I think the best long term solution for <100 Watts and limited space (but not severely limited) is a remote automatic antenna tuner at the feed point of whatever length wires you can put up.  it doesn't have to be perfectly symmetrical (or even close). There are some "particularly" bad lengths (you don't want the overall length of the antenna to be exactly 1 wavelength), but other than that, you get all band operation, coax feedline, minimum losses in that feedline (since it's always matched).    You can also put multiple wires of multiple lengths up, which will help the pattern on higher bands, if nothing else.

As others have noted, the other choices either have losses in the antenna (and how ever much loss there is in the autotuner, it's probably not 6dB, like the typical Terminated Folded Dipole) or require a low loss feedline and a tuner at the shack.

I've fooled with a variety of "no tune" or "multiband" antennas of one kind or another, and they're fine if you've got more time than money, because you're going to be doing a fair amount of fiddling around to get them to work right.

power density isn't as troublesome in the magnetics though. You can always get bigger cores, etc.   The problem with the transistors is that the dice are just so small, so that's where the big thermal issue is.

(After all, there are millions of liquid cooled transformers around already)

This is incorrect.  In fact, black things radiate better than other colors, although there isn't much difference.  What you're talking about is a property called emissivity, and most painted surfaces run somewhere around 0.9, regardless of color. Shiny things, though, have low emissivity in general (0.03 for silver and gold)

You need to be very careful, too, about comparing properties  in the visible spectrum (e.g. solar collectors) and far IR (where heat sinks radiate).  Well designed solar collectors are actually designed to have high absorption in the visible, but low radation in the far IR (heat).  Black paint is actually non-optimum for a solar collector.

Interestingly, shiny metal actually radiates worse.  (this is why those seat belt latches and shiny metal railings get hot in the sun..)  It has to do with the ratio of absorption and emission (so called alpha over epsilon ratio).

here's a table of a, e, and a/e: http://www.solarmirror.com/fom/fom-serve/cache/43.html

All that said, heat sinks generally cool by conduction to the surrounding air, not by radiation.  Even if there's no fan, they rely on hot air rising (fanless heatsinks should have the fins oriented vertically).  For conduction, thinner paint is better.  Most aluminum isn't painted, it's "chemical conversion" treated: anodizing, alodining, etc.

Most Tesla Coils (esp big ones) are in the 100 kHz range.
And these aren't spark gap transmitters.  They are basically a solid state power oscillator (usually using big IGBTs) driving an RF transformer, and actually have fairly well controlled RF spectrum. (google for DRSSTC)

However, given that they are operating in the 3km band, the "antenna" is pretty small and inefficient.

I've done quite a lot of analysis and tests on this kind of thing and the dominant RFI from them is actually not from the fundamental, but is the transients from the spark.  An individual spark is discharging  the capacitance of the top terminal (some 10s of pF, at a few hundred kV) over a loop that's several meters in cross section.  That is a pretty decent di/dt and generates a very nice waveform to kill victim equipment that's in the area.

Right now, liquid cooling is pretty complex and very individual design sensitive: that is, every design needs a different cooling design, either from a thermal standpoint or a packaging standpoint. That translates to expensive.  What's needed is something that is mass produced to meet a consumer need, which we as amateurs can then repurpose to our own needs.

There is some hope because the increasing power density the CPU and GPU world means that just about every laptop made these days has a heatpipe in it, and there are quasi-off-the-shelf liquid cooling schemes for PCs.  The cost of chillers has has also come down: either using Peltier devices (very inefficient) or just because of the inexpensive off-shore-manufacturing of simple mechanical devices (kegerators, bar fridges, etc.). You see chillers for aquariums at remarkably low prices compared to what we used to pay for a standard lab chiller.

Sooner or later some inventive ham is going to start looking at how do you bolt a set of FETs onto the "coldplate" of a CPU chiller, and still get the RF part of the design to work ok.  And gradually, the articles will change from "how to build a chimney and blower for your power grid tube"  and "how to make your own socket for that broadcast pull" to "how to turn an icemaker into a chiller" and "how to safely bend the heat pipe on a laptop CPU chiller".

What engineering basis do you have for recommending more rods?  As you point out "installing them too close... is poor economy", and I think the same would apply to number as well as spacing.

Not that more rods isn't better (it is, in general), but where is the cost/benefit tradeoff? 

Let's look at the physics.  There's three reasons to connect to the soil:
1) RF ground for an antenna where the soil is part of the radiating system
2) Electrical safety, in case a power line or other live conductor comes in contact with the antenna
3) Lightning energy dissipation

Ground rods in the middle of the bonding run do nothing for #1, because the AC resistance of the wire connecting them is sufficiently high at amateur HF frequencies that it's like hooking a big resistor between the "ground" and the antenna.

Ground rods in the middle of the bonding run do nothing for #2, because the DC/60Hz resistance of AWG6 (<0.5 ohms in 100 ft) is so low that the voltage drop along the wire is small in the event of a fault, so that all the equipment stays close to "bare feet" potential. (one of the reasons for NEC grounding)

Ground rods in the middle of a bonding run don't do much for #3, because the inductance of the 20 feet of wire to that middle rod is about 6-7 microhenries.  At the nominal 1 MHz for lightning, that's an impedance of 18 ohms, and at, say, 1 kA, that's a voltage drop of 18 kV.  Buzzing along to the next rod, the current is less, but the inductance is more... The net result is that the added rods don't add much to the effectiveness of a lightning ground.    If you're worried about reducing the impedance of your lightning ground, drive more rods near the antenna, or even better, bury some radials, or, if the antenna has a concrete base, make sure it's well connected to the grounding system.

Take home message:

The bonding rules aren't for lightning or RF.  They're for line frequency faults and shock hazard.  From that standpoint, it turns out that the bonding really doesn't need to be connected to "earth ground" very well at all. (and in fact, the NEC code makers are gradually changing the wording from grounding to bonding)

Long wires (unless buried) make terrible additions to RF grounding system.

Lightning is all about keeping the strike energy somewhere other than delicate stuff,  and making sure that all equipment goes up and down together.  The currents are so high and the rise time so fast that trying to keep the voltage "low" is impossible.

Ground rods worked great for Ben Franklin, because they replaced *nothing*.  But there's a lot of better ways to do it now, and we know a LOT more about the underlying physics and applications.

the answer to the question is the classic "it depends".

If it's a SDR approach where you're basically feeding bandlimited raw samples to a PC for processing, then USB is probably easier.  If it's an approach where you have some "smarts" in the box between sampler and interface, or where you might want to have multiple PCs and multiple sample streams, Ethernet is going to be more appropriate.

If it's an SDR (a'la Flex) where you are leveraging inexpensive high volume gear (audio interfaces), then you want whatever the inexpensive audio gear uses, today, that's USB 2.  Who knows where it will be in the future?  Thunderbolt? USB3? something new?

People have been professionally building, using, and fixing tube amps for almost 100 years, and solid state amps for a bit less than half that.  Hams are running somewhat behind that.. I would say that the fraction of hams who have built and modified and fixed solid state amps is perhaps a few percent of those who have done the same with tube rigs.

In another 30 years, I suspect it will be different.

...in 10 years.

There, in a nutshell, is the problem faced by anyone selling to the amateur radio market.  There is an expectation of 30-40 year life. Something that really doesn't exist in other industries, except perhaps spaceflight, and we all know how inexpensive it is to build spacecraft.

Well, yes, a chiller is a decent approach, but that's a big step: from cooling based on transfer to ambient environment (heat sinks) to mechanical refrigeration.  But maybe that's what it's going to take to get the heat out of high power density devices.  For all we know, in 10-15 years, you might be able to get it all as a reliable packaged assembly, sort of like ThermoElectric coolers today.

you can shorten your zip code search by using the first few digits.  900xx are all in LA, for instance.  911xx are Pasadena, and so forth.  (there are zip codes that span city boundaries, though, e.g. 91020, aka BHPO in real estate ads, so it can get a bit complex). 

I think there's someone who has some scripts that geocodes the ham license database and runs stats..

perhaps not so oddly, most ground rods are driven next to the foundation, because you want the path from "thing being grounded" to "ground" to be as short as possible.  Putting it next to the foundation also means that it's less likely to be disturbed (by mowers, brush trimming, painters dragging ladders, who knows what all). That doesn't change that ground rods, as a grounding means, are less than ideal.

Most of us don't have infinite money and time, so it's worth spending those resources wisely, in accordance with sound engineering and physics principles.  There's a fair amount of  bad advice out there.  There's also a whole lot of misconceptions out there that while not actively bad, cause people to spend time and effort in ways that are ineffective. Some are drawn from applications which are similar, but different in important ways (lightning grounds are not HF antenna grounds are not 60 Hz power line grounds).  Some are historical, are not recommended in light of modern knowledge, and deserve a peaceful retirement. 

Since you mention the IEEE, they actually have several books on grounding, but the so-called "emerald book" (IEEE-1100) is the one most applicable to hams.  It's fairly straightforward, has been reviewed by people whose life work is in the field, etc.. It covers all the kinds of grounding and bonding.  It's pricey to buy, but you can request it from your local library.

For free, I'd recommend the Mike Holt Low Voltage Grounding book.  Download it from http://www.mikeholt.com/ and it tells you pretty much everything you need to know about grounding for amateur radio installations (and low voltage control, like that rotor control) from an electrical code standpoint.  The actual electrical code (NEC - NFPA70) from the NFPA  isn't readily available on the web, but Carl Malamud has most of the codes online at https://public.resource.org/index.html or more specifically at http://bulk.resource.org/codes.gov/   (Florida doesn't have an electrical code apparently, but if you look at California's, it's basically a copy of the NEC).  It's just a giant pdf, and doesn't have all the nice notes and links you'd get if you forked over the $125 to NFPA.   You're interested in Article 250 (grounding), and the 800's (all the antenna stuff).

If your insurance requires it, then NFPA 780 is the standard for lightning protection.  There's a copy out on the web at New Mexico State University, I think.

And, of course, the ARRL handbook does have a section on grounding, recently revised.
 

on the other hand, depending on where the vapor barrier (if any) is, the concrete in the basement wall might be a good low impedance grounding connector, particularly if it has rebar
if your soil is rocky and dry, don't fool with ground rods, which have only the virtue of being easy to install, but aren't a very good way to connect anything to the soil.  Use a Ufer ground (concrete encased grounding electrode), specifically developed for dry rocky soil by Herb Ufer.

you might ask yourself why many jurisdictions these days don't allow rods to be the sole grounding means, and actually require a CEGR or simlar.

There is no such thing as "saturating" the ground around a rod.  Yes, if the current density is high enough and the stroke lasts long enough, you'll get a "smoking rod", but that's more likely to happen with something like a utility power line fault to your system (where you can get kiloamps for many minutes).

Megawatts means nothing.. what you are really concerned about is dissipating kilojoules or megajoules.  A megawatt for one microsecond is only one joule (about the energy from dropping an apple a meter).   Lightning strokes last about 50 microseconds.


You're confusing a lot of things here. Rods are cheap, but aren't a very good ground. Use them if you have nothing else.  Protecting your electronics doesn't depend on ground rods, it depends on good system design: making sure that lightning current doesn't flow "through" your electronics.  That's more a matter of paying attention to where things are connected (you want to the same point) rather than whether they happen to be at the same voltage as the soil.

First part is fine. the code requires (for good reason) that all grounds be bonded together (typically AWG6, but there are other ways).
Connecting your antenna ground to the power line ground doesn't necessarily help with transients, but it is required for other reasons.

A few comments..

bend radius isn't all that critical, don't obsess.  There's no good physics reason for bend radii greater than, say, 1" or so.  The code just says as direct as practicable.  Avoid a loop or half loop, though. Electromagnetic stresses will cause issues.

Pay attention to distance of lightning ground from "other stuff".  You don't want it running next to something important (like your coax?)

Strap isn't all that much better than wire for lightning or electrical safety ground.  If you were grounding a 40m vertical, strap is it, but other than that, other things dominate.

All grounds have to be bonded together with AWG6, if you're concerned about code compliance. The inspector wants to see round wire, not tubing, not strap, not braid, not anything other than round wire at least AWG6 or bigger. (there are weird exceptions if you're running the bonding conductor in a conduit and stuff like that)

A bigger question is how you're bringing the coax inside.  Code requires that the shield be bonded to the grounding system at the point of entry.  Good practice says all the "chassis grounds" of everything should be tied together, which implies that your AC wiring ground (green wire) should be connected to the same ground point as your coax shield, etc.

get yourself a free copy of Mike Holt's low voltage grounding book (it's online at the Mike Holt site).. it summarizes all the electrical code requirements nicely, so at least you'll have that part covered.  It doesn't deal with lightning rods or NFPA 780 grounding, if you need that (are you in the 4 call area? the southeast is notorious for high lightning rates.. compared to the very quiet 6 area, for instance)

Yes. although I went out and looked at mine.. The electrical service comes up in one stud bay where panel is, and all the rest come up in the next one over, where the cable TV hatches and the phone box is.  The ground wire from the slab (Ufer ground) comes up with the telco and cable tv, and goes through a hole in the stud to the electrical panel.

So you might need to do a bit of cutting and exploring.  Fortunately, you can get nice looking panels to cover the holes OR it's time to learn how to patch drywall.. It's pretty easy, and if it's in a garage, nobody will notice.  It's a very useful skill to have (putting a patch in, using the joint compound/spackle, and spraying it with the texture glop from a can.)  After that, you'll never fear changing a single gang to a double gang box inside, for instance.