In Reply to: RE: Let's reverse-engineer the PS Audio Noise Harvester posted by tomsyl1 on June 3, 2008 at 14:30:50:
Hi
You are right, a common mode choke is not the last word but….
From the standpoint of the equipment, one has a small hole in the chassis, through which passes 2 or 3 conductors. All the noise that can get in through the AC line, comes through that hole.
A cord laying somewhere is a conductive path of some length, running in some direction.
As such, it is an antenna also, a small voltage is imposed on the conductors (one end relative to the other) which is the sum of the RF field induced onto the conductors..
In this case, a shield has no impact on this induced signal as it is simply another parallel conductor on which the signal is induced..
Because the conductors are all very close together, the same Voltage signal is induced on each conductor, this signal is then “common modeâ€.
The common mode choke appears to be a very high impedance at RF and so the RF Voltage is not passed through effectively, it is a first order low pass filter.
Meanwhile, the AC power is a differential signal and so the choke effect is not present and a low impedance is presented for passing through power.
A common mode choke can be made more effective by having by pass caps to the chassis ground, now one has the series L of the parallel choke and then a C to ground making a 2nd order low pass filter.
Part “B†of the problem is that noise is also present in the differential signal, while not usually high frequencies like radio, it can have high components.
Here, the “obvious†solution would be to put a low pass filter in the differential portion as well.
As you suggest, one way this is done is using the leakage inductance between the primary and secondary of a transformer.
This appears as a series L, a first order low pass filter in the differential mode.
Where an output transformer is wound in layers, “multifilar†to reduce leakage inductance and increase the hf cutoff, isolation transformers are wound with much more physical separation. A typical normal power transformer in the 500W range (measured as a broad band device) will have a high frequency corner over 1KHz usually.
Modern designers sometimes seem unaware that a torroid transformer with a layered scatter winding is a wide band device and that isn’t always a good idea.
Anyway, if you think about and approach this in terms of filter design, with common and differential signals, you can design a filter to fix nearly any noise issue.
I had to design and build a PWM 100 Amp furnace regulator for a Shuttle flight experiment once (STS-7, STS-51a).
It was a real pain as the heating elements changed resistance by a factor of 5 as they heated, I could draw no more than 100 Amps (28V) and had to pass conducted and radiated emissions tests.
The bottom line is that it is the rate of change in current, which is the biggest problem in most noise situations, DI/DT.
Now days most of the toys around us are digital and radiate much more than “old time†analogue circuits. This is because they are digital, each transition from 1 to 0 requires a quick charge and discharge of various tiny capacitances, FAST changes in current.
Funny, analogue circuit theory still sets the upper limits of what can be done with digital.
Best,
Tom Danley
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- RE: Let's reverse-engineer the PS Audio Noise Harvester - tomservo 15:23:08 06/03/08 (6)
- How about balanced 120V power for noise reduction? - Atexanathome 18:22:12 06/04/08 (3)
- RE: How about balanced 120V power for noise reduction? - Dan Banquer 12:32:02 06/05/08 (2)
- There's a lot of truth in that... - Atexanathome 15:27:10 06/05/08 (1)
- RE: There's a lot of truth in that... - Dan Banquer 15:43:06 06/05/08 (0)
- RE: Let's reverse-engineer the PS Audio Noise Harvester - tomsyl1 16:05:02 06/03/08 (1)
- RE: Let's reverse-engineer the PS Audio Noise Harvester - chris_w 15:43:35 06/04/08 (0)