Joining(or not) of signal and chassis(safety) ground

Log in or register to post comments
Go To Last Post
17 posts / 0 new
Author
Message
Klemko
Klemko's picture
Level: Hangaround
Joined: Fri. Dec 4, 2015
Posts: 264 View posts
#1
Posted by Klemko: Tue. Mar 19, 2019 - 09:15 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello,

 

i have an enclosed device which incorporates some 5V custom PCB circuits that control some relays and LEDs and also an IntelNUC mini pc that runs off a 230V. The whole device case is grounded to safety ground and the custom PCB runs on a isolated 5V DC converter and also the PCB is isolated from chassy. I had some problems whit mosfets on PCB not opening and closing as they should, because there was one bolt touching the case (IntelNUC has signal ground tied to the mini pc case chassy), so any noise on chassy(safety ground) was changing signal ground levels, changing gate->source voltage on mosfet, causing them to not open or being stuck open. Now i removed that bolt, all works fine, but someone adviced me to put a capacitor beetwen my signal ground and chassy ground, to eliminate noise from ground coming to my signal ground, but still keeping them at same ground potentional. But i dont think this is important, since my 5V/12V pcb runs indepentend from the 230V device and only conenction to other stuff is via USB to intel nuc, which i also isoalted from chassy ground, since it uses signal groudn to protect itself from EMI and other stuff?

This topic has a solution.

Tags:

Learning and Information, General Electronics, solved
Last Edited: Thu. Mar 21, 2019 - 07:56 AM
Top
Moe123
Moe123's picture
Level: Resident
Joined: Tue. Aug 7, 2018
Posts: 711 View posts
Posted by Moe123: Tue. Mar 19, 2019 - 09:30 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

for these issues i recommend asking Brian Fairchild.

He have a very good knowledge in this area

 

Regards,

Moe

Last Edited: Tue. Mar 19, 2019 - 09:31 AM
Top
Kartman
Kartman's picture
Level: 10k+ Postman
Joined: Thu. Dec 30, 2004
Posts: 23184 View posts
Location: Melbourne,Australia
Posted by Kartman: Tue. Mar 19, 2019 - 10:10 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Your 5V isolated converter is only isolated at DC. At AC it is a different matter. This is why you want to add a capacitor between the grounds to form a path for the AC currents. Otherwise they will find their own way - which is usually not what you want.
This is how you spell chassis. It is very difficult to read your improper spelling. Especially ‘whit’. The wrong word.

 

[Cliff: I corrected the thread title]

Last Edited: Tue. Mar 19, 2019 - 01:32 PM
Top
Klemko
Klemko's picture
Level: Hangaround
Joined: Fri. Dec 4, 2015
Posts: 264 View posts
Posted by Klemko: Tue. Mar 19, 2019 - 11:32 AM (Reply to #3)
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for info. Yes my "with" and "whit" always plague me :P 

Top
Klemko
Klemko's picture
Level: Hangaround
Joined: Fri. Dec 4, 2015
Posts: 264 View posts
Posted by Klemko: Tue. Mar 19, 2019 - 11:43 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Below is crude ground schematic. To sum up: Intel nuc and PCB share same Signal ground (via usb cable connection). Chassis ground is connected to ground on wall socket. The bolt was changed for plastic one, because it shorted Signal and Chassis ground.

Last Edited: Tue. Mar 19, 2019 - 11:44 AM
Top
Kartman
Kartman's picture
Level: 10k+ Postman
Joined: Thu. Dec 30, 2004
Posts: 23184 View posts
Location: Melbourne,Australia
Posted by Kartman: Tue. Mar 19, 2019 - 12:13 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Where’s the rest of the stuff? As such, the problem isn’t with the two grounds - its everything else that is causing currents to flow in them.

Top
Klemko
Klemko's picture
Level: Hangaround
Joined: Fri. Dec 4, 2015
Posts: 264 View posts
Posted by Klemko: Tue. Mar 19, 2019 - 12:32 PM (Reply to #6)
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

No one said anything about current flowing. I said that there is noise on chassis ground, which connects to ground that entire object is connected to, so if there are motors or any other high power conductive or other components running, they can generate voltage on ground line due to coupling , static or any other form of radiation or bad insulation on those other devices. And those spikes can disrupt my 0V signal ground, that mosfets use to operate. The ones i use require 4.5V on gate to work, so if due to noise on chassis ground (which was shorted via bolt) the signal ground get 1 or 2V on it or any spikes, can disrupt my mosfets. Which is the LEAST of problems that can occur. Since Chassis ground is tied to signal ground, which the intel nuc has tied to usb gnd, the stronger spike or current on chassis (generated on any of other devices in the palce where my device will be installed), can even damage or destroy either my PCB or Intel NUC electronics. My question is not about any currents or anything, but if it is really necessary to join my signal ground to chassis ground via capacitors. Also many places have bad ground, whit constant noise on it.  So since currently i dont have any noise or disturbance in the device, why should i pus capacitor between signal and chassis ground?

Top
Klemko
Klemko's picture
Level: Hangaround
Joined: Fri. Dec 4, 2015
Posts: 264 View posts
Posted by Klemko: Tue. Mar 19, 2019 - 12:33 PM (Reply to #6)
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am trying to keep noise on chassis ground OUT of my PCB signal ground, not to get any noise on my PCB signal ground to chassis ground. The noise (there may be or may be not) is on the CHASSIS GROUND.

Last Edited: Tue. Mar 19, 2019 - 01:23 PM
Top
Kartman
Kartman's picture
Level: 10k+ Postman
Joined: Thu. Dec 30, 2004
Posts: 23184 View posts
Location: Melbourne,Australia
Posted by Kartman: Tue. Mar 19, 2019 - 08:28 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I don’t think you understand the problem. You mentioned noise. What is noise then? Have you measured it? Do you know where it comes from? If it affects your circuitry, obviously current has to flow unless you believe in magic. I gather your system has some switched mode power supplies. These have leakage currents. It is more likely these leakage currents are the issue rather than ‘noise’. The evidence you provide suggests this.

Top
Klemko
Klemko's picture
Level: Hangaround
Joined: Fri. Dec 4, 2015
Posts: 264 View posts
Posted by Klemko: Wed. Mar 20, 2019 - 09:59 AM (Reply to #9)
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I really appreciate you trying to help, but please, take 5min and read again stuff i wrote down.

 

-I dont use switching power supply, i said in first post i use ISOLATED power supply.

-Noise is coming from SAFETY EARTH. You cant measure that since it is random, it depends from other users in the place where my device is connected to power and from the quality of earth connection in that place. 

-AGAIN: there is NO ISSUE IN DEVICE. If my device is powered via my high quality lab power supply, there is NO PROBLEMS. 

 

-MY question was: SHOULD I PUT IN CAPACITOR BEETWEEN MY SIGNAL AND SAFETY/CHASSIS ground?

Top
Moe123
Moe123's picture
Level: Resident
Joined: Tue. Aug 7, 2018
Posts: 711 View posts
Posted by Moe123: Wed. Mar 20, 2019 - 10:12 AM (Reply to #10)
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

Klemko wrote:

 

-MY question was: SHOULD I PUT IN CAPACITOR BEETWEEN MY SIGNAL AND SAFETY/CHASSIS ground?

 

Klemko, bottom line is, no body can give you an exact answer if you didnt post the exact schematic for your project. you cant just post a screeshot in paint and expect people here to understand.

 

When you give more and clear information, then you get more and clear information. I understand that you dont want to post everything due to confidentiality of the info. however post only the relevant technical info and what you see that you might be able to share.

 

Thanks a lot for your understanding.

Top
Kartman
Kartman's picture
Level: 10k+ Postman
Joined: Thu. Dec 30, 2004
Posts: 23184 View posts
Location: Melbourne,Australia
Posted by Kartman: Wed. Mar 20, 2019 - 10:13 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The NUC uses a switching supply. How are your other supplies ‘isolated’? How is the noise coming from safety earth? Current must flow in a loop so if your system was actually fully isolated, the noise on the earth would have no effect.
Use a multimeter to measure the ac voltage between the two earths. What is the value? Better still, use an oscilloscope.
In #3 i suggested you put a capacitor. I explained the reasons. Of course, the exact placement of this capacitor is critical.

Top
Paulvdh
Paulvdh's picture
Level: Raving Lunatic
Joined: Fri. Aug 3, 2001
Posts: 2802 View posts
This reply has been marked as the solution. 
Posted by Paulvdh: Wed. Mar 20, 2019 - 06:36 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The usual way PC's are wired is to connect the chassis / safety earth to the "GND" of the power supply.

 

The screw holes of the NUC board are probably also connected to it's GND plane, and those are normally simply all bolted together, but not relied upon for electrical connection.

There are multiple GND wires.

I just tested another small PC (Cubie board) and the mounting holes are also hard connected to signal GND.

Also tested an old 3.5" HDD and it's 2 GND pins are also hard connected to the aluminimum chassis.

If you have an metal housing that is floating, then it just forms a capacitor with everything inside and it does not do much shielding.

If you connect the metal housing to signal GND then noise is attenuated / shorted to GND.

 

Just think of how you connect all the GND's together.

If you connect the metal housing to signal GND on a PCB, then make sure at least one of the connections is near the power connector.

 

This does not make sense to me:

Klemko wrote:
-I dont use switching power supply, i said in first post i use ISOLATED power supply.

Almost all power supplies are switching nowaday's, and whether it is an SMPS or an old fashoned low frequency metal plate transformer has nothing to do with routing of different GND / Earth wires.

 

Klemko wrote:
No one said anything about current flowing.
Then you should start thinking about currents flowing, and where it flows, and how all those currents influence your circuits.

 

Klemko wrote:
The ones i use require 4.5V on gate to work,

Such MOSfets are not suited to switch from a 5V rails. Such MOSfets need at least a few volts more (8V to 12V or so) to have decent noise margins to build a RELIABLE circuit.

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

Last Edited: Wed. Mar 20, 2019 - 06:42 PM
Top
Klemko
Klemko's picture
Level: Hangaround
Joined: Fri. Dec 4, 2015
Posts: 264 View posts
Posted by Klemko: Thu. Mar 21, 2019 - 08:06 AM (Reply to #13)
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The DC/DC 5V power supply is galvanicly isolated. Switching power supply usually have (as far as i know)  GND on input and output connected, isolated have a transformer in between, so there is no direct connection on GDN, which prevents the noise coming from input side, to the output side. Well since MOSFETs are have very high impedance input on gate, i assumed they only see voltage and current doesnt play a big role. So if i put 5V on the gate, and on the source, there is 1V or 2V isntead of true GND(0V), then the minimum Vth of 4.5V is not achieved. I forgot to tell that the Voltage threshold for my mosfet is 3V, 4.5V is the voltage where it fully opens(Rds is at minimum). 

 

I assume i could add some mosfet boost driver IC in the circuit that would put 10V or more on gate, to ensure that V threshold is always high enough.

Or i could leave everything as it is and just add an optocoupler on the usb line, that comes from Intel NUC to my pcb, which would isolated my PCB GND from whatever is happening on the Intel NUC side.

I will use one of this 2 solutions, thanks all for help :D 

 

Last Edited: Thu. Mar 21, 2019 - 08:10 AM
Top
Kartman
Kartman's picture
Level: 10k+ Postman
Joined: Thu. Dec 30, 2004
Posts: 23184 View posts
Location: Melbourne,Australia
Posted by Kartman: Thu. Mar 21, 2019 - 09:16 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Transformers have interwinding capacitance. This couples through noise. Switchmode supplies are less likely to be grounded each side. Your presumptions are most likely incorrect. Did you use your multimeter to measure the ac volts between the earths? The results might be surprising to you.

Top
Klemko
Klemko's picture
Level: Hangaround
Joined: Fri. Dec 4, 2015
Posts: 264 View posts
Posted by Klemko: Thu. Mar 21, 2019 - 10:54 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am using this DC/DC power supply  NCS1S2403SC. I connected an oscilloscope to PCB 5V power rail and PCB GND, then data logged everything for 10hours straight. There was almost no noise on 5V and GND rails (90mV was max spike on PCB GND and it only lasted few uS). Once i connected PCB GND and Chassis ground/earth together and left again to datalog for 10hours, there were several spikes up to 10V which lasted up to 1uS and a LOOT of smaller spikes ranging from 1V to 4V and lasting up to 2300 ms.

haven't tried measuring ac voltage between GND and earth. Will try this now.

Top
Kartman
Kartman's picture
Level: 10k+ Postman
Joined: Thu. Dec 30, 2004
Posts: 23184 View posts
Location: Melbourne,Australia
Posted by Kartman: Thu. Mar 21, 2019 - 11:19 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That power supply is switched mode. The transformer windings have capacitance. You can measure the leakage with an oscilloscope. Anyways, i’d say this power supply is the least of your troubles. Was your oscilloscope isolated?
The NUC power supply is more likely to be causing your issues.

Top