *most* importantly, both of these connectors are made to connect to a wire that has a *crimp* connector attached.
You might be a little angry at my 250amp+ connections that are all soldered …
Sooooo… What it the real difference in temperature in the joint on a PCB at 30 amps when the connection is 1.7 mohm-cm versus 3mohm-cm for half the distance?
Your figures really doesn’t support your claims unless you take them all the way and show the actual temperature rise. I’m sure it will be a difference, but if it ends up a degree or two higher it really doesn’t matter…
Why it matters is because the resistance of solder increases with temperature.
I’ll do the math another way if you tell me the power dissipation of the hot bed.
So how is the power MOSFET attached to the PCB? Is it crimped?
The FETs are reflow SMD so they have large contact surface area with a very thin slither of solder between.
Even though hole FETs are less of a problem as the FET leg size and hole size are predictable.
However it seems that the connector holes are large in an attempt to have one size fits all and that is a major problem with thinner connector pins.
I tried to download the board files so that I could give a better answer but they wont open on my computer.
There are a number of problems here:
– Screw terminal blocks inappropriate for the load current are used.
– 1oz copper is typically used on cheap products, although that’s probably not a big deal.
– Thermal reliefs are often used where they shouldn’t be, but that’s probably not the main problematic factor.
– The soldering is often crap.
– Excessively thin wires are used for high-current loads.
– People keep using terrible designs like RAMPS.
– People keep using the absolute cheapest junk that they can find sold for $5 on BangGood.
– Cheap crappy power FETs are used, made worse by the fact that the gate is usually driven directly from the MCU at only 5V or 3.3V, not driving the FET fully on, and leaving the drain-source on-state resistance higher than it needs to be, (and it’s already high, in a crap FET) therefore increasing power dissipation in high-current applications.
– PPTC thermistors are used for overcurrent protection, and they are terrible except in very low-voltage, low-current applications such as protecting USB ports.
– The stranded wire is poorly terminated into the screw terminal. The wire is loose, lots of bare wire is exposed from the terminal block, the wire is too thin to begin with and/or half the conductor strands have been cut off stripping the wire, half the strands aren’t inside the screw terminal, and no bootlace ferrule is used. This seems to be the main factor.
This is the biggest problem I see where 3D printer hobbyists have come to me with a melted connector on their 3D printer controller.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)