Should You Connect Grounds in an Isolated Power Supply?

[@andrasparanici5491]

I think these videos are a must for PCB Designers.

[@JohnScherer]

Thanks for all these videos you do and have done Zach! As a hobbyist doing PCB design mostly for the Amateur Radio community, its tough to find good information on best practices with regards to board layout, grounding, etc so most of the time I just wing it. That practice can lead to bad habits and your videos help correct some of that for me. Your videos are first rate and a big help for someone like me with little practical knowledge on the subject, they provide the answers to question that I wouldn't even know who to ask.. I wish you had a video series where we could submit a board layout to you and have you review it and provide feedback. That would really be a huge help!

[@kianisback2014]

Thanks because of you, I solved my problem

[@FailCommando]

Thank you very much for that video. I have actually asked myself that question a vew weeks ago. Also thank you very much for your great supportof students with the student license. It was a great help during my undergrad!

[@nishadrangnekar6933]

Thank you zach. This is the exact information i was seeking after for my Induction board design. I had faced issues with floating grounds earlier which i had used a y1 cap from my control ground to the PE. It had solved mostly my ground lift problems.

[@AltiumAcademy]

Glad it helped!

[@leeslevin7602]

Awesome 👌, thank you, my colleague and I were having a debate on this very topic, discussing if we should bond the chassis ground to 0V, Star configuration etc. Your explanation will put our minds to rest 😁.

[@muhammadkamran9152]

Nice and easy explanation, Thanks

[@AltiumAcademy]

Glad it was helpful!

[@shahzaibzahid665]

Great Video Man. Thanks a lot

[@alvarodemontes3818]

thanks, how do you calculate the value of the capacitor ?

[@Zachariah-Peterson]

The standard guideline is to use a cap that has a larger value than the leakage capacitance of the transformer. The idea is to have higher frequency noise propagate back to the power supply return, rather than radiating or passing to an exposed connector where it could pass through the user to ground, so it is up to you to determine what the lowest frequency in the noise spectrum is and size the capacitor accordingly to provide low impedance across that isolation gap. The capacitor is typically a Y-type capacitor (that's the most commonly recommended I have seen), but the DC leakage current across the cap can also be important as this could create an annoying ESD shock if the user touches any plugs or connectors exposed in the system. This might also cause your design to fail some higher-end EMC requirements, such as in medical equipment.

[@pouyaabdoollahy9218]

Very nice It was great, this amount of knowledge and information conveyed was unparalleled. I love your channel❤️❤️❤️❤️❤️

[@AltiumAcademy]

Thank you so much 😀

[@PartheevSaiBoddeda-u4o]

I have a question, why electrical tester led is ON only for few SMPS boards. Most of the smps adapters like mobile chargers are not detected by electrical tester. So what could be the issue?

[@atluntadimanthoni]

Hi Zach...Learning a lot from your videos...How to calculate the capacitors value....?

[@Zachariah-Peterson]

You just need to make sure the capacitance is larger than the copupling capacitance between the coils in the transformer. If you are using an isolated power regulator chip, then you need to know the capacitance across the isolation barrier in the internal portion of the chip, which can be quite small. When the capacitor is larger, the impedance for higher frequency noise will be lower, so it can more easily travel back across the isolation gap through the capacitor rather than the transformer.

[@crimpers5543]

an ESD event on the DC side ground will simply charge the capacitor. It's not going to "discharge" the ESD thru ground. You should use a resistor if you want to actually get rid of the excessive charge.

[@philiptong4978]

may be an inductor in series with the resistor to reduce that primary ground noise from getting in

[@yetanotherPC]

It will dissipate through the capacitor as leakage current over time

[@jluke6861]

great video. Thank you.

[@AltiumAcademy]

You are welcome!

[@uricohen5463]

you are amazing , i love your videos

[@Zachariah-Peterson]

Thank you very much!

[@luckyhanger1326]

When you ground the loads through capacitance and the input power through capacitance or EMI filter, how do you break the Current loop that set up?

[@Zachariah-Peterson]

The point is to control emissions at high frequencies by giving high frequency noise a path to complete a circuit. By directing that current where you want it you can reduce the inductance associated with that current loop and control coupling of noise into other circuits.

[@gbengaakeredolu7440]

hii, very insightful video indeed. How about if your system is purely battery powered, and there is no other power source around. There are analog inputs from field device. Does this require isolation?

[@mrechbreger]

I know this is pretty old but how would you classify the leakage current to eg. old applications like DVD players or VCRs? I usually measured like 80-90V on the enclosure with very low current of course. Enclosures are usually not grounded of those applications. The voltage will collapse as soon as it's properly grounded of course. Examples who made such applications are Sony, Panasonic, etc.

[@fadielnel]

Hi Zach, Thanks for the nice video. Just a question, if I may...let us assume an additional signal GND is needed on the secondary side, how do you deal with this? Do you use the same Y-rated CAP to connect it to earth, or do you use a net tie between itself and the secondary GND which is already connected to earth via a Y-rated CAP? Thanks for your insights!

[@Zachariah-Peterson]

You're kind of describing something like a fly-buck, where for example you might have multiple secondaries and a single primary where there is an AC input. I think you should be prepared to stitch those grounds together with Y caps in order to control radiated emissions. In terms of a prototype design, it's a simple matter to leave some through-holes on neighboring grounds and test with/without the Y caps.

[@PartheevSaiBoddeda-u4o]

Hello, Nice and clear explanation. BTW I have a question that I have an 12v 2Amp SMPS board in which the Y rated capacitor value is 102 and if I touch the DC output wires I am getting mild shocks. I have replaced this capacitor with 103 and now the situation got worsen than earlier. I want to apply the way you explained that I will keep the existing 102 capacitor in original position (with PGND and SGND) and then I will connect one terminal of additional 102 capacitor to PGND and other terminal of this capacitor to earth. Same I repeat for SGND as well. Is this correct approach? someone suggested that If you want to keep the EMC benefits of the capacitor, keep your output floating but divert leakage away from the output you should replace the existing 102 capacitor with two capacitors connected in series. Then connect the mid-point of the two capacitors to mains earth.

[@Zachariah-Peterson]

Without seeing your PCB layout I could not tell you with high confidence what is the problem. The two capacitors connected in series approach is normally a way to construct a filter on the AC input, not for connecting the grounds in an isolated power supply. But you can try it and see if it solves the output leakage problem.

[@Dragonmastur24]

Is there a recomended equation for capacitor sizing to guard against esd in these various situations? Any rules of thumb to follow? For an ideal design do you want a large inductor and a resistor in parallel with an esd cap?

[@Zachariah-Peterson]

There is not really a good rule of thumb other than to look at the capacitor and its ESL/ESR values as well as its parallel resistor (if present). Those will limit the rate at which the capacitor can charge/discharge given an input edge rate. If the expected transient you need to shunt is faster, then the time constant needs to be smaller to provide effective protection. Transients can be pretty slow but with high voltage, so the edge rate is still large. If you are connecting across two grounds (primary and secondary) in a power supply, then the cap just needs to be much larger than the leakage/winding capacitance in the transformer. However be careful with this because there is leakage current that can cross that boundary, and leakage current can reach the user at the output side of the power supply. Some industry EMC standards, particularly medical, limit the amount of current that can pass to a user via leakage.

[@setia2258]

Hi Zak, we have an internal power supply and connect the earth to the chassis. Then we have an MCU and a motor driver. Should we connect the mounting hole (GND DC) to the chassis that is already connected to Earth?

[@Zachariah-Peterson]

If the mounting hole is connected to the system ground then you should not do this. You may allow some return current to flow in the chassis, which could then create a shock to the user. Earth connections are also not intended to be used to carry return current, it is intended for safety only. Since you are bringing in earth I might assume you are also bringing in AC and you have an isolated switching converter, in that case the primary side ground must not connect to earth. There are exceptions to this, for example if you did not have an AC connection with earth, you would want to connect a metal chassis to your system ground to eliminate any floating conductor. I have seen this connection made with a small resistor (milliOhms value) between system ground and chassis ground via the mounting hole.

[@setia2258]

@@Zachariah-Peterson Thank you

[@beeslikehoney123]

Thank you, Zach! Super clear and helpful info. Just wondering though - is there ever a scenario to put a resistor in series or parallel to the capacitor that is connected between Secondary GND and Chassis GND? I seem to recall that being done on designs that I've come across. Thanks again.

[@crimpers5543]

Yes. A resistor will actual dissipate the charge. Otherwise a capacitor will simply hold the ESD charge.

[@sudheerkumar5966]

Very good sir

[@AltiumAcademy]

Thanks and welcome