Well, I never had the idea, but it sounds interesting: noise is plentiful on the mains, it will not register on the energy meter because of frequency, phase-shifts etc, and removing it will harm nobody, just the opposite in fact.
Somebody has taken it apart and came out with a block diagram:
PS Audio Noise Harvester - Page 3 - pink fish media
If the circuit looks like a typical LED power supply (and a half-wave rectified one at that), that is because it is. The difference is it is designed to be powered only with AC frequencies of >60Hz (hence the series capacitor) - the idea behind this "noise harvesting" is to consume power in the noise frequencies.
But:
- Half wave rectification means only noise on the positive half of the cycle is "cleaned up", the negative half remains on the mains, and now the leftover noise has a DC component.
- Capacitor power supply means noise current is only drawn from the mains if the noise voltage exceeds the voltage currently inside the capacitor.
The more logical route would be to use full wave rectification with a resistive load... or heck you can throw away the rectification since resistive loads won't need that. In other words, a light bulb with a series capacitor.
But why take the effort to dissipate the power when we can just short them to neutral or ground? So remove the entire circuit and keep the series capacitor. Or it should be called a shunt capacitor now.
PS Audio Noise Harvester - Page 3 - pink fish media
If the circuit looks like a typical LED power supply (and a half-wave rectified one at that), that is because it is. The difference is it is designed to be powered only with AC frequencies of >60Hz (hence the series capacitor) - the idea behind this "noise harvesting" is to consume power in the noise frequencies.
But:
- Half wave rectification means only noise on the positive half of the cycle is "cleaned up", the negative half remains on the mains, and now the leftover noise has a DC component.
- Capacitor power supply means noise current is only drawn from the mains if the noise voltage exceeds the voltage currently inside the capacitor.
The more logical route would be to use full wave rectification with a resistive load... or heck you can throw away the rectification since resistive loads won't need that. In other words, a light bulb with a series capacitor.
But why take the effort to dissipate the power when we can just short them to neutral or ground? So remove the entire circuit and keep the series capacitor. Or it should be called a shunt capacitor now.
Last edited:
If you want to be the local dumping point for all and sundry harmonic and noise currents then good on you 
That way you will sink voltage transient currents and lightning strike currents and so raise your own AC wiring well above 'ground', as well as add the noise and transient voltages to you local AC voltage that are due to external AC distribution resistance and impedances.
Better you than me (if I'm a neighbour nearby)!
That way you will sink voltage transient currents and lightning strike currents and so raise your own AC wiring well above 'ground', as well as add the noise and transient voltages to you local AC voltage that are due to external AC distribution resistance and impedances.
Better you than me (if I'm a neighbour nearby)!
I made a quick 'n dirty check, out of curiosity: I used a CMC as a transformer and a 0.33µF X cap. The inductance is 6.5mH, which puts the cutoff frequency at ~3.5KHz.
The diode bridge is made of big schottky's (BYS45) and the filter cap is 3µF mylar.
I connected a white LED as a load.
In my house, after the initial turn-on flash, nothing was visible, except a rare flash from time to time.
I measured 1.6V DC on average, ~open-circuit (10 megohm load), which explains why the white LED doesn't glow permanently.
The average short circuit current was ~400µA. This means that the internal resistance is ~4KΩ, and the maximum extractible power 160µW.
Clearly, you are not going to spare on your electricity bills.
By using a larger cap or inductance, it would be possible to harvest the mains harmonics: mine is quite distorted, with heavily flattened tops, problem is: with a larger cap the Tan δ losses may overcome the harvest, unless a good PP cap is used, and low frequency harmonics will be partially registered by the energy-meter
The diode bridge is made of big schottky's (BYS45) and the filter cap is 3µF mylar.
I connected a white LED as a load.
In my house, after the initial turn-on flash, nothing was visible, except a rare flash from time to time.
I measured 1.6V DC on average, ~open-circuit (10 megohm load), which explains why the white LED doesn't glow permanently.
The average short circuit current was ~400µA. This means that the internal resistance is ~4KΩ, and the maximum extractible power 160µW.
Clearly, you are not going to spare on your electricity bills.
By using a larger cap or inductance, it would be possible to harvest the mains harmonics: mine is quite distorted, with heavily flattened tops, problem is: with a larger cap the Tan δ losses may overcome the harvest, unless a good PP cap is used, and low frequency harmonics will be partially registered by the energy-meter
Most DIYers know enough to realise that flashing an LED does nothing useful in terms of mains filtering. However, there are a few genuine DIY filter designs, although I suspect most people simply buy a packaged filter as designing RF filters for poorly defined impedances is beyond most of them.
Because a filter has to be in line... The noise will bypass these so called harvesters either by inductive or capacitive coupling, hence why a filter has to be in line and laid out properly (a lot aren't especially for higher frequencies).
The problem is more problematic also as you really need to know what noise in on your mains and if you use power line communication you on to a looser to start with...
Like a lot of stuff sold for audio use they don't work and are there to generate cash nothing else, I would like to see the certification or compliance test results for a lot of this stuff (FCC or CE), also the new USB and Ethernet filters and associated products fall into this category, most filters not being designed properly and showing no measureable difference and the USB in line units being nothing more than a USB hub with fancy prices.
Look up mains filters and do the job properly...
The problem is more problematic also as you really need to know what noise in on your mains and if you use power line communication you on to a looser to start with...
Like a lot of stuff sold for audio use they don't work and are there to generate cash nothing else, I would like to see the certification or compliance test results for a lot of this stuff (FCC or CE), also the new USB and Ethernet filters and associated products fall into this category, most filters not being designed properly and showing no measureable difference and the USB in line units being nothing more than a USB hub with fancy prices.
Look up mains filters and do the job properly...
Why add any form of noise filter at all! Buy or make a well designed amp for starters.
A band-aid covers a sore. Try and understand the sore before applying liberal amounts of sticking plaster over it.
If you're worried about over-voltage damage then adding extra protection into the amp, or in line with the amp is a weak solution that may cause other collateral damage.
A band-aid covers a sore. Try and understand the sore before applying liberal amounts of sticking plaster over it.
If you're worried about over-voltage damage then adding extra protection into the amp, or in line with the amp is a weak solution that may cause other collateral damage.
The SLD and the tear down indicate it is trying to act as a high pass filter, attempting to clamp and snub some level of frequency dependant waveform peak voltage. Somewhat similar to clamping and snubbing a smps winding. The LED would indicate for a frequency dependant noise voltage.
Mains frequency voltage would be attenuated such that any stepped up voltage would then be below the clamp/snub level. A higher frequency limit would come in from the bandpass response.
Whether mains would ever experience noise levels that would end up being clipped - ?
Mains frequency voltage would be attenuated such that any stepped up voltage would then be below the clamp/snub level. A higher frequency limit would come in from the bandpass response.
Whether mains would ever experience noise levels that would end up being clipped - ?
An inline filter is not a band aid though, it is there to filter the noise that is superimposed on what should be a nice 230V sinusoidal waveform (Europe), using powerline communications adds EMC pollution to the waveform as do many other bits of gear hooked up to the mains, including energy saving bulbs... How else do you filter this noise or add overvoltage protection? This sort of thing is added to most commercial products that have to perform to a certain level and for UL I believe some form of over voltage protection has to be added.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.