Pano, the only actual exploration of this question that I know of, comes from Romy the Cat, and you know what you are in for there. To note, though, Romy is a Russian trained, high voltage electromechanical engineer, so there will be reasons attached to his rather forcefully presented ideas.
Bud
Bud
I am quite a bit farther down my learning curve for transformers, than I am for the other wild notions I posit here and there. It my profession, a magnetician, and I actually have application ideas that are far more odd, for transformers, than I do for speakers and ground side electrons. As with a few others in the world, I think of them as antenna events.
Bud
I do not know about North America power distribution, but in general you would like to have the grounded side of the primary windings as kind of shield next to the secondary windings - simply assuming that there the voltage noise levels are less due to its grounding (and also less affected by load over impedance).
You also could use a shield winding between primary and secondary but this I found to have drawbacks - at least in the power distribution system where I live and where neutral is grounded not in house but at the distribution transformer side.
No science just my thoughts...
Michael
Last edited:
Not knowing exactly what is involved, but I do recall there are times when two devices are pluged into the same power strip, pluging in one way will create a higher AC voltage between the chasses of the devices while swapping the polarity on one will reduce it significantly. There is audible difference of course.
Besides the deeeep insights Bud provided there are some plain measurements of different transformers linked at the DCX thread.
Especially Gary Pimm did a great work where you can look at the *technical* performance of quite a bunch of different coupling transformers.
My own interest in coupling transformers emerged from old measurements of mine and a discussion I had a veeery long time ago with Lynn in this thread regarding transformer distortion at low frequencies - but that might be slightly off to what you are after.
As Bud says, measurements and perception of sonics are not one and the same...
Michael
Last edited:
Here is a link to how the household circuit is connected in the USA. Guess where I found this link.......
Household Electric Circuits
Bud
Household Electric Circuits
Bud
correction:
Neutral here is grounded either side, at the distribution transformer and when entering the house.
But - in addition there is a "person protection breaker" obligatory - detecting current flow towards ground - which adds some impedance in the neutral line (haven't measured yet)
Michael
Ahh, quite different layout!
Hard choice indeed, but depending on the symmetry of the distribution transformer there might still be better and worse polarity.
Michael
Warning: Do Not let Bud drag you into this. He tried to warn me once that it is:"not a very deep pool of black magic, at least until you get to the pointy hat grotto....."
Did I listen?
No, I did not, and I now I have to wear a tin-foil hat all the time
Best regards,
TerryO
OK, this makes a bit more sense now that I've slept on it.
If the high capacitance side of the transformer is more likely to couple noise into the chassis, you would like the electrical supply wire closest to ground potential to be connected there. Thus less noise coupled into the chassis.
Of course both sides of the mains supply carry current, but the neutral is nominally at ground potential in many systems.
I do remember from the "old days" in the USA before polarized plugs (and also in France where they didn't use them), that swapping the plugs back and both could really result in lower noise. Getting the right combo was often tricky. In the USA we don't have the option now with polarized plugs.
.
If the high capacitance side of the transformer is more likely to couple noise into the chassis, you would like the electrical supply wire closest to ground potential to be connected there. Thus less noise coupled into the chassis.
Of course both sides of the mains supply carry current, but the neutral is nominally at ground potential in many systems.
I do remember from the "old days" in the USA before polarized plugs (and also in France where they didn't use them), that swapping the plugs back and both could really result in lower noise. Getting the right combo was often tricky. In the USA we don't have the option now with polarized plugs.
.
You can either use a heavy duty wire cutters or a Dremmel to remove the tabs on a polarized plug and then you can plug the plug into the receptacle either polarized or non-polarized.
OK, simplified, does that means that dot marked point on primary side of transformer should be connected to the phase and the not marked connector neutral?
So how about just telling us how large of a value change between the lines? That way we can get a feel for the amplitude variation.
Thanks,
Dan
More Measurements
This isn't an area where we have to guess or do a lot of audiophile listening - we can measure and get the right answer. All you need is a modern high-impedance DVM. (A sensitive capacitance meter is optional.)
Connect device A and B (say, a linestage and a power amplifier) to the AC wall socket. DO NOT interconnect them. Turn them both on. Measure the AC Volts potential between the two chassis. I've measured anywhere between 3VAC and 70VAC between pairs of chassis, on a North American 120VAC system. These voltages appear on two-prong (ungrounded) and three-prong (grounded) equipment, and are the result of small leakage currents (specifically, stray capacitance) between the primary of the power transformer and chassis.
Next, measure the potential between the chassis of device A and the ground pin of the wall socket. Ideally, float the chassis of the device by using a 3 to 2-prong "cheater" plug, although this poses a serious user hazard if there is a fault condition in the device. Wearing gloves is a good idea when doing things like defeating safety features.
Note the measured potential will NOT be zero. Repeat for device B. Keep the gloves on as long as the device is plugged into the "cheater" adaptor. (Common fault conditions include miswired or defective power switches.) The device with the higher AC potential between chassis and ground is most likely the one with the reversed primary winding. A more rigorous method is to directly measure the capacitance between the chassis and each side of the incoming AC line - with everything turned off and disconnected, of course.
Connect the two pieces of equipment with your favorite audiophile interconnect. (If using unbalanced RCA's, turn off the second device before connecting together, then turn it on again after connecting.) Measure the voltage potential between the two devices - it will probably be in the small millivolt range, or lower. Don't smile too quickly; that same voltage, which is very noisy AC, will appear as a common-mode signal at the input of the second device.
What causes this? There's a small leakage current, mostly capacitive, and thus tilted towards HF noise, between the incoming AC and the chassis of the device. The best way to reduce this leakage current is use specialized medical-grade low-leakage power transformers, but this rarely done due to cost and size considerations.
The shield, or "ground" side of the interconnect creates a metallic connection between the two chassis. Since it has a finite resistance (no, it's not a superconductor, despite what the advertising says), there's a voltage drop that appears across the length of the cable. This voltage drop translates the small chassis-to-chassis current flow into a voltage, which then appears at the (voltage-responsive) input of the second device.
How to improve this situation, which unfortunately applies to all interconnected pieces of equipment that are not transformer-isolated from each other? It may not be practical to rewire the offending devices, although that is the most permanent solution. An even better solution (and less practical) is to throw out the high-leakage power transformer and replace it with a medical-grade low-leakage transformer - these use foil-shielded windings that are on different sides of the core, thus no direct physical contact between the primary, core, case, secondary, or anything else. (A generic power transformer depends on the enamel coating of the primary wire for all shielding and isolation - think about it! All it takes is a pinhole in the enamel coating for things to start to go wrong very quickly.)
If you add a heavy chassis-to-chassis braid that covers and shields the audiophile interconnects and presents a lower impedance than the signal-return of the interconnect, that decreases the voltage drop between the chassis, and offers a lower-impedance path that does not carry audio currents. Crude-looking but effective.
Balanced XLR interconnects and circuits offer improvement, but as long as there is a metallic connection between the two chassis, AC noise currents will flow, and it is up to the CMRR rejection to the second device to get rid of this noise. In effect, you're using pair-matching, feedback and gain to get rid of a problem that shouldn't be there in the first place.
When you have to transport audio signals over long distances (20 meters or more), active CMRR solutions cannot handle the safety hazards of equipment plugged into dissimilar AC systems, nor can they reject the large amount of induced common-mode noise of long (balanced) cables. Think about what it takes to reduce induced noise in a large studio for television or sound movies that have a huge number of noise sources - motors, light dimmers, various gizmos of all different kinds, and long connections between audio gear. This is where transformers with high isolation ratios come in; to break the grounds, filter off RFI, and deliver very good balancing (on both send and receive) that is independent of the CMRR figures of active electronics.
Anyway, this is why I'm a little skeptical of cable comparisons in typical audiophile systems with hard connections between different devices. Low-level noise, particularly hashy noise from the incoming AC line, can mimic sonic colorations when it is below directly-audible thresholds. (In other words, you might have to be right next to the speaker to actually hear the hash or roughness, but it will be audible as a coloration when sitting further away.) There are good reasons for the pro world to go to the lengths they do to get AC noise as low as possible - and there's no reason we can't follow suit with our equipment.
This isn't an area where we have to guess or do a lot of audiophile listening - we can measure and get the right answer. All you need is a modern high-impedance DVM. (A sensitive capacitance meter is optional.)
Connect device A and B (say, a linestage and a power amplifier) to the AC wall socket. DO NOT interconnect them. Turn them both on. Measure the AC Volts potential between the two chassis. I've measured anywhere between 3VAC and 70VAC between pairs of chassis, on a North American 120VAC system. These voltages appear on two-prong (ungrounded) and three-prong (grounded) equipment, and are the result of small leakage currents (specifically, stray capacitance) between the primary of the power transformer and chassis.
Next, measure the potential between the chassis of device A and the ground pin of the wall socket. Ideally, float the chassis of the device by using a 3 to 2-prong "cheater" plug, although this poses a serious user hazard if there is a fault condition in the device. Wearing gloves is a good idea when doing things like defeating safety features.
Note the measured potential will NOT be zero. Repeat for device B. Keep the gloves on as long as the device is plugged into the "cheater" adaptor. (Common fault conditions include miswired or defective power switches.) The device with the higher AC potential between chassis and ground is most likely the one with the reversed primary winding. A more rigorous method is to directly measure the capacitance between the chassis and each side of the incoming AC line - with everything turned off and disconnected, of course.
Connect the two pieces of equipment with your favorite audiophile interconnect. (If using unbalanced RCA's, turn off the second device before connecting together, then turn it on again after connecting.) Measure the voltage potential between the two devices - it will probably be in the small millivolt range, or lower. Don't smile too quickly; that same voltage, which is very noisy AC, will appear as a common-mode signal at the input of the second device.
What causes this? There's a small leakage current, mostly capacitive, and thus tilted towards HF noise, between the incoming AC and the chassis of the device. The best way to reduce this leakage current is use specialized medical-grade low-leakage power transformers, but this rarely done due to cost and size considerations.
The shield, or "ground" side of the interconnect creates a metallic connection between the two chassis. Since it has a finite resistance (no, it's not a superconductor, despite what the advertising says), there's a voltage drop that appears across the length of the cable. This voltage drop translates the small chassis-to-chassis current flow into a voltage, which then appears at the (voltage-responsive) input of the second device.
How to improve this situation, which unfortunately applies to all interconnected pieces of equipment that are not transformer-isolated from each other? It may not be practical to rewire the offending devices, although that is the most permanent solution. An even better solution (and less practical) is to throw out the high-leakage power transformer and replace it with a medical-grade low-leakage transformer - these use foil-shielded windings that are on different sides of the core, thus no direct physical contact between the primary, core, case, secondary, or anything else. (A generic power transformer depends on the enamel coating of the primary wire for all shielding and isolation - think about it! All it takes is a pinhole in the enamel coating for things to start to go wrong very quickly.)
If you add a heavy chassis-to-chassis braid that covers and shields the audiophile interconnects and presents a lower impedance than the signal-return of the interconnect, that decreases the voltage drop between the chassis, and offers a lower-impedance path that does not carry audio currents. Crude-looking but effective.
Balanced XLR interconnects and circuits offer improvement, but as long as there is a metallic connection between the two chassis, AC noise currents will flow, and it is up to the CMRR rejection to the second device to get rid of this noise. In effect, you're using pair-matching, feedback and gain to get rid of a problem that shouldn't be there in the first place.
When you have to transport audio signals over long distances (20 meters or more), active CMRR solutions cannot handle the safety hazards of equipment plugged into dissimilar AC systems, nor can they reject the large amount of induced common-mode noise of long (balanced) cables. Think about what it takes to reduce induced noise in a large studio for television or sound movies that have a huge number of noise sources - motors, light dimmers, various gizmos of all different kinds, and long connections between audio gear. This is where transformers with high isolation ratios come in; to break the grounds, filter off RFI, and deliver very good balancing (on both send and receive) that is independent of the CMRR figures of active electronics.
Anyway, this is why I'm a little skeptical of cable comparisons in typical audiophile systems with hard connections between different devices. Low-level noise, particularly hashy noise from the incoming AC line, can mimic sonic colorations when it is below directly-audible thresholds. (In other words, you might have to be right next to the speaker to actually hear the hash or roughness, but it will be audible as a coloration when sitting further away.) There are good reasons for the pro world to go to the lengths they do to get AC noise as low as possible - and there's no reason we can't follow suit with our equipment.
Last edited:
By the way, I want to thank all of my transformer-building friends - Bud Purvine, Dave Slagle, Per Lundahl, Mike LeFevre, John Atwood, and many others for all the good info they've given me over the years. Also a big thank-you to Gary Pimm for educating me on instrumentation design and shielding techniques, and another big thank-you to John Atwood for vacuum-tube physics.
When you know more about how devices actually work, you don't need to resort to audiophile hand-waving. You know where to look, and what to measure for. All physical devices depart from the pretty little models; real engineering is knowing how the physical devices depart from the models, and how to design around that. (Tektronix 101, but I'm sure Bell Labs and Hewlett-Packard did things this way too.) Watching things getting built is very educational - you can then see why the device has the little peculiarities it does. Something as minor as the choice of voice-coil glue (doesn't sound like much, does it?) actually has an effect on the sonics of an entire audio system. Similarly, physically mapping out the capacitive leakage path from primary to chassis is very informative. Stray capacitance is everywhere, so you might as well know what it's doing to your circuit and the system as a whole.
When you know more about how devices actually work, you don't need to resort to audiophile hand-waving. You know where to look, and what to measure for. All physical devices depart from the pretty little models; real engineering is knowing how the physical devices depart from the models, and how to design around that. (Tektronix 101, but I'm sure Bell Labs and Hewlett-Packard did things this way too.) Watching things getting built is very educational - you can then see why the device has the little peculiarities it does. Something as minor as the choice of voice-coil glue (doesn't sound like much, does it?) actually has an effect on the sonics of an entire audio system. Similarly, physically mapping out the capacitive leakage path from primary to chassis is very informative. Stray capacitance is everywhere, so you might as well know what it's doing to your circuit and the system as a whole.
Last edited:
