Discussion of AC mains isolation transformers

[One and a half]

24 minutes ago, skatbelt said:

A question: after reading some earlier posts in this thread I measured the output of my Xentek IT. I was surprised to find out it is wired balanced. I always thought that in a 'normal' setup one of the secondaries was connected to ground so there would be a new hot and neutral. Output voltage between the two poles is (as expected) roughly equal to the input voltage (228V in, 234V out) but I measure approx. 72V between pole 1 and ground and 82V between pole 2 and ground. Is this normal? I would expect to measure approx. 117V on both sides. Anything to do with no load?

It appears the ground is not connected correctly. The ground points you’re measuring are the shields only most likely.

For a balanced system, the centre point x2 and x3 are joined , grounded to the frame and the incoming ground.

At the outlet of the transformer, each rail to ground should be really close to one another and only vary by 1V or less.

[Speedskater]

I don't believe that any of the primaries or secondaries are permanently hard wired to ground. That would be done at installation.

 

But remember that there are different grounds (for instance EGC and GEC) and in different set-ups different terminals would be connected to them (or not).

[look&listen]

I recall that if an iso transformer is wired in Balanced mode, it's power is cut in half. true ?

 

Also, this diagram is for a Dailtron IT, but i think they were the same as Topaz and other brands:

 

[Speedskater]

1 hour ago, look&listen said:

I recall that if an iso transformer is wired in Balanced mode, it's power is cut in half. true ?

For the most part, it's true.

The wire in each coil has a current limit. And power equals V*I=P, so if you drop the voltage to half (60V rather than 120V) then the power is cut in half.

But the transformer also has heat and saturation limits, so and half power neither of these is even approached.

So at the end of the day the power is reduced but not cut exactly in half.

[Speedskater]

1 hour ago, look&listen said:

Also, this diagram is for a Dailtron IT

If the Dailtron is wired as balanced then X1jumped to X4 would be the center tap Safety Ground.

[skatbelt]

18 hours ago, One and a half said:

It appears the ground is not connected correctly. The ground points you’re measuring are the shields only most likely.

For a balanced system, the centre point x2 and x3 are joined , grounded to the frame and the incoming ground.

At the outlet of the transformer, each rail to ground should be really close to one another and only vary by 1V or less.

 

This is what the output side looks like. Jumper between X2 and X3.

 

 

I think the ground is connected as it should on the input side and if I measure between the output line signals and the protective earth/ground pins in the double schuko sockets (IT in one socket, other one free) i get exactly the same (strange) reading. Remark: measured without any load on the IT.

 

Additional remark: the input side is wired for 240V input (options are 100, 120, 200, 220 and 240). I figured this is okay with a voltage of around 230V here in The Netherlands.

[CJH]

To answer skatbelt's question about his measurement at the input of each pole to ground--yes that is normal.

cjh

[Speedskater]

Unless a secondary winding has a connection to the Safety Ground/Protective Earth, any voltage reading from one of it's terminals to another circuit's SG/PE will have what electrician's call a phantom voltage reading.

[One and a half]

4 hours ago, skatbelt said:

 

This is what the output side looks like. Jumper between X2 and X3.

 

 

I think the ground is connected as it should on the input side and if I measure between the output line signals and the protective earth/ground pins in the double schuko sockets (IT in one socket, other one free) i get exactly the same (strange) reading. Remark: measured without any load on the IT.

 

Additional remark: the input side is wired for 240V input (options are 100, 120, 200, 220 and 240). I figured this is okay with a voltage of around 230V here in The Netherlands.

Yes, the green wire is the shield, it’s not true earth that’s why the volt to earth are weird.

Add a green wire from the primary earth to X2, and from the same point to the green terminal.

 

Also redo the outgoing cable by connecting the braid to earth as well. It’s going to be a little busy with the same spots having multiple connections to earth, suggest to make new cables from x2, green terminal, braid, outgoing earth and the main earth from the primary to one connection. Use eye type lugs and screw to hold in place with washers, spring washers and lots of tape.

it appears that the frame of the transformer is not earthed either unlike the Topaz. Please send photo of the primary, especially the earthing arrangement. It may be necessary to drill the frame for an earth point.

 

Voltage symmetry will improve, so will the noise reduction.

[One and a half]

21 hours ago, look&listen said:

I recall that if an iso transformer is wired in Balanced mode, it's power is cut in half. true ?

 

Also, this diagram is for a Dailtron IT, but i think they were the same as Topaz and other brands:

 

Oh, the connection H4 to S1 needs to be split so s1 goes to earth/ground directly. H4 is for the input neutral only and must not be connected to ground!

[skatbelt]

1 hour ago, One and a half said:

Yes, the green wire is the shield, it’s not true earth that’s why the volt to earth are weird.

Add a green wire from the primary earth to X2, and from the same point to the green terminal.

 

Also redo the outgoing cable by connecting the braid to earth as well. It’s going to be a little busy with the same spots having multiple connections to earth, suggest to make new cables from x2, green terminal, braid, outgoing earth and the main earth from the primary to one connection. Use eye type lugs and screw to hold in place with washers, spring washers and lots of tape.

it appears that the frame of the transformer is not earthed either unlike the Topaz. Please send photo of the primary, especially the earthing arrangement. It may be necessary to drill the frame for an earth point.

 

Voltage symmetry will improve, so will the noise reduction.

 

Pictures attached. The previous owner added the extra green/ground wire I suspect. Why a connection between X2 and ground? Thanks for all your time by the way!

 

 

 

[Speedskater]

1 hour ago, One and a half said:

Yes, the green wire is the shield

Clarification: the green wire is only connected to the shield (screen). The wire itself is not the shield.

In use it's connected to the approbate system's SG/PE.

Primary shield to primary system SG/PE.  Secondary shield to secondary system SG/PE.

[One and a half]

4 hours ago, skatbelt said:

 

Pictures attached. The previous owner added the extra green/ground wire I suspect. Why a connection between X2 and ground? Thanks for all your time by the way!

 

 

 

When the power is off, can you move the cable out of way to see the the nameplate. It appears the lower h are jumpers and the brown and blue need to be on h5 and h6.

 

The bluish cable is a Supra lo rad and should have its foil compressed against the frame. 

 

As for earthing x2, the whole idea is to have symmetry to within as close as possible to the same value between x2 x1 and x2 x4. According to the @Cornan school, this is floating, but I don’t attend that school, so x2 needs to be earthed. That’s a regulation for sure in NL.

[Cornan]

8 hours ago, One and a half said:

According to the @Cornan school, this is floating, but I don’t attend that school, so x2 needs to be earthed. That’s a regulation for sure in NL.

 

*LOL* I was´nt aware that I had started a school with my balanced isolation transformer with floating secondary. If anyone should have credit for that it would be @Abtr that originally told me about this simple and effective mod. I am extremely grateful to him for that tip. I am not going to change for any other solution...and I have tried them all. 

[skatbelt]

9 hours ago, One and a half said:

When the power is off, can you move the cable out of way to see the the nameplate. It appears the lower h are jumpers and the brown and blue need to be on h5 and h6.

 

The bluish cable is a Supra lo rad and should have its foil compressed against the frame. 

 

As for earthing x2, the whole idea is to have symmetry to within as close as possible to the same value between x2 x1 and x2 x4. According to the @Cornan school, this is floating, but I don’t attend that school, so x2 needs to be earthed. That’s a regulation for sure in NL.

 

So, you are saying that the way the instructions on the inside of the Xentek describes to connect for 240V in / 240V out is not allowed in the Netherlands? The frame is earthed by the way. I get continuity between incoming earth and all bare metal parts.

 

Order of the connection points:

 

H1       H3        H2        H4        H5       H6

 

 

The input instructions on the nameplate:

 

VAC    Jumper(s)            HOT      NEUTRAL

100     H1-H3, H5-H6       H6             H1

120     H1-H3, H2-H4       H4             H1

200     H3-H5                    H6             H1

220     H3-H2                    H6             H1

240     H3-H2                    H4             H1

 

At the moment I have hot on H4 and neutral on H1 and a jumper over H3 and H2, so wired for 240V input.

 

 

 

 

[One and a half]

1 hour ago, skatbelt said:

 

So, you are saying that the way the instructions on the inside of the Xentek describes to connect for 240V in / 240V out is not allowed in the Netherlands? The frame is earthed by the way. I get continuity between incoming earth and all bare metal parts.

 

Order of the connection points:

 

H1       H3        H2        H4        H5       H6

 

 

The input instructions on the nameplate:

 

VAC    Jumper(s)            HOT      NEUTRAL

100     H1-H3, H5-H6       H6             H1

120     H1-H3, H2-H4       H4             H1

200     H3-H5                    H6             H1

220     H3-H2                    H6             H1

240     H3-H2                    H4             H1

 

At the moment I have hot on H4 and neutral on H1 and a jumper over H3 and H2, so wired for 240V input.

 

 

 

 

Thanks for the table, the connections are clearer now.

some  corrections 

min the photo, the blue is neutral and it needs to go to H1. The hot is a brown and it needs to go to H4. The link at h3 + h2 can remain.

 

it is permissible to connect an earth at one point and take off another earth at the other provided the metal is a continuous piece not interrupted by mechanical fasteners.

example : one piece of brass, earth screwed in at one end and another earth taken at the other is permissible. 

Example: an enclosure that has the sides screwed to each other, earth is located on the bottom shelf and another earth taken from the side to provide an earth for a smps. Not permitted. The smps earth must connect directly to the main earth by a cable.

 

the Xentek is a construction of several parts, the two end pieces and the core held together by fasteners. The output earth must be connected by cable to the incoming earth. Sure you can measure continuity, but that method is not allowed now, because metal rusts and compromises the fault path impedance.

 

The output earth connection is a symmetrical/ balanced type. Again, to provide symmetry for reduction of noise, x2 must be connected to earth. This is how a symmetrical design works by cancelling currents of nuisance value. If the earth connection is not made, there’s no balance any more and the common mode noise is back polluting the audio system.

 

Another thing, the Lo rad primary earth wire has a conductive plastic, avoid placing it too close to the other conductors.

[Speedskater]

4 hours ago, Cornan said:

*LOL* I was´nt aware that I had started a school with my balanced isolation transformer with floating secondary.

No you didn't start that school, but you are an alumni.

[Cornan]

2 hours ago, Speedskater said:

No you didn't start that school, but you are an alumni.

 

Albtr´s balanced IT alumnus! I am fine with that!  

[lmitche]

1 hour ago, Cornan said:

 

Albtr´s balanced IT alumnus! I am fine with that!  

Nope, we know you are out of balance, and we like that about you!

[skatbelt]

6 hours ago, One and a half said:

Thanks for the table, the connections are clearer now.

some  corrections 

min the photo, the blue is neutral and it needs to go to H1. The hot is a brown and it needs to go to H4. The link at h3 + h2 can remain.

 

it is permissible to connect an earth at one point and take off another earth at the other provided the metal is a continuous piece not interrupted by mechanical fasteners.

example : one piece of brass, earth screwed in at one end and another earth taken at the other is permissible. 

Example: an enclosure that has the sides screwed to each other, earth is located on the bottom shelf and another earth taken from the side to provide an earth for a smps. Not permitted. The smps earth must connect directly to the main earth by a cable.

 

the Xentek is a construction of several parts, the two end pieces and the core held together by fasteners. The output earth must be connected by cable to the incoming earth. Sure you can measure continuity, but that method is not allowed now, because metal rusts and compromises the fault path impedance.

 

The output earth connection is a symmetrical/ balanced type. Again, to provide symmetry for reduction of noise, x2 must be connected to earth. This is how a symmetrical design works by cancelling currents of nuisance value. If the earth connection is not made, there’s no balance any more and the common mode noise is back polluting the audio system.

 

Another thing, the Lo rad primary earth wire has a conductive plastic, avoid placing it too close to the other conductors.

 

I will follow up your advice and report back.

 

If I go one step back, the reason I began doubting and measuring was because I did some tests with and without the Xentek in the chain and - although the difference was not huge - I preferred the situation without it. A bit more resolution, more grain free and dynamical. I really kept my test setup simple. So only Mac Mini/microRendu/LPS-1 all powered by Uptone JS-2 in one socket of the power strip and my Chord DAVE in a second socket. DAVE was connected to my Bakoon HPA-21 headphone amp (this HA operates on batteries) and listening was via Audeze LCD-3 headphones. No external incoming ethernet. I have a direct connection between Mac Mini and microRendu.

[One and a half]

9 hours ago, skatbelt said:

 

I will follow up your advice and report back.

 

If I go one step back, the reason I began doubting and measuring was because I did some tests with and without the Xentek in the chain and - although the difference was not huge - I preferred the situation without it. A bit more resolution, more grain free and dynamical. I really kept my test setup simple. So only Mac Mini/microRendu/LPS-1 all powered by Uptone JS-2 in one socket of the power strip and my Chord DAVE in a second socket. DAVE was connected to my Bakoon HPA-21 headphone amp (this HA operates on batteries) and listening was via Audeze LCD-3 headphones. No external incoming ethernet. I have a direct connection between Mac Mini and microRendu.

When the reference is lost, there's additive common mode noise, since the supply rails float to what they want to do. There's no cancellation either as explained before, can explain why the grain is appearing. The ultra isolation transformers are common mode killers, they do a great job at that, they don't add!

 

I tried two Curios cables and found them to add crud in a balanced system, rather than take it away, perhaps if you have a generic cable to try, this might be an eye opener. Also keep the polarity the same on the output for AC powered devices.

[jamesg11]

 

22 hours ago, One and a half said:

I tried two Curios cables and found them to add crud in a balanced system, rather than take it away, perhaps if you have a generic cable to try, this might be an eye opener

 

How so?

[One and a half]

6 minutes ago, jamesg11 said:

 

 

How so?

It's what I listened to, for a little while I tried to like them, just didn't work out. The vendor was OK in returning the Curious Cables.

[jamesg11]

Righto, so presumably it’d add crud in any setup then ...

[One and a half]

16 minutes ago, jamesg11 said:

Righto, so presumably it’d add crud in any setup then ...

A cable can’t really add anything but can propagate noise from device to device.