New project: Buzzardly 6 meter AM Amp.

[KD1SH]

  I've got a handful of brand new 4CX400A's and GS36B's hanging around, and I do a lot of 6 meter operating, both AM and SSB, so I figured I'd put some of those tubes to use. This is the progress on the RF deck so far. Note the spiffy brass anode clamps, courtesy of my milling machine. The Teflon chimneys were inside-bored on my lathe from a solid 2.5" round bar. Somewhat wasteful of material—made a huge pile of Teflon fluff—but I couldn't find any suitable Teflon tube. I'm thinking I'll wind up replacing the separate anode straps with one contiguous one—I don't really know motivated me to make it in two pieces.
  The B+ supply will be external, but the screen, bias, and filament voltages will be supplied internally, with provisions to kill the screen voltage should the B+ drop out for some reason. It'll be cathode driven. Cathode impedance is around 120 ohms for one 4CX400A, so two should give me around 60 ohms, but I'll give a tuned input, just for good design practice.
  It's somewhat overbuilt, appropriate for AM use: I used #20 wire on the plate choke, where #24 would have been sufficient, and the blower I dug up provides 0.3" H2O pressure differential on the Magnehelic, where Svetlana specs 0.2" for the full 400 watt dissipation per tube. I've got a larger blower that gives me 0.4", but at the expense of more noise and vibration.
  More pictures to follow as the project progresses.

[w8khk]

That is some extremely fine workmanship, Bill!  It looks much finer than many commercial implementations.  The value of having a well-equipped machine shop, including milling and lathe equipment, should never be underestimated!

I see nothing wrong with the two separate plate bus components.  They allow for slight alignment differences in component positions.  I would not change anything!

Please continue to share more photos of the development process, they are very motivating and informative.  I am sure it will be a pleasure to operate, once the project reaches completion.

[K1JJ]

 .... and the blower I dug up provides 0.3" H2O pressure differential on the Magnehelic, where Svetlana specs 0.2" for the full 400 watt dissipation per tube. I've got a larger blower that gives me 0.4", but at the expense of more noise and vibration.
  More pictures to follow as the project progresses.

Good show!  A single bander makes getting on the air with no tuneup a breeze. You will use it more often.

The only problem you may have is air noise.  Those smaller anodes like 4CX-300, etc., can make our signals sound like a machine shop in the background.  At full rated power / air, it is real important to locate this class of tube amplifier far away from the mike. A uni-directional mike will help as well as a noise gate.  But there is nothing like distance. The drawback is tuning the amp by having to get up and move around.  

In contrast, I run my homebrew YC-156 linear (about 5" diameter anode) with very low air noise using a Variac. Running it well below its power ratings can allow a much quieter air flow without exceeding rated seal and element temperatures.  

Rick has done some work with sound proofing the tube air system.   Wat sa Rick?

T  

[KD1SH]

  Thanks, Rick. Full disclosure: I'm a lifelong electronics guy who sometimes pretends to be a machinist. I've got a full-sized standup vertical mill and an old—probably older than me—South Bend lathe, but I'm no kind of real machinist. A number of years ago, I'd mentioned on my resume that I could run a milling machine and a lathe; not intending to imply that I was actually employable at it, only that I was versatile. My boss at the time came into the lab with a box of metal stock and some drawings and said, "Hey, Bill, can you make me a hundred of these parts? The regular machinist is out today." They were somewhat complex parts, with lots of cuts, angles, finely finished surfaces, and even a hole for a tight running-fit shaft. I had to explain to him that, sure, I could do that, but it would take me ten times longer than the real machinist could do it, and he'd be back long before I finished. Over the next few years the boss would come to me often for quick one-off and very small quantity stuff, but fortunately he understood that I wasn't a real production machinist who could quickly turn out quantities of precision parts.
  I got spoiled using the machine they had there, though: a Bridgeport with ball-screw power feeds and digital readouts, as well as the nice variable spindle speed crank on the head. Mine just has Acme-thread lead screws and dial wheels, and step-pulleys for changing spindle speeds.
  But yes, even though I'm just a home workshop hack, having those machines, and some basic skills, is a real blessing.
  And I probably will keep the separate plate straps. Like you say, they allow for alignment differences, and one tube did wind up about .05" closer to the plate choke terminal than the other—just the way the tolerance stack went.

That is some extremely fine workmanship, Bill!  It looks much finer than many commercial implementations.  The value of having a well-equipped machine shop, including milling and lathe equipment, should never be underestimated!

I see nothing wrong with the two separate plate bus components.  They allow for slight alignment differences in component positions.  I would not change anything!

Please continue to share more photos of the development process, they are very motivating and informative.  I am sure it will be a pleasure to operate, once the project reaches completion.

[KD1SH]

  The small blower I chose is impressively quiet and smooth, but I'm going to mount it on vibration dampers, just as added assurance that it won't use the chassis as a sounding box. But, I haven't ruled out using a larger blower and making the speed adjustable. This little blower exceeds the need, according to Svetlana specs, and I suspect that there's a rapidly diminishing returns curve as you increase the flow. Once I get the amp running, I'll be looking at the anode temperature with my IR thermometer to see what benefit the beyond-spec pressure actually gives me.
  My usual operating position is a good 7 to 8 feet away from the actual equipment. On HF AM my microphone is on a boom-mount attached to my desk on the other side of the shack, and on other bands I use a microphone with an 8 foot cord, still sitting in the same chair.

 .... and the blower I dug up provides 0.3" H2O pressure differential on the Magnehelic, where Svetlana specs 0.2" for the full 400 watt dissipation per tube. I've got a larger blower that gives me 0.4", but at the expense of more noise and vibration.
  More pictures to follow as the project progresses.

Good show!  A single bander makes getting on the air with no tuneup a breeze. You will use it more often.

The only problem you may have is air noise.  Those smaller anodes like 4CX-300, etc., can make our signals sound like a machine shop in the background.  At full rated power / air, it is real important to locate this class of tube amplifier far away from the mike. A uni-directional mike will help as well as a noise gate.  But there is nothing like distance. The drawback is tuning the amp by having to get up and move around.  

In contrast, I run my homebrew YC-156 linear (about 5" diameter anode) with very low air noise using a Variac. Running it well below its power ratings can allow a much quieter air flow without exceeding rated seal and element temperatures.  

Rick has done some work with sound proofing the tube air system.   Wat sa Rick?

T  

[W3SLK]

Just remember: Those 'old' South Bends, Cincinnati Milicrons, etc without the digital readouts, helped us win WWII and some their tolerances were so good that they were used to make Fatman and Little Boy! Huey, W3WAM(SK) (Nee KD3UI), had those in his shop. He claimed he could make stuff better than the 'new youngins' could. I can believe that since the reality of machine making lays in feeling and intuition! I wish I had a small table top to some work, (maybe a lathe with 3" jaw max.). But to paraphrase Norman Maclean, "Machining comes by grace and grace comes by art and art never comes easy!"

[KD1SH]

  My father worked at Pratt and Whitney, starting back in the early 50's. I remember him taking me to the plant when I was barely old enough to remember, and showing me the rows of Cincinnati horizontal mills and boring machines busily cranking out cylinder heads and connecting rods for those huge radial engines. Not a single digital readout in sight.
  My 9" swing South Bend Model A lathe has a three foot bed and can work 17" between centers, theoretically, but in reality I can probably turn a piece no more than 10" long with good tolerance. Over the course of years, a lathe's carriage spends most of its time traveling nearest the head-stock, and thus wears the ways more heavily in that area, with less wear occurring nearer to the tail-stock, so that the tolerance changes as the carriage moves toward the tail-stock. But in my case, the vast majority of my lathe work is only a few inches long, and almost always in the chuck rather than between centers, so it's not an issue. That old lathe is an honored veteran; it's done its time, and I won't hold a little wear against it.
  I always enjoy watching "real" machinists at work. We had this old Italian guy at one job; give him a block of metal and a drawing; he'd go over to the Bridgeport—whack...slap...clink—have that metal clamped in the vice and a tool in the collet and be cutting metal in moments. Taking deep cuts, too, because he had confidence in his settings. Me, I'm a cutting-tool wimp: make a run with the edge-finder twice—or maybe three times—to make sure I'm all zeroed; take a first very cautious cut and then back the tool out and mike the cut, just to make sure of my settings, and then make a series of cuts until it's done, checking the result each time.
  Yes, machining, an activity theoretically governed by inflexible hard mathematics—unyielding numbers and measurements—can involve a lot more feeling and intuition than one might think.

Just remember: Those 'old' South Bends, Cincinnati Milicrons, etc without the digital readouts, helped us win WWII and some their tolerances were so good that they were used to make Fatman and Little Boy! Huey, W3WAM(SK) (Nee KD3UI), had those in his shop. He claimed he could make stuff better than the 'new youngins' could. I can believe that since the reality of machine making lays in feeling and intuition! I wish I had a small table top to some work, (maybe a lathe with 3" jaw max.). But to paraphrase Norman Maclean, "Machining comes by grace and grace comes by art and art never comes easy!"

[KD1SH]

Continuing progress: as always, it seems that the physical layout of the parts is the most time consuming, and the most prone to redesign on the fly. I'd originally mounted both air variables—tune and load—hard up against the chassis front bulkhead, but then, after a bit of re-thinkage, I figured that it would be prudent, especially at 50mhz, to shift the tune cap nearer to the doorknob; thus reducing the conductor length from the doorknob to the stator terminal of the tune cap. So, I lathed up a couple of 3.5" standoffs, and made up a bracket to firmly anchor the tune cap, both electrically and mechanically, to the deck. The bracket was made from the standard anodized aluminum stock available at most hardware stores—note in the pictures that I gave the aluminum a quick hit with an end-mill to clean off the anodized finish around the mounting holes. I've often found that anodized finish to be surprisingly non-conductive. The #6 internal tooth ring lug on one of the bracket screws will connect to that little terminal on the tune cap that serves as a direct path to the rotor shaft, just as added assurance. The tank coil (next operation) will connect at the junction of the doorknob and the tune cap's stator terminal; run parallel to the cap's shaft, and wind up with its opposite end close to the load cap's stator terminal, again resulting in short connection lengths.

[KD1SH]

  Okay, so the upper side of the RF deck, complete with tank circuit, is all built. I'd originally intended to do all the interconnecting with brass strap, like I did with the tube plate leads (see picture) but, having second thoughts, I got nervous about introducing extraneous capacitance, so I went with some heavy silver plated wire instead. If this were an HF amp, I'd have gone with the brass, but extra capacitances sprout up like weeds at six meters.
  Which brings me to a finding that didn't entirely surprise me: I think I've got a minimum capacitance problem with the tune cap. Once I had all the tank components connected, I did the "cold tune" thing with my antenna analyzer. Using a 2.2K carbon-comp across the tank's input and the analyzer looking "backwards" into the tank's output end, I can sweep it and see resonance right where I wanted it, at 50.25mhz—a good thing—but I can't get the SWR any lower than 3.5:1. That might not be disastrous in itself, but the problem is that the tune cap must be completely un-meshed: anything at all above its minimum capacitance raises the indicated SWR.
  The tune cap has a minimum capacitance of 10pf, and with the addition of the capacitance of the tubes, there we go. Should have see that coming. Two options present themselves: the first—and easiest—is to take my Dremel and remove, or reduce, one or more of the stator plates, and the second—and most complicated—is to make my own air-variable. Not a horrible job—might even be fun—but I wish I had a more suitable ready made cap.

[K1JJ]

Wow - very nice looking build so far!

T

[KD1SH]

  Thanks, Tom. Coming from a master builder like you, I really appreciate that. It's a fun little project, and a bit of an eyeopener when it comes to the arcane nature of upper frequency behavior. It must be really fun at 144mhz or 432mhz!
  One observation I've made is that, unlike with a lower band amplifier, just plugging in tank component values from the tables in the West Coast Handbook—which is what I did to start—becomes problematic on the higher bands. Looking at various designs for 50mhz amps in the East Coast and West Coast handbooks, for example, and calculating out the actual tank inductor values that were used (based on the diameter, length, and turns as described), as opposed to the values that would typically be predicted in the tables or by the various online calculators, the actual values used in the construction almost always wind up much lower. The tube's internal capacitances and that of the associated connections really add up fast at 50mhz and above.

Wow - very nice looking build so far!

T

[K1JJ]

Bill,

We were recently discussing in another thread about minimum C1 + plate-cathode capacitance as being a problem on 10M.  But on 6M, forgetaboutit!  
A possible solution:  A neutralizing cap with a threaded shaft might work dandy as C1.  Add a turns counter and you're in full control.

** Edit:  Also, adding some L in series with the pi-net may do the trick with you exising C1.  I think KB1H has a lot of experience converting SB-220's over to 6M.  Probably knows exactly what to try here...  


T

[KD1SH]

  That's a good thought. I do have some of those neutralizing caps around, so I might just try something like that.

Bill,

We were recently discussing in another thread about minimum C1 + plate-cathode capacitance as being a problem on 10M.  But on 6M, forgetaboutit!  


A possible solution:  A neutralizing cap with a threaded shaft might work dandy as C1.  Add a turns counter and you're in full control.

T

[KD1SH]

Interesting. Do you mean increasing the value of the existing L2, or adding inductance immediately after the doorknob?

Bill,

We were recently discussing in another thread about minimum C1 + plate-cathode capacitance as being a problem on 10M.  But on 6M, forgetaboutit!  
A possible solution:  A neutralizing cap with a threaded shaft might work dandy as C1.  Add a turns counter and you're in full control.

** Edit:  Also, adding some L in series with the pi-net may do the trick with you exising C1.  I think KB1H has a lot of experience converting SB-220's over to 6M.  Probably knows exactly what to try here...  


T

[K1JJ]

Bill,

It seems to me the 11M big tube builders added a few turns between the coupling cap and the input to C1 but I'm am not sure - I've never used the technique.  But some of the guys know what I am talking about and will chime in...

T

[KD1SH]

Intriguing idea; at 50mhz it probably wouldn't take much series inductance to have some effect.

Bill,

It seems to me the 11M big tube builders added a few turns between the coupling cap and the input to C1 but I'm am not sure - I've never used the technique.  But some of the guys know what I am talking about and will chime in...

T

[K8DI]

I’m curious here.  When I’ve been playing around with a lower-frequency (14MHz) two variable caps and a roller pi network, doing the same thing…resistor for the plate load with tubes in place and a VNA looking back into the output….It seemed that if I got to minimum possible Tune capacitance , all I needed to do was reduce the inductance. I could literally see-saw L vs C and maintain resonance over some range.  

What happens if you short a turn (or part of one) of that inductor, instead of trying for a smaller C?  

Ed

[K1JJ]

What happens if you short a turn (or part of one) of that inductor, instead of trying for a smaller C?  

Ed

I was thinking the same thing, Ed.  But I figured Bill had already calculated the Q and wanted to keep L1 at a certain value, like a Q=12, etc.

When I get in a situation like this, I usually start experimenting with L1 turns and try to get a good output power with minimum coil heating.  Q can even go up to 20 or so and still remain efficient, so prune L1 and see if the C1 capacitor can remain in service.  Also, find out what the overall minimum C1 + tube plate capacitance is to know if pruning L1 is even worthwhile at all.

T

[KL7OF]

I built a mono band 29.1 mhz amp and I started out with a Vac cap on C1 and an air cap on C2.  In fact several different air caps on C2.  Back feeding the VNA I could get resonance but high VSWR. Wouldn't tune with power applied..... after much experimenting and no joy , I installed a 5-500 pf Vac cap and it came together.  At resonance (29.1) C2 is at approx 70 pf and at 28.3 it is at minimum so very little C needed .
     When you "backfeed" ,What resistance do you use for plate R?  the tube input R , or the tube output R?


  6M is a challenge .  You do very nice work...good luck   Steve

[KD1SH]

  I tried experimenting along those lines this afternoon, and pretty much found the sweet-spot on the inductor by shorting out exactly one turn. Now, with the analyzer looking back into the output, I can get the impedance to look exactly like 50 ohms; this with a 2.2k carbon-comp to ground on the tube side of the doorknob. Not only that, but I can actually add just a tad of C1 capacitance, with the effect of improving the transformation a bit.
  I had another thought, too: the bottom side of the tube sockets aren't wired yet: no cathode, screen, or grid connections, so my measurements are likely not representative of the true effects of the tube's capacitance. Really, since the amp will be cathode driven/grid biased, none of those elements will be directly tied to ground, so I wonder if there really is any way, outside of actual power-on testing, to get a handle on the capacitance of the tubes? Cold tuning is great ball-park stuff, but the proof will be in the actual live testing.
 

I’m curious here.  When I’ve been playing around with a lower-frequency (14MHz) two variable caps and a roller pi network, doing the same thing…resistor for the plate load with tubes in place and a VNA looking back into the output….It seemed that if I got to minimum possible Tune capacitance , all I needed to do was reduce the inductance. I could literally see-saw L vs C and maintain resonance over some range.  

What happens if you short a turn (or part of one) of that inductor, instead of trying for a smaller C?  

Ed

[KD1SH]

Based on my calculations, using a K-factor of 1.8 with several different assumptions as to how much plate current I'll be drawing—it will vary depending on how hard I chose to drive the amp—I figured a plate output resistance of between 1900 and 2400 ohms. Going with a middle-ground of 2200 ohms, I used a 2.2K carbon-comp resistor from the tube side of the doorknob to ground. Once I get a better handle on the minimum C1 issue, I'm going to try a spread of resistors within that range, to see if I can transform them all to 50 ohm.

I built a mono band 29.1 mhz amp and I started out with a Vac cap on C1 and an air cap on C2.  In fact several different air caps on C2.  Back feeding the VNA I could get resonance but high VSWR. Wouldn't tune with power applied..... after much experimenting and no joy , I installed a 5-500 pf Vac cap and it came together.  At resonance (29.1) C2 is at approx 70 pf and at 28.3 it is at minimum so very little C needed .
     When you "backfeed" ,What resistance do you use for plate R?  the tube input R , or the tube output R?


  6M is a challenge .  You do very nice work...good luck   Steve

[KD1SH]

  I was going for a Q of 10, but I'm not really going to split hairs over it. The dip looks reasonably sharp on my analyzer, so it shouldn't be too bad. When I first contemplated the project, I went through Tonne's, G3SEK, West Coast Handbook, and all that, only to find that when the rubber actually meets the road, there's a whole lot of fudge-factor at 50mhz. Another thing I am noticing is that my LCR meter, the DE-5000, doesn't inspire a whole lot of confidence when looking at decimal microhenry values. Setting it to the highest test frequency, 100khz, gives me some decent resolution, but the repeatability is shaky. When I shorted out one whole turn on my tank coil, the meter barely showed a change, while the effect on my cold-tuning was very significant.
  It seems that sometimes, despite our best efforts, we're reduced to educated dart-tosses in the end.

What happens if you short a turn (or part of one) of that inductor, instead of trying for a smaller C?  

Ed

I was thinking the same thing, Ed.  But I figured Bill had already calculated the Q and wanted to keep L1 at a certain value, like a Q=12, etc.

When I get in a situation like this, I usually start experimenting with L1 turns and try to get a good output power with minimum coil heating.  Q can even go up to 20 or so and still remain efficient, so prune L1 and see if the C1 capacitor can remain in service.  Also, find out what the overall minimum C1 + tube plate capacitance is to know if pruning L1 is even worthwhile at all.

T

[KD1SH]

Tomorrow I'm going to remove C1 and—temporarily, anyway—kluge in a neutralizing cap to get a better window on this minimum capacitance thing.

[Opcom]

one of the East Coast HB projects used two 813s on 10M, which they said presents problems and so the 10M coil was tapped close to the tubes and the tuning cap put there. The article was about a KW amplifier for each band.

[KL7OF]

one of the East Coast HB projects used two 813s on 10M, which they said presents problems and so the 10M coil was tapped close to the tubes and the tuning cap put there. The article was about a KW amplifier for each band.

Yes.. good call....Thats in an arrl handbook.   They tapped the air cap into the middle of the coil and used a small inductance coil in each plate cap lead that looked like parasitic chokes but I believe in reality they were just a necessary inductance to make the 813 work well at 10 M