Single ended PSU noise measurement. Strange grounding effects measurements.

[rwgast_lowlevellogicdesin]

I have some cheap Hanmatek HM305 switching power supplies. Out of curiosity I wanted to measure the ripple/noise. I found that when making a single ended measurement I get less noise if I connect a wire directly from a scope ground such as the bnc shell or cal ground, to the black ground post on the PSU. I don't even understand why this makes a difference since the scope probe is directly connected to the same black binding post. Im hoping someone can explain exactly what is happening here.

I uploaded some pictures of the scope without and with the extra ground lead connected.

[Conrad Hoffman]

Read the old Jim Williams articles about probes and probing. Getting a decent measurement from a switching supply is hard and you can't use a wire at all!

[rwgast_lowlevellogicdesin]

Even using a differential probe, or 2 channels in subtract mode?

Im not trying to get a crazy characterization of the supply, just a ballpark number within a few mv.

Like i said above though this is less about the measurements and more about why adding a wire between the ground of the psu output and the scopes signal ground has any effect on the measurement

[T3sl4co1l]

Seems to be a flyback converter with little or no EMI filtering components?

Consider a typical EMI equivalent circuit like:



There is some voltage across the isolation transformer, and thus through its isolation capacitance.  This is V1 and C3.  There is usually a shunt capacitor (between pri/sec GNDs), C7, which must be Y1 rated.  The secondary to ground/earth path, and primary to earth or secondary ground path, varies between devices (hence C6 is shown shorted: take your pick, sometimes it's direct connected, sometimes it's isolated and RF-grounded through a capacitor).  It's reasonable to assume there is no (or insufficient) secondary-to-chassis path here, given the amplitude shown; or that the primary-side choke is woefully insufficient.

The effect further extends to the scope probe.  Consider the ~0.2uH series inductance of the ground clip, through which CM current flows (up to the scope body, and perhaps onward to earth or mains, back through the wiring, to the PSU primary side, closing the CM loop).  Whereas CM current is NOT flowing through the probe tip (it's high impedance at these frequencies), thus the voltage drop on the ground clip is subtracted from the probe differential measurement, and you cannot measure the differential voltage alone, not with generic probe technique anyway.

Tim

[rwgast_lowlevellogicdesin]

That was a pretty good reply im not sure i understood it all, but the take away is these PSUs dont have an appropriate choke before the main transformer, or there is a bad ground between the case and secondary side of the circut?

[temperance]

Maybe this helps to understand the problem:

[rwgast_lowlevellogicdesin]

Thanks, that video was great. He was explaining the exact same thing i was doing. Im definately going to subscribe to his channel.