Simple Redundant AC Input Circuit Issue 3

[Kepco]

Hi Guys,
The attached circuit is an AC redundant input circuit for a chassis that runs several SBC cards. The AC powers four PSUs. For some reason when the AC is turned ON, occasionally the fuses on both AC inputs (one fuse each on each side) burns. Switching the AC inputs individually the circuit works normal. The fuse is 15A and the load is 8A max. The fuse is fast blow type. The PSUs are inrush current limited at 10A a piece.

The circuit works like this. When the top AC input is present the relay turns ON and connects the Top AC input to the PSUs. Now when the top AC input is turned OFF the relay turn OFF and the lower AC input powers the PSUs.

The attached scope shot shows the AC waveforms when both AC inputs are killed at exact same time manually. CH1 and CH2 are AC inputs and CH3 (magenta)is current. This shot is taken at the AC input side before the filter. The shots shows the high current condition happens for 1.25ms which is also current limited since it is flat at the top. Not sure why this is happening and how a fuse can respond to such a short duration of surge current.

Any ideas regarding this failure?
Thanks

 

[Kepco]

Attaching the circuit again.

 

[Kepco]

Attaching the scope shot again.

 

[IRstuff]

??? Your scope trace shows what appears to be a 46A negative and 27A positive spike; how is that not sufficient to pop a 15A fuse?

It seems to look like someone didn't bother to find out what the actual transient current from the power supplies looks like.

TTFN
faq731-376

 

[Kepco]

Thanks for the response.
The time is only roughly 1.25ms on 46A and 5ms on 27A. The fuse specs chart shows at 240A load current lasting 10ms min can blow the fuse. In this case it is 46A max and it will take about 500ms to blow the fuse per the spec. I did not made this clear earlier but the scope shot TEK0005 is shot on the PSU side at the relay common point and the TEK0004 shot is on the AC side. Not very convinced that this short duration can pop fuses but something is definitely going on and you may be right too.

 

[Skogsgurra]

It doesn't look like there's a current limit. The current trace (magenta) rather seems to hit a saturation limit. What kind of current transducer do you use? And what is its setting?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Kepco]

That was absolutely correct. I screwed up on the setting of the probe. Now, the probe is 100A max and I am measuring about 300A and then the probe saturates. So I guess now the reason why the fuse is blowing can be seen on the scope shot as attached. CH1 being AC1=208V CH2=208V and CH3=AC1 line current=300A

The primary question now is how this short circuit is taking place. Any ideas?

Thanks a lot.

 

[IRstuff]

Wow, is that even possible? The energy delivered appears to be humongous.

However, what exactly is happening from t=15ms to t=20ms? The waveform for CH2 looks like it got delayed by 1.25ms, which suggests that there is something going on with your line voltage.

TTFN
faq731-376

 

[VEBill]

Are L1 and L2 two phases of a three phase supply? Does the 120° phase discontinuity when switching from L1 to L2 cause all the PSUs to throw a hissy-fit and blow the fuses? Is it the same switching in either direction? If not, swap L1 and L2 to see if it also swaps the difference.

Another possibility: there's a miswire not shown on your sketch. It's not uncommon for the actual root cause to be NOT shown in the information provided in the thread.

 

[waross]

Take a look at the specs of your relay. It appears that you may be using a connection that is sometimes called opposite polarity. Many double throw switches and relays are not approved for opposite polarity. It is often specifically prohibited in the relay specs.
When the switch breaks contact, it may carry an arc that is extinguished at the next zero crossing.
Now, if the other throw of that pole is connected to a different phase, you have a phase to phase short through the arc.
Your scope trace seems to be showing a voltage drop on both phases which supports this conclusion.
To sum up, the issue may not be the PSUs at all but an arc carry-over in the relay. Not much clearance in those little relays.
Try a three pole relay: One pole for the neutral, one pole for one phase and one pole for the other phase.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Skogsgurra]

Bill is right. There is a short between the two phases and it is clearly indicated by yellow and blue/green touching each other (shorted) just where the current starts increasing. As contacts separate and current grows, the difference in voltage (arc voltage) is growing and after around 1 ms, the arc is disrupted and current starts to decrease at the same time as the voltages "kick back" and assume phase potentials again.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[ScottyUK]

Well done Bill, a problem which is obvious when highlighted but so easy to overlook.

 

[VEBill]

Good one.

 

[Kepco]

Excellent suggestion regarding arcing. There is a big pop when this happens confirming that belief. Using a three pole does not solve the problem because it seems that the arc create a short of long enough duration when breaking from AC1 to PSUs that when PSUs connects to AC2 by second pole during relay switchover the AC1 and PSUs connection remains active ( the ionized air stays ionized long enough) and thus creating a short between AC1,AC2 and PSUs. The solution may be to use a vacuum sealed relay.
Thanks a lot for valuable comments.

 

[Skogsgurra]

I think that two separate relays that are cross-interlocked can do the job. But make sure they have AC coils - a DC coil with snubber diode will delay the drop-out so you will still have a short.

Two standard relays will be a lot cheaper than a special vacuum DPDT relay - I don't even think they exist in the size you need.

If you look at pull in and drop out times for AC coil relays, you will find that drop out usually is about half the pull in time. So you will have something like 5 to 10 milliseconds safety. But make sure you check the timing before buing!

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[VEBill]

...runs several SBC cards...

Depending on several unstated variables (DC current, cost, size constraints), there may be a point where it would be simpler to double-up the number of PSUs, simplify the AC circuits back to normal, and then use diodes to provide the redundancy on the DC side.

If you start getting into vacuum relays, it'd probably be cheaper to buy another bunch of cheap PSUs. This is assuming that your "several SBC cards" aren't requiring (for example) 30A each. If the SBC cards just need a couple of amps, then such PSUs are extremely cheap. Vastly cheaper than a vacuum relay.

Just something to stop and think about.

 

[iop995]

This simple scheme assure redundancy at AC lines level, on DC side is more easy to assure that by paralleing and load sharing of 2 or more DC power supplies. Arcing at relay contacts level apear only at fail-back (restore main AC line) and not at fail-over, when main AC line have zero voltage. A relay with more contacts in series connected or already sugested 2 realy smay be good solutions to avoid such problem at fail-back.

 

[Kepco]

Skogsgurra,can you shed some light on using relay interlocking with a work around to the arcing? The switchover time cannot exceed the hold time of the PSUs, which is 20ms. The current relay has 20ms max make time and 10ms max break time. Also the neutral on top source and bottom source may not be from same source and has to be switched separately.

The PSUs are already redundant and ORed at the ouput. The input redundancy is a separate requirement.

Thanks.

 

[VEBill]

"The PSUs are already redundant and ORed at the ouput. The input redundancy is a separate requirement."

OMG. What's this for? ...No, just kidding (none of our business).

Of course we can't see your system spec from here, but one can sometimes meet multiple "separate" requirements with one approach. If the redundant PSUs were wired up directly to the separate AC inputs, this one approach might meet both requirements - providing AC supply redundancy and PSU redundancy (allowing either failure mode, not both at the same time).

Unless the system spec has another requirement that specifically excludes such a two-for-one approach (to allow for simultaneous failures of BOTH an AC input AND a PSU), the wording may allow this.

This may be disregarded if it's off the mark.

 

[waross]

You may be able to up-size the relay. A higher voltage relay will have greater clearances, and a higher current relay may have less arcing.
Another option may be to feed both relays from the same phase. This may not be a good idea if the relays are fed from different transformers.

Bill
--------------------
"Why not the best?"
Jimmy Carter