Discussion of AC mains isolation transformers

[One and a half]

22 hours ago, Talley4217 said:

Don't over think this it's just a transformer.  
 

i can tell you the Inrush on my 20kva isn't even measurable using a fluke meter on the inrush setting. This wiring through the 480v input with 120v to get 60/60 output.  
 

where the hit comes from is when you wire 240v input and get a 120v regular output and not having a center tap setup.  Where you parallel the output windings.  My 7.5 hits pretty hard.  I've got it on a 30a 2p breaker and it's fine.   
 

as long as it's remaining a center tap output you shouldn't see a hit. 

That information is good to know. For the primary, it can take 6000V in microseconds and not pass this onto the secondary, so that's a pretty steep dv/dt, so the change in current would oppose by the same. It seems fickle in comparison to parallel the outputs and that ability is compromised. So the output impedance is halved, gee quite a difference to the inrush current. A balanced output has further advantages that affect the primary. I wouldn't use anything but balanced on the output for audio, the noise just vanishes. 

 

For information of members, inrush current on a toroid transformer can be 11 times the primary rated current for 10ms (50Hz), iron cores are a little less when several kVA are concerned. 

[One and a half]

An effective way to filter Transverse (aka differential) and common mode noise is to use a standard EMC filter on the input, and why not on the output.

This pdf from Murata discusses what techniques are used in suppressing both noise parameters. Experiments with filters on the output of a Topaz cleaned up the noise a lot further. Note the EMC filter reflects the noise back to the source. The Topaz already shreds common mode noise, the EMC filter makes sure the noise is contained. At 50/60Hz frequencies the filter has no effect.

 

Most EMC filters have differential mode caps fitted from hot to neutral with a common ground centre point. In a system where the neutral is grounded, one half of these caps is hunted out. For a balanced network, the EMC filter then can actively use both caps They are typically about 1-10uF, sine there are few of them in the complete filter.

 

26to30.pdf

[One and a half]

19 hours ago, gererick said:

Sustained load probably about 1,100 watts as each monoblock draws ~425 watts at idle.  That is not including two Revel Performa B15 subs, which at present are connected via preamp but aren't powered by the IT.

 

How did you measure the 425 Watts or is it spec from amp manufacturer? Need to be a little careful here, amplifiers draw current in pulses, it's not linear. The duty cycle of current can be as bad as 25%, so the cable would have to cope with the derating and go lower down the AWG scale number. Since the amps draw 425W a piece, it's significant. The text below explains. 

 

I'm not familiar selecting cable for NEMA installs, since this depends on how the cable is installed, on a wall, on a wall spaced, in a conduit, how much space inside.... all sorts (for IEC anyway). Dimension the cable based on current draw... @Talley4217 or your local electrician can advise the correct derating to apply.

 

425 (Watts) x 2 (amplifiers) / 120 (Volts) = 7.08A / 0.25 (Derating) = 28.33A rated cable, plus voltage drop allowance.  For IEC in open air that would be 6mm (~ AWG10) , with voltage drop and enclosures, like conduits 10mm (AWG7) is more suitable.  

 

The subs have SMPS Class D amps inside them usually. Connect these to mains directly and not go through IT, there's enough noise as it is, so keep them off the IT supply. 

 

[One and a half]

21 hours ago, Talley4217 said:

And I respectfully disagree with your magnetic field induction.  Nothing works better than external grounding.  We use it on highly sensitive instruments that deal with milliamps as its signals.   Milliamps....  way more sensitive than audio equipment.

Regarding mA, the output of a CD player is rated at 2V 0db, into a 100k ohm load, that's 0.02mA. At quiet passages, that will drop further.

Gas plants traditionally rely on 4-20mA, which is far more robust than audio which are voltage source and pick up interference very easily due to high input impedances. 4-20mA usually 50 Ohms something, quite a low impedance in comparison.

Oddly PT100 cabling can run for some distance without too much drama, provided they stay away from high voltage equipment.

[One and a half]

On 2/12/2022 at 3:36 AM, oneguy said:

After applying power check with the voltmeter function:

H1 to H4 = 120V

X1 to ground = 60V*

X4 to ground = 60V*

X1 to X4 = 120V*

*The isolation transformer will slightly bump up the voltage approximately 1-3V from input so expect slight higher than half the inputs on X1 to ground and X4 to ground along with X1 to X4 being higher than H1 to H4.  

The reason why balanced AC power works is the symmetry between the X1 and X4 to Ground voltages. Connect a bar heater (on the low setting and no fan design) and the measure X1 to GND, then X4 to ground. Really good Topaz and its brethren, can measure to 1% or less between the two measurements.

 

The Topaz 2000VA 0.0005uF measured on a Fluke 289 just now, into an open circuit with low pass filtering. I'm not going to run a heater in 31C day. House voltage is 239V, so on no load the output voltage on the Topaz climbs to the addition of the two voltages, 242.28V H1, H4, in line with your post. House voltage is usually 230V, but on a Sunday afternoon, not much demand, so the volts climb a little.

 

X1 to Gnd 121.32V

X4 to Gnd 120.96V

0.29% variation

 

Adding caps on the input and output will increase this voltage.

 

Measuring an Equitech 1000VA downstream of the Topaz are

121.14 V

121.42 V

0.23% variation.

 

The audio system is changing using the Equitech for the source devices, CD, Lumin, Tuner, DAC.  Preamp, power amp ahead of the Equitech and downstream of the Topaz. I thought of using the Equitech connected to house power, but a problem with this the ground paths are different and would tend to rise. In this way, the grounds are the same between the Topaz and the Equitech.

 

 

 

[One and a half]

6 hours ago, oneguy said:

Both of my 5’s are very quiet. I had a 2kva -32 that had a really bad buzz so I suppose some of it is luck of the draw. What I can guarantee though is yours would be much louder not in balanced configuration.

 

You can put it inside a metal enclosure and it shouldn’t be an issue. I did that with a 2.4kva and a 1kva. Also, as previously shown, putting the capacitor inside the power strip case works too. 
 

I circled the capacitor in the 2.4kva build:

 

 

 

 

 

Nice job! Looking at the interior of the Equitech transformer, all grounds are star connected to a welded stud on the chassis as a star connection. I've copied this setup for a few builds including the internals of PCs where there's a HDD, keeps the grounds at the same potential. The Topaz has two ground connections in the primary box and the secondary box, and to the core. I don't know how they connect, nor can inspect this point for rust/corrosion. For this reason, I wire the grounds from each side to the same star point.

[One and a half]

15 hours ago, rah50 said:

 

Does it get any points for being "Balanced"?

When using off the shelf split phase toroid versus a Topaz or Equitech transformers the difference was dramatic.

Equitech use toroidal transformer,  it is bifilar wound, close coupled with a screen between. What happens is the heat still needs to escape, so the physical size increases to the next size up to dissipate the heat.

 

The other issue is to provide a true balanced output the either end of the output to the centre must be very close less than 2V within each other and the regulation also increases the size. Since amplifiers draw pulsed currents, regulation on the transformer becomes critical as to how the balancing occurs. All that is not simple.

 

From memory the Equitech 1000 VA transformer is about 25cm diameter, a regular 1500VA is 2/3 that.

[One and a half]

Transformer buzzing?  Read on!

 

Transformers can buzz, the EI types like a Topaz or even a toroid type like an Equitech, due to mechanical issues, but often it's electrical. I'm not going to entertain the idea of a DC blocker, they need a bit of care to build, then it's really hit and miss. Buzzing is caused by transfer of harmonics from the secondary to the primary, and with no place to go, the harmonics swirl around the primary in a loop and the buzz continues. The frequency is a little higher than 50/60Hz, and when the transformer is energised, it makes a loud growl then settles down. Well the buzz continued on....

 

The other day, my audio system is setup (on a temporary basis) in a different part of the house, mainly to check the room for rattles, hum, and reflections. The Equitech balanced transformer (1000 VA) plugged into the wall and made a very noticeable buzz and was really annoying. Everything was plugged into the Equitech even a small 230V/120V 100VA transformer. buzzed.

There's not much to fix mechanical wise, it has been a few years since this device was in service, maybe something wrong with it? Plod on.

Being downstairs, and one spare receptacle from the garage 2000 VA Topaz left, and a 20m extension cord was used to feed the Equitech.  Powered up everything & lo, the buzz and noise vanished 😀 Happy Days!

 

The long lead with the upstream Topaz found a nice lossy path and the room is quiet. In testing music, there's still a difference when running AC directly from the wall and with a balanced supply in that:

 

a) the bass is tighter, and the room doesn't rattle with Sade - Love Deluxe - 'No Ordinary Love'

b) It's easier to distinguish the spread-out soundstage, where instruments and vocals are easy to pick out

c) some gain with stage height

 

The effects are similar when USB or Ethernet has noise mitigation, applied although the bass improvement is solely due to the balanced transformer(s). The room is not that quiet, to detect a blacker background, so once other furnishings are in the room, this may show itself. In the meantime, there's the dedicated line to run from the Topaz, that's for the future as part of the renovations.

[One and a half]

While @Speedskater mentions higher than nominal voltage can affect the buzz, in thinking more on the buzz reduction the other day. By adding in the Topaz 2000VA transformer, must have provided enough inductance to slow down the current pulses from the various amplifiers and linear supplies, or literally take the edge off the pulses.

 

The voltage at my house is fairly stable at 230V +/- 3V, measured by a mixture of Kill a Watt, Monster Power conditioner and analog meters.

Used to be 240V, but in line with the euro/IEC standard now the 400V/230V system. To provide for various IT devices, the house 120V (50Hz) varies about the same, but half the 230V variations. 

 

When the Equitech was plugged directly into the wall, the low impedance of the house network has not enough to reduce the noise created by harmonics/current pulses. in the primary I do recall reading in the Equitech Q manual, that an isolation transformer fitted to the upstream of the Equitech is a 'highly recommended' practice, now I know why that it is so.

[One and a half]

6 hours ago, Rsbrsvp said:

Thank you John so much for the response.

 

One more mains purifier that has got literally hundereds of praises on the internet is the Puritin PSM156.  

 

I would be thankful if you could share with us what you think of it.  The reports are so so numerous and perhaps the approval rate is around 90%....  I cant help but wonder what they are doinbg inside of that box????

Every filter like for a DAC has a roll off slope either side of the frequency that is a little flexible. If you have several overlapping filters, as source impedance changes, and the level of noise the filter is trying to remove, it's possible to end up with a bit of mess... depends on how the filter is designed.

I looked at the Purtian PSM156, seems like a regular T network of chokes and caps, plus some controls. I would be tempted to buy, but the cost is a little high just to experiment. It's a drawback that the PSM156 doesn't have under or over voltage protection, this is simple to implement. Surges are a different story some magnitudes higher, but are shorter in duration like ms/us. The longer term where there is sustained overvolts (>260V) for 1/2s or longer, they do happen. Supply monitoring, like a lost neutral is also important., should also be included, and won't impact SQ.

 

In a few months, will either add a Furman sequencer and power conditioner, the regular pro audio kind. Both of these, when healthy can operate a relay which switches an Equitech output to the rest of the audio system. This way, the audio equipment is protected, the power conditioning section of the Furman remains not used, and full isolation from the regular AC.