Cable Modems

[Xymox]

Have you guys experimented with sectioning off your network into a clean and dirty side? It solves a lot of the problems mentioned here...
I wrote a little piece about it yesterday in this thread... https://www.whatsbestforum.com/threads/ethernet-tweak.36727/
Maybe it helps.

Cheers, Hans.

Thats what i make now.. And you can use any access point. I am leaning towards supplying a WiFi 5 based Aruba with a special linear and configured to very low power. WiFi 6 introduces a lot of jitter because of the way it works. I am talking about measurable jitter from a IT standpoint, wifi is just nasty for a connection of course, never run your audio packets thru it. The Aruba has very fine grained control of the RF ( radios ) and so I can set this very specifically. But best to use one a few gens back and also load a specific Aruba OS. The shielding on the Arubas is also really good.

I have also done modded cable modems ( pure modems like the Arris S33 ) with a sensing noise cancellation linear.

The switch, that was at Axpona, has been shipping since Axpona. So far everyone likes it and sells off their old switch. SO FAR. Of course it might not be for every one..

The Switch X

 

[vert]

@InfigoAudio If the A “dirty” side is powered by a low cost SMPS, and the A media converter is connected to the router, isn’t it going to pollute the entire system?

Anytime I have a network device powered by a low cost SMPS connected to my router, I get the glassy sound and splashy treble.

 

[MarcelNL]

unless that SMPS is connected to a different group in the house it'll for sure pollute other parts of the system IME.

 

[vert]

Thats what i make now.. And you can use any access point. I am leaning towards supplying a WiFi 5 based Aruba with a special linear and configured to very low power. WiFi 6 introduces a lot of jitter because of the way it works. I am talking about measurable jitter from a IT standpoint, wifi is just nasty for a connection of course, never run your audio packets thru it. The Aruba has very fine grained control of the RF ( radios ) and so I can set this very specifically.

This is very interesting and a meaningful discovery. The low power wifi 5 access point sounds better to me than my previous higher power wifi 6.

 

[vert]

The switch, that was at Axpona, has been shipping since Axpona. So far everyone likes it and sells off their old switch. SO FAR. Of course it might not be for every one..

If the clean island side has an access point just for control of music playback, how does the house get wifi from the dirty side?

Would another access point be used, connected to the dirty side?

 

[Xymox]

If the clean island side has an access point just for control of music playback, how does the house get wifi from the dirty side?

Would another access point be used, connected to the dirty side?

This assumes you already have a wifi system and network deployed for the house. This creates a isolated network.

 

[Xymox]

unless that SMPS is connected to a different group in the house it'll for sure pollute other parts of the system IME.

SMPSs are a horrible RF transmitter. Nasty PWM with zillions of harnonics and these vary second by second with the load on it as the PWM varies to regulate the voltage. The RF from these types of supplies will travel on anything metal and pollute other things and then will travel to the things these are connected to until you can measure it with a spectrum analyzer on any device that is connected with metal like wires and power cords. Or even feet thru a stack. You can measure RF from a SMPS someplace in the system on the cases of the gear.

It also radiates thru the air. With the wire from it, of from a device with one inside it, acting as a transmission antenna. Then any mental object around it like wires acting as a rcvr antenna. AND AGAIN you have wide spectrum nasty PWM devived RF on things.

Ethernet cables themselves also have a RF spectrum of noise they radiate. But the spectrum is more like a white noise and has no real harmonics.

Here is the scary part. Inside all your devices that are digital or have CPUs are "buck boost regulators" These use the same principle at SMPS to regulate internal voltages. Most CPU based devices have AT LEAST 3 and usally 5 of them. They drop the 12V down to things like 5v, 3.3v, 1v for use by the chips. These things are just as bad as SMPS. BUT.. They are right on the board. So they spray RF interference with zillions of harmonics all over all the circuits around them. ALSO they are switch mode with a specific base frequency and this interacts with the chips as they need juice. They are in everything these days. They are also not so good at regulation and have nasty spectum noise. These are one of the key things I address in all my gear. I use high end caps 100's of times bigger then called for by these devices and use caps made to supress RF up to Ghz ranges. This produces VAST reductions in radiated RF and leaves the surrounding circuits super clean. This has a vast affect on reducing jitter on everything as I have nearly eliminated a very nasty amount of noise, so phase noise on clocks is much cleaner and buses between chips is far better.

This is what the output of a std buck boost regulator looks like feeding a chip. With a after mods below. The scales and setup is a bit different, its just give you a basic idea.



After



If you zoom in on one of those nasty peaks you see its composed of just loads of other harmonics.. SMPS and SMRegulators are just nasty..



AND ALL THAT CR*P gets onto everything. Internally from the regulators and externally from a SMPS.

So one of the big things I spend a long time tuning in these mods is to KILL off the RF spray and transmission..

I was sad to see that the Taiko little box they are shipping comes with a SMPS. I suppose they know everyone will use a linear tho.

 

[Xymox]

All that said.. I am unsure if 2.4/5ghz from wifi really affects audio gear in a way that makes a audible difference ? Those freq are so high they dont really cause things to jitter electronically. For SURE there is a issue with stuff below say 50Mhz, but 2.4Ghz ? I donno. I think the RF spray from the SW Regulators inside the access points radiating is what is making for issues, not so much the 2.4/5/6 Ghz. BUT it IS a kinda higher power RF signal. But the spectrum of these is really well controlled.

As a ham radio guy tho, keying up 1000W on Mhz bands is so rank I can scramble audio gear with a CPU in less then 1 Sec and require a power cycle to get it back. The 40M Ham band produces havoc on most any electronics up close. SO, ZERO douby that audio gear can pick up RF and act badly.

 

[Kingrex]

Chris, very interesting talk on noise.
Your using caps on the board to filter smps noise. What is your understanding of Ground Boxes? Are they able to "Damp" this type of noise and pull it out of a system? Or are they doing something else?

Is this noise a vibration, or something else.

Why about an earth ground. If you have a very short, strait, unbroken connection of equipment neutral/ground to earth, will some noise be damped?

 

[MarcelNL]

it's one thing to know that caps (the right sort and amount) can combat noise due to the voltage regulators on a MB, without a way to adapt a MB it is a moot point IMO...

One way of dealing with that noise is not using metal cases, and using EMI damping material.

 

[Kingrex]

My Mojo Audio DAC was just rebuilt. Better Chokes, some vibration control and EMI paper. Let me also say my First Sound preamp was also upgraded. Yesterday an audiophile friend was over and shocked how good my whole system was. He particularly liked my vinyl. I sort of wonder if the DAC and the EMI damping helped rid noise overall that may have been creeping into the vinyl.

 

[MarcelNL]

All that said.. I am unsure if 2.4/5ghz from wifi really affects audio gear in a way that makes a audible difference ? Those freq are so high they dont really cause things to jitter electronically. For SURE there is a issue with stuff below say 50Mhz, but 2.4Ghz ? I donno. I think the RF spray from the SW Regulators inside the access points radiating is what is making for issues, not so much the 2.4/5/6 Ghz. BUT it IS a kinda higher power RF signal. But the spectrum of these is really well controlled.

As a ham radio guy tho, keying up 1000W on Mhz bands is so rank I can scramble audio gear with a CPU in less then 1 Sec and require a power cycle to get it back. The 40M Ham band produces havoc on most any electronics up close. SO, ZERO douby that audio gear can pick up RF and act badly.

I've been playing with shielding everything else around the AP and I do not find the WIFI radiation to effect sound a great deal, I concluded it's the stuff going on on the MB that has most effect, hence using a remote AP and shutting off internal WIFI in the router (and server) really helps

 

[Xymox]

it's one thing to know that caps (the right sort and amount) can combat noise due to the voltage regulators on a MB, without a way to adapt a MB it is a moot point IMO...

One way of dealing with that noise is not using metal cases, and using EMI damping material.

Yep... Warning, tech talk below..

The best way is to not use PWM based regulators in design and most designs for almost everything these days use the "reference design" from the chip companies, and these are all PWM regulation. Sadly. The issue with using linear regulators is usually heat. Heat tho can be minimized by doing real analog power supply design were the voltage before regulation is close to the desired output voltage so the voltage drop across the regulator is minimized and so is heat. Most designers today have completely forgotten how to do proper linear power supply design because the PWM regulators are so easy and efficient they can take in 12V and put out 1v with not much heat. Plus they don't think that a few PWM regulators just for the CPU would affect the analog sections. But they do.

The REAL way to do all this goes back to a better era of linear power supply design. Lambda made some killer linear supplies that did not even use a chip of any kind. Used all discrete components. Mil spec. I restore and recap these as these were great. Really fast regulation, great wide band low output impedance, low noise. Great impusle response. Well, I admit I also do some changes to them as well

The no holes barred way to make a power supply for current gear would be expensive and no one does this for any digital device in audio I know of. It would be to use a multi-output custom transformer. One output for each needed supply. Rectify this using schottky diodes. Tailor the transformer secondary windings so the resulting voltage, under load, is just above the deseried regulated voltage. Use all discrete components, transistors, ultra high precision zener or some lab grade voltage source and then use a serious output transistor. Lots of WIMA and the right caps. USE SENSING right at the chip input voltage pin to produce lab grade PPM regulation and exceptionally low noise. Each regulated output would need serious testing to be sure it was ultra low impeadance across all the freq the chip would pull and these CPU based chips produce all sorts of bursty low freq current draw. So you would need 5+ supplies.. This is all perfectly normal for gear prior to like 1985 when PWM came along and made regulation cheap and easy. The above is hard and requires a lot of actual work, not just copy past a reference design from a chip company data sheet. This also has the big advantage of each supply being isolated from the others. Like having a seperate linear for every supply needed inside the device.. BUT the device becomes bulky and tricky because the regulation needs to be very close to the chip to work well. I see this kind of design all the time in older test equipment. Tektronix made scopes, full analog, using these kinds of proper power supply design.

BUT YES,, if you have the right tools to look at the power supply rails to the chip and you have the right parts you can knock a PWM power supply down to reasonable amounts of noise and get good regulation. Yep, noise containment using various stuff like shielding and 3M products can help isolate the noise. BUT once a PWM regulator is on the board, you can't ever get rid of the noise completely unless you do like 1/2" thick sandwich solid metal blocks with engineered cavities for each section. Like a RF Deck in a high end spectrum analyzer where you need 130db of SN..

OK all that is pretty techy.. Lemme answer one of the other questions and make it easier to understand..

 

[MarcelNL]

thanks, yet before you continue allow me to clarify I was aiming at a real world motherboard, there is little I can do to fix how it's built...I'd LOVE to be able to custom order a MB but I dread the choices and price penalty..

 

[MarcelNL]

Also, for a CPU burst power is all important IMO...and if there is one thing these PWM convertors can do well it's that...
18 of them each capable of 90A...femtoseconds likely but that is what we're looking at @4Ghz or so.

 

[Xymox]

Chris, very interesting talk on noise.
Your using caps on the board to filter smps noise. What is your understanding of Ground Boxes? Are they able to "Damp" this type of noise and pull it out of a system? Or are they doing something else?

Is this noise a vibration, or something else.

Why about an earth ground. If you have a very short, strait, unbroken connection of equipment neutral/ground to earth, will some noise be damped?

RF noise is radio frequency. Its electromagnetic. Think of a bar magnet. That magnetic field that moves a compass is the force involved. BUT. Its flipping the N and S back and forth thousands ( khz ), millions ( Mhz ), or even billions ( Ghz ) of times a second. Its a very fast spinning magnet

This very fast magnetic field alternating back and forth millions of times a second can transmit radio, TV, satillite, cable tv, blutooth, wifi, and NOISE.. A wall wart SMPS generates these fields with TONS of **** and this gets transmitted thru the air and also travels down wires and even on cases.. RF in the air becomes really hard to block in audio gear because the wires pick it up acting like antennas and power supply cords and cases all also acting to pass it along.

This spinning bar magnet idea also means the faster its spinning the shorter the "wavelength". The higher the frequency, the shorter its wavelenght is. If you could stop time you would see the lower frequencies are longer in the air or on the case. The high frequencies are very short.



SO.. A grounding block, and grounding systems, very how effective they are by the frequency. The lower frequencies are long and get to the blocks and grounding system and can be suppressed. The higher freqs tho, they are so short by the time they get to the block they are less suppressed. Also there is resonance. It turns out that a wave has a specific speed in a wire or on a piece of metal. To be a effective "transmission line" your wire should be a multiple of the frequency. SO a grounding system wont always be perfect for all frequencies. BUT you can get close and these can be effective for sure on the lower freq stuff which affects audio the most. For example, a grounding system will have no affect on 2.4 Ghz but be near total absorption at 20Khz.

Earth ground. Hmmmmmmmm.... Hahaha.. Well.. Alignment with cosmic forces might be good I am always serious about insanity levels of real earth grounding. I did a system that has 14 40' deep chemical ground rods in a star pattern over a huge lot just for them. Insanity low impedance.. Why ? hmmmm... Technically I am unsure.. BUT why not. It does have a serious affect on ELF/VLF frequencies and so the room sits in a area that is very RF dead right into the super low freq band of RF. I donno, it aligns the audio with cosmic forces, hahaha.. I never hooked and unhooked it from the Torus system to see if I could hear it, I should..

 

[Xymox]

thanks, yet before you continue allow me to clarify I was aiming at a real world motherboard, there is little I can do to fix how it's built...I'd LOVE to be able to custom order a MB but I dread the choices and price penalty..

I am thinking about modding a Intel NUC. These are pretty small and I think I could supress the bad stuff far enough to make it worth while. These NUCs are kinda cute and can be used with Linux or windows in many applications. I like the small PCB because all the traces are short and so after all the mods the jitter tends to be lower then a large PCB.

 

[Xymox]

Also, for a CPU burst power is all important IMO...and if there is one thing these PWM convertors can do well it's that...
18 of them each capable of 90A...femtoseconds likely but that is what we're looking at @4Ghz or so.

WELLLLL......

A linear supply is way faster. Its only a few transistors. A PWM regulator is well known to have much slower responses. It has to do math and use op-amps and all sorts of stuff inside that chip. So there is latency. A linear supply done correctly is way faster. BUT way harder to do.

There is another subtle issue. The clock freq of the PWM regulator. Obviously it cant slew to a corrected voltage any faster then its clocking. So this gives a PWM supply a upper limit on speed. It gets REALLY complex if you measure what is goin on current wise. The CPU can only draw current during the posistive clock of the PWM regulation. This is smoothed out by a cap, but this then introduces all sorts of weirdness and noise because the CPU is not being feed everything it needs immd when it needs it. SURE it works OK, but, noise develops AND the CPU jitters some. You end up with noise on things like the busses, this noise ends up as phase noise and jitter on the data busses.

A full linear supply WOULD be better, but, REALLY hard to do. 1v @ 90A is really gnarly and nearly a art form doing it with a linear. The switch of current is also tricky. 1A-90A in ms.. Tricky for sure. Its why they just use PWM, its good enough for most uses and EASY to copy paste from a datasheet.

Taiko uses a linear as the main supply. You can bet they would love to extend this all the way to the chips, but, OMG, thats intense and really engineers today just don't do linear design. Its kinda a lost art form. Its a alchemy that has been lost in modern designs.

 

[Xymox]

One thing about grounding.. Higher frequencies can TRAVEL on the grounding system So the grounding system can spread higher freq RF to other devices..

SO grounding is tricky..

Best to make sure the RF is minized before it begins. NO SMPS, hahaha.. Wall Warts = bad...

 

[vert]

@Xymox You are absolutely right about wifi 6 access points having more jitter.

I had to remove the non wifi 6 TP link access point because it was dropping connections.

I put the wifi 6 access point back in and it sounds terrible, even though it’s only used to control music playback. My streamer is not connected to it.

I’m going to try Aruba access point next.