Wisely (I think) after connecting my DUT (Tektronix 2710 Spectrum Analyzer) to an isolation transformer, I measured AC voltage between the DUT chassis ground and my scope (Tektronix 2440) ground lead. I see around 19VAC as measured by a fluke multimeter. Therefore, I did NOT try to attach the scope ground to the DUT.

The 2710's service manual says to use an isolation transformer as its PSU is switched mode (and apparently not isolated), which can be a big problem during service.

It should be noted that the DUT GND connection in relation to the isolation transformer is floating. There's no "ground" on the isolation transformer and therefore none to the DUT. As I understand the concept, connecting an otherwise isolated device to mains (earth) GND would defeat the purpose of the isolation.

My question is: where is this 19V coming from? Would a leakage path be possible through the scope's input EMI filter? Would connecting the grounds together draw a huge amount of current? Or, is this perfectly normal?

Never been here or done this... and never discovered it either as I've happily connected scope GND to instruments (not on an isolation transformer, however), with no issues in the past.

enter image description here

  • \$\begingroup\$ Just to be clear, you're using a battery operated handheld DMM for the measurements right? Fluke 87 or similar? \$\endgroup\$ Dec 25, 2015 at 19:09
  • \$\begingroup\$ Touch your scope lead and you might see tens of volts but it doesn't mean there is some kind of fault. \$\endgroup\$
    – Andy aka
    Dec 25, 2015 at 19:33
  • \$\begingroup\$ Yes, a battery operated DMM. \$\endgroup\$ Dec 25, 2015 at 20:38

2 Answers 2


Leakage currents at high voltage but tiny current are common with older equipment and equipment with EMI filters. To answer all your questions, connect a 100k resistor between the two grounds, and measure the voltage again. If it's a lot lower, then it is a small leakage current, and they are safe to connect together.

  • 1
    \$\begingroup\$ AC leakage is not limited to older equipment and equipment with EMI filter. It applies as well to current ungrounded SMPSs. Any transformer has some capacitive leakage, and if a device is ungrounded (which is always the case after an insulation transformer), there is no way for the leakage current to go. \$\endgroup\$ Dec 25, 2015 at 19:39

Isolation transformers are good but they have capacitance between primary and secondary or in your case secondary and ground. This can be measured like so: -

enter image description here

For an isolation transformer of 1kVA capacitances of 1nF are not unheard of. So that's one half of the story and maybe you have a smaller isolation transformer that has 100pF to ground on the so-called isolated winding.

Next consider what you are connecting up to - it could be anything from your description (DUT) but it might have a chassis that is also "affected" by it's own in-built isolating transformer and then you have another 100pF. But what does this capacitance do to your chassis when it is left to "float"?

Consider that there is equal capacitance to the live connection and neutral (say 50pF each). Now that forms a potential divider and with the chassis unearthed you might measure half your AC power voltage with a DVM to earth.

So you have maybe 60V on the chassis being sourced by 100pF of capacitance from across its own transformer and, you have maybe 100pF to ground from your scope's isolation transformer - you will could see anything up to 60V RMS between these two points. You might even see a small spark if you brought the scope ground and chassis together and you'd probably feel a mild tingle if you touched both nodes (don't try that of course for safety reasons).

  • \$\begingroup\$ In the case of your illustration, earth ground remains common at both sides of the isolation transformer. I've read about this a bit and find some folks advocating a connection between neutral and earth on the DUT side and leaving earth floating, to connecting earth gnd all the way through. Your suggestion was a good one. Using a 10K I get 0.03V at the scope side of the resistor when the other end is attached to the DUT chassis. \$\endgroup\$ Dec 25, 2015 at 20:41
  • \$\begingroup\$ So wondering if it's smart to connect earth at both sides of the isolation transformer (rather than leave it to the scope leads). \$\endgroup\$ Dec 25, 2015 at 20:48
  • \$\begingroup\$ If you connect earth to both sides it fails to be an isolation transformer in many respects. Leaving off a hard-wire means that under a fault condition the max current that can flow into the isolation transformer back to earth (via the parasitic capacitance) is limited by that value of parasitic capacitance. Probably less than 1mA and regarded by many to be safe. \$\endgroup\$
    – Andy aka
    Dec 25, 2015 at 21:56
  • \$\begingroup\$ If earth is not connected, then the earth connection is made through the scope ground lead, leading to the same issue: not completely isolated. Given length (and size) of cabling, I bet this increases any voltage drop (increased resistance in earth lead). I need more information on why isolation is necessary when testing switch mode supplies, and how we deal with earth ground. \$\endgroup\$ Dec 26, 2015 at 18:37
  • \$\begingroup\$ @MarkRichards I think you need to draw a picture of your proposed setup with the options that have been discussed. \$\endgroup\$
    – Andy aka
    Dec 26, 2015 at 18:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.