As noted before, RF issues in Audio fall into the EMC agenda.
If I am allowed to quote from the book "The Circuit Designer’s Companion"
(Tim Williams. 2nd edition NEWNES ISBN 0 7506 6370 7 / 1st edition Butterworth-Heinemann ISBN 0 7506 1756 X) :
• Design for EMC from the beginning; know what performance you require
• Select components and circuits with EMC in mind:
• use slow and/or high-immunity logic
• use good RF decoupling of power supplies
• minimise signal bandwidths with RC filtering, maximise levels
• use resistor buffering on long clock or data lines
• incorporate a watchdog circuit on every microprocessor
• PCB layout:
• keep interference paths segregated from sensitive circuits
• minimise ground inductance with an unbroken ground plane or ground grid
• minimise loop areas in high-current or sensitive circuits
• minimise track and component leadout lengths
• Cables:
• avoid parallel runs of signal and power cables
• make sure that screens are 360˚ bonded through properly designed connectors
• use twisted pair for high-speed data or high-current switching
• run internal cables away from apertures in shielded enclosures
• use multiple ground wires or planes in ribbon or flexi cables
• Grounding:
• ensure adequate bonding of screens, connectors, filters, cabinets etc.
• ensure that bonding methods will not deteriorate in adverse environments
• mask paint from any intended conductive areas
• keep earth straps short and wide: aim for a length/width ratio less than 3:1
• route conductors to avoid common ground impedances
• Filters:
• apply a mains filter for both emissions and immuity: check its required current rating
• use correct components and filter configuration for I/O lines
• ensure a good interface ground return for each filter group
• ensure that filter input and output terminal wiring is kept separate
• apply filtering to interference sources, such as switches or motors
• Shielding:
• determine the type and extent of shielding required from the frequency range of interest
• enclose particularly sensitive or noisy areas with extra internal shielding
• avoid large or resonant apertures in the shield, or take measures to mitigate them
• use conductive gaskets where long (>l/20) gaps or seams are unavoidable
• Test and evaluate for EMC continuously as the design progresses
Each of these sentences is a whole chapter on each own, but every word is carefully (wisely) selected from Tim Williams in this sum-up. The book worth every cent.
Regards
George
If I am allowed to quote from the book "The Circuit Designer’s Companion"
(Tim Williams. 2nd edition NEWNES ISBN 0 7506 6370 7 / 1st edition Butterworth-Heinemann ISBN 0 7506 1756 X) :
EMC design checklist
• Design for EMC from the beginning; know what performance you require
• Select components and circuits with EMC in mind:
• use slow and/or high-immunity logic
• use good RF decoupling of power supplies
• minimise signal bandwidths with RC filtering, maximise levels
• use resistor buffering on long clock or data lines
• incorporate a watchdog circuit on every microprocessor
• PCB layout:
• keep interference paths segregated from sensitive circuits
• minimise ground inductance with an unbroken ground plane or ground grid
• minimise loop areas in high-current or sensitive circuits
• minimise track and component leadout lengths
• Cables:
• avoid parallel runs of signal and power cables
• make sure that screens are 360˚ bonded through properly designed connectors
• use twisted pair for high-speed data or high-current switching
• run internal cables away from apertures in shielded enclosures
• use multiple ground wires or planes in ribbon or flexi cables
• Grounding:
• ensure adequate bonding of screens, connectors, filters, cabinets etc.
• ensure that bonding methods will not deteriorate in adverse environments
• mask paint from any intended conductive areas
• keep earth straps short and wide: aim for a length/width ratio less than 3:1
• route conductors to avoid common ground impedances
• Filters:
• apply a mains filter for both emissions and immuity: check its required current rating
• use correct components and filter configuration for I/O lines
• ensure a good interface ground return for each filter group
• ensure that filter input and output terminal wiring is kept separate
• apply filtering to interference sources, such as switches or motors
• Shielding:
• determine the type and extent of shielding required from the frequency range of interest
• enclose particularly sensitive or noisy areas with extra internal shielding
• avoid large or resonant apertures in the shield, or take measures to mitigate them
• use conductive gaskets where long (>l/20) gaps or seams are unavoidable
• Test and evaluate for EMC continuously as the design progresses
Each of these sentences is a whole chapter on each own, but every word is carefully (wisely) selected from Tim Williams in this sum-up. The book worth every cent.
Regards
George
Last edited:
I accept that digital stuff could be a cause of RFI, but my guess is that most of it actually comes from external sources such as SMPS and broadcast, phones etc.
Full EMC precautions might be a bit over the top for audio, but some of the ideas are useful. At the very least, audio designers ought to be aware of the issue and remember that the active devices don't know they are only supposed to handle audio and ignore everything else.
PS Mods, I'm sorry I allowed someone to annoy me!
Full EMC precautions might be a bit over the top for audio, but some of the ideas are useful. At the very least, audio designers ought to be aware of the issue and remember that the active devices don't know they are only supposed to handle audio and ignore everything else.
PS Mods, I'm sorry I allowed someone to annoy me!
I agree, and my experience is when using guidelines as e.g. cited by gpapag, you save a lot of time and effort, also in audio design.
EMC and signal integrity are two sides of the same coin, EMC is external noise that effects your system, or what your system is putting out there, signal integrity is the noise effecting the circuitry that is creating the noise (cross talk, ground bounce etc).
Have a look at some of the class D designs (a modulated switch mode supply), some of the mods done to digital equipement, these are going to create noise.
Broadcast and phones are well known creaters of noise (transmission) in specific bands, SMPS create noise, digital boards are covered in them, HVAC is one of the best ways of polluting the mains (variable motor drives). It does not take a lot of common mode noise to pollute the atmosphere with EMC, equipement has wires on it (Think dipoles).
To say EMC is not realy relevant for Audio design is being rather naive, the electrical equivelant of burying your head in the sand. Your house will generally have wiring, and noise on the mains means your equipement is within the magenetic loop of that noise.
DIYers go to great lengths to "improve" and "achive" the perfect sound, yet EMC seems to not be of a concern in any of its forms.
Any form of high speed square wave switching, digital boards, SMPS, class D amplifiers, the simplest logic circuit has the ability to create a hell of a lot of noise, it is not just the switchin frequency its the harmonic content of square waves that causes the problems.
Sorry to go on but I spend my days doing PCB's for bespoke communication equipement, and have worked as a PCB designer for over 25 years, and have seen first hand how EMC and its related problems are increasing and how it effects products and equipement, and belive it is bad engineering to ignore tit (EMC).
Have a look at some of the class D designs (a modulated switch mode supply), some of the mods done to digital equipement, these are going to create noise.
Broadcast and phones are well known creaters of noise (transmission) in specific bands, SMPS create noise, digital boards are covered in them, HVAC is one of the best ways of polluting the mains (variable motor drives). It does not take a lot of common mode noise to pollute the atmosphere with EMC, equipement has wires on it (Think dipoles).
To say EMC is not realy relevant for Audio design is being rather naive, the electrical equivelant of burying your head in the sand. Your house will generally have wiring, and noise on the mains means your equipement is within the magenetic loop of that noise.
DIYers go to great lengths to "improve" and "achive" the perfect sound, yet EMC seems to not be of a concern in any of its forms.
Any form of high speed square wave switching, digital boards, SMPS, class D amplifiers, the simplest logic circuit has the ability to create a hell of a lot of noise, it is not just the switchin frequency its the harmonic content of square waves that causes the problems.
Sorry to go on but I spend my days doing PCB's for bespoke communication equipement, and have worked as a PCB designer for over 25 years, and have seen first hand how EMC and its related problems are increasing and how it effects products and equipement, and belive it is bad engineering to ignore tit (EMC).
Yes, absolutely.
Just because it's audible in your home doesn't mean it's measurable in my lab, as conditions may be quite different.
Then there's the other problem, which is that when it comes to measurements, it's very easy to not find what you're not looking for, or don't want to find.
EMC test is applicable to both conducted and radiated interference.
A device will be tested both for emission and immunity.
In the case of radiated emissions and immunity, testing may take place in the open air, but more commonly nowadays is conducted [sic] in an anechoic chamber. A calibrated antenna and spectrum analyser are used to examine the DUT in normal operation across a range of frequencies while it is rotated on a plinth, as the emissions are directional. The antenna may also have a variable elevation.
The same antenna (usually a log-periodic) is used with a high-power broadband amplifier to bombard the device with a variety of modulated and unmodulated signals depending on the particular standard the equipment must meet. The device must continue to perform to specification during the bombardment. The test can employ as much as a kilowatt of power focussed into a comparatively tight beam
Even a comparatively low-power, low frequency digital device can fail radiated test, and obviously the power of the device is irrelevant in immunity test; an induced bit error can cause a system to fall over.
Conducted emissions are tested with the device connected to a specialised piece of test gear called a LISN (line impedance stabilization network) which supplies power to the device through its normal power inlet. This isolates the device from incoming RF and provides a known impedance to which the measuring equipment can be connected.
Devices are normally tested for immunity to static discharge as part of the suite of tests. This consists of 'shooting' the device with an ESD test gun. If it survives in working order, it has passed.
Sometimes the devices under test are quite large, such a mobile phone basestations, and structural modifications may be necessary, as parts of the structure may act as antennas. The skills required to 'massage' items such as these through test are generally only gained over years of experience of RF phenomena.
RF signals are grist to the mill of general electronics, it is only on audio forums that they are treated as having some special significance. They should be kept out of electronics which is not intended to process them, but their presence or their effects are no less observable or tractable than any other real-world environmental influence (such as temperature) affecting the performance of any electronic device.
Anyone interested in EMC standards can find an entry point for research by googling 'CISPR' (Comité International Spécial des Perturbations Radioélectriques)
w
A device will be tested both for emission and immunity.
In the case of radiated emissions and immunity, testing may take place in the open air, but more commonly nowadays is conducted [sic] in an anechoic chamber. A calibrated antenna and spectrum analyser are used to examine the DUT in normal operation across a range of frequencies while it is rotated on a plinth, as the emissions are directional. The antenna may also have a variable elevation.
The same antenna (usually a log-periodic) is used with a high-power broadband amplifier to bombard the device with a variety of modulated and unmodulated signals depending on the particular standard the equipment must meet. The device must continue to perform to specification during the bombardment. The test can employ as much as a kilowatt of power focussed into a comparatively tight beam
Even a comparatively low-power, low frequency digital device can fail radiated test, and obviously the power of the device is irrelevant in immunity test; an induced bit error can cause a system to fall over.
Conducted emissions are tested with the device connected to a specialised piece of test gear called a LISN (line impedance stabilization network) which supplies power to the device through its normal power inlet. This isolates the device from incoming RF and provides a known impedance to which the measuring equipment can be connected.
Devices are normally tested for immunity to static discharge as part of the suite of tests. This consists of 'shooting' the device with an ESD test gun. If it survives in working order, it has passed.
Sometimes the devices under test are quite large, such a mobile phone basestations, and structural modifications may be necessary, as parts of the structure may act as antennas. The skills required to 'massage' items such as these through test are generally only gained over years of experience of RF phenomena.
RF signals are grist to the mill of general electronics, it is only on audio forums that they are treated as having some special significance. They should be kept out of electronics which is not intended to process them, but their presence or their effects are no less observable or tractable than any other real-world environmental influence (such as temperature) affecting the performance of any electronic device.
Anyone interested in EMC standards can find an entry point for research by googling 'CISPR' (Comité International Spécial des Perturbations Radioélectriques)
w
FYI
A good cheep way of adding additional screening,
Press Release [ Noise suppression sheets: Flexield offers outstanding performance for high-frequency applications ] | Corporate Profile | TDK
A good cheep way of adding additional screening,
Press Release [ Noise suppression sheets: Flexield offers outstanding performance for high-frequency applications ] | Corporate Profile | TDK
My belief is that it is foolish to just focus our measurements on the audio band when we are trying to evaluate our listening experience.
Agreed we hear what we hear but do our ears act as a brickwall filter & everything outside our concious hearing is blocked out? Does our ear/brain pick up other information which effect the experience of listening? This, to me, is an area of interest & one that is not really being touched on here but there are references to it throughout the forum.
It would be useful, I believe & was my intention in starting this thread, to try to gather this dispersed information together into this thread for future refernce. For instance JohnLoudb posted something along these lines just recently about a speaker. The whole Bybee thread is beginning to focus on RF & it's possible effect on the audio band. Mr Feedback states that according to the datasheets the effect of ferrite rings in the audio band would be expected to measure infinitesimally - yet I have often seen here claims about the gross audible effect of ferrite beads. Is this an area that there is some real expertise in & that could be shared here?
Agreed we hear what we hear but do our ears act as a brickwall filter & everything outside our concious hearing is blocked out? Does our ear/brain pick up other information which effect the experience of listening? This, to me, is an area of interest & one that is not really being touched on here but there are references to it throughout the forum.
It would be useful, I believe & was my intention in starting this thread, to try to gather this dispersed information together into this thread for future refernce. For instance JohnLoudb posted something along these lines just recently about a speaker. The whole Bybee thread is beginning to focus on RF & it's possible effect on the audio band. Mr Feedback states that according to the datasheets the effect of ferrite rings in the audio band would be expected to measure infinitesimally - yet I have often seen here claims about the gross audible effect of ferrite beads. Is this an area that there is some real expertise in & that could be shared here?
DF96,
I don't think any of the amplifiers were broken, or unstable. But, our amps were clearly susceptible to power line noise. I've also had noise from batteries used as a voltage reference hurt my ears, as well as a cheap computer amp with suspect power supply. In both those cases I could hear the noise and it would hurt just listening to the noise as well as listening to music through those systems.
The brain turns up the gain for some sounds that are unnatural.
I suspect that in a very quiet "near silent" environment you would perceive these sounds that hurt.
I know that when listening to test tones I can't hear anything above 9-10Khz, regardless of sound level, due to high frequency hearing loss.
Do you know how loud or high in frequency the "silent" alarm was you heard? It may not have been above what you could actually hear.
Also, what's the purpose of a silent alarm?
John
I don't know what the alarm operating frequency was. I am not talking about the sounder, but an ultrasonic standing wave detector - sort of audio radar. It detected movement in the room, or windows breaking.
I find it hard to believe that noise from a battery could be painful. I find it easy to believe that ultrasonic instability could be painful, and that this could be triggered by changes in the supply arrangements. Or, the amplifier has a sharp peak in response which happens to coincide with incoming interference. HF instability can be a problem. At the other end, LF instability is common in valve amps with negative feedback. Too much LF can be uncomfortable too.
I find it hard to believe that noise from a battery could be painful. I find it easy to believe that ultrasonic instability could be painful, and that this could be triggered by changes in the supply arrangements. Or, the amplifier has a sharp peak in response which happens to coincide with incoming interference. HF instability can be a problem. At the other end, LF instability is common in valve amps with negative feedback. Too much LF can be uncomfortable too.
Quite so. With every "new technology" device I hook up , I can see it's "garbage" (picture 1 - rail shot) , even on the common ground of the HT system. Turn off all the NT stuff , most of this goes away. I can't leave my stuff off , or no music. I tried this filter (picture 2 - schema 3) , it drastically reduced what I saw on the scope. The problem with pix 2 is that it is for a small SMPS and not for a 1KVA toriod , I plan to fix that and build the "real deal" (big RFI filter).
As far as sound , I think it sounds better without this "hash" in the PS. I have tried a battery powered MP3 player and wondered why it sounded cleaner than my $100 sound card (at times ****).
*** it is strange how the switching pulses "gang up" with multiple sources going simultaneously (TV/PC/Cablebox/HUB). Under these conditions I could hear a difference. On the CRO it was quite obvious.
OS
Attachments
Try stacking five 3 volt coin cells, and using it as a 15 volt reference, in a cascode preamp. Output noise was about -88dB. Anyway, after much trial and trouble, removing the coin cell reference and using voltage divider reference cured the noise problem and it no longer hurt my ears. However, just yesterday I measured the noise of some 12v sealed lead acid batteries I'm using for a preamp. I couldn't measure any noise, but I'm still going to regulate the output.
We bought 6 of those amps and they all caused us problems, until we purchased a power line conditioner. It's a very stable amp ( lots of phase margin ) and has excellent measurements.
But it's hard to understand how such quiet sounds can hurt, unless you've had hyperacusis. Think of nails scratching a chalk board - not a loud sound but it will drive most people crazy (a natural dislike or misophonia). That doesn't hurt, but there are some sounds that just hurt, because of the way our brain responds to the sound.
OS, I did similar scope analysis of PS rails in a USB audio device & found some ugly RF noise (can't find the scope shots at the moment). When I substituted battery supply for critical parts (clocks) of the device it showed significant RF noise reduction on scope measurements & sounded significantly better.
Johnloaudb, thanks for posting your experiences. Have you tried A123 LiFePo4 batteries - I find them the best I have used, without caps or ergs on the output (they have 8mOhm internal impedance already!). I believe there is a noise form batteries when they draw current if the internal impedance isn't sufficiently low enough to deliver the current. Don't know what was going on in your battery ref.
I would love to see some attempts at measuring these "audible" effects in the audio range - I wonder what noise floor level would be needed to reveal these effects in measurements? Any ideas?
Johnloaudb, thanks for posting your experiences. Have you tried A123 LiFePo4 batteries - I find them the best I have used, without caps or ergs on the output (they have 8mOhm internal impedance already!). I believe there is a noise form batteries when they draw current if the internal impedance isn't sufficiently low enough to deliver the current. Don't know what was going on in your battery ref.
I would love to see some attempts at measuring these "audible" effects in the audio range - I wonder what noise floor level would be needed to reveal these effects in measurements? Any ideas?
Last edited:
AVE...
Correct me, if I'm wrong, but putting audio device in metal, grounded box shouldn't remove all RF interferences?
Few weeks ago I made small headphone amp with high gain. Even when powered from battery pack it picked every RF signal in my room: from 50Hz noise of power cords across my desk, BLDC motor of fan I use to cool down my laptop, to SMPS and back-light module of my LCD screen. Each transistor acted as AM demodulator. So I placed this amp in small, metal box, connecting it to the ground of circuit. Blissful silence caressed my eardrums. From my other experiences I know that shielded cable stops any RF signals from entering the circuit through it as long as shield is connected with circuit's ground...
OS, I have 11 EMI filters of this type you presented. I got them from broken computer power supplies for free. However if there is no grounding connected, and you connect the grounding wire of filter to the case of device you placed the filter in, then the case will be constantly polarized with half of mains voltage. I had this problem since in good, old, communism times in my country they preferred two-wire installation with no grounding wire. I had to connect all my equipment to pipes...
Correct me, if I'm wrong, but putting audio device in metal, grounded box shouldn't remove all RF interferences?
Few weeks ago I made small headphone amp with high gain. Even when powered from battery pack it picked every RF signal in my room: from 50Hz noise of power cords across my desk, BLDC motor of fan I use to cool down my laptop, to SMPS and back-light module of my LCD screen. Each transistor acted as AM demodulator. So I placed this amp in small, metal box, connecting it to the ground of circuit. Blissful silence caressed my eardrums. From my other experiences I know that shielded cable stops any RF signals from entering the circuit through it as long as shield is connected with circuit's ground...
OS, I have 11 EMI filters of this type you presented. I got them from broken computer power supplies for free. However if there is no grounding connected, and you connect the grounding wire of filter to the case of device you placed the filter in, then the case will be constantly polarized with half of mains voltage. I had this problem since in good, old, communism times in my country they preferred two-wire installation with no grounding wire. I had to connect all my equipment to pipes...
Yes it would, provided you had no cables connecting to the inside of the box. Hard to imagine how that'd work as an audio system
Conducted RF is a problem.Are you sure they're the same? OS filter does not look to have the Y capacitors, only an X. The schematic shows filters with Y caps, definitely to be avoided unless you're very sure of the quality of your earth
AVE...
Abraxalito, I'm sure. I used the same filter design in my small SMPS, which will be finished when I get my toroid chokes installed...
The best grounding one can get is by connecting ground/earth wire to metal pipe. Grounding in my room is connected to heating system, and all other are connected to the water supply pipes. Water is very good conductor and all pipes are grounded by groundD)...
Shielded cable with grounded shield acts like unshielded cable in metal, grounded pipe. So if the shield is grounded, then there should not be any problems. I have some RF interference and EMI sources, ie. plasma lamp with unshielded, self-oscillating SMPS. I used them to test my designs against EMI and RF interferences. As long as there is grounded case, there should not be any problems...
However it's impossible to make amplifier with no holes, mostly because of heat dissipation concerns. It's OK to make holes, but every RF signal that has wavelength equal to 1/4 of hole diameter or less, would pass through it. That's why sometimes cellphone can interfere with shielded device...
Abraxalito, I'm sure. I used the same filter design in my small SMPS, which will be finished when I get my toroid chokes installed...
The best grounding one can get is by connecting ground/earth wire to metal pipe. Grounding in my room is connected to heating system, and all other are connected to the water supply pipes. Water is very good conductor and all pipes are grounded by ground
D)...Shielded cable with grounded shield acts like unshielded cable in metal, grounded pipe. So if the shield is grounded, then there should not be any problems. I have some RF interference and EMI sources, ie. plasma lamp with unshielded, self-oscillating SMPS. I used them to test my designs against EMI and RF interferences. As long as there is grounded case, there should not be any problems...
However it's impossible to make amplifier with no holes, mostly because of heat dissipation concerns. It's OK to make holes, but every RF signal that has wavelength equal to 1/4 of hole diameter or less, would pass through it. That's why sometimes cellphone can interfere with shielded device...
RF goes through gaps. It is not the width of the gap which matters, but the length of the gap. AM broadcast will be severely attenuated by the gaps in a filing cabinet, but as the signal strength is quite high enough will still get in. Alternatively, if the drawer is not properly earthed then it could act as a coupling capacitor, not an isolator.
Everything you would ever need to know about electromagnetic shielding calculation and shield effectiveness.
http://www.wovenwire.com/NASA-Contractor-Report-4784.pdf
Regards
George
http://www.wovenwire.com/NASA-Contractor-Report-4784.pdf
Regards
George
Maxwell?s Equations, Quantum Electrodynamics, and good installation practices for SI, PI and EMC
Have a look around the rest of the site, shielding is not simple, as DF96 mentioned above any slots, gaps etc and something will get in. We have the use of a ful EMC testing set up and still have problems.
Interesting the views on ultrasonics, some US Navy research has claimed we can percieve sounds up to 100KHz (also human echo-location). What I know from doing SMPS designs is some components can emit very high frequency sound, though you dont actively hear the sound, you persieve it, this can cause discomfort and headaches over time. Wide bandwith amps can also amplify HF noise and againg cause discomfort with the tweeters emiting this noise. Ultrasonic transducers work better. This effect has been tried out to stop teanagers hanging around in certain locations (the mosquito), and also in some military applications.
Have a look around the rest of the site, shielding is not simple, as DF96 mentioned above any slots, gaps etc and something will get in. We have the use of a ful EMC testing set up and still have problems.
Interesting the views on ultrasonics, some US Navy research has claimed we can percieve sounds up to 100KHz (also human echo-location). What I know from doing SMPS designs is some components can emit very high frequency sound, though you dont actively hear the sound, you persieve it, this can cause discomfort and headaches over time. Wide bandwith amps can also amplify HF noise and againg cause discomfort with the tweeters emiting this noise. Ultrasonic transducers work better. This effect has been tried out to stop teanagers hanging around in certain locations (the mosquito), and also in some military applications.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.