Generic Magsafe charger ghost touches

[cruiser_]

hi there,
I am trying to fix my generic magsafe charger so that my old macbook does not get 'ghost touches'. When I ground the case with my hand touchscreen works fine, but without it I am getting erratic movements. The charger is class II - no earthing, floating ground.

What I tried so far
- add more capacitors to the output (from nF to 2200uF)
- add RC filter to the output (different coils and capacitors)
- change the Y cap from default 2.2nF to 1nF, 33nF and remove
- bridge secondary and primary winding circuit with 4x10M resistors

unfortunately, that did not improve much - and some changes made the touchbar work even worse (e.g. removal of Y cap)

The voltage between macbook's ground and me is circa 40V AC.

I would like to understand what is it that original chargers do better - that does not lead to ghost touches. And why they occur in first place?

[stj]

Re: Generic Magsafe charger ghost touches

what happens if you earth the macbook?

[cruiser_]

Re: Generic Magsafe charger ghost touches

Well, I earth it by touching the metal case, then it works fine.

The mystery for me is why (and it's not just macbooks) some laptops are prone to this phenomenon either with genuine or generic chargers while some are not. And this for class II devices so no earthed connection.

[stj]

Re: Generic Magsafe charger ghost touches

the charger is transfering high voltage noise to the laptop body because it's not earthed

[redwire]

Re: Generic Magsafe charger ghost touches

The ghost touches are caused by a high level of common-mode noise at the Macbook. This is not 50/60Hz noise, but say 60kHz.

Cheapola chinese chargers have poorly made SMPS, missing EMI filters and ferrite beads, also the transformers in them wound in a way that ends up with too much primary-secondary capacitance. So they spew out tons of EMI. The Y-cap is the only thing keeping it down, dumping the noise back to mains.

You can try add a common-mode clamp-on ferrite to the charger's output cable, something with a few turns passing through it. Or on the mains cable.
Realistically, the charger is plain noisy. It would need a better primary CM choke/filter arrangement, a better transformer to be less noisy.

Also note the high level of noise... can imply bad insulation inside the charger.
Big Clive got one that had shorted out (pri-sec) and blew up the guy's Surface and monitor! Even the fuse was in the wrong spot and did not blow after the ground fault. There are no savings running expensive gear on a cheap charger.

[cruiser_]

Re: Generic Magsafe charger ghost touches

Thanks for all the information and video. I have seen it before and I am aware of the risks. This is driving a 'lab' 2009 macbook so by buying a genuine charger I would double the value of the set.

Now, on the adapter. It has some basic filtering on the mains side and there are proper gaps between the sides. It is driven by an IC too. I have seen worse adapters than this one for sure.

I forgot to mention that I have also tried clamp on ferrite on the output as well as coils in series on the positive low voltage output.

EMI / noise wise - I assume you are implying that the noise is carried over through the connecting wire? Wouldn't the low pass LC filter pick that up? I have been playing with different combinations 'down' to a 1kHz+ setup. Also, charger sits roughly 1.5m meters from the laptop. Would that not cut it out?

On the 'toy' oscilloscope, the output is pretty steady with Vpp of 0.3V/0.4V under load.

I am yet to try wrapping the charger in tinfoil... I won't be able to replace the transformer but anything in that charger I'm more than happy to play with.

At this point it is really a knowledge seeking / experimental quest so any suggestions are more than welcome.

[redwire]

Re: Generic Magsafe charger ghost touches

[...] there are proper gaps between the sides. [...]

I can see one spacings violation (pic related), your usual chinese screwup that would fail any safety certification.

I would say you need better common-mode filtering which can be done at the input or output or both.
Usually the primary CM choke is small inductance value for low cost, if you have another you can temporarily add it to the mains input of the adapter, like adding a second stage. Although it is a dangerous experiment. Just to narrow down if the noise can be lessened there. Or go to town with the ferrite clamp choke # turns to achieve the same result.
On the output side, adding a CM choke you get losses, a voltage drop so it is not the ideal place.

[cruiser_]

Re: Generic Magsafe charger ghost touches

The two pins are both on the primary side, I thought the worst were small gaps between primary and secondary?

On the input, this adapter has X rated 0.1uF capacitor, double coil (each has 1400uH) and 4.7nF with 10k resistor. My other coils are in the 0.x uH to hundreds uH range so I guess no point in adding them here? Pretty similar to the filtering section in

I used HT2629H ferrite clamp on either input or the output, no joy.
I even put the adapter in a metal bowl and covered it with a metal cover

I do not have the original charger for comparison unfortunately... but I hooked up some old Delta 19V adapter (II class too) and the touches were gone so it must be doable!

[momaka]

Re: Generic Magsafe charger ghost touches

How much space is there on the secondary side between the output filter caps? Could we see more picture of the whole adapter from different sides?

If there is a little bit of space on the output side, perhaps you can try adding a "PI" coil (rod core with wire wound around it) between or after the two output caps (if after, add another small electrolytic cap before the output wires, and possibly a ceramic too.) This may or may not reduce the noise enough. I know you mentioned you already tried an "RC" filter with coils and caps... but if you do it on the PCB very close to where the noise is being generated, that could help.

If there is more space, you could also add a small CM choke, as already suggest (only needs a few turns, so you don't get too much voltage drop.)

Finally, I don't see an RC snubber circuit before the secondary-side rectifier - that could be another source of HF noise. On the above Delta schematic, you can see this with R21 and C21 right next to the output rectifier.

[cruiser_]

Re: Generic Magsafe charger ghost touches

hi momaka,
attaching some more pictures.

Small correction on the lowpass RC filter - I meant LC filter (2nd order) - a coil with a capacitor. With 34uH coil and 2200uF cap it's supposed to filter anything above 580 Hz.

There seems to be a snubber on the double diode - I also temporarily added additional - used 2.2nF & 1k resistor (which got pretty hot..?)

Also, cut off the trace between two 'legs' of the diodes' input (found some reference to EMI reduction there).

My last attempt was to add small coil with double-winded cable, with it the small power board PCB in a laptop went with a puff of smoke

Is it possible that some high voltage was induced by this? No-load measurement showed 16.5V there. Or was it just a pure coincidence...

[cruiser_]

Re: Generic Magsafe charger ghost touches

... and above all, just going through the diagram of macbook's board (820-2530, with power board 820-2565-A) I noticed that DCIN is supposed to be at 18.5V, while all the 60W chargers are only providing 16.5V.

Anyhow, anyone with a power board schematic? Bear in mind there's a full laptop board under the same number (820-2565-A)...

[cruiser_]

Re: Generic Magsafe charger ghost touches

board fixed, must have been an accident - TVS diode RClamp2402B blew up.

going back to initial issue, any other suggestions as to what I could try to fix the charger? This is really puzzling me.

[stj]

Re: Generic Magsafe charger ghost touches

add an earth connection - change the figure-8 socket for a cloverleaf style

[cruiser_]

Re: Generic Magsafe charger ghost touches

This adapter doesnt have any of the connections for it unfortunately and it would require changes outside of its shell -> ugly look.

What's interesting, I took a regular 19V HP laptop adapter, it has ground connections but I connected it to an ungrounded socket, it still made ghost touches go away.

IF ghost touches are caused by floating ground, then why would they go away in such case?
and IF they were caused by common noise, why wouldn't filters/chokes make it go away?
and finally, IF they were caused by voltage ripples, why wouldn't capacitors solve it?

What could be another reason ?

[cruiser_]

Re: Generic Magsafe charger ghost touches

One more observation.

I've got the original macbook charger which, even if run off a non-grounded socket makes no ghost touches.

The difference that I saw so far is that the voltage between myself and the GND on the macbook is:
- 0V in case of original macbook charger
- 0V in case of HP charger that did not give any touch issues
- 70-80V AC on the charger in question
- 40-50V AC on another generic 12V charger I tried

Could this be issue? Which part of the circuit is responsible for this difference (thought of Y class between primary & secondary, but I tried a few different values)? How to rectify?

[sam_sam_sam]

Re: Generic Magsafe charger ghost touches

The only thing I can think of is one side of the rectifier is leaking ac voltage because in the circuit diagram it shows that there is a capacitor but it also shows four 10 megohm resistors on the negative side of the rectifier and on the secondary side of the power supply it shows that it goes to the ground terminal

If this circuit diagram is correct for the switching power supply in question then this might explain why you are having this issue

I could be wrong but this is the only place that I could possibly see you getting this voltage from

[cruiser_]

Re: Generic Magsafe charger ghost touches

Hi samsamsam
The diagram is for another, well-behaving adapter which I am trying to mimic in my 'chinesium' one.
I have added common choke to the output, also capacitors before it and after it. Sadly, no difference in ghost touches.

Thought the choke would get rid of common and differential noise. Perhaps this isn't the reason of this phenomenon?