Antiparallel diodes as power line filter : how do they work ??

[ygg-it]

Please check the following schema with two antiparallel diodes for both power line conductors

http://www.degenonline.de/hifi/index.html#dcfilter

The purpose is to create a power line audio conditioner.
I tryed by myself and it change dramatically the soundstage (not better, not worse, just change...)

But I don't understand:
1) how they works
2) do capacitor in parallell useful. Which value ?
3) the wave spectrogram of 50 Hz power line doen't change, nor the THD nor the SNR so I relly don't understand how the work.

Probably, I think they should not coduct in the crossover part (+- 0,7 volt)?? So in that time period the trasformer doesn't work??

Can you please clarify?

[amc184]

The capacitors C1 and C4 are to block any DC that may be present in the mains AC. The diodes D1 to D4 are to ensure that the caps don't get anywhere near their ratings. Once more than about 0.7v DC is present, the diodes will conduct, bypassing the capacitors and preventing a dangerous condition. I would want to use some caps with very good ripple ratings here.

In principle, this should be a good idea. DC causes vibration and heating in a transformer. Not sure how well it would work in practice however.

[janneman]

Originally posted by ygg-it
[snip]I tryed by myself and it change dramatically the soundstage (not better, not worse, just change...)
[snip]

Really? I'm sure you did a comprehensive duoble-blind test to find that out? Can you give details.

Jan Didden

[analog_sa]

Originally posted by ygg-it
I tryed by myself and it change dramatically the soundstage (not better, not worse, just change...)

Initially i really liked the effect of DC blocking on sound. Nice, liquid midrange but a bit softer bass. Then it turned out that those electrolytics are (as usually) quite audible. There was no real dc problem (mV) and the transformer was not toroidal so there wasn't much to be gained by the blocker anyway. Of course if one can't hear a difference between power cords, amplifiers, cd players, etc this will also be inaudible. If, otoh, one hears power cables, it's quite obvious that the introdiction of additional non linear reactance along the power lines will be audible.

There is also a pi-filter shown in the link. It can also be responsible for the change in soundstage.

[amc184]

As far as I know, this is more for physical noise, not for reducing electrical noise in the power supply. DC across a transformer can make it vibrate audibly, which can be damn annoying! I would be surprised if it made a difference to the actual output of the amplifier.

[ygg-it]

You wrote:

Once more than about 0.7v DC is present, the diodes will conduct, bypassing the capacitors and preventing a dangerous condition.


For me :

Once more than about 0.7v DC is present, the diodes will conduct and this can be dangerous for the trafo. In my opinion the filter just block less 0.7 DC. I'm right?

How do you calculate the capacitors? I saw also the same circuit without the cap, what does it happen to the trafo in this case??


PS: To people who asked: the soundstage became harsher

[amc184]

If there is more than 0.7v DC on the mains, then you have a seriously poor electricity supplier. This will probably heat the transformer and blow its thermal fuse. The diodes prevent the caps from blowing, covering the discussed circuit from a safety viewpoint.

In essence, this circuit guards against small DC voltages, but does not increase the susceptibility of the protected device to large (>0.7v) DC voltages.

[amc184]

As for calculating the size of the caps, C1 (and C4) and the resistance of the primary forms a high pass filter. The frequency of the filter, which needs to be well lower than 50Hz, is given by:



Since R (resistance of primary) is likely pretty small, it is important to make C (value of C1 and C4) big. The linked circuit diagram says anything larger than 2500uF is allowable, which sounds about right.

[richie00boy]

Look at Rod Elliots ESP site, there is a recent article covering this very subject.

[ygg-it]

But the author just put 10 nF (10000 pF...


And here there are no cap at all....:


http://www.patentstorm.us/patents/57...scription.html


Can you check why?