5

Every 480V variable frequency drive I've had the pleasure of dissecting has had two DC capacitors in series across its DC bus, each rated for ~400VDC. UL 508c testing as executed on one of my recent products requires that all components whose failure to short could cause damage be evaluated. At the time we had a single DC bus cap rated for 1200V. When that cap was shorted for testing, the full DC bus was placed across two diodes, which caused flame to escape the box.

To solve this, we had to put two capacitors in series across our DC bus, each rated for the full DC bus voltage. Obviously, drive manufacturers don't do that. If any one capacitor in the drive shorted, you'd have a ~650VDC bus across a ~400VDC cap. So how do they pass the tests? Does the cap pull enough current that the fuses open before the cap ruptures? Is this predictable by the surge rating of the cap? Or perhaps the cap ruptures, but the box is designed in such a way as to contain the flame? Or do they simply not run this test, and I'm misunderstanding something?

Stephen Collings
  • 17,373
  • 17
  • 92
  • 180
  • "the full DC bus was placed across two diodes" and no fuse? – pjc50 Jan 07 '14 at 17:34
  • Our goal was to pass without fusing the DC link. DC fuses rated for 1000V and 125A are rare and expensive and, in this application, unnecessary. In any real-world scenario, the DC feeding our unit comes from a rectified AC line, and the AC is fused. But you can't UL-list something with a caveat like that. If it's got a DC input, you test it with DC fusing or nothing. So we added that second cap so we could survive a cap shorting which will never short, even if we're on an unfused infinite DC source. Better that than require all the customers to buy a fuse they don't need! Paperwork. – Stephen Collings Jan 07 '14 at 17:41

2 Answers2

1

This type of cap is subjected to the Breakdown of Components test where one of the caps is short circuited.

If there are no flames / explosion, even if fuses blow then it can be Ok. The result must be no fire or shock risk... so if the door blows open its a failure if you can touch live parts.

This can be an expensive and exciting test... fire extinguishers at the ready!

Spoon
  • 1,292
  • 1
  • 8
  • 8
  • So... what happens when drives are tested? – Stephen Collings Jan 07 '14 at 17:16
  • a loud bang, some smoke, maybe flames(failure), fuses blow, if your lucky not all the fuses blow. If your unlucky a fire. – Spoon Jan 07 '14 at 18:34
  • Just in case you meant how is the test performed?...the drive is running during the test with the largest fuses the manufacturer wants to use within the range allowed by ul508c at the highest prospective current required (usually 5000A minimum) at the highest voltage required. – Spoon Jan 07 '14 at 18:39
  • This doesn't really explain how they pass the test. Are you saying that they just run it, catastrophic failure of the caps occur, but flame doesn't escape the box? – Stephen Collings Jan 07 '14 at 18:49
  • in short yes, along with the rest of the evaluation and the 4 times a year follow up inspection to make sure the construction stays the same as evaluated drive. What were you expecting? – Spoon Jan 07 '14 at 20:40
1

The fact that a capacitor has a 400VDC rating means that it will behave according to specifications for voltages up to 400 volts, and at higher voltages its behavior may diverge from what would be specified at lower voltages. While some capacitors will behave as specified at any voltage below a point where they fail shorted and/or explode, other capacitors may exhibit other anomalous behaviors under "mild" overvoltage conditions.

Electrolytic capacitors in particular are prone to have higher-than-normal leakage as the voltage across them increases. Further, current which flows as a consequence of such leakage is prone to increase the thickness of the dielectric, which will have the effect of (slowly) increasing the breakdown voltage of the cap, reducing the leakage at any given voltage, and reducing the capacitance. Given that capacitance is usually sized according to requirements, such behavior may be regarded as "damage" which leaves the capacitor unable to store the required amount of energy, but the product need not remain usable to satisfy the "product must not burst into flames" safety requirement.

I would be loath to rely upon capacitors not to burst into flames at voltage levels beyond their rating without substantial assurances that (1) current production caps will behave safely, and (2) I would receive ample notice if any future caps would no longer do so; I don't know that cap manufacturers would be eager to provide such assurance, though for a high-volume customer they might e.g. be willing to sell caps which were specified to work usefully at up to 400V and not burst into flames (nor fail shorted) below 700V, for a price between that of a "400V" cap and a "700V" cap.

supercat
  • 45,939
  • 2
  • 84
  • 143