Question regarding GND

To power up some small electronics I love to use the 5V USB chargers from China (here € 2.85, 1000mA rated but in realty limited to 600mA). Plug, LED, cable with plug - everything you need. Power dissipation as well as acoustical and electrical noise are OK.

But they have a huge drawback when it comes to some feet of wiring for sensors / actors: The output is floating and has a common mode voltage of 115VAC at 50Hz plus some mV at 40(?)kHz.

Because of the high internal resistance (>15MOhm) it’s not dangerous to touch the wires, however, it seems that touching or some (capacitive?) coupling to ground may interfere with the electronic circuit and cause malfunction. A mobile phone also can cause troubles in this situation.

Once the DC supply is grounded (in any form) all the troubles seem to vanish.

I’ve tried to find out why and how the floating supply is bad but measuring with my simple (USB) oscilloscope is useless because the scope is grounded so everything is fine when I connect it.

To clamp the DC to ground I have two variants in mind, one is a bit complicated but would immediately indicate any second ground connection (a loop, may be dangerous or cause other problems), the second variant is simple (but seems to work, too).

So my questions are:

  • What is the better solution in regards to HF - interference, what about element’s values or is this kind of grounding generally bad / to be improved?

  • What would I have to do to test if my circuit is “foolproof” in respect to household / small industry environment, I don’t mean against any professional RF jammer in the kW - area ;-) ?

I know that for everything there is a limit but it’s really bad if you don’t know how far you are away from “failed”.

Thus I’d appreciate any thoughts / hints / lights into my AC - darkness.

Regards, Sancho

by Sancho_P
March 06, 2013

Really, no one?

@Signality? No hint?

Regards, Sancho

by Sancho_P
March 11, 2013

Sorry Sancho, get back to you ASAP.

by signality
March 11, 2013

Not quite sure what's going on with your PSU yet.

If you plug in your PSU and small circuit but don't switch the PSU on, do you still see 115VAC 50Hz in the same way as when it is switched on?

If you have a 2 channel scope, what happens if you probe the 5V with one probe and the 0V with the other and subtract the two traces in the scope?

Make sure:

1) both probes are properly compensated:

http://www.picotech.com/applications/how-to-tune-x10-oscilloscope-probes.html

2) the channels are set to the same gain

3) neither channel is overloaded.

4) connect the probe ground clips to each other but to nowhere else.

Hmmmm.

I have an old IBM Thinkpad with a cheap Chinese after market switchmode charger. It works OK but it blots out my BT broadband hub if I try to run the laptop with the charger connected to charge the laptop with the power on.

If I charge tha laptop with the laptop itself off then all's well.

I suspect the problem with it is that it has no proper mains input filtering to stop switching noise being driven back up the mains wiring.

Not helped by the BT Hub2 having a (failry well publicised) conducted susceptibility problem, i.e. stuff noise on the wires on its mains input wires and it falls over.

I wonder if you have a similar problem that just looks like an earthing problem related to the 115VAC leakage signal ....

"I know that for everything there is a limit but it’s really bad if you don’t know how far you are away from “failed”."

Sadly that is how most electronics is supplied.

"One of the things that usually get lost in the cloud of dust we are left in as the deadlines stampede by is visibility of how much margin the product has before it no longer works”.

(A phrase I originated in an email I sent 21st Nov 2012)

And:

“However, design for SI and EMC is only one part of the process. The use of correct measurement techniques is crucial to SI. Poor measurement techniques can hide true signal behaviour, falsely indicate - or even be the cause of - bad SI in a circuit.”

from

http://signality.co.uk/siemc.html

by signality
March 11, 2013

Thanks, @Signality, so at first you focus on the PSU and the source of trouble it causes?

To clarify the model or type, there is no switch at these chargers, to give you an idea what they look like: My actual charger is very similar to http://www.amazon.es/König-Cargador-USB-para-iPod/dp/B001T6O10Y/ref=sr_1_252?ie=UTF8&qid=1363042357&sr=8-252 (here without any specs), respectively http://www.amazon.de/Ladegerät-Adapter-Netzadapter-Steckdose-Player/dp/B000OC0X4M/ref=sr_1_2?ie=UTF8&qid=1363041332&sr=8-2 (but mine is “specified” to 1000mA and has a mini USB cable attached)

They are very lightweight, I had a look inside, seems to be a step down with small transformer and optocoupler feedback.

Yes, it’s strange that the DC output is symmetrical to the 230VAC, but I’m afraid that’s a result from how to measure the voltage.

Using my digital multimeter it is 115VAC. My scope is a PicoScope 2205 with two channels, probes 1:1 / 1:10 (10M, 47/15pF), both ground clips are connected (DC resistance about 2Ohm ?) and are connected trough my iMac to (mains) ground (DC about 4Ohm).

I did the proposed compensation and trace subtract, the result is given as fairly 4.7VDC with quite some noise, the channels reading 10 and 14VAC (noisy). It doesn’t matter whether I connect the ground probes to ground or not (already grounded by my Mac).

In respect to ground my scope reads 20VAC from the +5V output and 25VAC from the 0V output - thus I think the reading depends on the load. Could be a capacitive coupling via the transformer?

I have heard about problems using such chargers with smart-phones when they are switched ON, starting / stopping apps or “touching” the screen so you can’t use the phone while charging. I think these effects result from capacitive coupling to the table or user’s body with these tiny handheld devices.

With your laptop and hub (+router?) you didn’t say how they are coupled, cable or wireless, and if the router remains operable for other computers (if it is a hub, too).

But both would be feasible, noise brought to the hub via the networking cable (if there is any) or mains, as the power spikes to run the computer will be high compared my small device which has practically no effect on the AC supply. You may check that when using a 50m drum of extension cord to supply hub or PSU.

I did not really understand your “I wonder if you have a similar problem that just looks like an earthing problem related to the 115VAC leakage signal ....” because I can’t see how my “drifting” 5VDC would go together with spikes on the supply.

Just now I realize a funny “crosstalk” from my PSU’s DC to the mouse of my Mac: When I touch the DC using my left hand and try to grab the mouse (a “mighty mouse” with USB cable) by my right hand a “click” can be heard from the mouse and Exposé jumps in ( a Mac feature, usually you’d have to squeeze the mouse).

… Oh, now I see, RFI, SI, EMC and power supply are your domain!? :-)))

Regards, Sancho

by Sancho_P
March 12, 2013

The 115V leakage is almost certainly just capacitive coupling from primary to secondary of the SMPS in your adapter.

I doubt that is the cause of the problem. What's more likely is that there's HF noise being blasted back up the mains out of the input pins of the adapter. This is probably also coupled to the output side and will be picked up on the USB cables as well as radiated out of any other mains cabling.

You touching the connected kit or other equipment may just be conducting more RFI from you as an aerial (antenna) picking up the rubbish and then coupling it into the victim kit (phone, ipod, etc.).

It only happens when you have your adapter plugged in or maybe only when it's plugged in, switched on and supplying a load so you only tend to think of the problem as affecting load you are powering from the adapter.

You could try using an extension cable and looping the mains cable to that through a ferrite ring (common mode choke):

http://en.wikipedia.org/wiki/Choke_(electronics)

http://www.murata.com/products/emc/knowhow/pdf/26to30.pdf

http://atnm.mcars.us/HomeBrew/WB4BYQ/CommonModeChokes/

Another idea is to split the L & N wires and put each through a ferrite ring separately (differential mode choke or two single ended line chokes).

That might reduce the noise being fed back into the mains.

"With your laptop and hub (+router?) you didn’t say how they are coupled, cable or wireless, and if the router remains operable for other computers (if it is a hub, too)."

Laptop connected to BT HomeHub2 by wireless.

Hub and laptop charger on same mains but widely separated wall sockets.

Broadband connection completely lost for all PCs etc. because the Hub is completely reset to power up state (power light only) and never progresses to re-establish connected status.

Situation persists until the laptop or the charger is switched off.

There is a nasty trick in certain goods to get the EC EMC approvals with a sample unit which has all the right filters etc. fitted and then to ship product with those essential bits not fitted and wire linked out!

Hey! Here's a thought. You're not using any Power Line Modems (PLM) for Ethernet are you?

PLMs are well documented as causing serious conducted EMI problems. They are allegedly the main reason BT developed the HomeHub3 (bearing in mind my own experience with Hub2) and then promoted it as being soooo much more immune to RFI etc. ... largely from their own PLMs!

http://www.ce-mag.com/archive/03/ARG/hansen1.html

http://www.ce-mag.com/archive/03/ARG/hansen2.html

" ... RFI, SI, EMC and power supply are your domain!?"

It's all just analogue to me mate!

Push any squeaky clean digital signal fast enough and you'll see it for what it all really is: analogue.

:)

by signality
March 13, 2013

Interesting and frightening reading, thanks. Fiber optics of course requires cables, however, I think it’s much better and cheaper - Although I lost quite some hairs already in such (digital) communication issues regarding your “it’s all just analogue” hint, especially (but not only) in the fiber optics!

I do not have PLM here but I live in an extreme countryside - environment with a supply “quality” you wouldn’t believe it’s Europe.

Um, you think the reason is HF noise from the PSU out into mains. OK, I’ve tried a small ferrite ring, first both wires (one winding), then single wires, but nothing changed. I tried a better quality charger, slightly better results (reliability of my circuit) and common voltage was 85VAC (115 with the chinese one). So I was back at the 50Hz AC idea.

In the meantime I have added the protection circuit with the LEDs (variant a) and found out that even 5m of unconnected ground wire to this circuit would help to make it more reliable. This brought me back to the HF-introduced-to-mains idea and I tried to use my scope with 2x1MOhm voltage divider to find HF at the PSU input. But I could not find any at the phase (nice sinus today) and a few at the N (about 15VAC 50Hz ! to ground) which are not from my PSU.

So I’m back at zero.

However, variant a) works perfectly when grounded. Not sure if I ever will find out why ... :-(

Regards, Sancho

by Sancho_P
March 13, 2013

On the perils of cheap PSU's:

http://www.edn.com/electronics-blogs/power-supply-notes/4409903/How-to-avoid-being-ripped-off-when-buying-power-supplies

by signality
March 14, 2013

Yep, one should know the product’s inner values. I agree that the price / brand may be kinda help. On the other hand, each of us already bought “quality” and was ripped off.

My PSU has all required labels. No one is responsible, of course. Neither importer nor politicians, only for taking money. That’s our system, called capitalistic democracy.

My “better” PSU (not much better results) still has 85VAC common mode and I don’t know if that’s the reason for the problems. However, I’m thankful that the cheap one made my circuit fail, because I found that my circuit was bad, incomplete.

And I’m happy that it happened on my workbench!

But back to my original question, would you say one variant is better than the other or could it be improved? (I took your ferrite ring anyway as an improvement, thanks!)

Regards, Sancho

by Sancho_P
March 14, 2013

More later ...

by signality
March 14, 2013

@Signality: Thinking about your proposal “Common Mode Choke 01” I modified the circuit to reflect my “reality” (the 115V leakage), especially which minimum coupling cap would be needed (at 50Hz sin) together with my (empirical) variant a) circuit to bring the voltage down to some millivolts.

Also the measured voltage increases when decreasing the short - cut caps (C1/C2) from 470nF to 220nF. I found similar results with this simulation circuit:

Finally I’ve connected 1nF to mains (230VAC), touched the cap and my “mighty mouse” - bingo, I heared the clicking sound coming from the mouse. It turns out there is a capacitive sensor in the mouse to distinguish between left and right mouse button, and a piezo speaker to give a better acoustical feedback of mouse clicks (genuine Apple ! ).

So it seems to me the problem is indeed from the 50Hz capacitive coupling, however, I don’t know if 2nF are a realistic value for my SMPS transformer (seems a bit high to me) and if that would be “normal”.

In the meantime the whole circuit is working perfectly at my “customer” (a friend of mine), which is really surprising, because the electrical GND (mains) there has 12 - 15VAC in respect to his water pipings …

Regards, Sancho

by Sancho_P
March 22, 2013

Post a Reply

Please sign in or create an account to comment.