Wire overheating

[lem0nhead]

Hi there!
More than 1 year ago I made this post while trying to make a project for a switch for a water heating element. Pretty simple stuff, but it was my first time dealing with high-power (220V 13A) on a PCB.

Turns out after some tries and high temperatures (even using just wires on the PCB, without tracks high-current) I stopped using a PCB, decided to get some double-pole 30A relay and wire them, controlled by a very simple PCB with a ESP32 controller.

Everything worked fine for about 3 months (temperatures ~55°C/130°F), then the box started to overheat (I know because I monitor the ESP32 temperature). I opened it and noticed obviously rusted/overheated wires, which are the ones connected to the 30A ACS712 current sensor. They were above 150°C/300°F when I opened.

From the image, is it obvious that I should just replace this ACS712 current sensor or perhaps I'm missing something?

[isometrik]

That little ACS712 current sensor board appears to have 0.1-inch (2.54 mm) traces between the IC and the screw-on terminal. Even with dual-sided traces, that does not look sufficient to handle that level of current without heating up significantly.

I would suggest you change your current sensing scheme to something like a split core current transformer. That type of sensor will also offer a much better isolation from mains.

As a final comment, I find your setup frankly dangerous (no protection fuse, high current going through slim PCB traces, etc.). It seems you disregarded many of the comments offered in your previous thread. If this setup ever causes a fire, I would not be surprised to see your insurances reject your claim.

[Daniel Perez LW1ECP]

An additional hint: I see you are using terminals to connect cables. This is not good practice: increasing the number of contacts current has to traverse decreases reliability: a bad contact breaks the circuit from there on. It is better to make a daisy chain without cutting the conductor, just removing the insulation. And apply good torque to the screws.

[lem0nhead]

As a final comment, I find your setup frankly dangerous (no protection fuse, high current going through slim PCB traces, etc.).

maybe I'm missing something, but there is no high current going through PCB traces (except in ACS712, if that's what you meant)
as for fuses, power comes from a fused socket switch

[lem0nhead]

I see you are using terminals to connect cables.

Do you mean the blue terminal? The cables are connected together there also because they need to supply (low-current) to the AC-DC Hi-link component that outputs 5V to power the ESP32

[james_s]

Those wussy little green terminal blocks aren't good for anywhere near 30A, maybe 3-4A tops. This is especially true of random cheap generic ones as you most likely have.

[andy3055]

It is clearly visible that you have very poor contact in that screw terminal. Use a heavy duty spade lug and crimp the two wires. Use a terminal like what you have at the top left in the picture. In fact, you should use spade lugs everywhere on that board that can be really tightened.

[Siwastaja]

You mention two 13A heaters, is it just 13A or really 26A going through that cheap-looking green terminal block?

[lem0nhead]

13A. The cheap looking green connector is for the current sensor, so it's one wire in, other wire out (the current sensor just intercepts the wire)

[Siwastaja]

And this damaged wire is the same brand / size wires as the others, yes?

If so, the heat has to be generated in the contact site at the screw terminal, and just conducts through the short wire, damaging the insulation throughout the whole length of it.

I would not trust that cheap green thing beyond 8-10A but I'm still a bit surprised if it causes so much damage at just 13A.

Make sure the wire strands are all fully under the spring/contact pressure plate/whatever it is called (I'm assuming screw is not directly pushing the wire down; if it is, then you must use ferrules), and the screws are tightened properly.

[james_s]

I'm not surprised at all. I've used those screw terminals for lots of things, they're handy for low current low voltage stuff. They're cheap and flimsy with thin contacts, I would never trust them to carry 13A, especially at mains voltage, and even more so powering a heater which is a nearly ideal ballast for sustaining an arc. I have seen daisy chained receptacles burn up powering a 1500W space heater down the line on multiple occasions and the terminals on those are a lot more robust than those green screw terminals.

[d-smes]

Sure looks like left screw head is much higher in the green connector body than the right screw head.  So... is the left screw even tightened?

[wizard69]

My perspective here is going to be a bit different, especially with respect to the first comment.  So"

• I consider it poor design to put 30 amp relays on a printed circuit board.   I know people will argue that it is perfectly fine if properly designed but I'd suggest that it is a bit silly to put a wear product like a relay on PCB, instead put a DIN rail in the control box and mount some DIN rail relays.
• None of the AC wires in this installation look like 14 ga or larger.
• Some codes or practices specify no more than one wire per screw.
• Terminal blocks have current ratings and sometimes those are not realistic, don't overload the terminals.
• Ring terminals, spade terminals and wire ferrules can be wonderful things.
• Proper torquing of electrical screw connections is required and frankly is a huge problem in the electrical world.    Get a torque wrench and learn what proper torque feels like for each terminal.   Reputable suppliers with have torque charts for every screw on their components.   Respect those torque values!!!!   In case you are wondering this is an area where regulators are focusing on because so much damage to electrical systems is caused by improper torque of the mechanical connections.
• I have to agree with others; an off board current transducer is probably a better idea.
• No fusing or other protections circuits??   

In a nut shell This looks to be very problematic design wise.   I have to wonder if there is any over temp protection anywhere in this implementation.

[Siwastaja]

Good comments;

1) even 100A is fine on PCB, but I suppose you meant it's not a good idea in a beginner circuit, or a Chinese Ebay module. This I do agree with.

Really, controlling HVAC water heaters should be done with DIN rail contactors because they are dirt cheap. There is no need for current sensing, the heater is nearly constant resistance and unlikely to fail. This simplifies the AC wiring. Install the DIN rail contactor in a proper distribution box and bring (double insulated) coil wiring out of the box for the Arduino-ESP-whatever control thing which runs off a good-quality wallwart.

3) Two same-size wires per screw is usually OK and allowed. If more than two, or different sizes, the smallest one(s) do not get decent pressure.

6) Yes, both under- and overtorquing is a problem.

[Doctorandus_P]

What version of the ACS712 do you have? It comes in versions for 5A, 20A and 30A.

As others have already mentioned, the little green PCB connector is underrate, and not fit for 13A.
The Blue connector on the same PCB (to the Hi-Link) is also underrated, because it carries the full 13A current to the relays.

The relay's themselves look good, and the black connectors on them also look good, but all connectors on the matrix board should be upgraded.

Or, if you connect the incoming 230Vac wire directly to the big connector on a relay board, then you can add another wire under the same screw for a lower current connection to the Hi-Link module.

I also advise to buy a box of ferrules and pliers for them.

[Gyro]

Just another observation, you haven't included any strain relief on the mains cables. From the sharp angle, those wires going to blue terminals are definitely under strain. The box has provision for fitting proper cable bushes but you haven't used them.  It's vital for safely that all mains wiring is properly secured.

[lem0nhead]

Thanks everyone for the amazing feedback!

I now see those are not sufficient, but just in my defense, some protections I took are:
1. As I mentioned, the mains comes from a fused socket switch, so at least there's some protection against high-current
2. There's no physical protection for high temperatures, but the microcontroller (which is installed inside the same box) is programmed to disconnect the relays and shutdown if temp > 70°C / 160°F (which I know is far from ideal - but that's what caught this issue)

Based on the feedback, I agree a din rail contactor would be better, but I've tried finding some before and couldn't find one controlled by 5V. Yeah, I could use small relays just to trigger the coil, but that adds complexity to the project that I planned on removing (and staying safe) by getting those sturdy relays.

For now I removed the ACS712 (it was the 30A version, but I think the manufacturer never though about changing the connectors, just the sensitivity) which indeed looks bad; and I'm now connecting live and neutral directly to the relays (without going through the blue connector first - they go to the blue connectors (low-current required) splitting from the sturdy connectors).

I'm considering adding some sort of thermal fuse, but I obviously need to do more research

[langwadt]

It is clearly visible that you have very poor contact in that screw terminal. Use a heavy duty spade lug and crimp the two wires. Use a terminal like what you have at the top left in the picture. In fact, you should use spade lugs everywhere on that board that can be really tightened.

https://www.elecbee.com/en-26223-ACS712-30A-Current-Sensor-Module-Board

[tooki]

I'm considering adding some sort of thermal fuse, but I obviously need to do more research

The overheating issues you had in the original thread would have been enough to stop me dead in my tracks. Sure, you then added temperature monitoring, but that’s really not good safety design: if you KNOW your design is overheating, the right thing to do is identify and rectify the cause of the heating, not to add a safety net that hopefully triggers. It did this time. But what if it hadn’t? You could have burned down your house, which you stated is an apartment, meaning that a fire could theoretically kill other people in the building. You’d be morally and likely criminally responsible for their deaths.

Is this going to the absolute extreme worst-case scenario? Absolutely. But it’s not impossible, and given the nature of problems you’ve had already, it’s not as unlikely as you apparently think. (I say apparently, because if you didn’t think it was unlikely, you would not have continued using the device.)

[Siwastaja]

I would strongly agree with the above. You should be playing with safety extra low voltage stuff, and not try to control your boiler.

And screw regulations and paperwork, I'm fine with ghetto breaking the law stuff. But if you do that, you need to know exactly what you are doing. You can't afford mistakes which can cost lives or serious property damage. My rule of thumb is, if you do something you are not licensed to do, you need to do better than the professionals!

Question every connection, every component in your design, and if you can't prove to yourself it is done to the code, and safe, don't do it.

[lem0nhead]

Just to clarify, I removed the ACS712 that caused the issue, now 13A is only carried through the sturdy relays. Mains are plugged to the blue connector, but it only takes a few milliamps that are converted to 5V.
Current measuring was a nice feature, but I can live without it (and since it's only on/off 13A, I suppose I can detect it easily with a non-contact current detector, but I'm not doing this now.

Before doing this, the alternative I considered was using SONOFF or Shelly (they're rated for >15A) and honestly my setup seems better than theirs (I opened a SONOFF, the current is basically carried through a soldered track and they use connectors similar to those blue ones).

I understand there are risks, but are those really higher than a professional solution such as those I mentioned?

[james_s]

Those Sonoff devices unfortunately have bullshit ratings too. I tried using a Sonoff Pow to switch a 14A load and it burned up in less than a week, I should have known better.

You really have to be careful with this stuff, mains voltage at tens of amps is no joke, houses burn down regularly due to electrical faults. You have to be very careful to make good quality connections and make sure that everything in the chain is good for the current you are dealing with. High quality crimp ring or spade terminals are a good idea here. If you want to measure current that is easily done with a current transformer that you simply pass one of the wires through.

[lem0nhead]

Based on the feedback here, I think I improved the design a bit (the loose wires are going to be connected to the controller board on the other side of the box).

Wires are now using a ring connector; and I'm using a wago 221 to splice the wires.

I know it's not perfect, but I think the other big improvement would be using DIN rails contactors, which would be a big project change

[Gyro]

That looks rather better. At least you no longer have high current connections joined by being side by side in low current terminals.

Without meaning to nitpick, you still have zero strain relief on the mains cables and case holes that you could drive a bus through. It would of course fail electrical inspection in Europe for access to live parts. It really wouldn't hurt you to use the proper cable bushes that the box is designed for.