Insulation appears as a bunch of capacitors. Good insulation appears almost constant with voltage. Deteriorated or contaminated insulation megger readings drop as voltage increases. The standard measurement is the dielectric absorption ratio which is the 10 minute divided by the 1 minute reading. Typically a good number is 3-5. Megger tests need a constant (NOT hand cranked) voltage source and take readings after one minute to be valid. And you need to ground (short) for 3 times that time if you've already taken readings once. The DAR test gives hint why. Repeatedly meggering shows an increase no matter what the voltage (even at the same voltage) if the system isn't grounded for 3 times the test interval in between tests.

A lot of guys just stick the megger on there and take less than a 10 second reading. That's not a valid reading but hey all I'm looking for is over 5 megaohms org 100 megaohms as per the current IEEE standard temperature corrected after 1 minute but if in 10 seconds I've already got 20+, I don't need to continue the test. So I do it too but the closer I get to 5 or 100 megaohms the more I start watching the test time to get a by-the-book reading if the test is marginal. So if what I'm doing gets passed on the next guy might think A 10 second reading is valid when it's not...it's just close enough for spot check. As I megger longer the reading will climb, fast at first then slowly eventually. If there is moisture or contamination it will be very erratic. With very accurate readings you can see stair steps as the cracks charge up one at a time. Watching the readings or graphing them can be just as useful as the number.

And you can't "increment" it except for tip up (step bvoltage) tests where DAR is a better test anyways so step voltage tests are unnecessary except by customer request. And ever try measuring each of the three leads on a 6 or 9 lead motor nd wonder why the readings "go up"? I've seen that done too. It also happens when I megger an open frame breaker but again...I'm doing pass fail. Precise readings aren't needed. Hand cranked is OK for that but not for accuracy. Not only that but at one time and cranked megger were $500 and digital battery powered ones were $5000. Now you can get a decent Amprobe (a division of Fluke) or Extech (a division of Flir) for under $200 new and the AEMCs nd Biddle/Megger brands falling in the range of around $300-500 so there is no justification for hand cranked anymore.

As to what is valid, IEEE 43 gives 500 V for everything under 1000 V. Up to 12.5 kV, 1000 V is plenty. So you can just use one of the cheap common meggers to test everything until you get out of motors nd generators and out deep into distribution voltage territory. Up to 600 V, all insulation by ICEA/NEMA/UL is actually tested (megger) to some crazy value like 3 times the rating plus some more so even 150 V insulation (the lowest quality rating that has to be special ordered) is still rated higher than what a megger puts out so there is no reason for true test to use 100 or 250 V, again using a 1 minute reading. Use either a built in timer or your phone to time it.

The exception is VFDs. Following the manufacturer instructions you inside before meggering. But knowing it has blocking in the forward nd reverse voltage directions I sometimes drop to 250 V on a 460 V motor or 100 V on a 230 V motor just so that I can save time on a spot check and megger without unwiring. If the drive/motor fails (shorted) I know I have a bad motor/cable/drive anyway and step two is unwiring it all and running diode tests on the DC and AC busses to test the antiparallel diodes for shorts. So I am not only cheating but directly ignoring manufacturers instructions that tell you that you will destroy a drive by meggering it. There is a voltage that is really close to the AC peak voltage where particularly with SCRs self commutation occurs. Even without voltage on the gates, a high enough voltage on a power semiconductor will cause it to start conducting nd once it starts with some of them, they won't stop until they re damaged or power is removed. Damage with a battery powered megger is probably impossible but it leaves the possibility open and shows a smart customer you don't know what you are doing.

So yeah I'm not surprised at all. First unless you have a special case there is no reason for a 100 or 250 V test. Second if you didn't discharge fully...and I mean shorting it with a ground clamp for 5 minutes or more between tests, you're just seeing residual charges building up as your test gradually looks more like the 10 minute result. I'll bet if you went down in voltage or started over you would continue to see increases. The way a high end meter tests for this is that it measures residual voltage and grounds the leads until it goes away. And you get just ground a little...it takes much longer to drain a capacitor at a very low voltage compared to charging it at a very high voltage where the voltage difference helps speed the process up. //