Interference, noise and the like
Interference, noise and the like
I am bringing this issue to the hardware thread were I think it should be.
We are currently testing two well known acquisition systems for the biomedical signals (high-resolution). One of the systems is the Biosemi active 2. We have tested this system in a salt water solution as described in the Biosemi instrutions. We have done the acquisition in a shielded room where the other system in test gave us no detectable 50 Hz component.
The spectrum of the bipolar signal from the biosemi system shows the following features:
1. Remarkable high level of 50Hz noise components, and its odd harmonics, that we would expect not to find, given the makers statement and I quote: "ActiveTwo system, the effect of interference currents in the electrode wires is practically eliminated by using active electrodes". the FFT was done on the bipolar signal.
2. A patern of waving frequency tones that seem to bear no relation with the 50 hz component, above the 1000Hz, and that doesn´t seem to follow the filter decay rule. Is this sigma-delta related?
Below i send a picture (the last one is a zoom of the low frequency band)o f the analysis we got, where the 50Hz are seen and its 3rd and 9th harmonics on 150Hz and 450Hz.
http://img56.imageshack.us/my.php?image=ffthj3.png
and here is a zoom:
http://img115.imageshack.us/my.php?image=fttzoomnp8.png
The sample rate used was 16384.
We would very much appreciate any input on this, as this is crucial for a high-resolution aplication such as ours (will not work with this level of 50 Hz interference), thats why we bought the Active2.
We are currently testing two well known acquisition systems for the biomedical signals (high-resolution). One of the systems is the Biosemi active 2. We have tested this system in a salt water solution as described in the Biosemi instrutions. We have done the acquisition in a shielded room where the other system in test gave us no detectable 50 Hz component.
The spectrum of the bipolar signal from the biosemi system shows the following features:
1. Remarkable high level of 50Hz noise components, and its odd harmonics, that we would expect not to find, given the makers statement and I quote: "ActiveTwo system, the effect of interference currents in the electrode wires is practically eliminated by using active electrodes". the FFT was done on the bipolar signal.
2. A patern of waving frequency tones that seem to bear no relation with the 50 hz component, above the 1000Hz, and that doesn´t seem to follow the filter decay rule. Is this sigma-delta related?
Below i send a picture (the last one is a zoom of the low frequency band)o f the analysis we got, where the 50Hz are seen and its 3rd and 9th harmonics on 150Hz and 450Hz.
http://img56.imageshack.us/my.php?image=ffthj3.png
and here is a zoom:
http://img115.imageshack.us/my.php?image=fttzoomnp8.png
The sample rate used was 16384.
We would very much appreciate any input on this, as this is crucial for a high-resolution aplication such as ours (will not work with this level of 50 Hz interference), thats why we bought the Active2.
-
- Site Admin
- Posts: 1141
- Joined: Fri Mar 26, 2004 7:00 pm
- Location: Amsterdam, Netherlands
- Contact:
A Faraday cage shields against electric fields. However, unless metal with a high magnetic permeability is used, the cage does not shield at all for magnetic fields. Therefore, only cages made of special a alloys (mu-metal made of iron and nickel) will effectively shield against magnetic interference. Example of magnetic shielding are the special rooms build for MEG measurements (a noise measureme with the amplifier in such a room will indeed not show any trace of 50 Hz). On the other hand, the typical Faraday cage made of copper (as found in many EEG laboratories) has no effect on magnetic fields at all.
50/60 Hz alternating magnetic fields are typically generated by electric motors, CRT monitors, transformers, and power lines carrying (high) currents.
Magnetic fields cause interference because currents are generated in the loops formed by the electrode wires. The amount of interference is proportional with the area included by the loop. The only practical method of reducing the amount of interference is to bundle/twist the electrode cables in order to minimize the included area of the loops as much as possible, also see http://www.biosemi.com/publications/pdf ... uction.pdf .
During review if the data, your should always choose a reference (a channels, or a combination of channels), and subtract this reference from all channels, see http://www.biosemi.com/faq/cms&drl.htm and viewtopic.php?t=33 . Only after this referencing step, you achieve the full Common Mode Rejection Ratio.
Best regards, Coen (BioSemi)
50/60 Hz alternating magnetic fields are typically generated by electric motors, CRT monitors, transformers, and power lines carrying (high) currents.
Magnetic fields cause interference because currents are generated in the loops formed by the electrode wires. The amount of interference is proportional with the area included by the loop. The only practical method of reducing the amount of interference is to bundle/twist the electrode cables in order to minimize the included area of the loops as much as possible, also see http://www.biosemi.com/publications/pdf ... uction.pdf .
During review if the data, your should always choose a reference (a channels, or a combination of channels), and subtract this reference from all channels, see http://www.biosemi.com/faq/cms&drl.htm and viewtopic.php?t=33 . Only after this referencing step, you achieve the full Common Mode Rejection Ratio.
Best regards, Coen (BioSemi)
Last edited by Coen on Mon Oct 20, 2008 5:53 pm, edited 1 time in total.
Granted there there must exist some degree of 50Hz noise in the shielded room but, what about the biosemi statement: e: "ActiveTwo system, the effect of interference currents in the electrode wires is practically eliminated by using active electrodes".
Isn´t the ActiveTwo suposed to reduced this 50Hz noise to levels compatible with EEG , HR-ECG resolutions?? Is this not its selling flag and and also justification for its price tag?
Isn´t the ActiveTwo suposed to reduced this 50Hz noise to levels compatible with EEG , HR-ECG resolutions?? Is this not its selling flag and and also justification for its price tag?
-
- Site Admin
- Posts: 1141
- Joined: Fri Mar 26, 2004 7:00 pm
- Location: Amsterdam, Netherlands
- Contact:
Active electrodes, nor any shieding of electrode cables or the experimentation room guards against the effects alternating magnetic fields. As explained, only the bundling of electrode cables has effect (as with any other system).
The unique setup of the ActiveTwo (active electrodes, battery power supply, fiber optic data transfer) offers the best possible rejection of interference caused by capacitive coupling of subject, cables and amplifier with power lines and other interference sources.
Best regards, Coen (BioSemi)
The unique setup of the ActiveTwo (active electrodes, battery power supply, fiber optic data transfer) offers the best possible rejection of interference caused by capacitive coupling of subject, cables and amplifier with power lines and other interference sources.
Best regards, Coen (BioSemi)
Our shielded room is professional (it is not a Faraday cage, which is other thing, a term mis-used in this context) it is a room with thick mu-metal walls with no 50 Hz power inside, everything running on bateries.
If the Active2 Biosemi system still picks up 50Hz noise in such a "high noise sterile level environment " I can only assume Active Two is not up to the makers stated specifications in this respect.
Sure our experience with other acquisition high-resolution systems is that the 50 Hz interference in our shielded room is well below the average value in any hospital setting or open research lab, to the point of the other system in test being unable to detect this interference.
So it boils down to the fact that your statement:
""ActiveTwo system, the effect of interference currents in the electrode wires is practically eliminated by using active electrodes"
is a case of misleading publicity, since as you declared above the system seems to be unable to remove that interference, and, in our case it is certainly remarkable low.
If the Active2 Biosemi system still picks up 50Hz noise in such a "high noise sterile level environment " I can only assume Active Two is not up to the makers stated specifications in this respect.
Sure our experience with other acquisition high-resolution systems is that the 50 Hz interference in our shielded room is well below the average value in any hospital setting or open research lab, to the point of the other system in test being unable to detect this interference.
So it boils down to the fact that your statement:
""ActiveTwo system, the effect of interference currents in the electrode wires is practically eliminated by using active electrodes"
is a case of misleading publicity, since as you declared above the system seems to be unable to remove that interference, and, in our case it is certainly remarkable low.
-
- Site Admin
- Posts: 1141
- Joined: Fri Mar 26, 2004 7:00 pm
- Location: Amsterdam, Netherlands
- Contact:
With the ActiveTwo system, it is not possible to pickup 50 Hz interference inside a properly shielded mu-metal room (provided the electrodes, AD-box and battery are inside the room, and the USB-receiver with PC are outside the room). Whether the electrodes are active or passive plays no role in this case, because no 50 Hz electric or magnetic files are present in a totally shielded room. The signal transfer via optic fiber out of the shielded room is fully digital. Therefore the (analog) 50 Hz interference sources outside the room cannot corrupt the signal either
In general, performing a noise measurement (on a microvolt level) with total absence of 50 Hz and its higher harmonics in a high resolution spectral plot is nearly impossible in most practical situations. In my experience it is only possible in a typical MEG room. Such a high quality shielding is not easily duplicated in the most laboratories. This is not a matter of limitations in interference suppression qualities of the ActiveTwo or any other system, it is a matter of limitations in the test setup given the all around presence of 50 Hz interference sources.
The ActiveTwo is designed reject inference better than competing systems in a real situation where electrode impedances are high and the subject's body is capacitive couples with interference sources. Tests in a shielded room are certainly useful to check the system and electrode noise level, but say nothing about the interference rejection capabilities of the system.
Best regards, Coen (BioSemi)
In general, performing a noise measurement (on a microvolt level) with total absence of 50 Hz and its higher harmonics in a high resolution spectral plot is nearly impossible in most practical situations. In my experience it is only possible in a typical MEG room. Such a high quality shielding is not easily duplicated in the most laboratories. This is not a matter of limitations in interference suppression qualities of the ActiveTwo or any other system, it is a matter of limitations in the test setup given the all around presence of 50 Hz interference sources.
The ActiveTwo is designed reject inference better than competing systems in a real situation where electrode impedances are high and the subject's body is capacitive couples with interference sources. Tests in a shielded room are certainly useful to check the system and electrode noise level, but say nothing about the interference rejection capabilities of the system.
Best regards, Coen (BioSemi)
[quote]With the ActiveTwo system, it is not possible to pickup 50 Hz interference inside a properly shielded mu-metal room (provided the electrodes, AD-box and battery are inside the room, and the USB-receiver with PC are outside the room). Whether the electrodes are active or passive plays no role in this case, because no 50 Hz electric or magnetic files are present in a totally shielded room. The signal transfer via optic fiber out of the shielded room is fully digital. Therefore the (analog) 50 Hz interference sources outside the room cannot corrupt the signal either [/quote]
That is true for any acquisition system: no 50 Hz noise, no 50Hz signal in the acquisition records. What are you trying to prove here? The obviuos?
My point is this claim in Biosemi site:
"The ActiveTwo system is the only commercially available system with active electrodes (first amplifier stage on the electrode), the use of active electrodes [b]eliminates[/b] 50 Hz interference pickup by the electrode wires, even when the electrode impedance is high (in the 100 kOhm range)...
This claim is in the first paragraph of:
http://www.biosemi.com/faq/skin_preparation.htm
This claim utterly false as you end up to demonstrate when you state:
[quote]The ActiveTwo is designed reject inference better than competing systems in a real situation where electrode impedances are high and the subject's body is capacitive couples with interference sources[/quote].
So, after all the Biosemi system does not "eliminate" 50 Hz noise as stated in the company site, it just "reject interference better then competing systems" as you state here.
Where do you stand then? Whom do you want to measlead with your tortuous incorrect arguments?
Moreover we are at the moment testing the a so called "competing system" along with the Biosemi one, and we might be up to a surprise.
I think that your customers, namely the ones involved in EEG research, high-Resolution electrocardiography and EMG should be informend that the Biosemi claims regarding "ellimination" of the 50Hz noise, and most common mode signals for that matter, is false, and they should be given the real interference levels in the output (and interference spectra) for a typical pre-determined setup. We have that data for our setup which is "a best scenario setup", so results will be much worth for the average research lab.
That is true for any acquisition system: no 50 Hz noise, no 50Hz signal in the acquisition records. What are you trying to prove here? The obviuos?
My point is this claim in Biosemi site:
"The ActiveTwo system is the only commercially available system with active electrodes (first amplifier stage on the electrode), the use of active electrodes [b]eliminates[/b] 50 Hz interference pickup by the electrode wires, even when the electrode impedance is high (in the 100 kOhm range)...
This claim is in the first paragraph of:
http://www.biosemi.com/faq/skin_preparation.htm
This claim utterly false as you end up to demonstrate when you state:
[quote]The ActiveTwo is designed reject inference better than competing systems in a real situation where electrode impedances are high and the subject's body is capacitive couples with interference sources[/quote].
So, after all the Biosemi system does not "eliminate" 50 Hz noise as stated in the company site, it just "reject interference better then competing systems" as you state here.
Where do you stand then? Whom do you want to measlead with your tortuous incorrect arguments?
Moreover we are at the moment testing the a so called "competing system" along with the Biosemi one, and we might be up to a surprise.
I think that your customers, namely the ones involved in EEG research, high-Resolution electrocardiography and EMG should be informend that the Biosemi claims regarding "ellimination" of the 50Hz noise, and most common mode signals for that matter, is false, and they should be given the real interference levels in the output (and interference spectra) for a typical pre-determined setup. We have that data for our setup which is "a best scenario setup", so results will be much worth for the average research lab.
-
- Site Admin
- Posts: 1141
- Joined: Fri Mar 26, 2004 7:00 pm
- Location: Amsterdam, Netherlands
- Contact:
The active electrodes eliminate interference by capacitive coupling of the electrode cables with interference sources. In most practical situations, this interference pickup by the electrode wires is the main cause for 50 Hz interference. Therefore, the advantage of active electrodes is appreciated by our users. In a test setting, this feature is readily tested by varying the electrode impedance with an inference source near the electrode wire. A system with passive electrodes and unshielded wires will show interference to be dependent on the electrode impedance, whereas a system with active electrodes will show no influence of the electrode impedances up to a quite high levels of impedance (as found with electrodes on the unprepared skin)
As explained in http://www.biosemi.com/publications/pdf ... uction.pdf , there are several other mechanisms besides pickup by electrode cables that may cause 50 Hz interference in the eventual data. The ActiveTwo system is designed to reduce all these influences as best as possible with the currently available technology. Of course, this does not mean that you will never see any residual 50 Hz interference under every circumstance. In other words: we do not claim that all 50 Hz is always totally eliminated (the "elimination of interference" statement cited above explicitly refers to the mechanism of pickup by capacitive coupling of electrode cables). For example, there can always be an exceptional Common Mode problem (e.g. subject very near a power line, AD-box placed on a large metal table, etc.), or a problem with a nearby source of AC magnetic fields (fans, transformers, etc.).
The existence of residual 50 Hz noise in your noise measurement does not contradict our statement about the effectiveness of active electrodes. As I have suggested, there may be other reasons why 50 Hz can be picked up with noise measurements in an improperly shielded room. To be specific, if the level of 50 Hz interference is independent of the electrode impedance, but dependent on orientation of the electrode wires then there is most probably a problem with AC magnetic fields in the test setup. When you compare systems, you will find that in such a case, the level of 50 Hz is only dependent on the area of the loop enclosed by the electrode wires, and not influenced by any other feature of the systems.
There can be many other reasons for 50 Hz pickup in your noise measurements, and for the differences between systems in your particular setup, but I can of course not comment on that without knowing further details of your test setup. When comparing systems, the measurent situation should be equal into detail: same lengths and orientation of the electrode wires, same electrode impedances, same location of the system, same bucket of water, etc..
Best regards, Coen (BioSemi)
As explained in http://www.biosemi.com/publications/pdf ... uction.pdf , there are several other mechanisms besides pickup by electrode cables that may cause 50 Hz interference in the eventual data. The ActiveTwo system is designed to reduce all these influences as best as possible with the currently available technology. Of course, this does not mean that you will never see any residual 50 Hz interference under every circumstance. In other words: we do not claim that all 50 Hz is always totally eliminated (the "elimination of interference" statement cited above explicitly refers to the mechanism of pickup by capacitive coupling of electrode cables). For example, there can always be an exceptional Common Mode problem (e.g. subject very near a power line, AD-box placed on a large metal table, etc.), or a problem with a nearby source of AC magnetic fields (fans, transformers, etc.).
The existence of residual 50 Hz noise in your noise measurement does not contradict our statement about the effectiveness of active electrodes. As I have suggested, there may be other reasons why 50 Hz can be picked up with noise measurements in an improperly shielded room. To be specific, if the level of 50 Hz interference is independent of the electrode impedance, but dependent on orientation of the electrode wires then there is most probably a problem with AC magnetic fields in the test setup. When you compare systems, you will find that in such a case, the level of 50 Hz is only dependent on the area of the loop enclosed by the electrode wires, and not influenced by any other feature of the systems.
There can be many other reasons for 50 Hz pickup in your noise measurements, and for the differences between systems in your particular setup, but I can of course not comment on that without knowing further details of your test setup. When comparing systems, the measurent situation should be equal into detail: same lengths and orientation of the electrode wires, same electrode impedances, same location of the system, same bucket of water, etc..
Best regards, Coen (BioSemi)
-
- Posts: 2
- Joined: Thu Jun 25, 2009 9:44 am
- Location: Herzliya, Israel
50 Hz and odd harmonics noise
We have observed the same phenomenon noted by Parabelo: quite strong noise at 50, 150, and 250 Hz (we are sampling at 512 Hz). While it is the case that we are not recording in a room with magnetic shielding, we have made eery effort to keep potential sources of noise as far as possible away from the subject. We, too, had hoped that the active system would be more resistant to such noise.
Technically, I simply do not understand why we would get the odd harmonics as well - can someone explain that?
Practically, I would like to unerstand how to do the "bundling of electrode cables" mentioned by Coen - what does that mean? Making them shorter by folding them? Wrapping something else around the braids? Coen - can you please explain what to do?
Thanks
Technically, I simply do not understand why we would get the odd harmonics as well - can someone explain that?
Practically, I would like to unerstand how to do the "bundling of electrode cables" mentioned by Coen - what does that mean? Making them shorter by folding them? Wrapping something else around the braids? Coen - can you please explain what to do?
Thanks
Re: 50 Hz and odd harmonics noise
The odd harmonics presence is due to the fact that your electrical company is not providng you with a pure 50Hz sinusoide: it is, in some way, distorted by power transformers and other power devices in its way to the consumer . This is generally the case, electrical companies are not suposed to deliver us a pure 50 Hz sinus wave.
Regarding noise have you tried to subtract the acquired signals, to obtain diferential signals: this reduces common mode noise such as 50 Hz.
What signal are you acquiring?
para
Regarding noise have you tried to subtract the acquired signals, to obtain diferential signals: this reduces common mode noise such as 50 Hz.
What signal are you acquiring?
para
-
- Posts: 2
- Joined: Thu Jun 25, 2009 9:44 am
- Location: Herzliya, Israel
50 Hz
Thanks, Parabelo, for the explanation about the odd harmonics being from the power lines...I wonder why only the odd harmonics?
We are recording EEG on 64+3 channels. This is what our spectral analysis looks like for the entire recording:
http://img338.imageshack.us/img338/7734 ... csnois.jpg
As far as subtraction: when we look at ERPs, subtraction of a prestimulus baseline from each trial should remove the 50 Hz noise. What baseline would you recommend for subtraction for spectral analysis?
In the meantime we are notch filtering using EEGLab at 48-52 Hz, but that removes all the energy in that window.
Daniel
We are recording EEG on 64+3 channels. This is what our spectral analysis looks like for the entire recording:
http://img338.imageshack.us/img338/7734 ... csnois.jpg
As far as subtraction: when we look at ERPs, subtraction of a prestimulus baseline from each trial should remove the 50 Hz noise. What baseline would you recommend for subtraction for spectral analysis?
In the meantime we are notch filtering using EEGLab at 48-52 Hz, but that removes all the energy in that window.
Daniel
-
- Site Admin
- Posts: 1141
- Joined: Fri Mar 26, 2004 7:00 pm
- Location: Amsterdam, Netherlands
- Contact:
Even if the mains voltage waveform is close to a sine wave, the resultant current waveforms may be more like square waves because of non-linear loads (e.g. fluorescent tubes). AC currents in mains supply wires generate magnetic fields. Interference looking like a 50/60 Hz square wave is typical for magnetically caused problems. The spectrum of a square wave consist of only odd harmonics.
Magnetic interference is proportional to the area enclosed by cable loops.Try to keep the area of enclosed loops minimal by bundling all electrode cables together (or even twine them, like the construction of UTP cables). Tape the flatcabels to each other, and to the EX and CMS/DRL cables (or even twist the EX and CMS/DRL around the flatcables) It is not useful to shorten the cables, or wrap shields around them.
As stated above, full CMRR is only achieved after a reference is chosen and subtracted for the other channels, see http://www.biosemi.com/faq/cms&drl.htm, Do not use the unreferenced (raw) signals for further analysis see viewtopic.php?t=33 and viewtopic.php?t=177
Best regards, Coen (BioSemi)
Magnetic interference is proportional to the area enclosed by cable loops.Try to keep the area of enclosed loops minimal by bundling all electrode cables together (or even twine them, like the construction of UTP cables). Tape the flatcabels to each other, and to the EX and CMS/DRL cables (or even twist the EX and CMS/DRL around the flatcables) It is not useful to shorten the cables, or wrap shields around them.
As stated above, full CMRR is only achieved after a reference is chosen and subtracted for the other channels, see http://www.biosemi.com/faq/cms&drl.htm, Do not use the unreferenced (raw) signals for further analysis see viewtopic.php?t=33 and viewtopic.php?t=177
Best regards, Coen (BioSemi)