Can someone explain "harmonic interference" please?

[Harlesden]

A line of text I found in Wikipedia.

Several 378s on the North London Line suffered shutdowns due to harmonic interference from Class 92s using the line for freight movement. The problem was fixed after Bombardier altered the interference tolerance settings.

 

[starrymarkb]

Ah, the infamous AC Traction Pack emissions. Eurostars have caused similar problems in the past with signalling

 

[Old Hill Bank]

Harmonic Interference, well how long have you got, it's a science all of it's own.
Start with the school experiment with the piece of wire and the iron filings and work up from there.
Every electrical device emits a magnetic field which other electrical conductor in close proximity will turn back into electrical current, this could be the trackside signalling equipment or a train on an adjacent line. In the early days of electrification BR had to do a lot of work on this and the engineering documentation must now run into thousands of pages, so sorry no quick answer but I hope this helps.

 

[High Dyke]

It's basically electromagnetic interference of the signalling systems on the railway. A bit like you get too close to your home stereo system with a mobile phone and you hear a burbling sound.

Seemingly the subject of Electromagnetic interference (EMI) has been discussed at worldwide level.

Published in:
Electromagnetic Compatibility, 2003. EMC '03. 2003 IEEE International Symposium on (Volume:1 )

This work shows an evaluation methodology for the occurrence of electromagnetic interference in signaling and control equipment of a DC electrified railway under contingency conditions in which the traction rectifying unit works with 6 pulses instead of the 12 pulses of its normal operation. Thus, electromagnetic interference can result from the direct interaction of the magnetic fields close to the rails as well as the resultant induction in the cables that interconnect the signaling and control equipment

Totally off topic, but as i looked for an easy answer for the OP i found this scientific report...

Influence of the 50 Hz magnetic field on the body color and the amount of urine of goldfish

The mind boggles! :roll:

 

[Nym]

We have a trade just for EMC engineers...

Engineers today also try to design out the problems with things like this from minute one.

The main issue with working on AC systems is that anything nasty is injected back into the grid (being AC) and is seen by everything else. I've seen this on a motor installation we did where the 3rd, 7th and 9th harmonics where already in a very poor condition in the area (Local DNO) so when we put our equipment in place (VSD, Motor, RLV Tranny and Sundries) becasue of DNO regs for our connection, we had to 'Not inject further harmonics below 51st' so we had to fit active filtering, as a result a large proportion of pre-existing harmonics is now warming up my site through the active filtering, we've improved grid quality in that area.

PS: This gets much worse between 3:30pm and 11pm (TVs!)

 

[Teaboy1]

Its a vast subject and to explain & understand you need to be science geek.

Most of us know the existence of the 50Hz AC power network we use at home work ... everywhere. 50Hz is known as the fundamental frequency, it's the base frequency out there in all the wires. When you add active devices like VSD, ReGen Braking, Invertors, Phase splitters .... even rectifiers then these things tend to make noise & frequencies NOT of 50 Hz but of higher in various multiples of 50 Hz. These are classed Harmonics and can be odd or even multiples of 50 Hz .... still with me?
So even harmonics tend to fit in the 50 Hz waveform while odd harmonics dont, odd harmonics tend to distort the nice sinusoidal AC waveform and give it an odd shape ... and anything with an odd shape will produce unwanted effects.
Harmonics can be filtered out but it costs and adds complication [anti KISS philosophy] but it keeps certain types of people in work. These tend to be capacitors and inductors 'tuned' to block certain frequencies but pass other frequencies.
As Nym states, these frequencies are real and consist of real power .... and so can do work and are usually a burden on the system. Putting a resistor in there will just make heat and cause losses [ ie money wasted] so the smart method is to not generate harmonics in the first place or keep them small and well suppressed.
You dont like noise [interference] on your HiFi, radio, TV signal, electrical networks dont like noise on the wires, its similar unwanted stuff!!
HTH and I didnt bore you to sleep !!! ZZZZzzzzzzzzz

 

[Nym]

I'm sure many of us can come up with a better answer when we're awake, but it's a hard thing to explain as a lot of people (no offence here) don't know some of the basics that those who tend to know what harmonics do take for granted.

By the way, any switching of any kind be it solid state, galvonic, or contact gaps will cause harmonics as it puts a square wave into the network...

 

[rebmcr]

square wave

Is that not an oxymoron? I seem to recall on some of my networking training (some time ago) the effort that Ethernet adaptors go to with Fourier transforms to make an "as square as you can get" wave, but it can never actually be square.

 

[PaxVobiscum]

Is that not an oxymoron? I seem to recall on some of my networking training (some time ago) the effort that Ethernet adaptors go to with Fourier transforms to make an "as square as you can get" wave, but it can never actually be square.

An ideal square wave requires instantaneous switching between a fixed positive voltage and a fixed equivalent negative voltage and, so it ain't gonna happen. We can but try, and reasonable approximations are possible at audio frequencies at least, as an oscilloscope will testify. Some of the old analogue synths could produce quite a clean square wave.
However, not really relevant to an electrified railway system where people presumably try hard NOT to generate them.

 

[edwin_m]

An ideal square wave requires instantaneous switching between a fixed positive voltage and a fixed equivalent negative voltage and, so it ain't gonna happen. We can but try, and reasonable approximations are possible at audio frequencies at least, as an oscilloscope will testify. Some of the old analogue synths could produce quite a clean square wave.
However, not really relevant to an electrified railway system where people presumably try hard NOT to generate them.

Electrics with thyristor drive (ie all EMUs from 314 to 322 I think) turn on the AC current part way through each pulse of the AC waveform. This generates numerous harmonics at multiples of 50Hz but fortunately any other railway equipment is designed to tolerate this. Trains with AC motors have variable frequency drives although they are designed to avoid sensitive signalling frequencies. However they also have monitors to shut down the drive if it is malfunctioning and producing harmonics at the wrong frequency, and presumably the 378 in the OP detected some such harmonics in the line produced by a 92 nearby and thought the 378 was itself producing them.

 

[D365]

Electrics with thyristor drive (ie all EMUs from 314 to 322 I think) turn on the AC current part way through each pulse of the AC waveform.

So all PEP and Mk3 (second-generation) EMUs?

This generates numerous harmonics at multiples of 50Hz but fortunately any other railway equipment is designed to tolerate this. Trains with AC motors have variable frequency drives although they are designed to avoid sensitive signalling frequencies. However they also have monitors to shut down the drive if it is malfunctioning and producing harmonics at the wrong frequency, and presumably the 378 in the OP detected some such harmonics in the line produced by a 92 nearby and thought the 378 was itself producing them.

Sounds like a fair reasoning.

 

[edwin_m]

So all PEP and Mk3 (second-generation) EMUs?

Only the AC-only ones. Until the late 80s the traction current through a thyristor could only be turned off by reversing its polarity, which AC does 100 times a second but DC doesn't, so a simple thyristor drive as I described above can't be used on DC.

Most DC units of this era were instead fitted with traditional camshafts and resistances like the slammers, including the 313 which is basically a DC unit with added transformer and rectifer. The exception was some 455s experimentally fitted with solid state chopper control - something which has a lot of potential to cause signalling interference. I think these were later replaced with camshafts to standardise the fleet.

Correcting my previous post, class 319 has solid state choppers using the then new Gate Turn Off thyristors, but still had DC motors. GTOs were later used to drive AC motors on Networkers and 323s.

All post-privatisation EMU designs use a different solid state power device (Insulated Gate Bipolar Transistor) which switches at frequencies well above those used by signalling equipment. Since the harmonics are always higher than the fundamental frequency, IGBT drives are intrinsically much less likely to interfere with signalling.

 

[samuelmorris]

I may well be getting out of my depth here, but something I wanted to know:

I think I eventually grasped roughly how GTO and IGBT systems work, and why the frequency varies. However, can anyone explain what the reasoning is behind:

Electrostars - why the fixed c. 2KHz pitch abruptly disappears above around 40mph, reappears below
Example: http://www.youtube.com/watch?v=Ai1KQCnGAT0&feature=player_detailpage#t=61
Desiros - why the cyclic tone produced from these units changes pitch abruptly at a certain speed (maybe 10-15mph?), and more to the point:
- why this wasn't the case when the units were first delivered [higher pitch was used from a stand on the 360/1s], and
- why AC and DC units behave differently (whereas to my knowledge electrostar behaviour is the same on AC vs DC in this regard) - i.e. increasing pitch with speed at low speed on 450s, different static pitch used at higher speed on DC.
DC:http://www.youtube.com/watch?v=VeDqSRGVqkY
Low increasing pitch: 22s-45s
High pitch: 45s onwards

AC:http://www.youtube.com/watch?v=tyCYHf8QOm4
Low static pitch: 1s-18s
High pitch: 18s onwards

It might seem like a really pointless thing to ask, but it fascinates me that obviously a design decision has been taken to make these changes, I'm curious to know if anyone knows the reason behind them.

 

[edwin_m]

I don't know the detailed answers to these questions but it is probably something to do with the safety case requiring different frequencies to be avoided at different times, or the detailed design of the traction electronics and the software that drives them.

For example the DC units must avoid generating any 50Hz, as that frequency is used for many track circuits in third rail areas, but 50Hz on an AC network is unsurprisingly not a problem. So it is possible that the software does different things depending on the supply voltage at the time (what do dual-voltage Desiros do on the diferent voltages?)

The AC waveform is generated by chopping the supply at a higher frequency and varying the mark-space ratio. The frequency of this variation results in the speed-dependent variable AC frequency that goes to the motors, whereas what you hear is the chopping frequency. It is possible that Electrostars use a fixed chopping frequency at low speeds and switch to some other method at higher speeds when the fixed frequency is too low to produce a useable AC waveform.

 

[Teaboy1]

I think thats the Doppler effect you are referring to ..... someone more informed will correct me if not. EMU's will have several powered wheel-sets and those traveling towards the observer [camera] will perceive an increase in frequency [pitch] while those traveling away from observer will perceive a decrease in frequency. So while at mid point on EMU, front half powered wheel-sets [or noise sources ]are traveling away from observer while that at backend, t'other half are still traveling towards observer.... so a bit of everything.
Naturally an observer riding on said EMU will perceive noise as constant cos she/he is moving at same rate as noise source.
Noise is usually iron laminated tranny cores 'moaning' ... all to do with magnetci characteristics of iron and the polarity flipping back & forth at high frequencies.
If you can crack that problem, you will be minted for life matey !! ie quite magnetic laminated cores with lo-permeability and lo-losses, etc
Pheeww !!

 

[59CosG95]

I myself have always wondered why Desiros and Electrostars (and Networkers too, for that matter) sound the way they do, and now I know why! I presume (if anyone in the BML area when LM sent 350s to Southern to cover the Electrostar shortfall knows otherwise, please tell me) that a dual-voltage 350 on DC sounds like a 450, while the same unit on AC sounds like a 360?

 

[swt_passenger]

I myself have always wondered why Desiros and Electrostars (and Networkers too, for that matter) sound the way they do, and now I know why! I presume (if anyone in the BML area when LM sent 350s to Southern to cover the Electrostar shortfall knows otherwise, please tell me) that a dual-voltage 350 on DC sounds like a 450, while the same unit on AC sounds like a 360?

There are quite a few videos around including 350s at Clapham Jn, (and elsewhere on the SN network) such as this one, about 1m 30s onward:

https://www.youtube.com/watch?v=FszAUPRrw14

 

[edwin_m]

I think thats the Doppler effect you are referring to ..... someone more informed will correct me if not. EMU's will have several powered wheel-sets and those traveling towards the observer [camera] will perceive an increase in frequency [pitch] while those traveling away from observer will perceive a decrease in frequency. So while at mid point on EMU, front half powered wheel-sets [or noise sources ]are traveling away from observer while that at backend, t'other half are still traveling towards observer.... so a bit of everything.
Naturally an observer riding on said EMU will perceive noise as constant cos she/he is moving at same rate as noise source.
Noise is usually iron laminated tranny cores 'moaning' ... all to do with magnetci characteristics of iron and the polarity flipping back & forth at high frequencies.
If you can crack that problem, you will be minted for life matey !! ie quite magnetic laminated cores with lo-permeability and lo-losses, etc
Pheeww !!

That's a very interesting thought. If the two traction packages (ahead and behind the observer) were running at precisely the same frequency then you'd hear a "beat", based on the Doppler-related difference in actual frequency. However if you were close to the track you'd here a constant tone when the whole unit was either approaching or going away.

 

[rebmcr]

That's a very interesting thought. If the two traction packages (ahead and behind the observer) were running at precisely the same frequency then you'd hear a "beat", based on the Doppler-related difference in actual frequency. However if you were close to the track you'd here a constant tone when the whole unit was either approaching or going away.

Assuming you could identify the two sound sources independently, they may interfere with each other. In rare cases they may actually line up perfectly and phase-cancel each other.

 

[Teaboy1]

Oh no ... what's happening ??? Run away quick !!

 

[samuelmorris]

I don't know the detailed answers to these questions but it is probably something to do with the safety case requiring different frequencies to be avoided at different times, or the detailed design of the traction electronics and the software that drives them.

For example the DC units must avoid generating any 50Hz, as that frequency is used for many track circuits in third rail areas, but 50Hz on an AC network is unsurprisingly not a problem. So it is possible that the software does different things depending on the supply voltage at the time (what do dual-voltage Desiros do on the diferent voltages?)

The AC waveform is generated by chopping the supply at a higher frequency and varying the mark-space ratio. The frequency of this variation results in the speed-dependent variable AC frequency that goes to the motors, whereas what you hear is the chopping frequency. It is possible that Electrostars use a fixed chopping frequency at low speeds and switch to some other method at higher speeds when the fixed frequency is too low to produce a useable AC waveform.

I was under the impression that DC units still had inverters producing 50Hz for the lighting etc. All the GEC Alsthom stuff tends to have quite loud inverters producing 50Hz, 100 and 300Hz harmonics, but that's mainly noticeable on the tube stock and the Mk4s. The only unit relevant to that particular discussion would be the 458 - not travelled on them often enough to know if they do it much.

I think thats the Doppler effect you are referring to ..... someone more informed will correct me if not. EMU's will have several powered wheel-sets and those traveling towards the observer [camera] will perceive an increase in frequency [pitch] while those traveling away from observer will perceive a decrease in frequency. So while at mid point on EMU, front half powered wheel-sets [or noise sources ]are traveling away from observer while that at backend, t'other half are still traveling towards observer.... so a bit of everything.
Naturally an observer riding on said EMU will perceive noise as constant cos she/he is moving at same rate as noise source.
Noise is usually iron laminated tranny cores 'moaning' ... all to do with magnetci characteristics of iron and the polarity flipping back & forth at high frequencies.
If you can crack that problem, you will be minted for life matey !! ie quite magnetic laminated cores with lo-permeability and lo-losses, etc
Pheeww !!

Nah, as if you stand on the platform watching it pass - the pitch gets a little higher for each motored car that passes you on the 450s/444s - not so with the 360s.

 

[edwin_m]

I was under the impression that DC units still had inverters producing 50Hz for the lighting etc. All the GEC Alsthom stuff tends to have quite loud inverters producing 50Hz, 100 and 300Hz harmonics, but that's mainly noticeable on the tube stock and the Mk4s. The only unit relevant to that particular discussion would be the 458 - not travelled on them often enough to know if they do it much.

I was discussing the traction electronics, which use large currents so any AC frequency generated by on-board inverters is also likely to appear in the supply and return current. 50Hz produced for auxiliaries would be tiny by comparison and relatively easy to filter out.

 

[Nym]

I was under the impression that DC units still had inverters producing 50Hz for the lighting etc. All the GEC Alsthom stuff tends to have quite loud inverters producing 50Hz, 100 and 300Hz harmonics, but that's mainly noticeable on the tube stock and the Mk4s. The only unit relevant to that particular discussion would be the 458 - not travelled on them often enough to know if they do it much.



Nah, as if you stand on the platform watching it pass - the pitch gets a little higher for each motored car that passes you on the 450s/444s - not so with the 360s.

850Hz on conventional stock, D Stock and 1973 Stock, and this usually isn't seen at the supply as it's isolated by the MA.

 

[samuelmorris]

Motor generators though - so that tends to vary by at least 10% either side in my experience, but yeah it's interesting (if unsurprising) to hear it's only traction side frequencies that cause any impact.

 

[Nym]

Motor generators though - so that tends to vary by at least 10% either side in my experience, but yeah it's interesting (if unsurprising) to hear it's only traction side frequencies that cause any impact.

Just as a quick note, Motor Alternators and Motor Generators are different things.
MGs are DC-DC only and MAs and DC-AC with a rectified DC from the AC signal.

The MARs or Motor Alternator Regulators do keep them pretty close to where they need to be, but the AC on most stock only really drives the lighting (when using an MA). Although I do know that D Stock has more MAs, with HVAC MAs to provide 415/400V AC 3p, being added later, for 72TS this 400VAC was provided by a static convertor.

 

[moggie]

Not my specialist area but from the comments above and depending on the traction package the issue was the not just the harmonic interference generated but the period of time the harmonic existed. Signalling systems will be susceptible to the current only if it is present for a sufficient period to allow the signalling circuit to respond. My understanding is that the traction engineer either has to design out the possibility of the harmonic frequency being generated under any condition or to be able to suppress the harmonic generation within a given time which prevents the signalling system responding. Hence the occasional shut down of the traction package when the time period is exceeded.

Am I right in thinking the software adjustments are changes to increase the allowable time period the harmonic can exist while still maintaining acceptable tolerance to satisfy the signal engineer. Hence the sensitivity of the traction package is reduced to more practical levels?

 

[Nym]

I beleive the timeouts would more be related to heat than issues relating to signalling systems.

 

[edwin_m]

Not my specialist area but from the comments above and depending on the traction package the issue was the not just the harmonic interference generated but the period of time the harmonic existed. Signalling systems will be susceptible to the current only if it is present for a sufficient period to allow the signalling circuit to respond. My understanding is that the traction engineer either has to design out the possibility of the harmonic frequency being generated under any condition or to be able to suppress the harmonic generation within a given time which prevents the signalling system responding. Hence the occasional shut down of the traction package when the time period is exceeded.

Am I right in thinking the software adjustments are changes to increase the allowable time period the harmonic can exist while still maintaining acceptable tolerance to satisfy the signal engineer. Hence the sensitivity of the traction package is reduced to more practical levels?

The critical time for signalling would be in the region of 1 second. Pick-up times on track circuit relays are around that.

 

[Nym]

The critical time for signalling would be in the region of 1 second. Pick-up times on track circuit relays are around that.

Ummmm....

I have DEV relays I'm working on at the moment that pick up in under 200ms.

 

[Teaboy1]

Getting a bit lost with the thread here ........ signal lamps are 110vdc arn't they??
What about track circuits? These much lower I would estimate cos you never get zapped just crossing a pair of normal rails.... say 10-12v. If thats the case, then what sort of interference is needed to swamp 10-12 vdc? The S-N ratio must be nearly half for an olde-schoole magnetic relay to pickup in that case so the crossover/ interence must be truely horrific for an old school DEV to respond.
Obviously electronic 'I_O' in my INDUSTRIAL CONTROLS world pick up range with a few volts upto 240vdc but we filter out any noise above 25Hz and most stuff is immune to AC anyway by design. I see 180 vac superimposed on 24 vdc wires and it just dont see it as a true signal. But them Im in a noisey old switch yard anyway.

My old gaffer use to say that anything below 13A was just a random flow of electrons !!! ..... in a HV world of 275kV etc !