RCD "blinding" by DC

[pc1966]

A recent thread was discussing having a delay RCD (type A or AC) for TT before a garage CU with a type B RCD for an EV charger. The question was would this be OK or not? Standard advice is the type B should be direct off supply (possible problem for TT) or fed from another delay type B (at eye-watering expense). But just how sensitive are type A RCD to a small DC component, below the type B trip point, in practice?

So from the department of "don't try this at home girls & boys" I bring you some test results based on this set-up:

Test devices are unused, a year old Hager ADA106U RCBO, and a many years old PowerBreaker H92MPS RCD-FCU. The big potentiometer served to provide a resistance for the current-limited PSU to reliably run at constant current when the RCD was closed, and a path for any mains test current when the RCD opened so it would not be totally forced back through the DC PSU. For testing at home (TN-C-S with RCBOs) I used a simulated TN-C-S test rig so the PSU negative and the MFT earth both go to the pre-RCD mains neutral.

Basically I force DC through the neutral path of the RCD while doing normal MFT tests.

TL;DR Testing RCD to DC effects

 

[pc1966]

Results for the Hager RCBO:

DC current (A)​	In/0 (ms)​	In/180​	5In/0​	5In/180​	Ramp/0 (mA)​	Ramp 180​
0.0​	9​	19​	9​	19​	24​	24​
​	9​	19​	9​	20​	24​	24​
​	9​	19​	9​	19​	24​	24​
​	​	​	​	​	​	​
0.1​	9​	19​	9​	19​	21​	24​
​	9​	19​	9​	19​	21​	24​
​	9​	19​	9​	19​	21​	24​
​	​	​	​	​	​	​
0.2​	28​	18​	9​	19​	24​	24​
​	30​	18​	9​	19​	24​	27​
​	28​	18​	9​	19​	24​	24​
​	​	​	​	​	​	​
0.3​	28​	18​	9​	19​	27​	30​
​	28​	18​	9​	19​	27​	30​
​	28​	18​	9​	19​	27​	30​
​	​	​	​	​	​	​
0.4​	Fail​	Fail​	9​	19​	33​	33​
​	Fail​	Fail​	9​	19​	33​	33​
​	Fail​	Fail​	9​	19​	33​	Fail​

 

[pc1966]

Results for the PowerBreaker RCD-FCU

DC current (A)​	In/0 (ms)​	In/180​	5In/0​	5In/180​	Ramp/0 (mA)​	Ramp 180​
0.0​	9​	18​	9​	19​	21​	21​
​	9​	18​	9​	19​	21​	21​
​	9​	18​	9​	19​	21​	21​
​	​	​	​	​	​	​
0.1​	9​	19​	9​	18​	24​	24​
​	10​	19​	9​	18​	24​	24​
​	9​	18​	9​	18​	24​	24​
​	​	​	​	​	​	​
0.2​	28​	19​	9​	18​	33​	33​
​	28​	99​	9​	18​	33​	33​
​	49​	239​	9​	18​	33​	33​
​	​	​	​	​	​	​
0.3​	Fail​	Fail​	9​	18​	Fail​	Fail​
​	Fail​	Fail​	9​	18​	Fail​	Fail​
​	Fail​	Fail​	9​	18​	Fail​	Fail​

 

[pc1966]

The conclusion? Well for the devices tested (both claim to be type A) there is:

• Negligible change at 100mA of DC (well above a EV's 30mA type B RCD trip current)
• Some obvious degradation of characteristics at 200mA
• By 300-400mA they are failing to trip at the 30mA threshold for "additional protection"

Over to the sparkys of this forum for comments!

 

[Vortigern]

That's fantastic bit of science and investigative work. Very informative and really clever work @pc1966

 

[Pretty Mouth]

Brilliant work @pc1966 ???

 

[suffolkspark]

@pc1966 do you have any type AC devices to compare against?

So from how I read it even type A devices here are being paralyzed by DC beyond a few mA?

 

[mattg4321]

@pc1966 do you have any type AC devices to compare against?

So from how I read it even type A devices here are being paralyzed by DC beyond a few mA?

200-300 is more than just a few!

The power breaker rcd will be type ac I expect?

 

[Baddegg]

Bit over my head at this time of day @pc1966 ?......but it’s things like this and people that take the time to explain that make this forum what it is!.....good man ?

 

[suffolkspark]

200-300 is more than just a few!

Yes.... my fault for reading it at 5.30am ?

 

[James]

300 IS JUST A FEW
My phone charger is 5000mA

If a fault made it push that into mains then many rcd would fail to work

 

[littlespark]

“Don’t try this at home...” I like that

From the same author;- “here’s one I melted earlier”

 

[pc1966]

@pc1966 do you have any type AC devices to compare against?

Unfortunately the only two RCD I had lying around were those two tested. Having something like a 100mA AC delay RCD would have been perfect but a bit much to spend £50+ out of curiosity!

So from how I read it even type A devices here are being paralyzed by DC beyond a few mA?

I suspect it is what the standard calls for (sort of "must work up to 6mA DC" requirement) and not what is actually achieved. When I did the tests it was using the AC waveform for a 30mA RCD on top of the DC bias. The "fail" was it not meeting the corresponding test limits of my DiLOG MFT.

Going beyond the 30mA spec and testing with my MFT set for 100/300/500mA RCD types' ramp tests I got this for the PowerBreaker RCD-FCU (was still on test rig this morning):

 

[newfutile]

you must have some phone to have a 5 amp charger ! ,its hard to consider a fault that could transfer that extra low voltage DC on to the mains without the mains being connected to the DC output , these things have areas of separation on the circuit board , sometimes with air gaps to prevent this.

 

[pc1966]

That graph is kind of what I would expect, that as long as you fault AC current is enough to drive the differential sense transformer back against the DC bias to get the "30mA" trip threshold in the secondary it will always fire.

So "blinded" is not quite right word, desensitised is more appropriate.

 

[pc1966]

As @newfutile has just said, you would need a very peculiar fault to allow a typical SELV supply to be driving N-E and not go bang! very shortly after.

Anyway for the purpose of the original discussion, that would a 30mA B-type RCD allow enough current to upset a 100mA incomer, the answer is very definitely no. Certainly so if the earth rod Ra will allow for much more than 100mA to flow as you are interested in hard-fault disconnection, not shock protection as such.

However, what is also clear is for 30mA shock protection you can't allow more than a 100mA or so of DC to be present so if the EV charger or PV inverter says 'type B' is needed, then it matters.

 

[pc1966]

The power breaker rcd will be type ac I expect?

It is pretty old (over 10 years at least) and does not explicitly say what type it is.

The data sheet I found on line for the nearest equivalent says "pulsed DC" which would be type A, but when I tested its in-use brothers a few months back using a 'type A' waveform it failed on In/2 by tripping (one way of 0/180 deg, but not the other, forget which).

So yes, I suspect it would be considered type AC.

 

[pc1966]

I also did high current ramp-test of the Hager RCBO for comparison, one data point looks out of place, but my MFT was complaining of being too hot by then to re-test!

Results below show less initial problems with the DC (as seen in the tables earlier) for the RCBO but once you get to higher DC bias it is essentially the same sort of slope of ramp-test trip current versus DC bias current as the RCD-FCU showed.

It might also be my MFT only has 50mA step sizes for ramp testing by then and it just falls on the threshold for the outlying point:

 

[Risteard]

@pc1966 do you have any type AC devices to compare against?

So from how I read it even type A devices here are being paralyzed by DC beyond a few mA?

A Type A is only designed to deal with pulsating DC currents up to 6mA, so anything beyond that and you're on your own! That's why the 6mA DC leakage detection and disconnection threshold is there for the use of a Type A RCD.

 

[nicebutdim]

The thread that begat this thread spawned all sorts of questions in my amateur mind. Most were fairly straightforward and easy enough to reconcile in my amateur mind, but the issue of discrimination between 'types' of earth leakage protection devices was one which I could not reconcile.

Two things I've taken away from all of this are:

1. If my understanding is correct, it would hypothetcially fine to use Type A time delayed protection upfront of Type AC devices or Type B time delayed (in a future where they exist) upfront of Type A or AC devices.

2. The Type AC time delayed RCD I place upstream of my parent's shiny new board, populated with Type A 1P+N RCBOs is far from ideal. This occurred to me after asking questions in the other thread, so I might as well admit this oversight. Not an immediate issue as those RCBOs are functioning as the manufacturer intended (at least they were prior to this lockdown), and there are very few devices in their house capable of causing DC current issues, but it needs to be corrected

Thanks for taking the time to investigate this further, @pc1966 ?



Edit: Note to any mods reading, the esteemed badge really needs to be removed from my profile.