Make a trasnformer hum 101


So this comes up repeatedly and I wanted to share this with you.

Sometimes perfectly good transformers mechanically hum. Often this is caused by having DC on the line. This is kind of a weird concept so let me take it step by step.

Normally a good AC line has a neutral which is essentially at ground potential, and a "hot" which we say is 120 VAC. That is, it has a sinusoidal waveform which repeats at 60 Hz (for America) or 60 times/second. The magnitude of the complete swing is measured in RMS. It’s an average which allows us to easily determine power dissipation. What this really means is that the voltage "swings" from about +85V to -85V. This is also called Peak to Peak or Vpp. An RMS measuring multi-meter will correctly say 120VAC, but if you look on an oscilloscope you will actually see the very top and very bottom of the waveform at around +- 85 V.

It is possible to have DC, AC or both on a conductor. In an incoming power line however it is NOT desirable to have anything but AC. What’s odd is that stupid little power supplies and dimmer switches can actually take an otherwise fine power line and distort it by adding DC to it.

This happens for a couple of reasons
  1. Some amount of impedance/resistance on the wire to the transformer, and through the transformers.
  2. A power supply that ONLY takes from one or the other voltage swing.

Consider a dimmer or laptop power supply that only takes current during the positive cycle. With a little resistance on the line, this pulls down the maximum positive voltage. Say from +85Vpk to +83Vpk, but the negative swing is still -85Vpk. Congrats, you’ve just added -2Vdc to your waveform. To make matters worse, these supplies may take current only for very short periods of time, causing the waveform to distort far from sinusoidal, and adding harmonics, or high frequency noise, to the incoming power where none existed before.

It is this DC voltage that will make otherwise fine functioning transformers develop a mechanical hum in the laminations.

The point I wanted to make is that these power supplies don’t really inject noise, so much as slurp power unevenly, causing a host of possible audible or visible effects.

Everything I’ve written here is scientifically verifiable with models, measurements and test gear, however it does not automatically justify spending $50k on power conditioners. How much of a problem you have and how much it’s worth to solve is entirely up to the music/movie lover.

I hope this is helpful,


E
erik_squires
Normally a good AC line has a neutral which is essentially at ground potential, and a "hot" which we say is 120 VAC. That is, it has a sinusoidal waveform which repeats at 60 Hz (for America) or 60 times/second. The magnitude of the complete swing is measured in RMS. It’s an average which allows us to easily determine power dissipation. What this really means is that the voltage "swings" from about +85V to -85V. This is also called Peak to Peak or Vpp. An RMS measuring multi-meter will correctly say 120VAC, but if you look on an oscilloscope you will actually see the very top and very bottom of the waveform at around +- 85 V.

That’s not correct. Voltage will swing +-170V. Peak to peak voltage will be 340V. Any meter will say 120V since non-RMS meters are scaled to show RMS voltage (for the sinewave) in spite of measuring average voltage.

Consider a dimmer or laptop power supply that only takes current during the positive cycle. With a little resistance on the line, this pulls down the maximum positive voltage. Say from +85Vpk to +83Vpk, but the negative swing is still -85Vpk. Congrats, you’ve just added -2Vdc to your waveform.

Dimmer or laptop won’t do it - you need much stronger load for that. In addition dimmer or laptop supply will draw identical current from positive and negative sides.

The point I wanted to make is that these power supplies don’t really inject noise, so much as slurp power unevenly, causing a host of possible audible or visible effects.

Most of linear power supplies draw current in narrow spikes of very high amplitude producing a lot of high frequency noise. Computer power supplies often use primitive SMPS that produces a lot of noise while dimmers chop sinewave also producing a lot of noise. Some of them have decent noise filter but many don’t. In addition they "slurp power" pretty much evenly.

Whoops, my math was wrong! :)

However, the problem is related to the line impedance. If it were perfect, then an uneven draw wouldn't affect anything.

I thought most SMPS were positive only draws? I must be confused then.

Best,

E
This brings interesting question - what causes DC on the AC line.  Is it uneven phase loading, VFD drives or anything else?

So a dedicated line from the panel will never cause hum? Can another line sharing the panel bleed noise into the dedicated line? How can we get a perfectly hum free power line? Can an aging grounding bar cause noise? Inquiring minds want to know... 
One issue then would be - what to do with SMPS, wallwarts or otherwise, that are now prevalent as part of an audio setup.
Post removed 
Post removed 
I try to keep my wall warts outside of my power conditioner.

For critical things like my DAC I purchased outboard linear regulated supply built like a tank. :)

That's as fussy as I'm going to be.

Best,

E
^^ How about a power conditioner for SMPS and a separate power conditioner for everything else?  :)
Would bolting Sorbothane under the transformer of one of those Carver "Magnetic Field" amps stop the bad hum that comes from them sometimes? (I'm guessing No, because the vibration is internal to the transformer.)
@joeylawn36111

Well, kind of sort of. ;) The transformer will still vibrate, but the panels vibrating along with it definitely make it louder. :)

You can experiment with your hands. If you put your hand on a panel to dampen it, you'll hear it. Then that's definitely something to pursue.

Best,

E
There are also some cases when the transformer is actually silent, but the panels vibrate! Not sure I've come across anything like that in decades though. :)
No Idea. You may want to check out some of the panels they use for damping PC's.

Best,

E
SMPS have bad rap from crude computer applications.  Jeff Rowland uses extremely quiet SMPS (cleaner than linear power supplies).  Advantage of SMPS is that it can operate from AC and DC (tolerates presence of DC) while many operates from wide voltage range. In addition they are line and load regulated, while linear supplies in power amps are not.
Post removed