The AM Forum

   Wondering, pondering, cogitating: in traditional plate modulation, we vary the voltage on the plate; the plate current changes - per square law - and we can tune and load in the traditional way since the output resistance of the tube remains the same under modulation. In traditional "efficiency" modulation, as in screen modulation, we alter the current flowing through the tube but the plate voltage remains the same, hence it doesn't follow square law and thus the effective output resistance swings all over creation, so we have to load the output differently.
   Okay so far, but what about series modulation? With the series modulating tube "south" of the PA cathode - between cathode and ground - we modulate the current flowing through the tube by effectively modulating the series resistance, while leaving the plate voltage alone. Cathode voltage on the PA swings with reference to ground, but it's not cathode modulation because the control grid is referenced to the cathode. Reference the control grid to ground and it would be cathode modulation.
   With the series modulating tube above the PA plate, the plate voltage would seem to be modulated, since the series resistance between the plate supply and the plate is being modulated. But, I'd always assumed that the choice of putting the series tube above or below the PA tube was pretty much irrelevant other than considerations regarding what parts of the circuit you want to "float".
   I got to thinking about this because I like both series and screen modulation, and made an assumption that series modulation would have an advantage over screen modulation in that you wouldn't have to do any fancy footwork with the loading control to get full modulation, like you do with screen modulation. But then I started to second-guess that assumption.

If the cathode to ground voltage is varied, the plate to ground voltage is varied.

With a resistance between cathode and ground, we can swing the cathode anywhere we want (within reason), but if the plate is tied right to B+ through the plate choke, where can it go? RF wise, of course, it's swinging at your output frequency, but at DC?
(talking about the RF PA tube, of course, not the series mod tube)

Bill, I believe you are correct on all counts, except for second-guessing your assumptions.

Of course, if the RF amplifier grid is referenced to ground, and you vary the resistance between the cathode and ground, you have cathode modulation, which is efficiency modulation, with all its disadvantages.

If you reference the grid to the cathode, instead of ground, then vary the resistance between cathode and ground, you have series modulation, and the performance and efficiency of the RF amplifier are equivalent to standard plate modulation, identical in fact to varying the resistance between the RF amplifier plate circuit and the plate supply.  The RF amplifier receives a varying voltage between its negative and positive supply terminals, the same way it would with Class-B plate modulation.

The down-side is of course the poor efficiency of the series tube.  In order to reach 100 percent modulation, it is necessary to have more than double the plate voltage required for Class-B plate modulation, due to losses in the series tube.  Even at saturation, there is a significant voltage drop in the series modulator.  And it is not possible to have linear modulation if the tube approaches saturation; it is necessary to run the bias such that the tube operates in the most linear portion of the transfer characteristic.  With Class-B plate modulation, the modulator adds to, and subtracts from, the plate voltage.  The series modulator simply varies the available plate voltage, and the positive modulation peaks must come from the plate supply.  So even more voltage is required in order to exceed 100 percent modulation on positive peaks. 

But the great feature of series modulation is that it requires no expensive iron, therefor it can be designed for a greater bandwidth of modulation frequencies than can a Class-B implementation with the transformer and modulation reactor.  The excess heat produced by the modulator tube can be an advantage when heating your cold winter shack

With regard to practical implementation issues, it is easier to float the RF stage than to float the series modulator.  If the series tube is placed between the plate supply and the RF stage, then it functions as a cathode follower, and the grid drive circuit becomes more complicated.  Coupling the wide range of audio frequencies to the offset grid circuit, and controlling bias is not simple to design and implement.  If the RF stage floats, providing a high-isolation filament transformer is the major challenge, as well as proper insulating paths for the entire circuit.  Either Pi or Link coupling can be used for the input and output, easily handling the high voltage offset.

If the series tube is placed below the RF amplifier, it is rather simple to drive with another high-voltage tube, either a triode or tetrode, with direct plate to grid coupling and a plate resistor.  Then bias to the drive tube is varied to set the operating point.

I am currently testing the latter approach, using an RCA 810 as modulator, and RCA 8000 as the RF amplifier, with an Eimac 4-65 as the driver tube.  The final rig will be using a 3CX3000F3 for the series modulator, and I will probably use a pair of Eimac 304-TL bottles for the final, and either a 4-250 or 4-400 for the driver.  That tube is overkill as far as dissipation is concerned, but the voltage requirements dictate using what I have in inventory.

Bill, are you contemplating building a series modulated rig?  If so, tell us more about your plans!

  Yes, Rick, I figured my first assumption was correct, but then - as often happens - the more I contemplated on it the more I complicated it. So Steve is essentially correct - we're changing the voltage on the cathode; changing the voltage *across* the tube, and thus effectively increasing the plate voltage. A more round-about way to achieve plate modulation, adhering to square-law and maintaining a constant output resistance but without the weight, expense, and audio restrictions that come with mod iron.
   I started working on a small - much smaller than yours - series mod transmitter in the fall of last year, but - the bane of my existence - I get easily distracted, and dropped it for other projects. It's still down in the shop, on a pine-board, ready to take up again. I only got as far as the oscillator and driver: a 6AG7 oscillator running into a 2E26 driver, which will drive an 809 triode. I only need about 15 watts to drive my amp. Why an 809? I guess just because there's something sort of pure and uncomplicated about triodes, and they're pretty cheap and available. No screen supply, either. Got to make sure my filament transformer is well isolated, though.
   I love screen modulation, too. I've been running a DX-60 with the WA1QIX mod, and I absolutely love it - getting awesome audio reports. Fantasizing - don't know if I'll ever do it - of building a 3X 4-400A screen mod rig. Got most of the parts.
   I'm really looking forward to hearing about your results with your huge series mod rig - that's some big-bore hardware compared to my little 809 peanut whistle!

The two voltage sources are in series.

Thevenin stuff - sources become vague, reference points become crucial. Yes indeed.

William,   There are various efficiencies .... Are you talking RF, Modulation, input-to-output, or overall ( A.C. to Output )  ?     I ask because of the various classes ( A, B, C, D , and AB1 ).  All have various  efficiencies that add ( or subtract ) from the overall efficiency.

Don W4DNR

 

Talking about "efficiency modulation", the term typically used to describe modulation methods such as screen grid modulation, control grid modulation, and cathode modulation, which result in the effective output resistance of the RF PA tube varying with the modulation envelope, thus requiring output loading adjustments that give very poor efficiency and high dissipation under conditions of unmodulated carrier.

 Many years ago, I installed and maintained a Broadcast Transmitter that used a 4CX35,000 RF Tube and it was modulated ( turned on and off ) by another 4CX35,000 "switch tube" ( or modulator tube).   I don't recall that there was an issue with higher dissipation with the unmodulated carrier.    As plate voltage doubled, so did plate current.      I'm not sure about screen modulation, but I would think that if you loaded a tube for maximum peak amplitude / linearity  under 100% modulated tone , the tube dissipation at carrier would be equal to or lower.     

Don W4DNR

  Sounds like that broadcast transmitter was pulse-width modulated - that modulator tube is either on or off - so its dissipation would be minimal. If you're doubling the plate voltage on the RF PA tube, that tube doesn't really care how you go about accomplishing that; its dissipation should be the same as if you were using a conventional modulation transformer setup.
   With my screen modulated rig, I first tune for maximum carrier, then apply a triangle wave at 1khz and increase the loading (capacitor plate less meshed) until I can get nice clean triangles up to about 150% positive. Before increasing the loading, the peaks of the triangles are hacked right off.

Excellent procedure and attitude!  We all like to tune for maximum balls to the walls, but this is the proper way, sacrificing some heat and power output... with a little extra time and care for tuning.

And the same thing works with linear amplifier tune ups.  I think the main reason for splatter from linear amps is being too lightly loaded and of course, overdriving.  A ham has to bite the bullet with screen modulation and linear amps (grid/cathode modulation) and accept the heavier loading making it less efficient and blasting more heat. That's just the trade off.

When running class C tube RF finals that's why I prefer plate modulation or even PDM or series modulation. The ability to tune the RF final up at just about any loading level and still get a nice waveform and great RF final efficiency is worth the extra effort to me.

T

** BTW, stay tuned for the newest arrival; "Baby Blue," a single, self-contained, 110 pound 3-500Z linear amplifier with a baby blue panel and black lettering - just wired and ready for some testing and baby pictures, coming soon...

Wow, Tom, I can't wait to see it!  Must be very pretty if you call it "Baby...".   

Now 110 pounds sounds good for a 3-500Z amplifier, with a serious power supply - not like the typical wimpy desktop leenyar that will blow off the operating desk in a mild breeze!

  My screen-mod rig is only exciter class, so the dissipation isn't really an issue, but carefully increasing the final loading to get full and clean modulation adds one more step to the tuning process. I love my screen-mod rig, but regular plate-mod is less trouble tuning. And, of course, if I ever take a notion to build a big-bore legal limit class screen-mod rig, dissipation becomes an issue.
  When tuning my linear for AM, I use the adjustable output capability of my rice box to tune the amp for roughly 4X the carrier, and then fine-tune it with the triangle wave.
   Now, with one of your big 2X 813 plate modulated rigs, all that is unnecessary. Maybe one of these days I'll go that route - I've got a nice big 500 watt Thordarson mod transformer, some pretty massive power iron, and a handful of 813's and 810's kicking around in the shop. A few 4-400's, too.

Yep, after you perfect this rig then move on to an 813 plate modulated project. Next logical step. We were talking the other day how ham radio motivation is often centered around new projects and like-minded friends.  

I now have seven! discrete homebrew rigs... four AM plate modulated rigs and three medium to big linears....   three built in just the last six months. They are all operating FB and switch-selected, except for the latest Baby Blue.  I just love building stuff and dreaming of what it will look and feel like.  There's something about going into the cellar to round up old parts and creating a working entity - and making it look pretty. I am running out of room in the shack and parts, but want to build a 20-30 watt PissWeaker AM rig next... maybe this late fall or winter.

Rick, BabBlue is a desktop amp with rubber feet. Very compact - in fact I could have used a few more inches of room inside.  It uses a pressurized 4" X 4" X 4" sub-chassis (small welded alum box) with an attached blower to cool the 3-500Z tube  and chimney for very efficient air flow. Window view, of course.  Variac to slow the blower down. Roller inductor and you're right - the power supply transformer is a beast... about 40 pounds itself. It uses a fully tuned input with three meters for plate, grid and HV.   I think you will like it.

I'm looking forward to your 3CX-3000A7 and 304TLs series modulated rig.  That is a rare setup and very simple design to give flawless audio.

Keep building stuff guys!

T

Bill, you have all the hard to get stuff, now you need a building plan.  Be very careful with the older Thordarson mod iron, it cannot take a beating with arcs, best to plan spark gaps on the primary.  You should have a fine sounding rig with all that heavy metal!

Tom, I look forward to seeing the Baby Blue.  The challenge with the blower-cooled bottles is noise.  I want my 3CX3000 to be as quiet as the 304-TLs.  I am designing a non-metallic blower housing and ductwork to abate the blast. A DC-powered automotive air conditioner blower, just fast enough to keep it cool, and lots of acoustic insulation in a 19 inch rack cabinet 33 inches deep.  I am testing with convection cooled bottles, but the 3CX wants lots of air.

The audio processor project is first priority right now, and I am almost finished with layout and
QC for the PCB artwork.  Once the tiny stuff is done, out comes the heavy iron.

Hi Rick,

FB on the blower noise precautions using the 3CX-3000A7.  I once used one but it's fin structure was one of the louder external anode wind-noise generators I've used.  In fact, mine even whistled a bit due to air/mechanical resonance of some kind.    They do need a good robust air flow if not only for the filament power and idle current. (and talking real fast and short...  ;D)  You will be generating your share of heat in class A feeding those 304TLs.

My solution for my present big linear was to slow the blower down with a Variac and get the blower rebuilt at a motor rebuilding shop for new bearings.  I was down maybe -30 DB at best for on-air S/N.  It needed a noise gate on ssb.  That is not good enuff on AM but OK on SSB.  Baby Blue will be real quiet but not the bigger linear amps.  The class C plate modulated rigs are all OK for reduced noise on AM with the class C RF finals and class B modulators being in the middle of the road for heat.

I even looked into an acoustic barrier of glass and other materials, but the only real cure would be putting the big noise offenders into a different room with wall insulation like Ken/W2DTC did with his 3X3 rig. He has his in the garage and his signal is dead quiet in the background.  

T


This pic shows Ken's garage installation from pics he posted on the web.. It is loud and sounds very FB with excellent audio and a quiet background. I especially like the wheels for servicing:

  Rick, I plan to be very careful indeed with the mod iron, seeing as how they're not easy to replace and not easy to have repaired anymore now that all the rewinding shops are going out of business.
  I wonder if you could use gas-discharge tubes to protect a mod transformer? Problem with them is that you can only fire them a finite number of times. Of course, with good design you shouldn't be firing them very often, I'd think.
  Besides the Thordarson, I've also got a UTC VM-4. That's 300 watts, I think. I picked both of them up for very good prices at Nearfests over the years, but I've haven't yet taken the time to check them out - for all I know they both could be blown wide open.

  Rick, is that the "Max" processor that you're working on with Clark, N1BCG? I was talking with Clark a few weeks back, on 3885, and he was running a prototype. Sounded great. I'm really interested in seeing how that turns out!

  As far as blower noise, my big-bore rigs/amps will be cooled by a remote blower, located down in the basement, and the exhaust from the rig will be returned to the basement from whence it came via another pipe, thus keeping the room quiet and cooler at the same time. Haven't worked out all the details yet, such as which type of pipe to use (I hear PVC can be noisy), but I've got a huge bouncy-house blower that could probably literally blow the tubes right out of their sockets.
  As it stands now, I've mounted a "booster" fan in front of the OEM fan on my SB-221, which kicks in via a contactor when I key the transmitter. It makes a lot more noise than the stock fan, but my audio processor has a noise-gate, and I sit about eight feet away from the amp when I'm operating, so nobody seems to hear it.

Suck out of the basement, exhaust outside.  If you're running big tubes, eventually you'll heat the basement up.

It's turbulence that causes increased noise and lower flow.  Smooth bored pvc is good for channeling air.


--Shane
KD6VXI

Bill, that is the standard calibration method for air flow.  If the tubes stay in their sockets, then you don't have too much air flow.  I would do some experiments with the bouncy fan to see if it is way too much for your transmitter.  You have to contend with the mechanical noise of the blower motor, the high-pitched whine of the squirrel cage, and the rush of air moving through the duct work and tube.  Having a smooth duct wall reduces turbulence, and having a large diameter pipe provides the same airflow in CFM with much slower air travel, hence it is more quiet. 

I have a pair of 4-400s that I ran for years at legal limit, SSB and RTTY, with just a four inch muffin fan blowing air in a hole in the bottom of the chassis.  I cut out a hole in the desktop rack cabinet (a little larger than the 32V3) for fan clearance.  I used standard Johnson ceramic sockets for the tubes, and Coleman gas lantern chimneys around the tube.  Those tubes still run full power, in use since 1967.  The fan noise was not even noticeable, impacting neither receiving or transmitting.

I think it will be much more of a challenge with the 3CX3000, with 300 watts of filament power and dissipating over a KW when idle as a series modulator.  I plan to use a combination of sheet styrofoam and very open foam rubber to make the ductwork, very large air path, and lots of insulation and many small holes for exit air, hopefully muffling the sound of the air going through the tube fins.  I could use another 3CX3000 for the RF final, but for legal limit AM (50s style) it would be seriously overkill.  What's more, I want to keep the final plate voltage down to 2 KV idle, and provide over 4 KV on voice peaks, which requires a 5 KV supply.  Much more than that gets very hairy regarding insulation and safety.  The 304-TL tubes have such reserve emission that they play very well at 2 KV, a pair should be ideal, and they really need no additional cooling if open for adequate convection.  I could also use the 250-TH or 460-TH, but I already have an AM rig using those tubes.  4-400s require much more plate voltage, and need air flow, so the 304-TL will probably be the right choice.

Testing on a pine board now with the 810 or a pair for the series modulator, 8000 for final, and a 4-65 as a low current high voltage grid driver for the series modulator.  If that plays well, the driver for the 3CX2000F3 series modulator will likely be a 4-250 or 4-400.

Let us know what you find when you test your blaster fan and start building that big AM rig!

Yes Bill, that is the audio processor we are working on.  Thank you for expressing your interest.
I have been spending many hours revising the circuit layout based upon feedback we have received from folks using the prototype.  In addition to Clark, there are four or five people who have built the kit and are testing it and providing more input.  

Right now I am almost finished with the artwork QC, and expect to send gerbers for new PCBs in a few days.  We have gone to a proprietary microphone preamplifier chip instead of using the typical dual audio op-amp, to reduce standby noise to an absolute minimum.  The processor has a feature we call a "Gain Gate", which monitors voice level, and when the operator pauses, it holds the compressor gain, preventing the amplification rise that would normally occur when there is no input from the microphone.  But even with this feature, we want to make the microphone preamp as clean and quiet as possible.  

We have been working on the design for almost two years, and after the fourth board revision, we think we can make version five a winner.   If you would like more information on this endeavor, feel free to visit:

internetwork.com/MAX/

Availability of version five will be posted there in the near future.  Thank you again for your interest.

  Heating the basement is fine during the winter months, when I'm running a kerosene heater down there, anyway. In the summer, though, my dehumidifier already adds its share of heat.
  The problem with discharging the air to outside is that it would almost certainly cause a negative pressure differential between the basement and the outside, which could potentially effect the flue draft on my oil furnace and propane hot water heater; maybe even pulling combustion gasses back down the chimney and into the basement.
  I suppose I could pull from the outside and discharge back outside. Some sort of filter would be a good idea - nothing like cramming your RF enclosure full of insects.

  Thanks, Rick, for the link - I remember Clark saying that there was a website, but I must have forgotten to write it down.
  Very impressive indeed! I particularly like the fact that it's got both a limiter and a clipper. Right now I'm running a DBX-286S with my AM rigs, and while I absolutely love it - particularly the HF and LF "detail" enhancer - I wish it had a limiter and clipper.

Gotcha on the negative pressure diff.  We don't typically have that issue here in the left coast.  Almost everywhere I've worked at or lived we have the heater in the attic.

Problem with utilizing exterior air in the loop is moisture.  I've seen a couple TX that where decommissioned that the blower wheels where starting to rot, metal was rusted out, etc. from using exterior non-conditioned air. There is a reason well heeled radio stations used AC or chiller units!

Winter time seems you have the best option.  Summer, maybe it won't be an issue?

--Shane
KD6VXI

The  cooling requirements for a 3CX-3000A7  can be  greatly reduced IF the  ambient air intake temp is below  40 deg C  (104F).

The  67  cfm  @  1.2"   is  for  40 deg C.  (  104 deg F)
That drops to 56 cfm   for  30 deg C ( 86 deg F)
down to 51 cfm  @  25 deg C  ( 77 deg F)
Down to 47 cfm  @  20 deg C  (68 deg F)

The above is for  4000  watts  CCS anode diss.   Nobody is running anywhere near  4 kw ccs diss.  Even with 5 kw out, anode diss is only  2.4 kw...(68%  eff on 160-40m,  5 kv under load @ 1.5A).  So  actual air requirements are  even lower....much lower. (36  CFM for 2.5 kw CCS diss...with 40 deg C (104 deg F) intake air.

Drops  to 30  cfm  @ 30 deg C  (86F)
down to 27.5  cfm  @  25  deg C  (77F)
Down to 25  cfm  @  20  deg C  (68F).

A Dayton blower,  model  6DKY9  will  easily  do  80  CFM  @  1.2".   It's speed can be reduced  with a small variac, etc.
https://www.grainger.com/product/DAYTON-OEM-Blower-6DKY9

Yep, the Dayton blower has a great reputation for doing the job.

I tried to find one for my own big amp years ago but they were never available used on eBay and the new price was $300 to $600+... ridiculous.    But the current Grainger listing for $118 is quite a good deal.  I earmarked it for future use. A Variac to slow them down somewhat works well. I don't remember the actual Dayton model, but it was popular with BC stations and the big amp builders. The 6DKY9 looks ample enuff.

I once drew cooling air from a blower outside the house.  During the winter at 20F it was amazing to feel the tube air outlet at luke warm during full strap.  But the moisture eventually caused its effects.  Presently I rarely use a linear for AM. My AM rigs are all class C and the modulators p-p class B. This makes a big difference in managing air noise and heat.

T

Here is a suggested Cathode Modulated 809 circuit using a MOSFET.


Phil - AC0OB

Bill, unfortunately family and medical issues have significantly delayed the development of the new All-in-one MAX Audio Processor.   The good news is that the gerbers were sent to Shenzhen the end of last week, and the board manufacturing was completed on Sunday.  The copper-layered fiberglass artwork is now somewhere above the pacific pond, and will be arriving in Georgia sometime later this week.  With several hundred individual components on the PCB, it will take some time to populate an complete stand-alone and on-the-air validation of many significant updates.  Once those tasks are completed, I will be making more information available to all those folks who are interested.

Rick, thanks for the update; I truly hope all is well regarding family and medical issues!
I keep meaning to contact via the website regarding specifics that are more appropriately discussed there, like pricing and all that, but something else always comes up. I am very interested and looking forward to getting more information.

Looks like a fun board to build. All thru-hole - it's been many years since I've populated a thru-hole board that size. Pretty much everything I do at work is SMD.

Bill, we very much appreciate your interest in this project.  We will not be putting anything new on the web until we run it through its tests and make sure it is ready for prime time. 

Thanks also for the well wishes.  By the way, it does have just ONE SMD chip, just to keep you feeling comfortable.  Perhaps in the future we may produce an all-SMD version, with the components applied by the PCB fab facility, but right now, the through-hole gives many folks the opportunity to put it together and take pride in their accomplishments; it also has lots of options for customization, which would not be quite as open with SMD. 

It is not really a kit in the Heath or Elecraft sense, as there are no step-by-step instructions for each individual component.  But the instructions do group all the tasks to make them run smoothly, and a spreadsheet of all the components is used to easily populate all the resistors, capacitors, etc.  Sorted by component value, it is very easy to insert and solder all like values as a group, so it does go quickly without undue stress.

If you would like more details or have specific questions, use PM or email; I am good on QRZ, and we can even use the landline (cell) if need be.  We will post on the forums when new information is available on the web.

73, Rick

internetwork.com/MAX/

   Rick, personally I think that making the board thru-hole is fine business; many hams are vastly more comfortable with it, and it makes modding easier. I'm comfortable with either, really. Everything I scratch-build at home, out of necessity, is thru-hole, since putting SMD parts on a regular proto-board is problematic, and spinning SMD boards for one-off projects isn't practical. Maybe someday I'll try that, though: prove out a circuit on a breadboard; have a board-house spin up  five or ten boards or so, then keep a couple for myself and try to sell the remainder at hamfests.
   For those who've never done any SMD work, it's not nearly the horror that you'd think, at least as long as you keep the parts on the large side; 1206 resistors and caps, and even 0805's, are a breeze with only modest optical assistance, like the standard magnifier lamp, or even with naked eyeballs - my eyes are sixty-four years young, and I can solder the larger SMD parts by eye with no problem (although I prefer the scope). The larger pitch IC's are fine, too; many small signal FET's and bipolars, and diodes, are available in SOT packages that are easily manageable under the magnifier lamp.
   SMD can be faster, too; put a dab of solder on the pad, stick the part on with the tweezers and hit it with the iron, and you're done, no messing with lead-formers and then making sure the part doesn't fall out when you turn the board over to solder the other side.
   Sooner or later, for better or worse, SMD will be forced on us all; I'm finding it increasingly difficult to find the parts I want in thru-hole packages.

The boards were not lost while traversing the Pacific Pond.  Time to sniff some solder smoke and do the final QC testing.   Stay tuned!

There will soon be updates here:

www.internetwork.com/MAX/

Very nice looking board, Rick. Curious: you fan-out the traces from the SOIC-14 to plated thru-holes; is that for easy access to the inputs/outputs of the chip? Would something like a Berg header go into those PTH's for some reason, or maybe a DIP socket just in case someone came across a thru-hole version of the chip?

Thanks for the kind words, Bill.

Some folks may not want to solder in the SMD chip, but it is really not that hard.  We can either install the chip on the board for them, or a carrier may be plugged into a couple Dupont 7-pin female headers to address that concern.  I have a couple of the DIP versions of the compressor chip, but the spacing between the rows is too wide to handle that option. I suppose a carrier could be made for a through-hole DIP, but the SMD is less expensive.

It is also very handy to have easily accessible test points for the chip, not only after assembly, but it makes it very easy to verify the pins on the SMD  are properly soldered to the pads during assembly. 

None of the calibration procedures require measurements at the chip, but for troubleshooting it is much easier to place a probe in the through-hole than to balance it on one of the chip leads.  At my age that is almost a necessity!

Later on, we may shrink the board and go all-SMD, but right now that kinda takes the joy out of the kit-building experience.

  I would say stick with thru-hole until component availability forces the change to SMD. You'll most certainly sell more kits that way, and space saving isn't really a big concern with this sort of station accessory.  A while back I built a 2 meter transverter as a kit (Down East Microwave). It was all SMD, but for VHF stuff there's a definite advantage with SMD.