Transformer Capacitive Coupling. Impact on AC/DC Conversion.

[Ray CI]

Hi Folks...

I have a DC output from a simple rectifier circuit (see schematic) that is riding on an AC signal. The transformer is a 10:1 toroid.


The AC signal appears to be coming from capacitive coupling in the transformer (top left picture).  Two scope probes are connected between points C & D and ground.


To get rid of this, I can easily tie the DC output to Earth ground (top right picture) but I want to keep the DC floating. 

This happens with all transformers I have laying around.  Has anyone used any transformers with capacitive shielding between the windings?  If so, what was the outcome and did it help significantly?   Does anyone know of a way to remove the AC coupling without introducing Earth ground on the DC output? 

Thanks

Ray

[Benta]

It can usually be ignored completely. Your scope has a high input impedance and will also show a signal when you place your finger on the tip of the probe (pickup from the surroundings). Forget it, don't overthink.

[coppice]

Most applications can live with that kind of 50/60Hz pickup, as its quite high impedance and common mode. If you can't, and don't want to ground one of the DC rails, there are transformers with an electrostatic screen between the primary and secondary. If you ground that you really cut the capacitive coupling. Another option is a type of pseudo-toroidal transformer, which has a rectangular core, with the primary wound on a plastic former on one leg, and the secondary wound on a separate plastic former of the opposite leg. That really cuts the capacitive coupling.

[Weston]

Do you not want to galvanically ground your output, or do you not want any relation to ground?

If you add a Y capacitor from the rectifier (or any point on the circuit really) to ground, there is a return path for the coupled signal. Its a voltage divider with the impedance of the winding capacitance and the impedance of the Y capacitor. Same concept as the use of the Y capacitor in a SMPS.

If you don't want any reference to ground, its not an issue in the first place. This signal will not show up with a purely differential measurement on the output.

For SMPS with a two prong plug the Y capacitor is returned to the neutral line (which could happen to actually be the live side of the AC line so there is a limit on the maximum capacitor value).

With a two prong plug, if you trust your neutral to be neutral you will get reduction in the induced voltage with a Y cap. If neutral is actually the hot side of AC line the capacitor would make the problem worse. The same issue would also apply if you used a shielded transformer though.

[iMo]

[jonpaul]

read up on Faraday Shield and single/double/trple shielded transformers.

Measure Pp-s = Pri ><sec capacitacnce.

Short all primaries, to LCR meter or bridge high

Short all secondaries, to LCR meter or bridge low.

Place DUT on thick plastic or glass to float off bench.

Capacitance will be 20..1000 pF

Worse for toroids, better for dual or split bobbin EI.

Now you can easily model the test you make and what the stray common mode hum/noise will be


Jon

[David Hess]

Has anyone used any transformers with capacitive shielding between the windings?  If so, what was the outcome and did it help significantly?

Transformers which include electrostatic shielding between the primary and secondary to minimize capacitive coupling are common in off-line applications.  If the shield is not brought out to a separate pin, then it will be tied to the metal frame of the transformer.  The electrostatic shielding used this way is very effective.

Does anyone know of a way to remove the AC coupling without introducing Earth ground on the DC output?

An electrostatic shield will work, as described above.  Alternatively at greater cost, two transformers can be used in series with the floating winding which couples them grounded.

There are specialty transformers which provide lower capacitive coupling.  This may involve simply moving the windings apart from each other and the core, or transformers using other technologies like acoustical/piezoelectric transformers.  A motor/generator or lamp/solar cell can transfer power with very little capacitive coupling.

[Ray CI]

To all who have responded...  Thank you very much!

I'm building a 3-channel benchtop supply to power some handheld meters and a differential probe amplifier.  My handheld meters and probe amp have native USB charging capability but, the readings are noticeably impacted when used while the charger is plugged-in.  This is especially true of the probe amp.  After trying several "better quality" USB charging warts, I decided to build my own PS -mainly as an exercise.   BTW:  One of my meters will not work properly if the charger does not have well isolated ground.   The board I made is finished and the last detail was to try and clean-up that common signal due to transformer capacitance.  Minus this ugly signal (which does not impact the final outcome), the PS is very clean.

I have tried Y-caps (and basically, every possible thing I can think of) to pull that signal out of there.   Yes, I know that the floating signal does not really impact the devices being powered but, I was on a "mission" to figure this out and come-up with a clean solution.  I've narrowed my choices down to three different solutions and all of them seem to work.  I'll update this thread when I make the final move.

Here's a pic of the board.    Obviously, I don't do this line of work professionally so, please don't tear me apart.  In all honesty, this is the first board I've ever designed, built and sent-out for fabrication.  I had to teach myself KiCad to pull this off.   The board was sized to fit inside an enclosure.  I'm building the connector cables now.  I am an engineer (semi-retired) but worked in a totally different field.

If this thread is not bashed to pieces, I'll post more as I go. 

[ Specified attachment is not available ]

Regards

Ray

PS:  I hope the correct image is displaying.  The thumbnail image is not right but, clicking on it shows the board with onboard SMPS and filter circuits.

 

[Ray CI]

I was able to spend a few hours on this today and it's coming along well.  It fits nicely in the enclosure and the light-pipes are visible in daylight.  Green indicates the channel is turned-on and yellow indicates status of the fuse for that channel.  The external cables are shielded with braided wire and have single ground point at the respective terminal block. 

At first glance, the power output was still fairly clean.  I only checked one of the channels that was at 9V, 1.5A (pure resistive load).  The single-ended noise measurement under load was around 700uV RMS and 6mV pkpk.

Ray

[Ray CI]

This is coming-out better than expected.  It's far from being a reference grade power supply but, this was done as a learning experience with the dual goal of powering some simple benchtop devices.

The traces below were taken under almost full (designed) load.   Unloaded voltage was set to typical USB level (5.25V).  With a 2A resistive load, the voltage sagged to 4.9V but, the signal looks pretty decent given it's being driven by a $1.00 buck converter.   [I added a common-mode and differential filter at the output and it enormously improves the output noise].

The 2 big spikes in the top trace is ever-present noise in my home wiring.   The bottom trace was at 200ns to get a close view of the switcher oscillations. 

It was able to run like this for several hours and it maintained this level of performance.  The little heat sinks that were added made a big difference.

Ray