An isolation transformer is used to create a power source that is not electrically, but magnetically, connected to the source. It is primarily used for electrical safety where any exposed electrical conductor can be touched without the risk of electric shock. In certain circles, the isolation transformer is heralded as an answer to power quality issues and this article is intended to highlight why this is not and cannot be the case by a detailed technical look at the claims made by some power quality specialists.
History
Putting isolation transformers into a UPS system is not new. In fact, some technologies relied on the transformer to create the power waveform (ferro-resonance). In most cases the need for the transformer was due to it’s ability to step up an AC voltage. This allowed an inverter to output a low AC voltage due to limitations on the voltage capacity on the power electronic components. More recently advances in power electronics have enabled the transformer to be removed and the inverter output connected directly to the load. This is called a high frequency design and is the basis for modern online double conversion UPS.
The ½V 10V Myth
Some UPS manufacturers include an isolation transformer to allow the Neutral to be bonded to the earth thereby creating a new N-E bond and eliminating any Neutral to Earth Voltage. Claims have been made that common mode voltage (and that means a voltage referenced to a common source – in this case the ground or earth conductor) of over ½V or normal mode (between live and neutral) of over 10V can cause equipment malfunction.
However, Neutral to Earth voltage does not cause malfunction with modern electrical equipment. This can be easily demonstrated by going to Germany and plugging something into the wall socket. You can then unplug it, turn the Schuko lead around by 180 degrees and plug it in again. Hey presto, what was once live is now neutral and vice versa and guess what – the electrical equipment works just the same without any problems at all, despite the neutral to earth voltage being 230V. Or you could simply rewire your mains lead to demonstrate.
In the UK, the neutral and earth are bonded at the consumer unit in any case, and any significant N-E voltage is an infra-structure issue that would require addressing by an electrician. Now perhaps such claims were meant for higher frequencies and so the use of the isolation transformer is to eliminate noise, after all I’ve heard such claims that the isolation transformer “removes noise from the earth, and since many electronic devices use earth as a logic reference any noise of over ½V here can cause logic errors“. Noise on the earth is a misleading phrase. Electrical Noise is simply a high frequency voltage which is measured between two points. You cannot have noise on the earth in isolation, it must be between earth and some other point and in this case the claim is between Neutral and Earth.
This would mean that for any malfunction to occur with as little as ½V of N-E noise, the Neutral conductor must, in some way be used within the logic of the electronic apparatus. Now many years ago this could have been the case, but nothing manufactured over around the last 20 years would have this drawback. Electrical safety standards require that the Live & Neutral conductor are separated from the earth with a high degree of isolation (tested at up to 3000 Volts). To do this they use internal isolation. All computer devices use DC power derived in the main from a device known as a switched mode power supply (SMPS). This contains an isolation transformer which operates at a high frequency which allows it to be substantially smaller than a mains frequency equivalent. The output DC logic levels of the SMPS are completely isolated from the input AC Live and Neutral conductors. Devices that do not utilise SMPS often use rectifiers from a low level AC source, derived from a step down isolation transformer.
This internal isolation negates the need for an external isolation transformer as Neutral to Earth Noise on the input cannot propagate to the DC levels used by the computer logic except in extreme cases.
This is not to say that noise on the AC power line can not cause problems, but there are other ways of addressing this such as simple filtering or improved earthing. The isolation transformer also requires additional filtering to deal with noise and is effective only as a filter in this scenario. The EMC directive has also required equipment to be more robust to the effects of electrical noise.
The Schuko Lead. Proof that N-E Voltage isn't an issue.
Voltage Regulation
An issue with transformers is that they have an output voltage that is dependent upon the loading on the transformer. This is called Regulation. The output voltage of an isolation transformer under no-load is higher than the output voltage under full load.
This causes the terminal voltage to vary with changing loads. It also means that there is the potential to have dangerously high or too low voltages on the output. The input and output voltages of a transformer are dictated by the ratio of the number of windings on the primary side to the number of windings on the secondary side. In order to overcome the regulation issue with transformers operating at close to their capacity is to slightly step up the voltage. This is so as the transformer becomes loaded, the output voltage falls. So for a 230V input at 0% load, the output may be a few % higher than a nominal, and at 100% load the output would be a few % lower. How big this % is depends upon how good the design of the transformer is. The problem is, if your input voltage is on the high side (the specification for the UK is 230V±10%, so that could be 253V) adding a few % more makes the output voltage dangerously high and may cause equipment damage. Equally if your mains voltage is on the low side (207V) then a few % lower may cause your equipment to stop operating.
Some UPS overcome this with buck and boost circuits, however I have known for the output voltage from a transformer based UPS, even with a buck function to be at 260V.
RCDs Don’t Trip
It is perhaps one of the most disturbing aspects that many people installing a power protection device containing an isolation transformer are unaware that any residual current device that was put into the infrastructure to provide fast disconnection in the event of a fault will not operate. Of course, large fault currents as created during short circuits will cause other protection measures to operate, but small fault currents that RCDs are designed to protect against will not cause the RCD to trip.
This is because an RCD operates by detecting an imbalance between the Live and Neutral conductors. Since the isolation transformer isolates the secondary side from the primary, any fault current to earth -no matter how large- causes no deviation from the incoming Live and Neutral current balance as the fault current loop (secondary Live to earth) is contained entirely within the secondary side of the transformer.
This is, in fact, why isolation transformers are used for safety, and are also used in critical power applications such as operating theatres, intensive care wards and chemical plants etc., where an earth fault should not cause disconnection of the supply, and ensures that personnel are safe from electric shock even with a fault. However, this is only true if the output conductors are floating with respect to earth, that is, there is no Neutral to Earth Bond.
If your infrastructure has your circuit protected with an RCD due to the risk of electric shock caused by low fault currents (as recommended where water is nearby) then you should be aware that the RCD will never operate.
This drawback is used however for other purposes – where there is a requirement to remove earth leakage and this is one of the situations where the isolation transformer comes into it’s own.
Power Conditioning
It is interesting to note that for all the recognised power quality problems as defined by the Leonardo Power Quality Initiative (http://www.leonardo-energy.org/) not one of the solutions recommends the use of an isolation transformer. This despite the fact that UK Copper Development Association is one of the members and has some great guides on power quality http://www.copperinfo.co.uk/power-quality/power-quality-guide.shtml that again do not call for the use of an isolation transformer.
In fact, most power quality issues that are noise related can be addressed with improved earthing and many other problems such as harmonic content cannot be addressed with isolation transformers.
Ground Loops
A ground loop is where an earth current flows from one earthed point to another usually through some unexpected path such as data lines and causes malfunction and damage. Now an isolation transformer can stop a ground loop for one piece of attached equipment, if the earth is isolated and the output left floating. You could of course, just cut the earth wire which would have the same effect but this of course is electrically unsafe. As discussed above the isolation transformer can be used for safety purposes and this how in can be used in this context to prevent a ground loop from occurring by effectively disconnecting the earth.
However, if the isolation transformer has the earth straight through, as is the case with transformer based UPS then ground loops will not be prevented.
Transformer Hum
Transformers are known to hum and this is due to poor construction, poor AC power quality or both. Transformer hum increases inefficiency and is of course annoying.
Succesful Trials
One of the methods of selling a power quality solution is to offer to trial and see the benefits. This is all well and good but is not particularly scientific. Personally I knew one large corporation who spent millions of pounds on transformer based power protection solutions due to a successful trial. However this trial was not without it’s critics, who suggested to me that the problem with the trial is that they were comparing doing nothing with fitting one type of power protection solution. Other avenues to address their power quality issues were not explored.
The key is to understand what power quality problems exist and how to eradicate them effectively.
Demonstrations
I’ve witnessed demonstrations of power quality improvements using bespoke equipment that prints out endless reams of paper, and by using an oscilloscope display attached to some sort of box connected to the mains supply. You can flick the lights on and off and wow, an endless stream of spikes appear on the oscilloscope. You then put an isolation transformer based power protection solution in the way, flick the lights on and off again and lo and behold this time no spikes.
It’s a fairly impressive demonstration but has some serious flaws. The first one is fairly obvious. If all these damaging spikes occur just by switching on the lights, then surely computers would be falling over up and down the country. There’d be outcries against manufacturers producing substandard equipment that is susceptible to a light switch.
I’m not saying these spikes are not there, but they’re not what they are made out to be. The intensity of them is extremely low, and they carry no destructive threat. The reason you can see them is because you’re viewing through a very high impedance source of over 1M?. Add any sort of load to the output and you would see these spikes disappear. They’re a bit like radio signals and don’t do anything.
Another concern is that nobody knows how this magic box works. Is it active or passive? Is it amplifying any signals that happen to be there, and if so by how much? Does 1V noise on the oscilloscope equate to 1V noise on the mains? Is it calibrated? If so, how and by who?
That said, you can clearly see the isolation transformer remove the apparent noise, showing that it is doing something. However I’ve also seen the same effect using nothing more than a transorb costing a few pence in a device created by a rival power protection firm.
Like the trials, it is an understanding of real power quality issues that is required.
Switching On Lights - Power Quality Issue?
International Standards on Power Quality
There are several international standards that deal with power quality problems and acceptable limits. Not one of these standards raises a concern about a half volt of electrical noise on the mains supply. Power monitoring equipment manufactured by some of the most respected test equipment manufacturers does not contain measurement tools for such levels of noise either.
It begs the question where do these limits come from? The answer goes way back to a conference in California some time in the 1980′s to do with semiconductor test equipment. It was a paper that was presented and that’s it. There has been no recognised international standard written to endorse or otherwise substantiate the claims presented in that paper.
About Me
I started my UPS career with a company specialising in isolation transformer based power protection solutions. After several years I moved to a competitor company and spent several more years there. I was an isolation transformer fan. As a degree qualified engineer I was never comfortable with the “smoke and mirrors” demonstration and never demonstrated it, instead focussing on the benefits of power protection per se and how the isolation transformer can help overcome these issues. I left the organisation and got another view of the world and started to question the validity of these power protection claims. I have come to the conclusion that the isolation transformer has it’s place as a power protection solution but that there are other technologies that provide better level of protection at a fraction of the cost. Transformerless Online Double Conversion UPS provide protection that is more than adequate for 99.9% of applications, covering all recognised power quality issues.
At my company we provide excellent power protection solutions for a wide variety of applications, from simple battery backup solutions to small data centres in lightning prone Sierra Leone.
As mentioned in this article, education is the key and we’re open and honest about technology and what is can and cannot achieve.
Recent Comments