Hunt for RFI offenders

[One and a half]

On thoughts of reducing RFI and EMC offenders in the home and their influence on digital audio, the victim has enough resilience to ignore that interference, but why make it easy for noise to propagate further.

There are so many SMPS and RF generators in the home where will it ever stop, it's just not possible to remove them all, but at least identify them and route audio power and signal cables away from them to minimise coupling. Distance is very good, ferrites are a good weapon. Armed with a portable LW radio tuned around the 153kHz mark (the closest to the audio band as I could tune) lets set on a hunt to see which noise maker is the worst.

 

List of items tested

 

1. HP Z800 tower style workstation
2. Lenovo mini computer i7
3. HP Laserjet M252 network printer
4. Topaz 1000VA transformer
5. TPlink 24 way network switch
6. Generic 12V 2A SMPS
7. Dell 24in monitor
8. Mac mini 2010 vintage
9. QNAP 419P NAS
10. Asus wireless repeater
11. Mutec MC-1.2 USB to S/PDIF converter
12. Accuphase E-450 integrated amplifier
13. iPhone 8 plus 

 

Based what I heard on the radio, some items were disconnected immediately. All devices were in their normal operating condition or in standby mode like the printer.

 

Have a guess as to which devices were the worst and which were the least noisiest. No need to grade them all, just the worst or best will be fine. I was quite stunned as to which were really quiet.

[asdf1000]

11 minutes ago, One and a half said:

Have a guess as to which devices were the worst and which were the least noisiest. No need to grade them all, just the worst or best will be fine. I was quite stunned as to which were really quiet.

 

You'll need to reveal the rankings for us later!

 

But my guess for worst offender is the Asus wireless repeater? The best of the offenders being the SPDIF converter?

 

[One and a half]

Yes, I'll leave the post up for a few days - week then reveal!

[look&listen]

BTW - Magnetic, Electric, and Radio/microwave Detection All in One Package:

Trifield 100XE EMF Meter

[Spacehound]

 

Most of it should not matter.

And remember you tuned your radio to where there was just noise, which will be at a much lower level than a radio or TV signal, yet you don't seem to be worried about those.

 

Almost all of our stuff is in grounded metal boxes (Faraday cages - remember a few holes in a Faraday cage don't matter, chicken wire ones work fine) and usually all cables except maybe the speaker cables are screened.

And there is nothing preventing you screening those if you suffer from OCD

 

If RFI doesn't mess with a moving coil phono cartridge and its lead to the preamp  it certainly won't hurt anything else.

[mourip]

 

Looking forward to your results...

[Speedskater]

Jim Brown is the go-to expert on noise & interference.  He is a past AES committee co-chair on EMI/RFI and a Ham radio operator. He has about 50 papers & Power Points many on noise and interference:

http://www.audiosystemsgroup.com/publish.htm

 

[Speedskater]

While the TriField is a good meter, an old-fashioned battery powered AM radio can find many of the offenders. Tune the radio to a quite spot near the low end of the dial then go to the areas that you suspect.

 

Also the TriField can give high noise reading that have no impact on your audio system.  As good hi-fi designers know that we live in a noisy world and take preventive measures.

 

Noise should be attacked at it's source, it that's not possible then at the susceptible hi-fi component, not at a power conditioner 6 feet away.

[esldude]

I have done the same thing at different frequencies with multi-band radios.  I've kicked around the idea of getting one of those inexpensive software defined radios.  The advantage would be you listen (via radio waves) once and can let software examine the bands of interest.  No need to tune a single frequency and check then do it all over again at another frequency. 

 

I never did that for the reason stated above.  I could tune in some nasty sounding noise next to gear.  A nearby in frequency radio station might drown it out.  In the worst cases the gear might overpower radio broadcasts, but 18 inches from the gear radio transmissions were able to dominate.  So the emissions from gear, if not in cabinets close to other sensitive gear (like a tubed MC phono stage next to a Mac Mini) just don't amount to enough to do anything. 

 

Also gave lie to the idea various tweaks in OS software could tune such systems for less emissions and better sound quality.  The same exact OS on various PC devices had dramatically different signatures.  And different OS on the same device had dramatically different emission signatures.  The whole thing in that direction is a fool's errand. 

 

 

[Ralf11]

I agree with much of the above.

 

What we really need is a way to examine noise inside the cases of our gear, or - better - inside the circuits.

[mansr]

47 minutes ago, Ralf11 said:

What we really need is a way to examine noise inside the cases of our gear, or - better - inside the circuits.

I had an old computer that was so noisy inside that ATA drives wouldn't work at the top speed, not even with a fancy shielded 80-wire cable. SATA, which was new at the time, worked perfectly.

[One and a half]

The noise picked up by the radio is energy that's escaped the Faraday cages, so fat lot of good that does, and the 'best of intentions' as far as EMC Class B emissions for conducted noise. The reason for the 'best of intentions' statement, for testing EMC from equipment statements is dubious to non existent in the real world. Ever noticed the first part of an IT device packaging that has printed material is not the user manual, but the obligatory EMC statement, that the device complies with FCC et al. Throw this in the recycle bin amid all the other packaging materials. 

If that energy is picked up by the radio, how much of that energy is capacitively coupled to the ground plane at the source, where the energy is lot more intense. 

Some may argue that RF frequencies don't matter for audio bands, however, audio relies on very small voltages to extract minute details of recordings, and if coupled RF/common mode noise/differential mode noise is added to signals, then RF emissions do matter. SMPS are a classic noise generator, they have no place near audio systems. Sure, some SMPS are just as quiet as a linear (like one in a million designs), the piece of paper shipped with it says so, hahahahaha! Pull the other one.

 

Another classic example is to use a laptop -> (any) USB cable -> DAC and listen. Add in a (ISO)Regen/Intona/Micro 3.0/sOTMxxx to the chain and listen again. There's a reason why USB fixers still sell. Ethernet is not off the hook either.

The Jim Brown site is quite a good reference, it will take some time to digest. Hams are very keen to remove RF as it impacts on transmissions, however 'small' these usually IT devices are, they create a lot of havoc. I doubt an SMPS is in the same radio shack as the rest of the gear they use.

[mansr]

5 minutes ago, One and a half said:

If that energy is picked up by the radio

Remember that a radio is specifically designed to pick up these signals. A device designed to not pick them up ought to be influenced to a lesser degree.

[One and a half]

4 minutes ago, mansr said:

Remember that a radio is specifically designed to pick up these signals. A device designed to not pick them up ought to be influenced to a lesser degree.

Yes, it's the back door that's the main concern for one and how much RF is ignored.

[Speedskater]

3 hours ago, mansr said:

Remember that a radio is specifically designed to pick up these signals. A device designed to not pick them up ought to be influenced to a lesser degree.

"ought" is the operative word!

Unfortunately some hi-fi equipment still picks-up too much.

And in the US, noise & interference rules are so far done the priority list, that it is seldom checked in new appliances, high-tech lighting and wall-worts.

[Spacehound]

4 hours ago, One and a half said:

The noise picked up by the radio is energy that's escaped the Faraday cages, so fat lot of good that does, and the 'best of intentions' as far as EMC Class B emissions for conducted noise. The reason for the 'best of intentions' statement, for testing EMC from equipment statements is dubious to non existent in the real world. Ever noticed the first part of an IT device packaging that has printed material is not the user manual, but the obligatory EMC statement, that the device complies with FCC et al. Throw this in the recycle bin amid all the other packaging materials. 

If that energy is picked up by the radio, how much of that energy is capacitively coupled to the ground plane at the source, where the energy is lot more intense. 

Some may argue that RF frequencies don't matter for audio bands, however, audio relies on very small voltages to extract minute details of recordings, and if coupled RF/common mode noise/differential mode noise is added to signals, then RF emissions do matter. SMPS are a classic noise generator, they have no place near audio systems. Sure, some SMPS are just as quiet as a linear (like one in a million designs), the piece of paper shipped with it says so, hahahahaha! Pull the other one.

 

Another classic example is to use a laptop -> (any) USB cable -> DAC and listen. Add in a (ISO)Regen/Intona/Micro 3.0/sOTMxxx to the chain and listen again. There's a reason why USB fixers still sell. Ethernet is not off the hook either.

The Jim Brown site is quite a good reference, it will take some time to digest. Hams are very keen to remove RF as it impacts on transmissions, however 'small' these usually IT devices are, they create a lot of havoc. I doubt an SMPS is in the same radio shack as the rest of the gear they use.

Well yes, if you are talking about noise generated inside  each

[Spacehound]

4 hours ago, One and a half said:

The noise picked up by the radio is energy that's escaped the Faraday cages, so fat lot of good that does, and the 'best of intentions' as far as EMC Class B emissions for conducted noise. The reason for the 'best of intentions' statement, for testing EMC from equipment statements is dubious to non existent in the real world. Ever noticed the first part of an IT device packaging that has printed material is not the user manual, but the obligatory EMC statement, that the device complies with FCC et al. Throw this in the recycle bin amid all the other packaging materials. 

If that energy is picked up by the radio, how much of that energy is capacitively coupled to the ground plane at the source, where the energy is lot more intense. 

Some may argue that RF frequencies don't matter for audio bands, however, audio relies on very small voltages to extract minute details of recordings, and if coupled RF/common mode noise/differential mode noise is added to signals, then RF emissions do matter. SMPS are a classic noise generator, they have no place near audio systems. Sure, some SMPS are just as quiet as a linear (like one in a million designs), the piece of paper shipped with it says so, hahahahaha! Pull the other one.

 

Another classic example is to use a laptop -> (any) USB cable -> DAC and listen. Add in a (ISO)Regen/Intona/Micro 3.0/sOTMxxx to the chain and listen again. There's a reason why USB fixers still sell. Ethernet is not off the hook either.

The Jim Brown site is quite a good reference, it will take some time to digest. Hams are very keen to remove RF as it impacts on transmissions, however 'small' these usually IT devices are, they create a lot of havoc. I doubt an SMPS is in the same radio shack as the rest of the gear they use.

Sorry, my post above  screwed up and I can no longer edit it.

 

Well yes, if you are talking about noise generated inside  each box then a Faraday cage won't stop it interfering with itself.

And as has been pointed out, RFI is so far outside the audio band, and the usual frequency response of amplifiers etc. that it simply won't have any effect.

 

Regarding 'minute details' the only 'minute detail' involved is the output from a moving coil cartridge, which I have already covered.

 

As we are mostly talking about computer audio, a sample of a 'minute detail' is just as large voltagewise  as a sample of an orchestra at full throttle  with the 1812 cannons going off  at the same time.

 

A personal example:  I use a totally 'untweaked' Lenovo tower PC with an i7 processor, three spinning disks, and a fancy graphics card, so it is quite high on the 'RFI scale'. It's in the next room and the cables go though a hole in the wall so  it's inaudible. There are no  little 'improvers' anywhere and it's connected by USB  and Ethernet  cables from the local supermarket as my thirty years plus in the computer industry tells me that 'special' cables for digital working are a complete waste of money.

I use  big Tannoy dual concentric speakers so if take the grille off and  put an ear in one  I can hear all frequencies.

And there's a permanent WiFi signal  from the internet router that covers the whole house and most of the garden, maybe all of it.  And a couple from the neighbors too.

The 'worst case' for a test is JRiver playing 'silence' (which produces the same bit stream voltagewise  as a full orchestra or whatever at full throttle) and the amp at full volume. I hear nothing at all  from the speaker my ear is in  (and my hearing is  tested regularly for 'professional' reasons.)

[asdf1000]

5 hours ago, One and a half said:

The noise picked up by the radio is energy that's escaped the Faraday cages, so fat lot of good that does

 

I was told by someone much smarter than me, that just seeing a metal box doesn't automatically mean the housing approximates a Faraday cage... It's obviously a good start, of course.

 

"Devices with metal enclosure do not approximate faraday cages–remember, bluetooth, wifi, and air are passing in and out–at least. Computers are designed to be permeable."

 

[Ralf11]

Has anyone ever done listening tests by injecting noise into a device at different levels?

[Spacehound]

53 minutes ago, Em2016 said:

 

I was told by someone much smarter than me, that just seeing a metal box doesn't automatically mean the housing approximates a Faraday cage... It's obviously a good start, of course.

 

"Devices with metal enclosure do not approximate faraday cages–remember, bluetooth, wifi, and air are passing in and out–at least. Computers are designed to be permeable."

 

There's nothing special about a Faraday cage, it's just a grounded metal screen in the shape of a box so it works against interference from any direction. Holes don't matter as long as they are smaller across  than the shortest wavelength you want to exclude. Which is why chicken wire works fine in  most instances.

 

I did a lot of 'space' stuff. Some of our cages were big enough for what we were testing, test instruments, and several of us. Obviously we had to be careful about what instruments we brought into the cage. Some we left outside and just fed the probe leads in.

 

Computer aren't usually permeable.

Except for the fan inlets and the air exits. And as the fan sucks stuff in it usually has a fine grille over it often made of metal,  and the outlets are usually small slots. In most laptops the inside of the case (if made of plastic) including the part the keys poke through is sprayed with paint containing copper to make a shield..

[One and a half]

3 hours ago, Spacehound said:

Sorry, my post above  screwed up and I can no longer edit it.

 

Well yes, if you are talking about noise generated inside  each box then a Faraday cage won't stop it interfering with itself.

And as has been pointed out, RFI is so far outside the audio band, and the usual frequency response of amplifiers etc. that it simply won't have any effect.

 

Regarding 'minute details' the only 'minute detail' involved is the output from a moving coil cartridge, which I have already covered.

 

Here's a table taken from a you tube video from Hans Beekhuyzen "Audio Hygiene 3 Interconnects".

 

This table, very simply stated is a dynamic range table and the bits required to create that range. We only have 2V nominal to work with as a line out, so here's the able to the bare minimum of 20bits, of course most hires audio files are 24bit PCM.

 

 

Fine details are specs of 'dust', and the audible ones, often discussed here at which level perception starts is another issue. At -72db these are mV and -96db in the microvolt ranges. For a DAC to produce these levels, it would infer that these values are clear of noise to be reproduced by the DAC correctly. When RF is free to propagate at will, what guarantee do you have that these voltages are out of the audio band and won't cause playback issues in a DAC?  A DAC of course is where D meets A.

 

[esldude]

2 hours ago, Ralf11 said:

Has anyone ever done listening tests by injecting noise into a device at different levels?

I have.  I posted files of music with noise added at different levels asking people to listen at their normal level and say when it disappeared.  Almost everyone heard it until -70 to -75 db and couldn't hear it below that. 

[Don Hills]

2 hours ago, Spacehound said:

...

Computer aren't usually permeable.

Except for the fan inlets and the air exits. And as the fan sucks stuff in it usually has a fine grille over it often made of metal,  and the outlets are usually small slots. ...

 

Computers usually used in domestic situations are exceptionally "permeable". 

The biggest problem is the aforementioned "slots". Slots are good aerials / antennas, acting as dipole radiators. A slot will form anywhere there is not continuous metallic contact between panels. (*)

Side access panels make poor contact along most of their edges, forming slots.

Grilles are usually only bonded to the cabinet at the corners, forming slots along the edges. The metal covers on the back panel covering the card slots usually only make good contact with the case at the ends. And ventilation slots are, well, slots.

 

(*) Old timers may remember when equipment such as computers used to have metal "finger" springs along the edge of the panels, which formed a continuous connection along the joins when the panels were closed. 

[asdf1000]

I guess a very related discussion (in parallel) is the effects of RFI when listening to music.

 

Just one data point but from one of the better DAC designers (Rob Watts):

 

"the warmer, smoother or softer sounding is the more transparent, as the mechanism for changing the sound is RF noise creating noise floor modulation - and more noise modulation always sounds brighter. Moreover, it's very easy to confuse a bright sound with more transparency."

 

Some other learnings from Rob on RFI:

 

1. battery mode operation should mean no ground loops, so no current flow into the ground planes, then no RF noise pick-up in the DAC and so no problem… This has been validated by listening tests.

 

2. As to external RF, it only matters if there is a ground loop, so battery operation should mean no currents flowing into the DAC ground plane…

 

3. We need current through the ground plane to set up voltages - and it is these voltages that the analogue electronics pick up. So no current, no voltage on the ground plane, no pick-up… The common mode noise, won’t affect the analogue electronics at all.

 

So I guess these are some of the extra benefits of blocking ground loops and leakage current loops, via the DAC anyway.

 

There's plenty of his thoughts on RFI all over the Head-Fi forums:

 

https://www.head-fi.org/threads/chord-electronics-dave.766517/page-629#post-13889772

 

https://www.head-fi.org/threads/chord-electronics-hugo-2-the-official-thread.831345/page-663#post-13961394

 

https://www.head-fi.org/threads/watts-up.800264/page-38#post-13968918

 

 

[Spacehound]

28 minutes ago, One and a half said:

 

Here's a table taken from a you tube video from Hans Beekhuyzen "Audio Hygiene 3 Interconnects".

 

This table, very simply stated is a dynamic range table and the bits required to create that range. We only have 2V nominal to work with as a line out, so here's the able to the bare minimum of 20bits, of course most hires audio files are 24bit PCM.

 

 

Fine details are specs of 'dust', and the audible ones, often discussed here at which level perception starts is another issue. At -72db these are mV and -96db in the microvolt ranges. For a DAC to produce these levels, it would infer that these values are clear of noise to be reproduced by the DAC correctly. When RF is free to propagate at will, what guarantee do you have that these voltages are out of the audio band and won't cause playback issues in a DAC?  A DAC of course is where D meets A.

 

That chart is interesting (and it explains something I never fully understood - the  bit position/decimal value's relationship to the DAC output voltage).

 

And incidentally, very few dacs do better than 20 bits resolution, no matter what they claim. It's certainly rare with dacs using an IC DAC chip.

 

But it comes down to that the dac.   We need good RF isolation from the 'outside world' and at a minimum the internal isolation of the USB receiver from the rest of the dac.  With a 'chip' dac it's near impossible to well-isolate  the digital parts from the 'conversion' as it's all on the same die.

 

None of this matters inside the computer. They HAVE to be designed to work fine in their self-created internal high RF environment  or they wouldn't work at all. And that's quite easy as neither internal 'bit' voltages nor timing  have to be particularly accurate, by design. Voltages aren't critical and the bit 0 or 1 value  is detected where the middle of the 'pulse' is expected to occur, not at the edges. The detector and the pulse chain both use the same clock and  there is a 'fake' reversal every so often to make sure the detector  has not moved  too far away from the middle of  the pulses.

 

I'm not sure I believe in the effectiveness of what you are doing but in an hour or so  I'm off to the local big supermarket to see if they have a very cheap radio with AM on  it