AC Filtering, Grounding Boxes, Linear PSU and Balanced Power.

[YashN]

All right, got 3 filter lines on today and working solidly for a couple of hours now. I have the iMac, the external USB HDD and the amp connected.

 

Big sound, tall soundstage, lots of presence, with probably a little of roll-off at the high-end, more noticeable with native DSD (still don't know why that would be the case), quite satisfactory!

 

As a controversial aside, before that, I tested the two 'All Night Long's without filter, and there's still a difference between these two on playback. Re-tested it just now with the filter on, and there's no getting out of this: the up-conversion with the filter is more punchy and clearer, the other one is flatter and thinner in comparison.

 

Now, coming back to the filter at hand, I am wondering how it will sound if I add two more lines and then plug the cable modem and Wi-Fi router into them.

 

Additionally, I could also try removing the amp from a filter line and re-use it on its own outlet just in case the filter line is impacting its dynamics.

 

Can't wait to have my girlfriend listen to her Diana Krall, Holly Cole and Cassandra Wilson when she comes back.

[YashN]

 

Lionel Richie is nagging me from my Millenium Collection CD (I've started a little collection of them):

Are you gonna do that re-rip test, Bro?!!

[Speedskater]

The Tripoint Troy is a very, very expensive and beautiful box! But I don't know just what it hopes to accomplish. In another forum, there have been long threads about the Tripoint and the somewhat similar Entrec. All I can guess is that they are huge boxes filled with a secret material and connected to audio components in a rather random fashion. It seems that the users think (incorrectly) that bad noise & interference currents go into the box and just disappear. A heavy copper terminal strip or bus bar would do a much better job of connecting the chassis's of all your audio components together.

 

 

In all fields of electricty and electronics there are still some people that believe the Mother Earth acts as a sump or sink for noise & interference currents. This incorrect idea is extended to these boxes with their secret materials.

 

 

While in fact noise and interference currents only interest is to get back to their voltage source. What is their voltage source? Why it's the Neutral of that big power company transformer down the street from your house.

[sandyk]

While in fact noise and interference currents only interest is to get back to their voltage source. What is their voltage source? Why it's the Neutral of that big power company transformer down the street from your house.

 

Some of those currents also go back through mother Earth ! Chatswood Telephone Exchange in Sydney is right alongside Chatswood Railway station. Until they improved the Old Exchange earthing, occasional problems happened to trip the large 52V 2,000A Rectifiers. The same happened occasionally when Westfield shopping centre, which is 100s of metres away down the road, started up it's escalators , main lighting etc. just before opening for the day . No, the incoming A.C. voltage didn't drop much when that happened.

[YashN]

The Tripoint Troy is a very, very expensive and beautiful box! But I don't know just what it hopes to accomplish.

 

Speedskater, the Tripoint Troy is meant for connecting each of your components' chassis to it. The device also connects to mains Earth.

 

Isn't that similar to what you proposed in Post 37? The only difference I see is that the Tripoint device asks for chassis to its box individually rather than all together.

 

These should be doing the same thing ideally, no?

 

and the somewhat similar Entrec.

 

The difference is that the Entreq wasn't planned for chassis grounding at all, just signal clean-up. Moreover, the box doesn't connect to mains Earth. In this respect, it's more intriguing than the above.

 

Now, these manufacturers are playing catch-up a little seeing the positive reports from users of both camps, including those benefiting from both boxes.

 

A heavy copper terminal strip or bus bar would do a much better job of connecting the chassis's of all your audio components together.

 

Better, I don't know, but I think that would be a similar thing to what the Tripoint devices do, but again, not the Entreq.

 

While in fact noise and interference currents only interest is to get back to their voltage source. What is their voltage source? Why it's the Neutral of that big power company transformer down the street from your house.

 

I think it's more correct to say that they will preferentially try to go back there with the lowest impedance route. It is also important to consider energy conservation and energy conversion.

 

Thus, would a silver terminal strip and bus bar, with silver ground leads do better than an all-copper solution?

[YashN]

I would really need to listen and the effectiveness totally depends on the rest of the system and AC treatment.

 

Indeed, some people have reported no gain using these in their systems, but I think the mish-mash of different implementations we end up with in most systems (including very expensive ones), make these solutions worthwhile.

[YashN]

The Shunyata Hydra Triton v2 also sports a chassis grounding system. I am hoping my tests will be positive with a small chassis-grounding scheme, and if so that I may build it into my AC filter box similarly to this (great reviews):

 

Shunyata is careful about claiming all systems can benefit:

 

What should I connect to the CGS?

 

All components in the audio system should be connected to the CGS grounding system if possible. Additionally, you may connect an equipment rack that is made from metal.

I have an amplifier(s) connected to a different dedicated line — should I connect them to the CGS?

Generally speaking, you should try it. This may reduce ground loop problems and improve sound quality. However, with some amplifiers connecting the ground to the CGS may produce no results or may degrade the sound quality. It is important to test each of the ground connections independently.

 

Moreover,

 

If a component’s power cord is connected to the TRITON — do I need to run wire to the CGS?

 

If a component is connected to the TRITON with a power cord where the ground wire and contact are functional then the chassis of the component will be connected to the internal grounding system of the TRITON. You do not need to make another connection to the CGS terminal. However, some components may benefit from an additional chassis connection especially if you are using common stock power cords. Try it and see if performance is improved.

 

More info on the CGS (Chassis Grounding System) at WBF.

[Speedskater]

Speedskater, the Tripoint Troy is meant for connecting each of your components' chassis to it. The device also connects to mains Earth.

Isn't that similar to what you proposed in Post 37? The only difference I see is that the Tripoint device asks for chassis to its box individually rather than all together.

These should be doing the same thing ideally, no?

If the box has very low resistance from connector to connector, than it could act the same way a heavy copper terminal strip does.

 

The difference is that the Entreq wasn't planned for chassis grounding at all, just signal clean-up. Moreover, the box doesn't connect to mains Earth. In this respect, it's more intriguing than the above.

Any thoughts on what secret method they use for this signal clean-up?

 

Now, these manufacturers are playing catch-up a little seeing the positive reports from users of both camps, including those benefiting from both boxes.

Better, I don't know, but I think that would be a similar thing to what the Tripoint devices do, but again, not the Entreq.

I find it a bit challenging to keep track of just which box and which special cable does what. Not much in the way of spec sheets or instructions for these products.

 

I think it's more correct to say that they will preferentially try to go back there with the lowest impedance route. It is also important to consider energy conservation and energy conversion.

Could you expand on this thought, I'm not sure of where you are going.

 

Thus, would a silver terminal strip and bus bar, with silver ground leads do better than an all-copper solution?

At the low frequencies involved, not enough difference between silver and copper to even think about it.

But silver plating could cut down on corrosion.

[YashN]

If the box has very low resistance from connector to connector, than it could act the same way a heavy copper terminal strip does.

 

That's what I thought, so I do not think the Tripoint devices are that mysterious although, and it's understandable, he doesn't want to reveal what's inside.

 

Here again, what's if it's silver rather than copper?

 

Any thoughts on what secret method they use for this signal clean-up?

 

I can only speculate as have no idea, nor have I tried one, but if we're talking of the Entreq, I think it could be inducing losses through energy conversion within the box. Whether this is technically feasible, I don't even know, and I'm not even sure I can experiment with something similar at home. Well, I could always try.

 

 

 

I find it a bit challenging to keep track of just which box and which special cable does what.

 

Indeed, I found the same issue: not much details on the company sites, and a slew of products and product lines (especially Entreq), but the posts on Tripoint and Entreq at WBF and Devialet helped clarify the benefits of each.

 

Could you expand on this thought, I'm not sure of where you are going.

 

Not considering impedance is a mistake.

 

From Ott's 'Grounding of Mixed Signal PCBs':

 

High frequency digital ground return currents want to return directly underneath the signal trace. This is the lowest impedance (lowest inductance) path

 

At the low frequencies involved, not enough difference between silver and copper to even think about it.

But silver plating could cut down on corrosion.

 

Now, for Entreq, the box specifically is said by the manufacturer to deal with 'stray currents which are low voltage but high-frequencies', not low frequency.

[YashN]

I had the good idea of measuring the chassis potential differences if any between the iMac chassis, the iFi iDSD Nano chassis and the Amp chassis before embarking on some DIY. This may legitimate the DIY chassis grounding, but also gives me more time to listen to the 3 filter setup currently playing since yesterday.

 

Large soundstage, quite powerful sound, loud and clear enough even at -40dB on the amp (night), and -35dB or so (day). The sense of space is better, and you can definitely hear the mid-range better now. There may be a slight roll-off at higher-frequencies.

 

Here are the measurements, taking into consideration that in my setup, only the iMac is directly grounded:

 

iMac chassis - Amp chassis : 46mV (130mV earlier)

iMac chassis - iFi chassis : 45mV

iFi chassis - Amp chassis : 4 to 5 mV

 

Looks like there's something to be done here.

[Speedskater]

Here again, what's if it's silver rather than copper?

No need for silver (or silver plating) at these low frequencies.

 

I can only speculate as have no idea, nor have I tried one, but if we're talking of the Entreq, I think it could be inducing losses through energy conversion within the box. Whether this is technically feasible, I don't even know, and I'm not even sure I can experiment with something similar at home. Well, I could always try.

If he puts a resistive load in the box, then more of the noise currents will find other paths, like your interconnect cables. What we want to to is get all of the chassis's to the same reference plane voltage.

 

Not considering impedance is a mistake.

..........................................................

Now, for Entreq, the box specifically is said by the manufacturer to deal with 'stray currents which are low voltage but high-frequencies', not low frequency.

Yes, we have to consider the impedance (not just the resistance) of the conductors in this system.

No matter how big we make round wires, they don't do well with increasing frequency.

What does much better are wide flat straps and wide flat braids.

[Speedskater]

Some of those currents also go back through mother Earth ! Chatswood Telephone Exchange in Sydney is right alongside Chatswood Railway station. Until they improved the Old Exchange earthing, occasional problems happened to trip the large 52V 2,000A Rectifiers. The same happened occasionally when Westfield shopping centre, which is 100s of metres away down the road, started up it's escalators , main lighting etc. just before opening for the day . No, the incoming A.C. voltage didn't drop much when that happened.

Yes current is happy to use Earth as a path back to it's source. Earth has a point to point resistance of nearly zero Ohms even if those points are separated by many miles. The problem is making good connections to Earth, those connections may cost tens of thousands of dollars each. Even in the AM broadcast radio band, much of the signal travels through the Earth. But about the only current that sees Earth as a destination (rather than a path) is lightning.

[YashN]

No need for silver (or silver plating) at these low frequencies.

 

I am not sure here if Tripoint deals only with low frequencies or if it also deals with higher ones - cannot recall seeing a reference in a post somewhere about the type of frequency.

 

If he puts a resistive load in the box, then more of the noise currents will find other paths, like your interconnect cables. What we want to to is get all of the chassis's to the same reference plane voltage.

 

That's supposing the use of a resistor, but what if it's something else?

 

No matter how big we make round wires, they don't do well with increasing frequency.

What does much better are wide flat straps and wide flat braids.

 

Already did a flat braided ground for the amp, iDSD Nano and the iMac today and tested it: the measurement still reads rather high, so I'm not sure it's effective (I was expecting 0 here, right?), and on auditory testing, although I believe I perceive a larger soundstage and better micro-details like vibrato on bass that I wasn't hearing before, and generally a cleaner mid-range, my girlfriend isn't so sure.

[JohnSwenson]

The return current takes the lowest impedance path. At low frequencies (good old 60/50Hz) the dominating factor is usually the DC resistance of metals, so silver would be better than copper, IF you had the cross sectional area of both metals. But making the copper wire a little thicker does the same thing for a much lower cost.

 

As the frequency increases the inductance of the path starts becoming more important than the DC resistance, this transition happens at fairly low frequencies, it starts at about 1KHz and by 15KHz inductance is significantly more important than DC resistance. From 15KHz up to several MHz the cross section of the metal and dielectric are important in the inductance. Getting above that skin effect comes into play and the geometry of the outer surface and dielectric become of prime importance.

 

So things like buss bars and thick wires may be important for noise issues with low frequency components, they are not that important for higher frequency components.

 

John S.

[YashN]

John,

 

What do you think of the findings of some people saying they get results with these two boxes, even combined in their systems?

 

Supposedly, one is dealing with chassis grounding, which we have concurred can help if they weren't at the same potential in the system previously.

 

The other one apparently only connects in such a way as to clean up the signal (they use a free RCA or XLR, and purportedly only connect the ground in these).

 

Another quick question if I may: connecting my component chassis, I should expect the potentials measured between each pair of these 3 components to drop to 0, correct?

[YashN]

Those frequency-dependent characteristics are interesting: are the stray currents that supposedly accumulate on chassis surfaces (if I indeed got that right) usually within he audio band, and if so, lower end or higher end? Or are these above the audio band (here again, higher, much higher?).

 

Is the explanation of Entreq about their signal cleaning product dealing with 'low voltage, high frequency stray currents' realistic?

 

As the frequency increases the inductance of the path starts becoming more important than the DC resistance, this transition happens at fairly low frequencies, it starts at about 1KHz and by 15KHz inductance is significantly more important than DC resistance. From 15KHz up to several MHz the cross section of the metal and dielectric are important in the inductance. Getting above that skin effect comes into play and the geometry of the outer surface and dielectric become of prime importance.

 

So things like buss bars and thick wires may be important for noise issues with low frequency components, they are not that important for higher frequency components.

[YashN]

The sound currently is very good, with the 3 filters on: the impression of a lot of power even at lower volumes, big soundstage, and clarity in the mid-range stays for now.

 

The chassis-grounding test has yet to convince me. What I did was draw a single flat braid from the Amp's phono stage GND, to the iFi's outer casing, to the iMac's chassis.

 

I am not sure yet the connection is well-made at the iFi casing and at the iMac chassis. The iMac does have a screw but it's underneath the screen frame, so it's not easy to maipulate. I might have to unscrew one of the DAC faceplate screws to hook up the flat braid there as well.

 

Since there have been many changes to my system recently (3 filters, cup-and ball float of the speakers at speaker ends, under the external HDD, and the DAC although the latter is not very effective yet), I think it's going to be worthwhile to let the filter settle in for a while, and also to let myself get acquainted with the new sound until I do a proper re-test of the chassis grounding.

[YashN]

Found this technical article 'JPS Labs Superconductor Interconnects: A Technical Discussion' which includes a section on inter-chassis potentials, which could be very interesting.

 

I think a lot of the issues we have are because of the prevalence of unbalanced power and unbalanced interconnects, so this will be an interesting read.

 

This extract, in particular mentions different Y capacitors, and my filter prototype is probably currently guilty of that:

 

The noise current in this loop flows in response to an interchassis potential generated by the interaction of each chassis' differing line and neutral leakage currents with the common-mode impedance of their respective power cords. The current drawn by each power supply, through its interaction with the differential-mode impedance of its power cord and AC mains wiring, can introduce an additional interchassis potential in certain special cases. RF fields, intercepted by the loop, also generate a potential around the loop that induces an additional noise current.

 

 

 

Causes

 

 

The following discussion will center on the causes of interchassis potential, first through consideration of leakage currents and second through power-supply interactions, and the techniques by which it may be minimized. Other sources of noise currents, nonlinearities in active stages, their impact on effective noise-floor characteristics, and the unbalanced interconnect's role in converting noise current to a noise voltage are discussed in relation to interconnect design.

 

 

 

The "Y" capacitors in Figures 1a and 1b, labeled Cys and Cyd, generate a leakage current that flows in the power cord's ground and neutral wires. These capacitances may or may not exist as actual parts in the equipment as the winding method of the power transformer and the power wiring in the equipment forms parasitics that would appear in the schematic in exactly the same manner. These caps are also part of any input EMI filter and add with the parasitics of the unit's construction. Designers, however, often eliminate the cap between line and chassis, as in Figure 1b, when there is no EMI filter or three-wire cord installed, or in equipment where an internal high gain stage might generate too much noise and hum. A resistor is sometimes added between neutral and chassis to form a local reference for the signal returns that are then typically connected to chassis. This added resistor, however, is occasionally left out of the design under the assumption that the destination equipment, a preamp for example, will provide the reference to ground, as in most consumer equipment. Safety is still guaranteed, however, because the open circuit potential, while high, is not dangerous owing to the high impedance of the "Y" caps that limits ground-fault current to a safe level.

 

 

 

Power-supply currents, as drawn from the wall, can also create a chassis potential above ground, but only if the two "Y" caps are not of equal value, as shown in Figure 1b. A simple network analysis shows this effect to be true while also showing that if the two caps are equal, as shown in Figure 1a, the symmetry in the network causes cancellation of any chassis potential arising from power-supply input currents flowing through the power feed's differential-mode impedance.

 

[YashN]

A cool trick to try from the above-mentioned article if you have two-prong equipment like my amp and if you don't have balanced power:

 

When standard, unbalanced power is used to power a system, however, a voltmeter can be used to reduce chassis potentials due to asymmetrical leakage currents by allowing the user to optimally orient the power-cord polarity. This is actually a useful test in that if the system noise floor seems to drop in response to power-cord polarity changes, the two "Y" caps are then usually not equal in value and often different between on and off states of the equipment to boot, making it likely unsuitable for use on balanced power. However, if the equipment has a non-medical-rated EMI filter or discrete "Y" caps of equal value, this technique will have virtually no effect on the noise floor and is therefore likely to be suited for use with balanced power. Unfortunately, orienting the polarity of the AC power plug is generally the only available method to the audiophile for reduction of interchassis potential when balanced power is not an option.

[Speedskater]

No need for silver (or silver plating) at these low frequencies.

I am not sure here if Tripoint deals only with low frequencies or if it also deals with higher ones - cannot recall seeing a reference in a post somewhere about the type of frequency.

In this case high frequencies would be about 100MHz. Some FM transmitters have parts of their circuits silver plated.

But in this case we are dealing with low frequency noise & interference currents. Almost all of which are well under 100kHz.

 

I can only speculate as have no idea, nor have I tried one, but if we're talking of the Entreq, I think it could be inducing losses through energy conversion within the box. Whether this is technically feasible,

That's supposing the use of a resistor, but what if it's something else?

If the box converts it to heat, then in effect that's a resistor. But maybe the box converts it mechanical motion, or maybe?

[Speedskater]

The return current takes the lowest impedance path. At low frequencies (good old 60/50Hz) the dominating factor is usually the DC resistance of metals, so silver would be better than copper, IF you had the cross sectional area of both metals. But making the copper wire a little thicker does the same thing for a much lower cost.

As the frequency increases the inductance of the path starts becoming more important than the DC resistance, this transition happens at fairly low frequencies, it starts at about 1KHz and by 15KHz inductance is significantly more important than DC resistance. From 15KHz up to several MHz the cross section of the metal and dielectric are important in the inductance. Getting above that skin effect comes into play and the geometry of the outer surface and dielectric become of prime importance.

So things like buss bars and thick wires may be important for noise issues with low frequency components, they are not that important for higher frequency components.

John S.

Some nits to pick:

 

The return current takes the lowest impedance path.

The return takes all available paths! In inverse proportion to their impedance. And therein lays the ground-loop problem.

So let's make it:

Most of the return current takes the lowest impedance path.

 

so silver would be better than copper

But it would be a poor engineering decision to do so.

 

the geometry of the outer surface and dielectric become of prime importance.

Yes, flat, wide, thin copper strips (or maybe braid) are good choices.

 

So things like buss bars and thick wires may be important for noise issues with low frequency components, they are not that important for higher frequency components.

True about the thick wires, but bus bars are an ideal choice at higher frequencies.

 

Note that Buss is a fuse company.

[YashN]

In this case high frequencies would be about 100MHz.

 

Agreed.

 

But in this case we are dealing with low frequency noise & interference currents. Almost all of which are well under 100kHz.

 

Are these the ones involved in the chassis?

 

If the box converts it to heat, then in effect that's a resistor. But maybe the box converts it mechanical motion, or maybe?

 

Yes, something like that was what I was thinking, along the line of:

 

[Very Low Impedance Path] -> [EM Field] -> [Kinetic Energy] -> [Heat Loss]

 

Realistic?

[YashN]

so silver would be better than copper

But it would be a poor engineering decision to do so.

 

Poor engineering? Why?

[Speedskater]

Poor engineering? Why?

Why:

a] High cost

b] Lack of availability.

c] No significant engineering gains.

Just use a one size larger copper item.

[Speedskater]

Are these the ones involved in the chassis?

 

In audio systems, the noise, leakage & interference currents that we are most concerned with are all power line related. The are generated by power supplies or small circuits that we can think of as power supplies.

 

Almost anything that you buy today and plug into an AC outlet has some sort of power supply (including CFL & LED lights). Any of these power supplies can generate noise. But most of these noise currents are under 100kHz. The noise voltages over 100kHz can be filtered with small capacitors.