[TenTec] On Switching Power Supplies

[Jim Brown]

On 1/18/2014 12:34 AM, Rick - DJ0IP / NJ0IP wrote:
> And even the oscillations are often caused by grounding issues!
>
> If the regulator's ground connection is not VERY short and identical with
> the ground of the electrolytic capacitors, oscillation can occur,

Yes. These are the circuit layout errors of which I wrote. EMC expert 
Henry Ott (a ham, of course) talks in his tutorial workshops of the 
importance of keeping track of where the current is flowing, and he 
talks about "the invisible schematic hiding behind the ground symbol."  
And he means ALL of the current, not just the intended current.  In a 
return circuit, DC and RF often follow a very different path. DC follows 
the path of the least resistance.  AC follows the path of lowest impdance.

As a simple example -- when setting up a station on a county expedition 
for the California QSO Party a few years ago, I noted that one of our 
team had run a single conductor from the rig to the power amp for the 
keying line, relying on the shield of the coax carrying the RF drive to 
serve as the return. His logic was that since it was only DC, it didn't 
matter. What he forgot was that there are antennas around, radiating 
500W of RF, so in addition to that DC, RF was also present, and the 
large loop consisting of the single wire and the coax shield functioned 
both as a receive antenna and as a magnetic loop to couple the RF to the 
rig.

Another point with respect to power supplies and "grounding" -- power 
supplies, both linear and switchers, often suffer from "the Pin One 
Problem, and both designers and users think a connection to dirt is 
necessary for things to work right, be noise free, and so on. Nothing 
could be further from the truth. The sole function of a connection to 
earth is LIGHTNING PROTECTION -- the earth is NOT a sink into which 
noise, RFI, and so on can be poured. There is no good reason to bond DC- 
to the chassis of a power supply, and there are very good reasons to NOT 
make this connection.  On the other hand, for lightning protection and 
electrical safety, both good engineering practice and the electrical 
building codes in most of the developed world require that the chassis 
of anything connected to the power line be bonded to the third wire in 
the power cord -- "the green wire" in the US, the protective earth in EU 
-- and that all "grounds" in a premises be bonded together by short, fat 
copper.

Most of the noise we associate with power systems is carried on the 
green wire in the form of leakage current, and without proper bonding, 
this noise (hum, buzz, RFI, even broadband noise) can show up in our 
equipment.

Several years ago, I opened up the three big Astron linear supplies that 
I owned (I've since sold one) and studied their bonding. None were 
properly bonded. The green wire from the power cord was soldered to the 
mounting leg of a terminal strip, which was insulated from the chassis 
by paint. As if that wasn't enough, the two major sections of the 
chassis were also insulated from each other by paint.  This 
manufacturing defect is VERY common. I've seen it in older Elecraft 
products (they have long since fixed this), in N8LP's products, and it's 
been widely reported in the controllers that SteppIR provides for their 
antennas.

73, Jim K9YC