3
\$\begingroup\$

The electric strength test OR dielectric withstand test OR HIPOT test voltage level as specified in EN 60730 standard works out to be 2860VAC for a class II radio product I am developing. The PCB has an isolated switched mode power supply with a 2n2F Y-cap bridging the primary and secondary for radiated emissions control reasons. The PCB also has an SMA connector that sticks out of the plastic enclosure making things difficult.

The product fails HIPOT testing above 1600VAC! Remove the Y-cap and it passes at 2860VAC. I narrowed the failure down to the reactance of the Y-cap and through iterative trials, got a pass at 2860VAC with a Y-cap value of 100pF. However, this caused the product to fail radiated emissions. Following this, I did some research and found several articles advising that you can either remove the Y-cap for HIPOT testing or use DC-equivalent voltage which in this case will be 4044VDC (2860 * 1.414). I tested at 4044VDC and got a pass with the original Y-cap of 2n2F.

I went back to the 60730 standard and found in 'notes' that in fact you can remove electronic components that may render the test impractical and also that a DC potential equivalent to 1.414 times the test voltage may be applied.

I am not convinced because in a real-life scenario in the field where an overvoltage of 2860VAC appears on the power supply, a person who is touching the RF connector could potentially get a shock rendering the product unsafe. So, what is the rationale behind product approvals standards advising removal of Y-cap before HIPOT testing or using DC voltage to get a pass when in fact in practice, a high potential AC overvoltage could appear on power lines and pose the risk of electrocution? Am I missing any key principles?

Many thanks in anticipation of your help in understanding the issue.

\$\endgroup\$
1
\$\begingroup\$

I should point out that along with Tony Stewarts analysis there is the fact that a 2.2 nF capacitor has 22 times the surface area of a 100 pF capacitor. That offers much more leakage current, even at 50/60 HZ. Also much better chance of breaking down and arcing internally, so the cap has to be replaced.

It makes sense that under certain conditions a high-value Y capacitor will not survive a designated hi-pot test that a capacitor with the same maximum voltage ratings but much less capacitance would survive (AC test).

If I built a test fixture I often used a cheap 5 nF 3 Kv rated ceramic disc capacitor for the Y circuit. It passed the usual 1.5 kVAC hi-pot test but not a 3 kVAC test.

You are allowed to remove a Y rated capacitor if its capacitive value alone will cause it to fail hi-pot testing. The MUST pass by DC voltage rating still holds true for high value capacitors, using the rated DC *1.414 method. Small value capacitors MUST pass the AC test per the required standards. This avoids an engineering conundrum where the chief engineer must decide between a good filter or one that passes any hi-pot test.

\$\endgroup\$
  • \$\begingroup\$ I respectfully disagree with 1st paragraph. Breakdown Voltage level depends on margin above Voltage (withstanding) Rating, not necessarily the C value nor the current if specs are not exceeded. Smaller cap values may survive may , but due to other reasons such as ESR. Lower AC losses and less heat. Also it is possible to modify any HIPOT so that a leakage failure is non-destructive, by removing tester storage capacitance energy with an appropriate series resistance. \$\endgroup\$ – Sunnyskyguy EE75 Jan 24 '18 at 5:20
  • \$\begingroup\$ You make it sound like a free for all. Too many what it's, then do this. So many test options seem to negate any failed test. \$\endgroup\$ – Sparky256 Jan 24 '18 at 5:26
  • \$\begingroup\$ At some point a valid test that a capacitor fails must be accepted for what it is. \$\endgroup\$ – Sparky256 Jan 24 '18 at 5:44
  • \$\begingroup\$ HIPOT is steady state BDV test to ( >=1s) to simulate a transient of unknown duration except that it is << 1s. A valid test is correctly defined in the standard. Y caps may be removed from the test if using AC , but they must be legitimate caps with specs "kept on file" or submitted to UL/IEC as rated for Y cap usage and ratings. \$\endgroup\$ – Sunnyskyguy EE75 Jan 24 '18 at 5:47
  • \$\begingroup\$ It is late and I am using my cell phone. \$\endgroup\$ – Sparky256 Jan 24 '18 at 6:08
1
\$\begingroup\$

The real risk is an impulse breakdown at elevated voltage. It is tested for 60 s or raised 10% for 1 s.

schematic

simulate this circuit – Schematic created using CircuitLab

The rationale is that every product must pass the dielectric HIPOT test to ensure the leakage does not exceed 250uA at the rated voltage. The default test duration is 1 minute , which may be reduced to 1 second by a further 10% increase in the voltage. It is not possible for the grid to sustain these Voltage levels which may occur due to an inductive machine transient, or an induced from nearby lightning. The real event duration would be short, but this test is to stress the insulation in case ionization breakdown leads to failures.

The Y caps however induce a steady current at line voltages in order to shunt SMPS conducted noise which would cause HIPOT failure by this steady HIPOT test.

The HIPOT test is a dielectric breakdown threshold test and not intended to test line filter Y cap current at elevated steady-state voltages. It is intended to stress manufacturing process breakdown (pun intended) due to insufficient clearance gaps or lack of silicone added insulation in the process.

I once had qualified a PowerOne OEM supply that was later transferred to Mexico production. I also startd testing all HIPOT with DC secondary earth grounded to reflect application stresses. When this is grounded, the isolation transformer which has a small coupling capacitance but at switched rates is now a much lower impedance than the stray capacitance of primary gaps at line frequency. This raises the voltage across those gaps and stresses the primary gaps with more voltage with secondary DC grounded. It resulted in 30% failures in HIPOT with more than 6 weak spots and supplier was disqualified until corrected.

NB to all those Test Engineers.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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