I have heard people say you shouldn’t Megger test transformers because the high voltages used might cause insulation between windings and between windings and the core to breakdown. Is this true?

Suspect it will depend on the voltage actually used.

Assuming you should Megger test transformers, what kind of resistance should you see between windings and windings and the core, for the transformer to be deemed safe?

  • 1
    \$\begingroup\$ When connecting 1000V to the transformer it tends to confirm the insulation is in good condition... \$\endgroup\$
    – Solar Mike
    Commented Sep 5, 2019 at 11:40
  • 1
    \$\begingroup\$ 1500V is a very common withstand spec for primary to core or secondary. There are other figures depending on use cases. \$\endgroup\$
    – Neil_UK
    Commented Sep 5, 2019 at 12:46

3 Answers 3


You have my upvote, because I am often testing motors, but never had a neat specification where the lower limit of resistance is. You can megger as much you want if you are below the specifications.

If I remember from my university days, the test consists of applying a single pulse at 1.4 x max rated voltage and you may never repeat the test again, short answer. Well it is more complicated like that: they start at lower voltages, calibrating instruments, then rising voltage 3 pulses, finally one single pulse.

For test purposes you actually measure the insulation at lower voltages than max. declared.


Yes you should.

In many cases a transformer is what provides isolation from the main AC supply (or some other source). Meg or Hipot testing checks for potentially weakened parts of the isolation which may cause an unsafe condition during a line transient or similar. My company usually has our transformer manufacturers test to 3000VAC from primary to secondary. Double, or even triple, insulated wire can easily stand up to that.

  • \$\begingroup\$ Agree, test to the specification. But, I would avoid increasing until you find the failure point. \$\endgroup\$
    – Mattman944
    Commented Sep 5, 2019 at 12:28

Yes you must do a “Megger Test” on components and HIPOT on all AC powered Systems in production to guarantee performance. ESD tests are usually only done to validate the design.

A DMM should indicate > 20 MOhm or OL beyond max reading value of portable meter. But this does not measure breakdown threshold for lightning impulses or ESD static charge buildup.

For bigger transformers https://electrical-engineering-portal.com/measuring-insulation-resistance-transformer

This is (should be) a non-destructive high voltage insulation test which must be done on every system to ensure there is adequate margin in the “insulation system”. There may be testers which are destructive, but you can modify to make non-destructive.

It can be destructive to AC/DC power supplies when the HIPOT tester has too much capacitance. I have successfully modified these testers to make the Probe non-destructive. I used ~ x00k range series R’s to limit current and spread voltage drop on 500V rated 1/4W R’s. I still allowed 100uA discharges to trigger a test failure but not much more to cause component failures when insulation gaps were compromised.

One can do (achieve) BDV ratings by design, but it must be verified by non-destructive testing, which is simply limiting the current after ionization breakdown.

A better test which is the precursor to BDV tests is called a Partial Discharge (PD) test which is optional, but ought to be mandatory in my opinion. It can be done cheaply and in a minute for any component, cable, transformer or insulation system to determine the Inception Voltage or PDIV threshold which always occurs below the breakdown threshold. The difference between PDIV and BDV determines how much variation you can expect on a BDV test that determines the BreakDown Voltage threshold.

The closer the PDIV to BDV level, the better.. E.g. if it breaks down at 10kV and PDIV is 9.9kV then every arc test will occur at 10kV. But if PDIV is 5kV, then you can expect BDV to have a wide range in results. Low variation means fewer moving contaminants or voids. This means lower partial capacitance from the contamination and induced breakdown across that small capacitance and may be the cause for wide variations in breakdown voltage between ramp tests. Thus the initial conditions change due to partial leakage charge voltages from previous tests due to contaminants which may be insulators or conductors or even magnetic dust..

The PDIV test is basically wait for an RF noise burst on an AM radio while ramping up the insulation voltage and expect a Relaxation Oscillator effect starting from 1 pulse per minute to 10 Hz to maybe several pulses per grid cycle rate (50,60 Hz) just before ionization and full breakdown. It would sound like a long distant lightning stroke between channels on AM or SW band but can also be observed with a 300MHz scope as something like a 10W 1~10ns pulse into 50 Ohms using just a wire loop to measure excitation current into good 50 ohm coax and terminated with 50 Ohms.

Although more expensive PD measurement systems measure after calibration , the value in pC for every discharge for continuous operation as well as phase of discharge, or maybe use inexpensive handheld units.

I prefer the PDIV method to see what voltage margin there is above max operating voltage before contamination is excited enough to bridge the insulation gap. Normally 15% voltage margin is tested on transformers for coil saturation currents and Impulse surge voltage discharge tests are done to stress the insulation.

As I recall, they don’t measure insulation margin, rather just go/nogo tests for discharge pulse shape unpowered.

The PDIV margin test is a simple test and a better indicator of insulation health in my opinion.

Ramp up the insulation voltage slowly until you hear it tick on the AM band on a quiet channel with no follow-on AC power due to current limiting so the high voltage source is very low power limited only by the load capacitance.


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.