Question for BobZ and others skilled in design, especially at higher voltage levels.
The attached is a circuit I’m laying out… that portion of the board in the vicinity of J3 is at mains voltage level. Circuit-wise, how far back would the ‘danger’ zone extend?
It appears to me, that I need to take extra care as far back as R1, R60, and K1 common. Am I correct?
This coupling section is a necessary part of the design, but I want to get it right so I don’t risk electrical or personal damage…
if I understand this right you are limiting the operating voltages coming out of a capacitor sypply.
However limiting these voltages does not mean you are limiting potentials against earth (the user) ! Your whole circuit is potentially dangerous to the users touch.
But if you only refer to functional isolation the voltages may be uncritically after the zeners.
glad there is an expert in these matters. Let me take advantage of your expertise. How about that:
well, it is difficult to judge this without a deeper insight into the project. But it seems you have taken precautions seriously.
it all depends on if someone is allowed/capable of touching one of the connectors while another one might be AC powered !
Are your cutouts and the remaining distances specified according to an official and up to date regulation (it is up to you to define and find the regulations for your product. I often had to personally visit VDE for guidance in these matters) ?
The “Netz 12V” may come out of / lead to an unprotected circuit ? It may need better insulation against “Netz 230V” ?
You are using big upright THT resistors to bridge the gap. They can easily be bend and touch each other. I sometimes had to insert an isolating foam or similar inbetween to GUARANTEE correct assembly.
Such resistors should be “fusible” ones. when damaged they will turn into an open state without excessive heat, fire or smoke. Fusibles are not well established and more expensive …
That’s understood. I just wanted to see if there is something severely wrong.
It’s a personal project. Only person handling that is me. Of course that won’t make much of a difference regarding insurances in case of e.g. a fire.
So there we go with another cutout…
Even if they did, this would mean 150K between live and neutral which is well within the resistor’s power ratings (and a HUGE margin).
Thanks for the hint aboute fusible resistors, I did not know about them (MCU and processor boards guy here )
No. But as this is no product, I’ll go with it. The copper area on the board is GND and will be tied to earth, so any excess creepage will trigger the FI ( security device in the mains installation, I do not know the englisch term).
Also check voltage rating. A 0.5 W Yageo metal film might have a maximum working voltage of 300 V
Thx for the heads-up. Vitrohm 0414 2w rated 500VDC. There are 2 in series. Even if one shorts out, it should be ok with a margin.
That is a better choice. My point is that some resistors have surprisingly low voltage ratings. Always check the datasheets when you have high voltages, especially with SMD
That’s what I learned, too. So no SMD here. Not even 2512. They sit directly on the board as well, so if something goes wrong, they cook the PCB.
These are resistors in series.
(now I understand why ‘thread hijacking’ is frowned upon)
SMD capacitors rated for several kV do exist, but I don’t know if they exist with for example an X2 rating.
SMD capacitors are quite brittle, and can break when a PCB is abused (For example bent, when pushing with a screw driver into some connector). Ceramic SMD capacitors also tend to fail shorted, which is not nice for Mains voltages. Even ceramics can smoke (or the PCB around it).
The solution for the bent PCB is to place mounting holes (& hardware) in such a way that the PCB does not get bent when poking the connector with a screw driver, and also keep the capacitors far enough away. There are also guidelines for the distance between ceramic SMT components (also resistors), and V-grooves because of this.
To keep it simple, indeed, don’t use SMT capacitors for mains voltages on PCB’s.
On my board it was about fat 0414 resistors mounted vertically …
No cap anywhere in the 230v part.
I’m not sure I get the PCB pictures correctly but:
Assuming the relays and connectors (not visible on the screenshot) will be assembled, are you sure you gain something on isolation by placing cutouts? Plastic body of component will “short” the air gap on top layer. Soldermask or air gap <1mm are not increasing creepage, so if plastic component touches or is close to PCB the creepage path is measured along the body, no matter you add the cutout below or not (except if the plastic has much better CTI and cutout is wide enough take advantage of it)
If overvoltage category is II, pollution degree 2 and Vpeak is not much above 1.41*Vrms then I guess the cutouts are not needed at 230V (check standards).
Didn’t think of that, thanks. How about the cutouts underneath the relays? Paranoid?
check the specific part, most parts that are expected to work at higher voltage have little molded standoffs for exactly this reason so that the physical part doesn’t mitigate anything added to help the creepage.
It also allows conformal coating to flood under the part, or bonding agent
Always learning. I checked, and even the Aliexpress relays I ordered have those bumps.
I’ve really learned a lot from the community’s contributions.
Thanks to all!
I guess there was nothing to particularly call my attention to this post, and I just happened to find it.
Even though I have designed many power supplies I will not claim to be a safety expert. I will take a conservative tack on this. I see everything in this circuit more or less referenced to one ground, and unless proven otherwise I will take the entire circuit to be in the danger zone. Also I do not see that R1 or R60 form any sort of a safety barrier.
Thank you for that information.
I do have a schematic of an alternate coupling arrangement using a toroidal coupling transformer instead of the capacitor at C33. Then I could make the hot / cold split underneath the toroid. I assume that would be a better approach?
This is just for my own use - making a ‘copy’ of an obscure piece of test equipment with updated parts.