8

I am working to lower the leakage current in an isolated power supply.
The power supply uses an isolation transformer to step down from 120V/240V to 6.5V. The voltage is then rectified and run through five TO-220 LM2941 regulators to generate individual supplies for the different parts of the final circuit. The LM2941's are attached to an aluminum heatsink with Thermafilm insulators in between the regulator and heatsink and thermal compound on top and bottom of the mica. The screws used to attach the LM2941's have shoulder washers. The heat sink is attached to the enclosure of the power supply which is connected to earth ground.

If I lift all of the regulators off of the heat sink, I see a drop in the leakage current of about 10uA, indicating that the contribution of the regulator-heat sink interface to overall leakage current of the power supply is 10uA. I would like to reduce or eliminate the 10uA of current if possible. Is this amount of leakage current normal for this configuration, or should I be able to reduce it by using a different insulator or mounting configuration?

Background:

I am measuring the leakage using a Dale 601 Electrical Safety Analyzer. The specific test is the mains on applied part test where mains voltage (120/240V) is applied to the isolated side of the power supply and the leakage current to the protective earth ground conductor is measured.

Engineman
  • 81
  • 2
  • 3
    Is there a reason this tiny amount of earth leakage is critical, especially considering it's coming from the secondary side of the supply? – ThreePhaseEel Jul 16 '16 at 00:38
  • 3
    That is being picky! Medical grade equipment allows for 50uA of leakage current. –  Jul 16 '16 at 01:14
  • If you are not building a medical device, you are allowed to have up to 250 uA of leakage. Sounds high though with 10 uA if you have it isolated with bushings for the screws and mica. – winny Jul 16 '16 at 09:40
  • 3
    This is a medical device. I'm allowed 50uA total leakage. I currently measure about 60uA of leakage from the secondary side. 40uA is coming from the transformer, 10 is coming through a set of surge absorbers and 10 through the heat sink. I have a lower leakage transformer on order. I'm looking for ways to lower the other two places in case the better transformer isn't enough by itself. – Engineman Jul 16 '16 at 14:40
  • In that case, you are worse off. You really need to minimize transformer capacitance, use thick insulators and keep Y-caps to a minimum. – winny Sep 14 '16 at 13:37
  • I agree. It appears that most of the leakage is to the RF shield around the transformer. I am currently looking at ways to improve the leakage to the RF shield. – Engineman Sep 14 '16 at 13:41
  • Can you float the heatsink from ground? Maybe not the safest thing to do, but may help a bit in this marginal design. A better transformer helps all other efforts. – glen_geek Sep 14 '16 at 16:01
  • @glen_geek I have tried floating the heat sink using nylon washers. This is a Class I power supply so unfortunately the mechanical connection to the enclosure is necessarily to ground. – Engineman Sep 14 '16 at 16:23
  • @Engineman The only way to reduce the heatsink capacitance is to reduce the area of the heatsink (SMD) or increase the gap. Either way you probably have to reduce the voltage drop with a more carefully regulated input so that no heatsink is required. A shield will only increase it slightly. and will not help For ways of making it more efficient to not require a heatsink, you need to supply all in/out ranges and tolerances The xx pFcan easily be computed for area and gap and represents 6.5V/10uA= 650kOhm reactance at 100/120Hz? which means a few 100 pF – Tony Stewart EE75 Nov 22 '16 at 19:59

1 Answers1

1

240V AC at 50 Hz and 10 uA means an impedance of 24 Mohms. But almost certainly this is due to capacitance so, at 50 Hz this is a capacitance of 133 pF and probably not anything to do with real resistive leakage.

So, estimate how much capacitance 5x T0220 bodies have when bolted down to a large large heatsink via insulators.

Is this a problem? No.

Andy aka
  • 434,556
  • 28
  • 351
  • 777
  • Thanks for the reply. If it is coming from the capacitance alone then I should see improvements by lowering the capacitance. To lower the capacitance the two variables that are easiest to tweak would be the dielectric constant and the distance d. The Thermafilm has a fairly low dielectric constant of about 3. I'm not aware of anything much lower that has low thermal resistance. Stacking multiple sheets should increase the d but I don't see much change in the leakage. I did some more testing and if I stagger the sheets by placing a sheet between adjacent devices I see 5uA lower leakage. – Engineman Jul 16 '16 at 14:50
  • This could be because I've added an air gap that lowered the dielectric constant or it could be because the leakage is primarily along the surface even though I try to keep everything fairly clean wearing gloves, etc. – Engineman Jul 16 '16 at 14:52
  • 1
    To eliminate the effect of the capacitance (at least a part of it) it might be possible to also isolate the heatsink from the enclosure. – Decapod Sep 20 '16 at 07:22