Big linear bench PSU trips RCD at turn on or off

[Chris Wilson]

I got immense help fixing up my old Farnell 60V @ 50A variable bench linear supply and it's keeping my shack warm as I type, nightly powering a 1kW amp TX'ing on LF

It has one annoying trait though. Quite often at power on or power off from the dual pole panel mains switch it trips the mains distribution panel 30mA RCD. I suspect it ay be something to do with the filter caps and huge choke on the mains input to the supply. Is there a fix, it's a PITA both running up and downstairs to rest the RCD, and for the earache I get from the wife as her "whatever TV trash of the moment" is interrupted    I attach the relevant part of the schematic below, the input caps values are to the right, thanks. Supply is on UK mains 230V at 50 Hz.

[Freelander]

Hi Chris,
Try adding a small/medium load onto the output at power on.
Used to have a similar issue with a BFO ISO transformer on the 'yott'...
Give it a whirl and see what happens.

ps, hope you aint got the Kilo on  on Topband 

[jaycee]

If it's the RCD tripping then it's due to earth leakage, not inrush current. My suspicion would be on the two Y-class capacitors connected from L and N to E. Try removing them and see if the problem goes away. If it does, replace them with new Y-class rated caps.

[donmr]

Assuming your RCD is also the overload protector (circuit breaker in US power terms), it might be the surge current.

I had a similar problem with a saw and found that they make "High Magnetic" breakers which allow a larger short term surge, but have the same long term limits.  I checked that the wire was bigger than the minimum and fairly short before installing one.

[Chris Wilson]

Assuming your RCD is also the overload protector (circuit breaker in US power terms), it might be the surge current.

I had a similar problem with a saw and found that they make "High Magnetic" breakers which allow a larger short term surge, but have the same long term limits.  I checked that the wire was bigger than the minimum and fairly short before installing one.

No, the distribution panel has separate over current breakers, they don't trip, just the RCD one. Thanks for the idea though, cheers.

[Chris Wilson]

If it's the RCD tripping then it's due to earth leakage, not inrush current. My suspicion would be on the two Y-class capacitors connected from L and N to E. Try removing them and see if the problem goes away. If it does, replace them with new Y-class rated caps.

I have just received some new Y rated caps from RS so as it's a challenge wrestling the brute over to open it up i will just change them all out and see what happens. Thanks for the advice!

[oldway]

Do not replace C21 and C22, simply remove them.

[BrianHG]

Worse comes to worse, leave the supply switch on and use an external zero-crossing solid state relay to power the unit on and off.  This might help with surge.

Or, a soft start circuit would solve your power-up, but wont fix the power off.

[jaycee]

Do not replace C21 and C22, simply remove them.

He might prefer to keep the switching crap from the SCR's off his mains supply.

Since as many people have commented, this is not inrush related. An RCD does not detect current overload. It detects current imbalance (what Americans call a GFCI). This means whatever is causing the tripping is causing it by a L/N imbalance. The obvious path for this is leakage to earth.

In the UK, a device which detects both overcurrent and current imbalance is called an RCBO. These are used on individual circuits and are normally much more expensive units, so only used in particular cases. Much more typical is to have a "split load" consumer unit, with circuits requiring RCD protection (all sockets, electric shower) on the RCD side of the board.

[Cerebus]

Worse comes to worse, leave the supply switch on and use an external zero-crossing solid state relay to power the unit on and off.  This might help with surge.

Or, a soft start circuit would solve your power-up, but wont fix the power off.

A zero crossing switch would just make things worse. To minimise inrush current to a transformer you have to switch on at peak mains voltage. Non-intuitive, I know. If you switch at zero volts you apply maximum ^dv/_dt, switch at peak voltage and you apply minimum (i.e. zero) ^dv/_dt.

[floobydust]

I find it strange the on/off switch disconnects both line and neutral.
Electrical standards in North America don't allow that. Only L-L on polyphase gear.

Which gets connected or disconnected first - L or N (as switches never have identical contact action) I think might be aggravating things.

[CopperCone]

Yes, in this case, it sounds like you need the following circuit

1) leave mains earth connected to the chassis, for safety reasons.
2) turn on the Y capacitors, which filter, after a time delay

turn off is more troublesome, you would need to disconnect the caps before turning off, then connect them. Pain in the ass since you need mechanical sequencing or a super cap

or

get a RCD that is industrial, which has a greater trip requirement (but is dangerous)

or

add some resistors between the Y caps and the circuit to limit current.

Your safety is #1, the EMI performance of the thing only matters to legislative bodies.

your Y caps I think basically offer a better path for energy then radiation resistance

*****
personally, I just have a non GFIC outlet in my lab, for unruly devices I don't use often. Some of my microwave amps trigger them.

[C]

Hi Chris,

Some have commented about large inrush current.

If that power supply has huge cap's a smart designer would have the SCR control board turning on the SCR initially at the 180 degree point and gradually reduce the delay to the 90 degree point. Taking several cycles would have a low inrush current.

Notice that only part of line goes through the RFI filter.
Try something simple, just plug the power supply into a power strip with a good RFI filter.

The result would be something like this for power input

[Chris Wilson]

Removing the ground from the filter caps from L and N appears to have fixed this, but I am loathe to leave it like that due to what it powers and the supply's method of SCR switching the primary. I will replace all the caps with new Y caps which i have purchased and see if that helps. Thanks for all the replies, from which i have learned a lot!

[oldway]

Do not replace C21 and C22, simply remove them.

Repeating....
Whe have had the same problem with No Break's.....
The right solution is not to protect this kind of power supply by 30mA RCD.
Use a 100mA one or none....

If you can't modify your main's power distribution, than simply remove C21 and C22....

[jaycee]

Repeating....
Whe have had the same problem with No Break's.....
The right solution is not to protect this kind of power supply by 30mA RCD.
Use a 100mA one or none....

If you can't modify your main's power distribution, than simply remove C21 and C22....

No, the correct solution is to solve what is causing excessive leakage to earth... in this case the Y capacitors being bad. In a device of this age they are probably the old RIFA capacitors known to go bad with age. Modern polyester or polypropylene types should be fine.

[cowana]

A zero crossing switch would just make things worse. To minimise inrush current to a transformer you have to switch on at peak mains voltage. Non-intuitive, I know. If you switch at zero volts you apply maximum ^dv/_dt, switch at peak voltage and you apply minimum (i.e. zero) ^dv/_dt.

That's a really interesting comment, can you go into more detail?

I had always assumed that the dt involved was the switch closing time. This gives the two scenarios of:

Switch at zero crossing:

• Zero crossing occurs
• SSR switches on
• Instantaneous voltage is zero (dv = 0), so inrush current is close to zero
• Voltage and current ramp up to maximum over 5ms of quarter sine-wave (@ 50Hz)
• Device powered as usual

Switch at peak voltage:

• Peak voltage occurs
• SSR switches on in a very short time (dt)
• Instantaneous voltage (dv) is maximum, so inrush current is very high (dv/dt)
• Current decreases to normal maximum level once transformer has been energised
• Device powered as usual

[Freelander]

A zero crossing switch would just make things worse. To minimise inrush current to a transformer you have to switch on at peak mains voltage. Non-intuitive, I know. If you switch at zero volts you apply maximum ^dv/_dt, switch at peak voltage and you apply minimum (i.e. zero) ^dv/_dt.

That's a really interesting comment, can you go into more detail?

I had always assumed that the dt involved was the switch closing time. This gives the two scenarios of:

Switch at zero crossing:

• Zero crossing occurs
• SSR switches on
• Instantaneous voltage is zero (dv = 0), so inrush current is close to zero
• Voltage and current ramp up to maximum over 5ms of quarter sine-wave (@ 50Hz)
• Device powered as usual

Switch at peak voltage:

• Peak voltage occurs
• SSR switches on in a very short time (dt)
• Instantaneous voltage (dv) is maximum, so inrush current is very high (dv/dt)
• Current decreases to normal maximum level once transformer has been energised
• Device powered as usual

https://www.ametherm.com/inrush-current/transformer-inrush-current.html
Below might be more informative - see graph.
https://etrical.blogspot.pt/2016/05/transformer-inrush-current.html

[Cerebus]

https://www.ametherm.com/inrush-current/transformer-inrush-current.html
Below might be more informative - see graph.
https://etrical.blogspot.pt/2016/05/transformer-inrush-current.html

Well done that man. I put off answering that question earlier as I knew it meant a lot of typing, or a lot of hunting for a link. I came back with good intentions of having a crack, and discovered you'd save me the effort. Cheers!

[Freelander]

https://www.ametherm.com/inrush-current/transformer-inrush-current.html
Below might be more informative - see graph.
https://etrical.blogspot.pt/2016/05/transformer-inrush-current.html

Well done that man. I put off answering that question earlier as I knew it meant a lot of typing, or a lot of hunting for a link. I came back with good intentions of having a crack, and discovered you'd save me the effort. Cheers!

No worries,
It is a REALLY good site. ! - the math is sometimes over my (aging) head but the info is excellent.

[Chris Wilson]

Repeating....
Whe have had the same problem with No Break's.....
The right solution is not to protect this kind of power supply by 30mA RCD.
Use a 100mA one or none....

If you can't modify your main's power distribution, than simply remove C21 and C22....

No, the correct solution is to solve what is causing excessive leakage to earth... in this case the Y capacitors being bad. In a device of this age they are probably the old RIFA capacitors known to go bad with age. Modern polyester or polypropylene types should be fine.

I have polypropylene caps to swap to, and i think I may have a 100mA RCD somewhere, thanks!

[oldway]

Repeating....
Whe have had the same problem with No Break's.....
The right solution is not to protect this kind of power supply by 30mA RCD.
Use a 100mA one or none....

If you can't modify your main's power distribution, than simply remove C21 and C22....

No, the correct solution is to solve what is causing excessive leakage to earth... in this case the Y capacitors being bad. In a device of this age they are probably the old RIFA capacitors known to go bad with age. Modern polyester or polypropylene types should be fine.

This is not a problem of faulty capacitors ... If you pay a little more attention, you will find that the RCD triggers on the transient current when the power supply is switched on or switched off.
If the capacitors were faulty, it would no longer be possible to operate the power supply because the RCD would trigger each time the power was turned on.

A 100 mA RCD does not protect against risks of electrocution and can't be used for power outlets for TV's and other appliances.

[Freelander]

Repeating....
Whe have had the same problem with No Break's.....
The right solution is not to protect this kind of power supply by 30mA RCD.
Use a 100mA one or none....

If you can't modify your main's power distribution, than simply remove C21 and C22....

No, the correct solution is to solve what is causing excessive leakage to earth... in this case the Y capacitors being bad. In a device of this age they are probably the old RIFA capacitors known to go bad with age. Modern polyester or polypropylene types should be fine.

This is not a problem of faulty capacitors ... If you pay a little more attention, you will find that the RCD triggers on the transient current when the power supply is switched on or switched off.
If the capacitors were faulty, it would no longer be possible to operate the power supply because the RCD would trigger each time the power was turned on.

A 100 mA RCD does not protect against risks of electrocution and can't be used for power outlets for TV's and other appliances.

[mikeselectricstuff]

It may be worth measuring the current (due to the Y caps) in the ground lead to see if it's anwhere close to the RCD rating.
 

[jaycee]

This is not a problem of faulty capacitors ... If you pay a little more attention, you will find that the RCD triggers on the transient current when the power supply is switched on or switched off.
If the capacitors were faulty, it would no longer be possible to operate the power supply because the RCD would trigger each time the power was turned on.

I suppose it's completely irrelevant that I have seen this kind of fault before and repaired it by doing exactly what I've suggested?

Remember that the RCD detects imbalance between L and N, and that there is a double pole mains switch. One of the contacts is going to engage/disengage before the other one. So there can be a brief period where L is connected and N is not. The small leakage current, instead of flowing through the two leaking Y capacitors, flows to earth instead. Trip.

Also, why would there be an "inrush current" when the device is turned off ? If it was only tripping on power ON, i would agree. The fact that it trips at power OFF shows that it is a leakage issue.