Yes. I agree with this. There are situations where there is excessive differential or common mode noise on your mains supply. I wouldn't doubt that in many cases this can and will cause audible noise unless dealt with. Even basic things like ground loops would be included in this category.
Actually, as I said before, it's often harder to keep the power off the audio signal. The trouble is, nobody measures their gear relative to any fixed independant reference to see this. You might see a lovely stable, noise free 1Hz sinewave on your speaker wires, but if you earth reference the scope all you might see is a 60Hz 100VAC mains wave. What gives? Your amp power supply has reference current leakage is what. In a single instance it will make no odds, the speakers don't care if they are handling the 12V between 110V and 122V to them it's still just 12V + or -. However, once it's on your system, it can start to get complicated when it encounters an earth referenced bit of equipment which promptly dumps that leakage current to ground, creating current flow through the grounds... resulting in that nasty hum appearing where you don't expect it and a reinvestigation of your grounds and power systems.
Consider this. If you take a mains input into a breadboard and use appropriate transformers you can AM modulate the 50Hz/60Hz with pure audio. Then at the other end with another appropriate transformer demodulate the audio and play it. If you use the right components, you won't notice. It'll be noisy as hell though, I expect.
I'm Canadian but I have friends down under through my other avocation, electoral reform who pointed these out to me.
Canada and Australia have multi-party systems but your problems are similar.
You can find more at: https://www.youtube.com/channel/UCKRw8GAAtm27q4R3Q0kst_g
Sorry, bad phrasing. Should have used "significant measurable difference." Of course, "significant" is a slippery value all in of itself 🤔
It is surprisingly difficult to "do" clean power these days. In my space I'm not even worried about mains noise. I run as much of my audio gear off a battery. I have a 100Ah battery charged by a 50W solar panel. It runs from the garage up a 6mm^2 cable into the office/lab. A small quiet boost/buck converter gives me a stable quiet 14V. (the boost buck artefacts are in the 300kHz and up range).
However. You now take a custom PCB and put a whole mish-mash of digital audio signals, clocks and MCU signalling and your lovely 14V DC now looks like someone has been at it with a shovel and hammer drill. You can decouple and terminate signal lines all over the place, but digital noise is really nasty and really hard to get rid of. You don't want it on your opamp or amp rails. "phssstctctctctctcphssssstctctctctctc"
Sure if you have a degree in electrical engineering it's your bread and butter, but for a novice it's absolutely riddled with "traps for young players".
Consider the best possible DC filter is a LPF with a 0Hz cut off freq. However if you put in 1Hz into most calculators they tell you a 220uF and 1KOhm resistor will give you a, RC LPF with something like that. Woohoo!!!! You plop it into a bread board and measure it with your scope and ... indeed it's perfectly flat DC again. Oh.... but... that 1kOhm resistor has now limited your current to about 12mA and created a voltage drop. Damn. Then you discover "series pass transistors" and add one of them and again you get a stable supply, even when you add a 200mA load it remains stable. Oh.... but.... that transistor has dropped your voltage a fair bit. You no longer have 14V, you have 12.6V for some reason.
Playing with LC filters and the traps can not only mess your voltage and current like above, but come with the lovely added feature of LC tank circuit oscillations, or insanely high voltage inductive spike.
And that's only getting clean DC power for a 200mA headphone amp. Doing the same for a 200W power amplifier requires a lot more engineering and a multitude of stages, which is how I'm solving mine. A lightly filtered DC RC filter with series pass Darlington providing +/-12V @ 200mA to the main OPA amplification stage. Then that is additionally filtered by a much more aggressive RC filter w/ series pass Darlington to feed a Baxandall active gain pre-amp stage with ultra low noise, but completely powerless 20mA DC. I used to wonder why people did multiple stages in amplification, sometimes with 2 or 3 stages or more. It's almost certainly due to managing the power supply requirements and noise.
Yes!
Unfortunately, because they're less convenient to measure, common mode noise currents are very often ignored.
If you measure a device using Audio Precision measurement systems, you are measuring the differential mode performance. AP has done a superb job of creating a test environment where most common mode signals are rejected by the test equipment. That's exactly what they promise, too. But, are actual components like preamps used in an environment as pristine as what an AP system presents? Not very likely. If nothing else, the mesh of interconnected devices is much more complex than the simple in and out plus power in the AP set-up.
This isn't just a problem with audio equipment, either.
Here's my contribution for 2023.
One of the things often ignored in the application of filters is the actual impedances of the source and the load. In the case of an AC mains filter, the source impedance (the line) is hardly a single value.
In one of his books Ralph Morrison showed a plot of power line impedance versus frequency. At very, very low frequencies, the impedance is pretty much just the resistance of the wire. As you go higher, the impedance settles to some middle value out to forever. At least, on average. He explains why that is.
Line Impedance Stabilization Networks (LISNs) used to measure various conducted emissions create an arbitrary simulation of the AC mains impedance, with 50 Ohms used as the high frequency middle value. Pretty much like Morrison's graph.
Is 50 Ohms always the precise value? No, but it's probably in the middle of what's found in the field, so why not? Standards are usually arbitrary anyway. (Note: I'm not sure what the standards or actual values are for systems outside the US.)
So, take the most simple case of placing a suitable capacitor between line and neutral or whatever it's called outside NA.
I've attached an LTspice circuit that simulates that "simple case". There's a balanced equivalent of the NA AC mains impedance for the source followed by that cap and a resistive load representing the, ahh, load. Try for yourself. Simple, eh? What you expected?
Also note that this simulation is only for differential mode signals, nothing at all for common mode.
Done for the year already! Exhausting...
Last edited:
Not entirely. I just have a 12VDC feed capable of about 3 Amps with reasonable voltage drop and up to 10 amps (fused).
The whole system is floating. None earthe. I don't currently (bad pun) use an inverter for mains power from it. Having an outdoor floating, metal, device connected into the house and ultimately into my headphones does make me ponder my understanding of static voltages and electrical storms and usually remove the headphones and switch the 12V feed off when there is lightening around.
I just ordered replacement batteries, moving up to 105Ah of LiFePO4 cells from AliExpress this spring. The 3 year old marine start lead acid I've been using behaves more like a 20Ah battery these days and required another DC PSU in the garage to support it in December and I give it a "fizzing" charge and a week on float each winter. Its done. It's been charged and discharged about 10-40% every day for the last 3 years.
With the new cells the next upgrade it to a pro-quality 36V 330W panel on the garage roof. Lifting the system voltage up to 24V and I might be able to run a quiet true sine invertor to provide mains. There are SERIOUS considerations here however. A floating DC solar system producing 240V @ 50Hz output MUST NEVER, EVER be commoned with a live mains feed from the grid. I wouldn't even trust a grid-tie invertor with a sense line. In fairness though a good invertor should have protection should it suddenly fine itself out of phase.
On noise aspects on that rail, if it's isolated without the mains powered amp being involved, it's silent. The OPA551s in my headphone amp for a fixed gain of 11.5 and it is silent for my ears. That change when you connect it to a non-earth referenced amplifier because the whole thing then floats at 100VAC.... unless you earth the setup by connecting it to a PC USB port or similar, which adds all that lovely ground noise and your back to square one. I operate in a high noise environment, I have many problems to solve to get clean DC power.
Indeed. The other thing most people forget is that the whole circuit is involved all at the same time (excluding any digital domain decoupling). So the circuit you are working with and testing is generating it's own noise and adverse currents. Even more so if you are trying to feed it something it doesn't like, such as frequencies far, far higher than audio band caused by unstable opamps or LC resonance.
The relevant certifications (CE etal) and RF/EMF noise emitted from devices only tests the device effect on the exterior world. Such that your 1Mhz oscillating opamp or LC circuit would get picked up in a compliance test as it likely leaks out onto the mains or gets broadcast by some loop antenna on the PCB. etc. etc. However it says nothing about how noisy the device can be internally. It also really says very little about what happens if you mix and match all different devices and then amplify it.
For all you know your big ole iron valve amp might be why the little old lady next door can't get Wifi or use her mobile phone while it's on and you are wondering why you keep hearing a hum when the other neighbour is using their microwave. Pretty sure if you bring an old valve power amp into an EMC test centre they will just laugh and point to the door.
and so they (audiophile) chase ghosts and the get chased by the ghosts they create.
On the topic of non-audio equipment. Running high speed digital circuits on a breadboard and sometimes I discover something on the scope that terrifies me. Like a total of 14V peak to peak on a digital transmission line caused by inductive resonance on the square wave edges. That's going to put a lot of wear and tear on the ESD diodes in 3.3V digital inputs and its certainly not going to help anything. This is mostly caused by the big antenna line breadboard rails and the lovely looped antenna like jumper wires with multi-Mhz signals on them. All that inductance can create some pretty high voltages. You definately don't want those on your analouge side of things.
Last edited:
Pano, you found exactly what Shunyata has talked about for years, that normal power line filters reflect the noise generated by the equipment’s power supplies back into the equipment. Not a good thing.
Thanks for performing your power line noise tests!
Ah ha! Happy to be in good company.
The results were unexpected (by me) but so clear that other researchers must also have seen them.
I do know that in many uses keeping noise out of the power lines is the goal, but it can come at a local noise penalty.
Years ago Mitch Cotter sold grounding bars to which one would connect your equipment and from which a wire connected to a suitable ground (earth). Depending on where/how you grounded the bar, the line filtering could filter the AC line and the internal noise could be shunted to an internal ground sink.
One could do as my old dad specified for the lab building electrical systems he designed and drive a ground rod (or rods) -- Copperweld as I recall -- for a reliable reference.
I imagine shunt regulators that both source and sink, feeding individual stages would make sense too. Not sure how regulators such as the Super Regulator might help as well.
There is too much mentioning of possible mains and PS noise in this thread. PS as a noise source? Yes, if we are talking about SMPS. No in case of ordinary transformer + rectifier + CRC power supply, as seen on measurements I attached previously (#415, #497).
Standard mains input RF filters are tuned to be functional common mode noise filters from 10 kHz (10 - 20 dB reduction) up to 60 - 80 dB at higher frequencies. Differential mode filtering picks from 100 kHz.
Basic mains RF filter + ground loop breaker + regulated PS provides perfect isolation from mains noise or any hypothetical power cords influence. It can be seen from measurements at post #107. No premium power cord has a chance there. Snake oil supporter’s argument that ‘you simply don’t have ability to hear’ is funny I can’t hear something that doesn’t exist. Marketing videos with sound examples from various power cords all have clear differences in sound. I doubt that power cords are the reason.
Good shunt or super regulator PS for a preamplifier will have 1 – 2 uV total noise in audio bandwidth with noise floor at 10 - 15 nV/rtHz and PSRR above 100 dB Example with Salas UBiB from my preamplifier:
It is similar with regulated PS for power amplifier. Example from my amplifier:
Regulated power supplies also provide low output impedance with very small rails modulation and harmonics under load.
Last edited:
Much of the discussion in this thread lacks an understanding of power supplies. In a modern well designed commercial amplifier the power supply usually has a toroidal transformer, a rectifier section to convert AC to DC, and a large bank of filter capacitors to smooth it out. Any amount of AC that remains after that is so small, as tombo56 clearly demonstrates above, that it is inconsequential and will not be audible in the music.
The idea that the type of power cord delivering AC from the wall outlet has any affect on the DC voltage delivered to the amplifier output section or on the music itself that you hear is without any basis whatsoever.
And arguments that the various grounding possibilities are equally meaningless except in the few cases where actual hum or static noise are present. Those represent another matter entirely and need to be resolved on an individually basis. They are not common for most people.
The idea that the type of power cord delivering AC from the wall outlet has any affect on the DC voltage delivered to the amplifier output section or on the music itself that you hear is without any basis whatsoever.
And arguments that the various grounding possibilities are equally meaningless except in the few cases where actual hum or static noise are present. Those represent another matter entirely and need to be resolved on an individually basis. They are not common for most people.
I was responding to Pano's observation that significant AC filtering would keep noise in as well as noise out. I conjecture that some of that effect might occur with the line filter section of the Galo supply.
Paulca noted that digital circuits can and do generate a lot of noise internally. But mostly I'm building preamps and phono stages.
If there isn’t any noise to be kept in the PSU, well then there isn’t any noise, is there? No worries.
But have you measured the noise?
But have you measured the noise?
...Has wide bandwidth and can allow ingress of AC line noise by way of stray coupling right past rectifier and filter cap circuitry. Its been known to happen.
https://physics.stackexchange.com/q...oes-energy-flow-in-a-circuit-which-is-correct
Last edited:
Douglas Self pretty much explained most of this in his book “Audio Power Amplifier Design Handbook." The portion concerning power supplies can be found republished (and possibly expanded upon) here:
https://www.eetimes.com/Audio-ampli...er-supply-types---transformer-considerations/
Hal
https://www.eetimes.com/Audio-ampli...er-supply-types---transformer-considerations/
Hal
No. There are two ratings on line cords. Long-term, and just long enough to blow the fuse before fire starts.
Most homes now have 20Amp circuits.
US lampcord used to be #20, but was revised to #18 several decades back as 20A circuits got common.
#18 is usually rated 5A steady, but has often proven itself to blow 20A breakers without starting a fire. It is considered acceptable for the lampcord to bubble and die.
- Status
- Not open for further replies.