A novel way to massively improve the SQ of computer audio streaming

[JohnSwenson]

Sounds like you have the recipe for the perfect switch, Alex. I wish I knew some of this before I sent out this cheap Trend Net switch for clock modification. It indeed has this "GREENnet" power saving feature and I was suspicious of it but went with it anyway since I already owned this switch and had it lying around. I'm always looking forward to anything you and John come up with. I hope your audiophile switch becomes a reality sooner than later.

 

Good to know about the Broadcoms. Both my LAN ports (Mac Mini + Thunderbolt LAN) are Broadcoms. Definitely no complaints.

 

One interesting thing to note is that old switches tend to sound better than new ones. Modern switch chips tend to turn off most of the circuitry in between packets, this produces large transients on the power/ground planes. The old chips were designed in an era where going for low power was not the highest priority in the chip design, thus they usually do not generate nearly as much PG noise as newer designs.

 

I'm running my microRendu from an old 100Mb switch that is at least 10 years old and has a linear power supply, with this driving the microRendu I don't seem to have much of the issues all of you are having with different cables etc.

 

(Uh Oh, I think I just started a buying frenzy for old switches on ebay)

 

John S.

[JohnSwenson]

6 hours ago, austinpop said:

 On this point of clock quality - while I've found @romaz's experiments fascinating and educational, one aspect really puzzles me. Perhaps someone with a deep technical knowledge of digital audio clocking can educate me/us?

 

The question is this - unlike S/PDIF and AES/EBU that are synchronous, most DACs these days implement an asynchronous USB interface. I guess the pedants would say isochronous USB using the asynchronous transfer mode. Whatever.

 

So for synchronous interfaces, I can certainly see how the quality of the clocks upstream of the DAC come into play.

 

But what about the async USB case? Here, the clocking is controlled by the DAC, isn't it? So how and why does the quality or accuracy of the clocks in the upstream components come into play?

 

I wouldn't be surprised if @JohnSwenson has a treatise on this somewhere on CA, but apologies, I couldn't find one.

 

I have covered this fairly extensively in my series at Audiostream.

http://www.audiostream.com/content/qa-john-swenson-part-1-what-digital#35WEoHuMXmJ00tIX.97

http://www.audiostream.com/content/qa-john-swenson-part-2-are-bits-just-bits#Fz4mY1iS5fqvIUkV.97

http://www.audiostream.com/content/qa-john-swenson-part-3-how-bit-perfect-software-can-affect-sound#6yFiDB63CIzx5f3V.97

 

John S.

[JohnSwenson]

7 hours ago, austinpop said:

 

 

 

Having personally experienced the improvement wrought by this Ethernet regeneration, the next question is, indeed, where do manufacturers go from here?

 

@JohnSwenson, who is consistently honest and candid in these forums, has stated that this is an area he is interested in, and the science of what we're hearing is not well understood. I hope I paraphrased his views correctly.

 

Even if the science behind these improvements on the Ethernet is not yet baked, several questions arise going forward:

1. what makes the SOtM sCLK-EX so special, as to surpass previous reclocked TCXO/OCXO switch offerings like the Pang, the AQvox, etc, that Roy had heard?
2. how will SOtM modularize their offering? Right now, they have a single clock board - sCLK-EX - with 4 taps that they distribute both internally and externally. This approach is inherently limited by the length of the clock wire. What would make sense is to figure out a way to embed a fit-for-purpose sCLK-EX in each component. The current Ultra approach is wasteful and expensive - i.e. embed a full-size sCLK-EX board in each Ultra component.
3. how will other manufacturers respond? And will we see new designs that compete with, and even exceed the SOtM solution, at equal or lower prices? As consumers, I sure hope so!

As I often like to say - it's an exciting time to be a computer audiophile!

I have not measured any of these clocks so I can't comment on their relative performance. But I can give some general observations.

 

First a clock in a real system can be very different than a clock being measured by itself. The phase noise of ALL clocks will change depending on noise on the power and ground network feeding the clock. So what is happening with other circuitry around a clock can significantly degrade its output from what you see in a spec sheet.

 

The lowest phase noise clocks are OCXOs, but not all OCXOs have very low phase noise. There are quite a few inexpensive OCXOs (less than say $150) that have worse phase noise than a 575.

 

The result is that just because two devices have OCXOs, does not mean they have identical jitter performance.

 

John S.

[JohnSwenson]

7 minutes ago, Daudio said:

 

Does "OCXO" stand for something we might learn from, or at least of interest ?

 

 

OCXO stands for Oven Controlled Crystal Oscillator.

 

The hierarchy of crystal Oscillators:

 

XO  Crystal Oscilator

TCXO Temperature Compensated Crystal Oscilator

OCXO Oven Controlled Crystal Oscillator

 

What all this has to do with is how the frequency of an oscillator behaves as the temperature of the crystal varies. All crystals will change their frequency due to temperature changes. If you map these changes you get a curve that at low temperatures has a large change per small temperature delta, and at high temperatures a large change per small temperature delta, but at some intermediate temperature the "temperature coefficient" (TEMPCO) is zero.

 

If the operating temperature inside the equipment is not at this point, changes in temperature will change the frequency. The TCXO and OCXO are ways to make this better than a normal "uncompensated" crystal.

 

Note that these frequency changes are long term changes, sometimes referred to as drift. They are usually NOT the short term variations we call jitter. Thus optimizing for TEMPCO is not necessarily a good thing for audio.

 

The TCXO is an electronic circuit added to the oscillator. It turns out that you can change the frequency of an XO by changing the capacitance across it. The TCXO contains some form of temperature sensor that feeds a device that changes the capacitance across the crystal, the change in capacitance cancels out the change in temperature thus significantly lowering the TEMPCO, . But in doing so it increases the jitter, that temperature sensor and cpacitance changing circuit generates noise. This increases the jitter of the crystal. Thus TCXOs are usually not a good thing in digital audio, they are optimizing the wrong thing.

 

The operation of the OCXO is usually described as maintaining the temperature of the crystal so that the frequency doesn't depend on what is happening outside, the oven keeps the temperature stable. But this is not the whole story. The REAL purpose of the OCXO is to maintain the crystal at that temperature where the TEMPCO is zero, not just any old temperature. If the crystal is at the zero TEMPCO point then the temperature stability doesn't matter that much.

 

There is a big problem with this. The normal way a chunk of quartz is cut is called the AT cut, this has a zero TEMPCO temperature right around room temperature. This is fortuitous for non compensated crystals, but a real pain in the neck for OCXOs. If the crystal temperature is too high the system has to COOL the crystal, if too low heat it. This is quite complex and very expensive to do well.

 

The solution is to use a crystal cut called the SC cut. This has a zero TEMPCO point much higher than room temperature, so all you need is a heater, which is really easy to do (current through a resistor). This is why it is called an oven, it is always heating.

 

A side effect of the SC cut is that IF it is done right you get lower phase noise. But the inexpensive OCXOs put all the money into making the oven and skimp on the crystal quality, so they don't get particularly great phase noise. The ones that DO put money into the crystal can have the lowest phase noise of any oscillator, period.

 

So why doesn't anybody use an SC cut crystal without the oven? At room temperature the TEMPCO is SO far from the zero point that it would be almost useless.

 

I hope this gives a little insight into the world of crystal oscillators.

 

John S.

 

 

 

 

[JohnSwenson]

36 minutes ago, austinpop said:

 

 

Thanks for these insights!

 

I cannot tell from the sCLK-EX Press release or its product page any mention of TCXO or OCXO, so one has to presume the claims come from the overall board design, not the actual crystal oscillator.

 

@JohnSwenson - where does the oscillator used in the ISO-Regen (the Crystek 575?) fit into the XO/TCXO/OCXO taxonomy?

The 575 is an XO, but one of the best ones out there.

 

John S.

[JohnSwenson]

13 minutes ago, rickca said:

John, where does VCXO fit in the hierarchy you described?

 

The specs of my Jeff Rowland Aeris say

20 bit dynamic range Voltage Controlled Crystal Oscillators (VCXO)
44.1 kHz and 48 kHz    < 1 picosecond RMS jitter

A VCXO is an XO with a circuit that varies the frequency from an externally applied voltage. Again noise on that voltage and the noise of the extra circuitry increase the jitter.

 

As mentioned before a single jitter number is almost useless. The number is an integration of the area under the phase noise curve. When doing the integration you have to choose the frequency range to use for the integration, for exactly the same oscillator changing the range can vary the "jitter number" by a factor 100!

 

Thus specifying a number without the lower and upper frequency limits is pretty much useless. You might be able to compare oscillators if the limits are exactly the same, but even then a true phase noise plot is much better.

 

The manufacturers love to choose the points to get the lowest number of course. If an oscillator really is good the manufacturer will usually specify the range to show it off. If not specified that is usually good indication that the device may not be the best and the manufacturer is playing some "specmenship" games.

 

John S.

[JohnSwenson]

2 hours ago, austinpop said:

 

Sadek,

 

I'm not Paul, but I can give a layman's explanation.

 

A regulator is a circuit that accepts a range of input voltages and outputs a fixed voltage. So the higher the difference between the input and the output voltage, the higher the voltage across the regulator. Linear regulators achieve this "regulation" by dissipating heat.

 

The amount of heat the regulator has to dissipate is related to:

1. The current flowing through the regulator, let's call it I
2. The voltage difference between input and output, let's call it (Vin – Vout)

By simple Ohm's Law, the power dissipated across the regulator is then (Vin – Vout) x I.

 

Note that I here is the current that is drawn by the connected device at the Vout voltage. This is what Paul is referring to here. If you require a low value of Vout (in your case, 5V) AND a high current draw, this will cause a lot of heat dissipation, and may exceed the rating of the regulator(s).

 

To answer your actual question: No, it will not degrade the quality of the power supply in any way. All it means is that the 5V output will need to be rated to deliver a lower current - whatever Paul deems is achievable with the regulators he uses.

 

I'm sure others will correct me if I got anything wrong here. 

What you say is correct, just a little bit of more verbiage around it.

 

A linear supply has a "raw supply" which is the transformer into diodes then the capacitor bank, this voltage is usually fixed. Thus as you change the output voltage of the regulator you change the voltage across the regulator. This voltage times the current flowing through the output is the heat the regulator has to dissipate. As the heat the heatsink has to dissipate increases the temperature of the heatsink and the regulator itself increases, at some point the regulator stops working. (ie "burns up" or "fries"). Many regulators have a "thermal shutdown" circuit that shuts the regulator down when the temperature gets too high to prevent actual damage to the regulator, but at that point the heat sink is going to be VERY hot, probably hot enough to boil water, that is normally not considered good for the longevity of the circuit.

 

The PS under discussion seems to have been designed for 9v - 12V, the heatsink was designed to keep the temperature reasonable with the rated current. If you go to 5V output and the rated current, the heatsink is going to have to dissipate almost twice as much heat, which will cause things to get too hot. The solution is to only run it at no more than half the rated current.

 

In reality things get a little more complicated, the voltage of the raw supply usually decreases somewhat as the load current increases. This means the unloaded voltage has to be higher than you would normally think. This has a weird consequence, as the output current goes up, the raw voltage goes down so the voltage across the regulator decreases which can compensate for the increased current. The result is that a large increase in current can result in a fairly small increase in the heatsink temperature.

 

Designing the parameters of a linear supply that will work over a wide output voltage range is not an easy task, there are a lot of things to trade off, the higher voltages tend to dissipate very little, the droop of the raw voltage is the limiting factor. On the low voltage end heatsink temperature is the limiting factor (actually regulator temperature). in-between voltages are the sweet spot, the raw voltage can drop a lot and you can pull a LOT of current before the temperature gets too high.

 

I know, more than you ever wanted to know about linear power supply design.

 

John S.

[JohnSwenson]

17 hours ago, austinpop said:

 

You're right - I will have to try this out for myself. For now, I have been taking the LPS-1 claim of leakage current isolation at face value, unless there is a published metric I missed.

 

 

 

I have built a power supply leakage tester, again a single number is not nearly as useful as a leakage VS frequency graph. When I put the LPS-1 on the tester it shows a flat line right at the absolute floor of the tester.

 

Alex has a policy of not showing measurements of his products VS named products of other companies so I will not say how that compares to other specific supplies. I can say that nothing else I measured comes close.

 

I don't have a good way right now to calibrate these graphs to come up exactly what those leakage currents are, but I CAN say the floor of the tester is about 500,000 times lower than the maximum leakage I measured, the LPS-1 is going to be less than that.

 

John S.

[JohnSwenson]

On 8/21/2017 at 6:21 PM, austinpop said:

 

I am not a clock expert, but I think the answer is no.

 

Word clock is usually used to refer to clocks that are equal to or (sub-) multiples of the audio data rates of 44.1 or 48 KHz.

 

The sCLK-EX mods actually supply a 25MHz clock to the switch. As I understand it, this is neither a conventional word clock, nor a reference clock. At this point, I only know of SOtM providing this kind of clock improvement.

In this case the 25 MHz IS a reference clock. To my mind a reference clock is one that is used as the reference to a PLL, frequently a PLL used in a frequency synthesizer of some sort. In the case of an Ethernet switch chip the 25MHz feeds a PLL which generates the 125MHz clock which is actually used for the Ethernet data rate. (both 100Mb and 1Gb use 125 Mega Samples per Second on the wire).

 

So Yes it is being used as a reference clock. MANY clocks in digital systems are used this way. USB chips definitely do this (you feed 24MHz in and the data rate is at 480Mb/s), most of the clocks in a motherboard are generated by PLLs from lower frequency "reference clocks".

 

A fair number of DAC chips also have PLLs that generate the internal clocks from an external reference clock.

 

The Sotm clock board has a reference oscillator and a PLL based frequency synthesizer  that can generate 4 independent frequencies. If you want to you can also use it with an external reference clock. 

 

John S.

[JohnSwenson]

Those of you getting 75 ohm BNC cables just make SURE you get a cable with a REAL 75 ohm BNC plug. A LOT of "75 ohm" BNC cables actually have 50 ohm plugs on the 75 ohm cables. I don't really know why this happens, probably the purchasing department notices that 50 ohm plugs cost a lot less than 75 ohm plugs and swaps them without telling the engineers. I have a hard time believing that designers working at cable companies don't know the difference between 50 and 75 ohm plugs.  PHYSICALLY they are interchangeable, you can plug a 50 ohm into a 75 ohm jack so you can't rely on that. A 75 ohm plug usually has much less insulation around the central pin, so you can usually spot them.

 

And this is NOT just audiophile companies, I found quite a few well known cable manufacturers that cater to pro video and audio companies use the wrong connectors as well.

 

I not going to give a list, I don't remember the names, just be sure you get the right ones.

 

John S.

[JohnSwenson]

16 minutes ago, rickca said:

My solution is just using USB3 bus power for the iGalvanic3.0 and its GI USB NIC.  Will it get better with cleaner power?  We do have the option of using iFi's bundled iDefender with either an LPS-1 or a 5V power bank.

 

My question to @JohnSwenson is this ... you said leakage currents can go right through Ethernet transceivers.  Would the iGalvanic3.0 stop this on its own, or do we need the bus powered by iDefender/LPS-1?

 

As an alternative, we could introduce a switch powered by an LPS-1 or 5V power bank between the USB3 ethernet NIC and the microRendu.  This would be especially helpful if it also had a good clock.

 

So I'm trying to extend this thread's original idea with John's comments in this post

https://www.computeraudiophile.com/forums/topic/35129-a-novel-way-to-massively-improve-the-sq-of-the-microrendu-ultrarendu/?do=findComment&comment=713210

 

You can use this same approach to galvanically isolate your music files.  Just substitute an external SATA SSD like a Samsung T5 or a bus-powered USB3 hard drive enclosure (with 2.5" drives) for the USB3 ethernet adapter.  I haven't tried this yet.  I'm a bit concerned whether sufficient current will get through the iGalvanic3.0 to satisfy these bus-powered devices.

 

Many thanks to @lmitche for coming up with these creative ideas with his Adnaco S3B solution.  I hope my adaptation hasn't totally messed them up.

Most SMPS have a fairly large amount of high frequency leakage current that is very high in frequency, at very high impedance. An Ethernet transformer has to pass high frequencies (that's what it DOES), the attenuation of lower frequencies is not a brick wall, it increases as the frequency goes down. Thus the higher frequencies of the leakage current will not be attenuated very much and the lower frequencies will be attenuated a fair amount.

 

I have no idea how this works with an iGalvanic. I don't have one to test.

 

The main takeaway is that linear supplies always have much lower high frequency leakage than SMPS, so when in doubt, use a linear supply and you will almost always be better off than trying to clean up the mess from an SMPS. The qulity of regulator etc has NOTHING to do with the leakage, so even a cheap one will have much lower leakage than an SMPS.

 

I have a Baaske on order so maybe late this week I can do some tests and see what it does.

 

John S.

[JohnSwenson]

28 minutes ago, austinpop said:

 

Hi John,

 

Just following up on your point about 50 ohm vs 75 ohm - is there a standard, or a specification that one can reference, that applies to both variants? For example, if I have understood this right, 50 ohm cable is supposed to conform to the RG58 specification, and 75 ohm  cable to RG59. Is there a similar designation for 50 vs. 75 ohm BNC plugs?

There are 50 ohm and 75 ohm cables and 50 ohm and 75 ohm plugs, just make sure you match them.

 

RG-58 is a 50 ohm cable, RG-59 is a 75 ohm cable. They do not "define the spec" for 50 or 75 ohm cables, they are just particular implementations of a 50 ohm or 75 ohm cable. There are hundreds of different 50 and 75 ohm cables (more 50 than 75), they are all the correct impedance but vary in other parameters such as diameter, flexibility, attenuation, power handling capability (for transmitters) and of course cost. If you want to overwhelmed take a look a www.pasternak.com, they have a very BROAD range of coax cables. You can get pretty much anything you could possibly ever want from them.

 

BTW if you want what is possibly the best coax for handling a clock you should try a semi rigid assembly from them. It is essential thin walled copper tubing with a center conductor, you "form" it to the path you want, it stays in that configuration until you "form" it again to a different path. And not cheap, they are in the $10 per foot range or more.

 

John S.

 

 

[JohnSwenson]

On 9/4/2017 at 8:33 AM, auricgoldfinger said:

 

 

Both the LPS-1 and sPS-500 are more transparent and detailed with better separation relative to the JS-2.  They improved the macro- and micro-dynamics of the microRendu.  With the stock sPS-500 A/C power cable, the two supplies are comparable, but the LPS-1 does seem to have a slight edge.  Given the cost differential, if I only wanted to power a single 3.3V, 5V, or 7V device, the LPS-1 is possibly the better choice.  (In my case, I was fortunate to buy a 1-month old sPS-500 for the price of a new LPS-1.)

 

The Pangea A/C power cable greatly enhances soundstage dimensionality and imparts a sense of realism that makes the LPS-1 sound lean in comparison.  Please note that this particular Pangea cable is relatively inexpensive and would never be confused with a reference power cable.

 

I am not the first person to have noticed the importance of the A/C power cable: 

 

 

 

What are you powering the LPS-1 from? Have you tried powering it from the JS-2? If you use the Mean Well you can get high frequency noise getting into your system back through the AC mains. Some systems are sensitive to this and others are not.

 

John S.

[JohnSwenson]

There are two aspects to sound quality from a network connection:

1) clocking, this includes the phase noise of the clock in the switch, but it also includes the phase noise from other network devices connected to the switch,

 

2) leakage current. This includes the leakage of the power supply feeding the switch, but it also includes the leakage from power supplies powering other network devices. This leakage current can go right through Ethernet transformers.

 

The "audiophile" switches, use a much better clock for the local switch, but don't do anything about the phase noise coming in over the wire from other equipment or the network leakage.

 

The mentioned recommendation of mine is for dealing with #2, leakage current. What I found is that it is easy to get rid of the leakage current from the local power supply ( the PS for the switch)) by grounding the negative of the PS output. This will get rid of its leakage current. But you still have to deal with the leakage coming in over the network. IF you use the FS108 or FS105 (see above for the models) AND you ground the negative of the supply on the FS108 oe FS105 you will block the leakage from both the local PS AND all the others coming into the switch.

 

Using the switch without the grounding does not improve anything. Using the grounding with other models switches gets rid of the leakage from the local PS, BUT still lets all the others through.

 

Note this does not get rid of the clocking issues, just the leakage.

 

Personally I would get one of these switches and use it just to drive the computer/streamer whatever is the last point where a network connection is made to the audio system. Ground the negative of its output and you have blocked network leakage from getting into  your system.

 

See the SMPS grounding thread in the UpTone forum for details on how to do this grounding.

 

The fun part is going to be figuring out how to deal with the network clocking issues, that is much more difficult, but I am definitely working on how to deal with that.

 

John S.

[JohnSwenson]

5 hours ago, thyname said:

 

Excellent write up!

 

For the quoted phrase above, are you saying that when connecting the endpoint / streamer (such as microRendu or sMS-200) to that switch, nothing else should go to that switch (i.e. Apple TV, Xbox, etc)?

The switches mentioned are only 1/100 switches so cannot handle gigabit Ethernet. If you need gigabit for other devices go ahead and run them through a separate gigabit switch, using the FS108 or FS105 on a connection to the audio device.

 

I did not spend a lot of time trying all kinds of different configurations, in my tests I had the FS105 etc with just two connections, one to my main gigabit switch and one to the streamer. That does not mean that is the only configuration that will work, just that I didn't test any other configurations.

 

One thing I did find is that it is important to leave an empty port next to the one going to the endpoint. For example if the endpoint is on port 1, don't plug anything into port 2. (its actually more complex than that, sometimes you CAN plug something in next to the endpoint connection, but always leaving the ports next to it free will guarantee you don't have a problem).

 

John S.

[JohnSwenson]

4 hours ago, Forehaven said:

Thank you so much John!! NOW I see why the emphasis of adding another switch and ps.  Very clear and concise that even I understood lol

 

I'll def work on this aspect.

 

The 5 port/7.5vdc ps...will the LPS=1 at 7 work here John?

An LPS-1 will probably work with this. BUT the blocking of leakage from other devices on the network ONLY works if you ground the negative of the supply feeding the switch. The LPS-1 is specifically designed so it does NOT ground the output. So if you use the LPS-1 you must also ground the output of the LPS-1 going into the switch.

 

The grounding of the negative of the supply feeding the switch is the critical part here, no matter WHAT that supply is.

 

Grounding the supply will shunt the high impedance leakage from other devices on the network, the transformers in these particular switches will block the low impedance leakage thus there is no leakage to modulate the clock in the switch and no leakage to transfer to your endpoint.

 

The transformers in other switches do not seem to block the low impedance leakage nearly as well as these do.

 

John S.

[JohnSwenson]

5 hours ago, austinpop said:

 

John,

 

Is grounding necessary if using an LPS-1 to power the switch?

YES, the grounding on the negative of the supply feeding the switch shunts the high impedance leakage coming from the network. So if using an LPS-1 you must ground the OUTPUT so the negative going into the switch is properly grounded.

 

John S.

[JohnSwenson]

5 hours ago, austinpop said:

 

Ah - you mean the SMPS energizing supply for the LPS-1 - aka Meanwell?

 

Also, @JohnSwenson - do you know what the FS105/FS108 are doing that is unique to suppress leakage through the transformers in the signal path? 

No I don't know what is special about these. Someday I MIGHT look into that, but it is complicated by many of these switches using magnetics that are very hard to find data on. The large companies frequently get custom magnetics from manufacturers in China making this task almost impossible.

 

John S.

[JohnSwenson]

6 hours ago, thyname said:

 

Thanks!

 

This one?

 

https://www.amazon.com/NETGEAR-ProSAFE-FS108NA-8-Port-Ethernet/dp/B00002EQCS/ref=sr_1_1?ie=UTF8&qid=1508161224&sr=8-1&keywords=netgear+switch+fs108

 

Then, either:

 

1 - use a 12v linear power supply, or

2 - do the grounding think on existing power supply it comes with as @JohnSwenson describes

 

Correct?

 

Then since I am using an sms-200 (soon to be upgraded to either ultraRendu or sms-200ultra), the clocking issue it taken care of by these streamers / endpoints?

 

Thanks!

 

3 hours ago, toetapaudio said:

 

https://www.computeraudiophile.com/forums/topic/35191-review-dc-power-supplies-for-aqvox-audio-grade-switch/

 

The clocking part of this is a bit unknown at this point. I am currently working on the analysis of this, which means I have to build some of my own test equipment.

 

At this point I'm not ready to give full blown details on this, but preliminary results are indicating that clocking of any data stream coming into a digital device brings along the phase noise of the clock used to produce that stream. Packet systems such as Ethernet and USB complicate this dramatically because the data comes in packets, in between the packets there is no data to pass clock phase noise. Note that phase noise is another way to look at jitter, I'm NOT talking about amplitude noise here.

 

The implication of this is that putting a very low phase noise clock on the last switch is not sufficient. The effect of the phase noise of whatever else is going into the switch also makes its way into the clocking of the switch. So just improving the local clock helps, but is not all there is to this. Ideally you should get rid of the clock influence from other sources as well.

 

Please don't ask for details about this, I have a LOT more research to do before I'm willing get into more detail.

 

John S.

[JohnSwenson]

1 minute ago, rickca said:

@JohnSwenson why don't they make most SMPS with negative output grounded in the first place?  Is there a safety or cost issue?

 

Grounding the output opens the possibility of a good old fashioned ground loop (magnetic induction from hot wires to ground wire causing voltage difference between branches of the electrical wiring or long distances on the same circuit). But not grounding lets the high impedance leakage pass through to the output.

 

Nobody else even seems to be looking at high impedance leakage, I didn't even know about it until two months ago. It is so high an impedance that normal test equipment shorts it to ground and you never see it. The problem is that its effects ARE showing up in audio systems (not all and not in the same way in all systems, of course it can't be anything simple). I had to build my own test equipment to even see it.

 

A good part of the problem is that almost all the original work on leakage currents took place many years before the advent of SMPS so they were just looking at linear power supplies, which don't have the high impedance leakage. Its a fairly recent issue, significantly exacerbated by computers and computer networks involved in our audio systems. (since these almost always use SMPS)

 

John S.

[JohnSwenson]

7 minutes ago, austinpop said:

 

I feel like I'm hearing conflicting things. At the risk of sounding dense, John, please help me understand. In my case, the chain is:

• AC mains > Meanwell SMPS (7.5V) > LPS-1 (7V) > switch

Should I ground the output of the Meanwell, the LPS-1, or both?

To get the full advantage of using the FS105/FS108 you need to ground the output of the LSP-1 (or any other power supply you use). Grounding the INPUT to the LPS-1 will not make any difference.  You can do it if you want to, but it will not improve anything in this scenario.

 

John S.

[JohnSwenson]

2 minutes ago, auricgoldfinger said:

 

My question, exactly.  Plus, how does one ground the output of the LPS-1?

The same way as any other supply, just use one of the methods shown in the SMPS grounding thread, but in this case apply it to the output cable from the LPS-1 rather than an SMPS.

 

John S.

[JohnSwenson]

3 hours ago, Summit said:

 

Do you have any explanation or theory to why it’s better to leave an empty port next to the one going to the endpoint?

I don't have details on this, but it seems to be related to many switches using dual magnetics modules. These have two channels in one box. If the two connections are on the same magnetics box then leakage is not blocked. If the connections are on different magnetics boxes then leakage IS blocked.

 

A possible mechanism is that a box has one connection from transformer center taps to the ground plane. If going between magnetics boxes the leakage current HAS to go through whatever method is used to connect the box to the ground plane. If the connections are on the same magnetics box then this network that connects to the ground plane is bypassed.

 

Which port to port connections get properly leakage blocked depends on which ports go through which magnetic boxes. You don't really know unless you open up the switch and look.

 

FREQUENTLY, but not always, port 1 and 2 are on the same, port 2 and 3 are on the same etc. For a 5 port it may be different, 1 and 2 may be on the same OR 4 and 5 may be on the same, which means 1 and 2 are on different magnetics boxes.

 

The way to make sure you don't into this issue is to keep the ports next to the one connected to the audio system unconnected. So if you put the cable going to the audio system on the first or last port and put the one next to it unconnected you are almost certainly good.

 

John S.

[JohnSwenson]

It seems a lot of people are getting confused by my posts on the subject of network leakage, I will try and state things in a more concrete manor.

 

This is relating to a switch which is what your network endpoint into your audio system is connected, this may be streamer (microRendu etc) a laptop or other computer Mac mini, PC etc). We shall call this the audio endpoint (AE)

 

Leakage current can get into the AE through the switch in two ways, from the power supply powering the switch, or from the cable connected to the rest of the network. The leakage coming from the network comes from the SMPS powering THOSE devices.

 

If the switch connected to the AE is powered by an SMPS, grounding the negative of output of the SMPS will shunt the leakage from that supply, but the leakage from the NETWORK will still go through. There is one exception, see the next section.

 

IF the switch is one of FS105 and FS108, grounding the negative of the supply will get rid of BOTH the leakage from the SMPS AND the network leakage. NOTE, this ONLY happens for these two switch types. Grounding the supply to a different switch type does NOT block network leakage.

 

If you are powering a switch from a linear supply, this gets rid of the leakage going through the PS of the switch, but NOT the network leakage. The only way to get rid of the network leakage is to use one of the above switches AND ground the negative of the supply powering the switch, no matter WHAT that supply might be. (linear, SMPS, LPS-1 etc)

 

If you are using an LPS-1 to power the switch, see the above rules for ANY supply. ANY supply includes the LPS-1. Thus IF you have one of the two named switches and you are powering the switch from an LPS-1, you must ground the output of the LPS-1 in order to block the network leakage. This will only work with one of those two switches. Grounding the output of the LPS-1 will NOT block leakage if you are using some other switch. It will not help if the LPS-1 is driving some other type of device. Thus there is no reason to ground the output of an LPS-1 if it is NOT driving one of the above named switches.

 

There is one exception to the last point. IF the LPS-1 is driving an ISO REGEN there can be a situation where the whole audio system is floating with respect to earth ground and a charge can build up which can show up as clicks and pops. ONE earth ground in such a system can alleviate this. ONE way to do this is to ground the negative of the supply powering the ISO REGEN. If this supply is an LPS-1 then you can try grounding the output of the LPS-1 to see if it fixes the clicks and pops.

 

Grounding the INPUT to the LPS-1 can help in other situations by shunting the high impedance leakage.

 

I hope this makes things clear, I'm running out of ways to say this.

 

John S.

 

 

 

 

 

 

[JohnSwenson]

3 hours ago, Forehaven said:

Careful guys. I just bought this from the Amazon link, and I was sent FS105NA Version 3 that uses 12v's.  Only Version 2 uses 7.5v.  So I gotta return this one and buy elsewhere I guess.

No need, both the 7.5V and the 12V ones work fine. I originally tested the 7.5V since that was all I had, then I bought the 12V ones and tried them and they did the same thing.

 

The problem here is that I have posted information on this in at least three different threads and information seems to get dissipated rather rapidly.

 

John S.