Advice please. Bench safety isolation transformer - input to output grounding ?

[Freelander]

Hello all,
I wonder if you could offer some advice please ?.
I have a Powervar 750 (ABC-750-22MED) power conditioner / isolation device.
I want to use the device as a bench safety isolation transformer for testing and repair. - not as a 'power conditioner'
I have removed the Neutral Ground bond that the device was originally fitted with. (they produce an ABC-750-22ISO without this but hey - this was only 40 quid and I believe the 'link' from Neutral to Ground is the only difference ) .
So, at the moment, the device is completely isolated on the secondary.
My question is this  ?

Regarding the AC Ground (Earth). The unit is obviously fully protected on the input side and case properly grounded. On the output side though, should the AC in ground be connected to the AC output ground ?. (the output will eventually be a single front mounted full Euro socket as opposed to the 6 rear IEC connectors). - I want it front mounted for ease of use on the bench and the EU socket is due to the fact that I am in Portugal at the moment.

I cannot see a need for the AC in ground to be connected to the AC out ground ?. Also, I cannot, on the face of it, see any issue ?. What is the 'normal' way of doing this for a bench safety device please ?.

Should I leave the output ground disconnected or connect it to the AC Input (and case) ground ? - are there any pitfalls  /  benefits in either method ?

I have attached the spec sheet and other data for the device if of use. It is quality bit of kit and the TXFMR is a joy to behold 
(To clarify, I already have removed completely the existing Neutral Ground bond !)

Many thanks.

FL.

[Mark Hennessy]

Leave the output earth disconnected.

This was discussed a while back on another forum I frequent. As this - and many other - conversation shows, there are many misconceptions about isolation transformers.

http://golbornevintageradio.co.uk/forum/showthread.php?tid=4593

[alanb]

This may be of interest

https://youtu.be/byrDzJtR7Fs

[Ian.M]

Although the usual repair bench usage of an isolating transformer is with the output socket ground not connected (fully floating), depending on what you are doing, you may need to tie output Ground to output Neutral (so down-stream RCDs work if there is leakage to chassis) or to input mains Ground (so Live and Neutral are fully floating for minimal leakage, but the chassis is grounded).   The most versatile configuration is to have a row of five 4mm sockets on the front panel in addition to the Schuko socket: Input Ground,  Output Ground, Output Neutral, Output Live1, Output Live2.    The two output lives are normally connected by an insulated jumper^* - remove it to connect an ammeter in series with the output, or just put a current clamp round the jumper.   You *should* use shrouded 4mm sockets (and shrouded 4mm plugs on the jumper) as 4mm binding posts aren't touch-safe at mains voltage. 

* Its also a good place to hook up an appropriate incandescent bulb in series with the load as a 'dim bulb' tester. See: https://www.repairfaq.org/sam/tshoot.htm#tshslt.  N.B. SMPSUs with power factor correction *ARE* *NOT* *COMPATIBLE* with 'dim bulb' testing.

[Freelander]

Hey guys, superb info and good food for thought. ! - really appreciated. What a super wealth of knowledge here

[HalFET]

You can get some nice shrouded and recessed 4 mm jacks rated up to 1 kV if you look around a bit. I'll check if I can find the Farnell or RS order code again.

[Freelander]

This may be of interest

https://youtu.be/byrDzJtR7Fs

Hi Alan, very interesting video, thanks for the link, !
Something I am uncertain of (well, there are LOADS of thing lol) - but specifically here... in the video, at around 6:08, he discusses the 'terra neutral combined separated' system where the neutral and live are fed to the house, then, it appears that the neutral is connected to ground at the consumer unit ? - is this correct ? - this is not a system I have seen before, when I was in the uk it was 3 feed to the house - earth live neutral.

Am I understanding the vid correctly and that is what is happening ? the house is just fed by L / N with N / E connected at the substation and then N connected to E at the consumer unit ?

This seem a bit alien to me ?. For example - how would an RCD work in this instance as there is a direct connection between E and N ?.  (sorry for the lack of brain cells) - perhaps should be another topic....

[Freelander]

Although the usual repair bench usage of an isolating transformer is with the output socket ground not connected (fully floating), depending on what you are doing, you may need to tie output Ground to output Neutral (so down-stream RCDs work if there is leakage to chassis) or to input mains Ground (so Live and Neutral are fully floating for minimal leakage, but the chassis is grounded).   The most versatile configuration is to have a row of five 4mm sockets on the front panel in addition to the Schuko socket: Input Ground,  Output Ground, Output Neutral, Output Live1, Output Live2.    The two output lives are normally connected by an insulated jumper^* - remove it to connect an ammeter in series with the output, or just put a current clamp round the jumper.   You *should* use shrouded 4mm sockets (and shrouded 4mm plugs on the jumper) as 4mm binding posts aren't touch-safe at mains voltage. 

* Its also a good place to hook up an appropriate incandescent bulb in series with the load as a 'dim bulb' tester. See: https://www.repairfaq.org/sam/tshoot.htm#tshslt.  N.B. SMPSUs with power factor correction *ARE* *NOT* *COMPATIBLE* with 'dim bulb' testing.

Hi Ian, many thanks and very informative - great link. I must admit to always having a lack of grasp of 'power factor'. It would be great if Dave could do a video on that topic. It goes a tad over my head - about the height a 747 flies at... I searched the videos and cannot see anything. Is there a way to request a topic for Dave to present ? . There are other articles about but I like Dave's presentation methods.

[nctnico]

Hello all,
I wonder if you could offer some advice please ?.
I have a Powervar 750 (ABC-750-22MED) power conditioner / isolation device.
I want to use the device as a bench safety isolation transformer for testing and repair. - not as a 'power conditioner'

Stop right here and buy differential probes with a safety rating.
An isolation transformer only works if you are trained to use one and if your work environment has been setup for using devices powered from an isolation transformer. If not then it is very easy to make a simple mistake and the DUT will be grounded without any protection from a GFI and/or the earth. For example: the outer shells on the BNC connectors on an oscilloscope are connected directly to the mains earth. If you clip the probe ground to your device then your device becomes grounded again and thus bypasses the isolation transformer AND GFI you may have installed somewhere upstream.

[Freelander]

Hello all,
I wonder if you could offer some advice please ?.
I have a Powervar 750 (ABC-750-22MED) power conditioner / isolation device.
I want to use the device as a bench safety isolation transformer for testing and repair. - not as a 'power conditioner'

Stop right here and buy differential probes with a safety rating.
An isolation transformer only works if you are trained to use one and if your work environment has been setup for using devices powered from an isolation transformer. If not then it is very easy to make a simple mistake and the DUT will be grounded without any protection from a GFI and/or the earth. For example: the outer shells on the BNC connectors on an oscilloscope are connected directly to the mains earth. If you clip the probe ground to your device then your device becomes grounded again and thus bypasses the isolation transformer AND GFI you may have installed somewhere upstream.

Hi, thanks for that, I am fully aware of the potential issues and how to safely use the unit. Also, I am not fixated on using a scope only and have no desire to purchase a differential probe  . But I really appreciate the input. 

[Red Squirrel]

One of the main ideas too is that if you are using a scope, if nothing in your device is attached to earth ground you don't have to worry at all about accidentally putting the ground probe int he wrong place.  You want to measure two sides of a high side shunt or other component that is not ground referenced, no problem, you can.  If you connect the ground to earth ground then you still risk messing this up by error.

What I would maybe do though is make the output a GFCI.

Though I think a scope that is battery operated is probably a good idea too for this sort of thing if you don't have a differential probe setup.  Still good it isolate mains though, less chance of shocks. Either way if working with mains it requires some very special attention and disconnect power any time you make a change such as move a probe around.

[Freelander]

What I would maybe do though is make the output a GFCI.

Hi RS, thanks for the input.  You could not make the output function with a GFCI (RCD). There could be no pathway to ground from an RCD on an isolated output. All current going out = current returning, nothing can pass to ground causing a reduction in balance for the rcd to operate.
Even with gnd connected to the output there is no possibility of any functional rcd in the output as the isolated txfmr output is floating and gnd has no actual meaning or pathway. No pathway, no imbalance on the rcd. No protection.

Or did you mean something else ?. apologies if I misread your intention.

[Ian.M]

+1

A GFCI (RCD) on the isolation transformer output will only give you a false sense of security that may cause you to make a lethal mistake!
 
As all the current that goes into one end of the secondary *must* come out the other end, (neglecting inter-winding leakage capacitance)  there can never be a significant imbalance between the isolated Live and Neutral through the GFCI so it will never trip.   

If you are testing a GFCI equipped device, and need to actually test the GFCI, you can tap isolated Neutral before the GFCI for a local chassis 'Ground' and then the GFCI will trip on excess leakage to chassis, but if you attach the real Ground of a scope probe to the negative side of a SMPSU after the bridge rectifier, you can still get a lethal shock from the positive side without tripping any GFCIs before or after the isolating transformer.   Also clipping Ground to the bridge rectifier negative would immediately make the chassis lethally dangerous if it was tied to the isolated Neutral.

[nctnico]

What I would maybe do though is make the output a GFCI.

  If you put a GFI on the output of an isolation transformer then you need to make the output ground referenced again and thus completely bridging the isolation transformer.

[Red Squirrel]

Won't the GFCI still detect that some of the current is not going through the neutral and instead taking another path?  In older homes where there is no ground, they tell you to put a GFCI there and mark is as ungrounded.  Should current travel between the hot and real ground, since that current is not going to the neutral, the GFCI will trip. 

Of course if you touch the neutral and hot at the same time you will get a shock but this goes with a regular grounded GFCI too.

I just tested this and oddly it does not work like I expected but if current does bypass one of the prongs of the GFCI it does trip, but if it goes to real ground or even fake (the GFCI) ground then nothing happens.  The bulb won't light but it also won't trip.

Either way if working with mains you have to be extra careful.  Treat everything as a potential shock hazard, turn things off before doing anything that could cause accidental contact, etc. A GFCI is an added layer of safety but it's not 100%.  It will still provide lethal power in the right condition (otherwise it would not work at all to provide power).

[Freelander]

Hi RS,
No, it would'nt work.
Think of it like this (hopefully a good way to explain)
Using the familiar terms, L(live) and N(neutral), then normally (without an isolation tfmr) all current flowing between the 2 points must be equal (kirchhoff's law). What goes in, must come out.
At the substation or the house, the Neutral is connected to Ground(E) - (always) this forms the reference point for the AC.
The rcd is sensing this equilibrium down to a tiny tolerance.
'Dave' comes strolling along,,,  (a different 'Dave - 'cos I knew a dave who blew a hole in his finger with a bank of 1kv caps but that's another story)'  ) and touches the 'Live' side - or indeed a point of high enough Potential Difference to N, now there is a path to ground to form a circuit. (Because ground is tied to Neutral !!). As a circuit can be made from L to N so it can ALSO be made from L to E(Gnd) as gnd(E) and N are tied. (there is a slight difference in potential between gnd(E) and N due to the wiring path resistance  from either the substation tie or the house rods / entry point tie). an amount of current now flows from L to Gnd(E). this diminishes the balance from L to N. The rcd is set to detect an imbalance over a certain limit. (say 30mA (and react in a certain time - say 30mS) for example). The rcd trips. Prior to tripping Kirchhoff's law is still true (as it has to be) as the total current entering equals the total current leaving (in this case if X is the current entering, Y is the current returning to N and Z is the current passing to Gnd(E) via 'Dave' then X = Y + Z. ok ?
(this can only occur due to the Neutral to Gnd(E) tie. ) . The Ground(E) reference in use is essential as a common ref point standard is needed for multiple circuits connected together.

Ok, that is basically the NORMAL system.
Now we  add our isolation transformer. Let us call the output of our transformer T1 and T2. The Ground(E)  is NOT connected to either T1 or T2 !. (this is the whole point of isolation). neither is T1 or T2 physically connected to input L and N. Our isolated outputs T1 and T2 are a whole new entity in relation to references.
If 'Dave' now touches part of the circuit (any part!) then no current at all can possibly flow from him to gnd(E) as there is simply no pathway. The current does not originate or return to anything to do with E or our input L and N !. No current flows through 'Dave'. No RCD trip.

It is vitally important to understand that Dave can still die or have a nasty shock though if he touches 2 points in the circuit between T1 and T2 that are of a sufficient potential difference to enable sufficient current to  flow through him ! - he becomes part of the circuit.  .

No RCD can recognise this flow as current can ONLY flow between T1 and T2 and will still follow Kirchhoff's even with dave as an ' added circuit component' between T1 and T2 - note flow THROUGH him as part of the T1 T2 circuit, not FROM him to Ground.!

This is why a transformer classed as a 'power conditioner' with  'isolation' with N bonded to Gnd is NOT a true isolation transformer in a workshop 'safety' (if used correctly!) sense as we are discussing here ! - it is a power conditioner. And offers absolutely bugger all / Zero isolation from hot to earth ! If one bonds either T1 and T2 to Gnd - or indeed ANY input then the circuit now becomes reference and tied to ground(E) and a totally non isolated situation is re-instated

Regarding the house scenario you mentioned- in a non isolated system  - if no 'earth' was fitted. the RCD can still operate due to the mandatory and necessary N to E bond at the substation. (Actually - it is VITAL to consider and understand that  'Neutral' only becomes 'N' (Neutral) due to this bonding - without it there simply is no L and N - just L and L so to speak. ) then Current can pass through 'Dave' to ground and the rcd senses this. Kirchoff's law is true again.

Hopefully as you can see, adding the isolation transformer creates a fully independent  flow pathway from T1 to T2 that is not in anyway referenced to anything else on the input side. this is the absolutely key to understanding how isolation works, and why an RCD would not work.

Hope that makes sense
Joe

ps - This Dave used to work on the busses as before he snuffed it he was a 'conductor' ................ (showing my age ) ...
edited - cannot splllelll...........

[Red Squirrel]

Actually I see what you mean now, as if you do touch a "hot" from an isolated source you won't get a shock anyway so what a GFCI would normally protect you from won't do anything in this case.  But when I tested I did get the isolated ungrounded GFCI to trip in some situations but guess those would be hard to replicate in a device test setup.

I would have figured that you'd still want to tie the isolated "neutral" to isolated "ground" though to follow mains convention.  Wouldn't it still be considered isolated as it's not referencing earth ground?   At least that's what I was thinking of doing for the isolated AC PSU I will be building, so that the output outlets work exactly the same as a household setup just that they are isolated.  In my case the transformer has a centre tap so probably set it up like a split phase system.   If I plug a piece of equipment into it that has a fault where it's using the ground as a return path it should in theory trip the GFCI indicating there is something wrong with that device being tested though I have to test that now as maybe I'm wrong...

[Freelander]

Actually I see what you mean now, as if you do touch a "hot" from an isolated source you won't get a shock anyway so what a GFCI would normally protect you from won't do anything in this case.  But when I tested I did get the isolated ungrounded GFCI to trip in some situations but guess those would be hard to replicate in a device test setup.

I would have figured that you'd still want to tie the isolated "neutral" to isolated "ground" though to follow mains convention.  Wouldn't it still be considered isolated as it's not referencing earth ground?   At least that's what I was thinking of doing for the isolated AC PSU I will be building, so that the output outlets work exactly the same as a household setup just that they are isolated.  In my case the transformer has a centre tap so probably set it up like a split phase system.   If I plug a piece of equipment into it that has a fault where it's using the ground as a return path it should in theory trip the GFCI indicating there is something wrong with that device being tested though I have to test that now as maybe I'm wrong...

Hi RS,
Nooooooooooooo. no neutral to ground bond at the output... no no no....

You need to consider things in more detail.

If you produce electricity, there is a current flowing from one point to another. If you have a pterol driven generator - for example - and it is not connected to earth in any way, and is sat on a rubber mat (in case the neutral is conencted to the chassis) then you have an electricity producing device. this is a completely independent source of energy to any other electricity souce. (current source) and not conencted in any way. That is the key - it is completely independent.

Another example. Sit 2  batteries on the desk, one 12v and one 6V. nothing connecting them together. Now connect an led or bulb or your DVM on V  between positive on one battery and negative on the other (or any terminal on one to any terminal on the other. Will the led light or the bulb light ?. (Clue - - no it will not. Will there be any reading on the voltmeter - no there will not.  Why ? becuase there is no circuit - no current can flow. Each battery is a power source, but each battery is completely independent of each other.  They are completely  ISOLATED energy sources.
Now, connect the negatives together. Then use the bulb or voltmeter, will the bulb light ? will there be a reading on the voltmenter ?.
Yes, in both cases. - there is now a PD of 6v  (due to one battery being 12v and the other 6v) between the two positive terminals on the batteries as the negatives form a 'common' point. - the 6V will be negative or positive on your meter depending which way it is connected to the positive terminals. - the circuits are no longer ISOLATED.
If you do the same with 2 x 12v batteries then no PD is present across the postives with the negatives connected together.

Ok. when you use an isolation transformer, there is no PHYSICAL connection between the primary and the seconday. The current is induced in the secondary as an INDEPENDENT source of power. - INDEPENDENT is the key - it has no physical connection or reference to the input. Like the batteries - unless you have a common reference / connection - then no current can or will flow beween the two.

As I said before, in a normal mains supply, the Neutral is connected by tying it to ground EARTH at the substation or the house. If forms a reference point. There would be no such thing as 'Neutral' at all unless the power company did this.

As neutral is tied to ground, current can flow between L and N and also L and E.

Now, take our transformer output - we have two terminals, they are NOT live and Neutral !. They are not 'referenced' to anything. It is impossible for current to flow between either terminal and Ground.  Ground / the earth is not a source or sink of electricity. It is only that the N is tied to ground as to make it a common reference point.
Our tranformer has a full independent output that is NOT connected to L N or E.
No matter what you touch on the output, no current can flow to the input side L N or E. Not possible. - like the two batteries sat on the desk not connected to each other. the can be no current flow from one terminal on one battery to any terminal on the other battery.
Earth electrical Ground is not some magical connection that exists for all electrical sources. It is CREATED by the power company -  it does not exist in nature before this - It is man made. - out TXFMR output does not have this hence no current can flow to ground in any circumstances.  Current can ONLY flow between the two out terminals of the transformer. THIS IS WHAT WE WANT. We want an ISOLATED output.

IF you connect one of the terminals to ground (to create a 'Neutral') - as neither output terminal IS a 'Neutral' before you do this- (Neutral is a concept until it is created.) then you have completely removed ALL isolation. The output now is referenced to INPUT Neutral / and therefore Ground(E). there is NOW a path from L to Ground - just the same as without isolation and therefore completely useless.
---------------------------------

The isolated output is NOT and should no be referenced to input Neutral or Earth. It should have no possible current path to earth, we do not want any current path to earth.

---------------------------------

as for the RCD on the input in a house without a ground....
In simple terms, it is possible for an rcd on your input to trip without a house ground due to the neutral ground tie at the substation. (it will usually need more current than normal and be hit and miss unless specified accordingly)
---------------------------------
on another note -
IN the UK, electrical connections are keyed at the socket - as in Canada I believe , and the states. so you should (unless you have a wiring fault) always have a known L on the plug lead to equipment and a known N.
I do not know if you are aware of this, but in most of Europe (I am in Portugal at the moment) the plug can be connected to the electrical socket either way around. Earth is always in the same place, L and N is not !. that is perfectly normal here. A lot of techies from the UK and our colonial cousins scratch their heads here..... what is it Dave says - a 'trap for young players' lol
That is how it is - it is perfectly normal and safe.. but that is another topic really... But, hopefully it will give you food for thought that, even at a plug socket and connecting lead to the equipment, the concept of a /neutral/  (if needed) has to be created - in our case - at the equipment. - not at the socket.  However, the consumer unit DOES have a Neutral - and E bond - it just does not carry through to the equipment.
It makes one sit and think when you look at it.
All good fun.

[Ian.M]

For safety working on the isolated D.U.T. you don't want a Neutral to 'Ground' bond on the derived supply.  However if you need to test a RCD equipped DUT, or the DUT has a switched mode PSU and is creating excessive EMI because its Y filter is missing its ground, or excessive common mode switching noise is breaking through on its output, then and only then do you link the isolated Neutral and 'Ground', with the full realisation that this increases your personal risk.    That's why I did *NOT* suggest a hard link or even a switch.  If you are jumpering the isolated Ground to *anything* it should only be via a clearly visible plug-in jumper, preferably with a big yellow CAUTION! tag permanently cable-tied to the jumper.   

Also I didn't mention this earlier but the 4mm socket spacing should be such that the short grounding jumper cant reach the Live sockets, and the short default Live current measurement jumper cant reach the Neutral or Ground.   Plug in longer test leads at your own risk!

One end of each jumper should be a stackable shrouded plug so you can make on-load voltage measurements if you need to, even with the derived Neutral and 'Ground' linked, and the Live jumper in situ.

[Freelander]

For safety working on the isolated D.U.T. you don't want a Neutral to 'Ground' bond on the derived supply.  However if you need to test a RCD equipped DUT, or the DUT has a switched mode PSU and is creating excessive EMI because its Y filter is missing its ground, or excessive common mode switching noise is breaking through on its output, then and only then do you link the isolated Neutral and 'Ground', with the full realisation that this increases your personal risk.    That's why I did *NOT* suggest a hard link or even a switch.  If you are jumpering the isolated Ground to *anything* it should only be via a clearly visible plug-in jumper, preferably with a big yellow CAUTION! tag permanently cable-tied to the jumper.   

Also I didn't mention this earlier but the 4mm socket spacing should be such that the short grounding jumper cant reach the Live sockets, and the short default Live current measurement jumper cant reach the Neutral or Ground.  Plug in longer test leads at your own risk!

+1 . some really good points and suggestions there Ian.