Jim,
Thanks for your comments. The entrance panels for the two buildings
have separate meters so they are correctly installed with local bonding.
NEC, if I understand it correctly, does clarify one important point,
that is that permanently installed outdoor generators can be connected
to an auxiliary grounding electrode . Even if I didn't have a separate
electrode the gen frame is bolted to a reinforced 60 sq ft concrete pad.
So I have 3 code correct grounding electrode connections.
My logic for the tower shed panel bonding was to shunt to the radial
field a large amount of common mode RFI (S9+40) from PWM 240 VAC mains
amplifiers driving the rotator 180v DC 1/2hp motors. Even though the
16KHz PWM amps are 240v sourced they have a bolted frame connection.
These amps use IGBT's that switch really fast. I have a shielded
isolation transformer between them and the panel and balanced 50KHz
cutoff filters I designed installed on the output of the amps. This
bonding is not per code, and if there is downside, I'd like to know what
that is. Fortunately, all this suppresses the rotator moving RFI to
S1-2 for 20m and above, but I need another 60+ db on 40m. The challenge
is a filter design that doesn't cause the PWM amp protection circuits to
trip. This problem is a drift off topic but if others have solved it,
I'm eager to learn. The configuration is Green Heron controllers with
remote power amp turning K0XG rings. The rotator cable is inside the
tower, 16ga twisted pair, not shielded per manufacturer's advice (bad
advice, they should be shielded). The 180 VDC motors are filtered at
their frames and can barely be heard on all bands when driven from a
clean DC source.
Grant KZ1W
On 7/26/2016 13:41 PM, Jim Brown wrote:
Hi Grant,
I won't quote your entire post, but only some key elements that need a
comment.
On Tue,7/26/2016 11:21 AM, Grant Saviers wrote:
"Ground loops" are an interesting question. The problem is a
"single point ground" is often not feasible and is impossible at my
QTH. My decision was to connect everything together.
In general, that's the only right (safe) way to do it.
At my QTH the conductor paths are:
1. Buried HV feed to mains transformer which has a Ufer vault ground,
which then feeds
2. 200a house service with 2 ground rods at entry
3. 400a shop/shack entry panels with 2 ground rods and bonding to
structural steel
4. Charger and heater at a backup generator on a concrete pad,
which then feeds back underground to 200a transfer switches at house
and shop, next to mains entry panels
5. The shop/shack foundation is a perimeter Ufer and the structural
steel is grounded to it, also the main shack ground is to the Ufer
6. The 3 towers have Ufer bases and ground rod+radial fields
7. The shack coax entry panel is bonded to the steel structure, coax
is bonded to towers top and for the largest tower to a shed at base
entry panel, all cables are in buried conduits to the shack
All of this is REALLY GOOD
8. Underground control and coax to a 160m wire vertical T with a
ground rod and shunt inductor from 8 elevated 125' long radials
9. Ethernet, RG6, and control cables are in conduit from shop to
house (not protected at either end - needs fixed)
Yes, this is a recipe for destruction, needs serious attention.
10. The Comcast cable entrance is tied to ground rod at house,
telephone entry is there also, so the house on its own is near single
point grounding.
11. The equipment shed at the base of the largest tower has the coax
entry panel bonded to the rod and radial field. The shed mains power
panel is also bonded to the radial field.
Thus, there are a large number of "loops".
Of course. The concept of "ground loops" is fundamentally WRONG, and
causes us to do dumb things. The ONLY context in which a loop is a bad
thing is magnetic induction.
This is probably more complex than most, but I think it is not
uncommon to have loops. Code requires much of the above, and white
(neutral) is connected to yel/grn (earth) at several places.
TRAIN WRECK! Good engineering practice, and virtually ALL building
codes in NA, require that neutral be bonded to ground ONLY at 1) the
service entrance (where power enters a premises); and 2) where a "new
system" is established. A new system is established by a transformer.
It is NOT established by a feed to another building from the main feed
to the premises. The ONLY time there should be more than one bond in
two buildings is if 1) the two buildings have their own service (that
is, a separate metered connection from the power company; or 2) if the
second building is fed from the first, and ground is not carried
between the two buildings. This second scheme is no longer permitted
by NEC
SO -- if you have neutral bonded to ground at more than one place, you
need to change that. It's a VERY big deal.
So my strategy was to bury a large ground conductor below the conduit
runs when I could, to tie stuff together. I didn't do this initially
but wish I had as later DC ground resistance measurements showed the
buried wires were about 1 ground rod equivalent at about 100' of
buried bare copper wire (#6). I note that the Andrew lightning
protection guide advises against tying towers to building entry
panels with a separate buried conductor. Rather they let the voltage
surge be equalized on the coax and control wires, I think the theory
is that the differential voltages are less as a result. If my code
knowledge is correct, it requires towers to be bonded to "house
ground", if there is AC at the tower as is true for my motorized
crank ups.
NEC requires that the power system ground MUST be carried to all
outlets and loads, and it MUST be carried with the phase and neutral
conductors (in the same conduit or other cable).
I think my shop foundation Ufer plus structural steel frame makes a
low inductance path from the shack on the opposite corner from the
mains entry and transfer switch panels. The tower foundation Ufers
and rod/radial fields are other "good" low ohms and L grounds with
lots of surface area. The code required pair of mains entry panel
rods are poor grounds in comparison (180 sq ft of concrete per tower
in earth contact vs 0.1 sq ft per rod. 1000 sq ft of concrete
surface in the shop Ufer).
So if "SPG" means a single point ground at shack entry panel for
coax, control cables, and rigs, then what I have might qualify.
Otherwise, it is impossible to achieve.
Yes, that's the real world.
An "RF ground" for lightning as the Andrew guide explains is a ground
rod + radials field that distributes the RF energy of a strike over a
large area both capacitively and via conduction.
Erase the words "RF ground" from your memory bank -- it is a fiction
that has no meaning. Connections to earth, and bonding between those
connections is ONLY for lightning safety. The only relationship
between those connections and "RF" is to realize that lightning is an
RF event, not a DC event, so that the impedance of those connections
and that bonding at RF, which is where the energy in lightning is, is
what matters.
So the 5 ohm DC ground resistance target has merit as does the radial
wire and ground rod field size. Both are needed.
A radial field serves as a low resistance return for antenna current,
in place of earth, which is a big resistor. That radial field SHIELDS
the antenna from the lossy earth. And from the point of view of
lightning protection, if it is bonded to facilities grounds, it
provides capacitive coupling to the earth to reduce the impedance to
earth at RF.
While buried conductors benefit from the shunt earth conductivity,
the wire inductance limits the useful length of the radials to about
50ft according to Andrew. The K1TTT analysis shows why tower top and
bottom coax shield bonding is needed and why elevated coax should be
avoided if at all possible.
Yes on all counts.
Fortunately, Western Washington has a very low strike frequency, but
with a tower top 40' above the 110' tree line and on a ridge, my
attention to lightning protection has significantly increased.
Lightning enters our premises on MANY conductors -- power line, TELCO,
CATV -- as well as antennas.
"There is no such thing as ground" from Vonada's Engineering Maxims.
Now we're getting closer to the real world. :)
73, Jim K9YC
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|