The AM Forum

I am building a 813 HF amp as described in a February 1969 issue of 73 Magazine: https://archive.org/stream/73-magazine-1969-02/02_February_1969#page/n11/mode/1up

Just for fun and curiosity, I am building exactly down to the exact uuf and winding count.  I completed the tank coil and filament choke.  I will admit that making the tank coil spacer was rather unpleasant and this was my first attempt.  My Boonton L/C meter shows just over 15 uH.  I had also had difficulty "threading" the tank coil as it wanted to bind--several hours of work were spent just on the tank.   I secured all other parts with exception to the chassis and cabinet.  I will post my progress!  

It may have taken hours,  and it looks it,  and it's going to be a really good looking rig I'm betting!

What did you use for the coil spacer?

--Shane
KD6VXI

Thank you.  I wanted this to be a true "junk box" home brew amp but some substitutions were necessary.  I used acrylic rod for the spacer but did originally try pieces of plastic and glue (epoxy) as the author noted.  It was a total mess and I simply couldn't get the spacing correct (I also tried tile spacers).  Here is a photo of some of the other components.  Finding a decent chassis and cabinet is proving to be difficult and potentially expensive.  I even wound the L3 choke on the HV with a wooden dowel and the exact AWG wire as did the author.  I am experimenting by following the article as closely as possible--final result should be interesting!

The power supply is also coming from an other article and I will also replicate: https://archive.org/stream/73-magazine-1965-05/05_May_1965#page/n39/mode/1up

Looks like a neat linear amplifier project.... simple and will work well.

One suggestion:  To achieve a 1:1 swr input on all bands AND improve the IMD products by about 5 dB, add a small variable pi-network to the RF input.   C-L-C  or even an L-C will work.  There is very little voltage there at 50 ohms, so small spacing will work fine.  Initially you can clip lead the parts in and test to see what works best.

And, a little breeze from below chassis will help. I have also used lantern glass as chimneys.   Even air directed down from above the tubes using a muffin fan will produce a nice, even air flow across the whole tube.

Tom, K1JJ

Great suggestions, Tom.  I have a 100uf variable I can use and will wind a coil for the LC.

Hi Chris -

Any small coil that can be tapped with a small rotary switch will work fine.
I usually use a small roller inductor for the job - this makes precise tuning easy.

That 100 pF cap will be too small in value. This is a 50 ohm circuit in and out, so you will need at least 300 pF and even 500 to cover 160M.  I usually use one of those  common multi-section (3 sections paralleled) broadcast receiver variable caps. But the best thing to do is build the amp and then clip lead in some  C-L-C  or L-C parts and see what works best. Then mount them permanently.

You can roughly test the input tuning on all bands when the rig is off using an MFJ-259 meter or equivalent connected to the input and a non-inductive 50 ohm resistor from fil to ground. (output)

Yes, since this is a linear amplifier, the tubes will really appreciate the extra cooling, especially on AM. The carbon plate 813s will actually go from a glowing red plate (bad) to a dark plate when using a lantern chimney and some air from below. Highly recommended for linear AM.  Run as much plate voltage as you can get away with (up to 3500V)  for best positive peaks, efficiency and best IMD. Load it heavily. (more C2 plate loading) to drop power 7% after peak tune for best IMD.

Add a string of regular diodes in the filament CT lead to increase bias until the tubes idle where you want with a given HV, if needed.

T

Thank you again, Tom.  Your recommendations are greatly appreciated.

The author indicates he neutralized the 813 with an Arc-5 antenna connector in proximity to one of the 813s.  I have limited experience with the 813 and am wondering if this is necessary or recommended--say for 20 meters?  He does question the necessity of neutralization of 813s and my research online suggests it is not required.

Chris

Chris,

I built a similar amplifier and it works great.  I used a pi net at the input so the exciter, an SB-101, sees a 50 ohm load.  That way, I can switch back and forth with the amp, or barefoot, without retuning the SB-101 final.  One thing about the 813 amplifier though, it only takes about 65 watts to drive it.  I had trouble keeping the drive down, and the grid current would run too high.  I solved the problem by lowering the drive to the control grid. This is an old trick that has been published a few times.  Instead of connecting the control grid to ground, it is connected to a tap on the filament choke.  I get about 600 watts out using a 2500 volt 400 ma power supply.  Grid current runs 100 ma.  Good luck.

Best regards,
Fox

You might also want to look at the Electric Radio article of June 2016, issue #325 by Bohn, The 2X-813 Linear Amplifier


Phil - AC0OB

Thank you, Phil and Fox.  I will look for the article.  I plan to use this amp with my Lettine so the drive will likely be 30-35 watts or so.  My power supply (500 ma transformer) should deliver 3KVDC no load.  

My chassis arrived today... will start mapping out components and drilling/punching.  Photos are pending!

Chris, N4JOY

Hi Chris. A tip of the hat on the tank coil. I did the same construction tank for a pair of 811a's. It did not turn out as nice as yours but it worked. However, not real pretty, so I found a 3 inch ceramic form ribbed at a hamfest for a couple of bucks and swapped out my ugly one.  :)

Tnx for the reference to the 73' paper. One note not mentioned by the experts here, the plate choke. I see you mentioned the choke on the supply side of the bypass cap, but not the primary choke at the plate. What do you plan to use there? This one is somewhat critical to any SRF present and the 73 article identifies a unit I am not familiar.Care on the choke and it needs to be checked out ideally in place in your final layout. Tnx again!

Alan W4AMV

Thanks, Alan.  I have given up on perfect and am more tolerant to imperfections when working on my projects; but I likely spent more time than needed on the tank.  I used a micrometer to measure spacing but winding the forms too the most time--totally unexpected.  In the spirit of the article, I am calling this a "junk drawer amp" so it won't be a show queen.  :)  

Good question on the plate choke.  You can see the ceramic core plate choke in the photo above, which I had in my parts box.  My Boonton LC meter says 105uH and the wire size appears right.  I will double check the gauge but it looks to be about 22 AWG.

Great Chris. That is quite sufficient inductance. The real key issue is the self resonant frequencies, there are at least 2 key ones, present. Ideal, they should be out of band. You can search the internet on the topic and there are excellent articles on using a GDO to find the series and parallel resonant points. As a lesson on the topic, I picked up the famous National R154 choke which had a significant problem operating on the 15 meter band. It was soon modified by National and became the R154A. I think those are the right numbers. I have both and for fun I measured both with the GDO and sure enough verified the issue that was discovered by folks back then...  :)

Hi Chris -

I see the author fed some negative RF feedback into the fil choke via a winding.  Hopefully he cold-neutralized it by adjusting the pickup's proximity to the tube plate. Unless we adjust the feedback carefully using a scope, it can actually make things worse.

No, there is usually no need to neutralize most linear grounded grid amplifiers.  The grounded shield that the grid and screen form between the plate and filament (cathode) is enough to minimize RF feedthrough that can cause instability on the higher bands..   Just be sure to employ good practices that keep the input components from seeing the output.   The antenna relay can become a troublesome point of coupling in/out, so look it over too.

813s in GG and grid driven have always run well for me. It is a monster of a tube - very under-rated.

In contrast, a grid driven (cathode grounded) 813  (tetrode/pentode)  may need neutralization when run class C, class B or whatever on the higher bands. There is no natural shield action by the grid. Though, with a super good layout, many hams have gotten away without neutralization since the screen is at RF ground.  

BTW, when using a a grid driven triode final stage at shortwave frequencies, it is mandatory to neutralize .

T

Great information as always--thank you!  I will use the grid dipper and double check the plate choke.  I am fairly certain I purchased it as a 1-30 MHz plate choke, 1 amp.  I appreciate the clarification on neutralization.

I am getting ready to punch holes in the chassis--lots of careful measuring.  I picked up the radio chassis punches on eBay and they appear unused or rarely used.  My so called "chassis" is actually a 14x14 industrial cake pan.  It is thick, very sturdy, and perfectly square (not tapered).  I could probably stand on it!  There is enough depth/clearance underside for components and wiring.  I simply wasn't going to spend an arm and a leg for a large Hammond or Bud chassis.  I don't have sheet metal tools and/or experience to fabricate my own.  I will use angle bar and pre-cut aluminum panels to construct the front, side, and top panels.

Work in progress... ::)

Hi Chris,

That is a good looking collection of hardware for your amplifier, and the chassis should work fine for you.

As I look at your simulated layout, I would like to offer one suggestion.  The angle brackets to mount the plate tune capacitor may be left on the stator side of the capacitor, mounting the stator close to the chassis, or they may be moved to the top, or rotor side of the capacitor.  If you mount the capacitor the other side up, with the stator and connections up, away from the chassis, you will realize three benefits:  1) lower capacity between the capacitor and chassis ground, allowing a greater degree of control of the minimum capacity of the tune cap.  2) better protection against high-voltage arc between stator and chassis, due to increased spacing, and 3) shorter, lower loss (shorter wiring) between the plate circuit of the 813 and the tank coil and tune capacitor.  You may be able to do the same thing with the loading capacitor, but that is not clear in the picture.  Hope this helps....

GL on your project!  73, RIck

To add to Rick's comments. I wrestled with layout for days! However, a combination of what is in the books combined with folks who built them and they worked perfect is worth it's weight. I ended up with the tank band switch close to the plate tune cap if for no other reason than keeping the coil taps as short as reasonable. The plate load cap wound up towards the right side of the chassis obviously close to the antenna port. There are comments in the handbooks about keeping the High Q tank inductor at right angles to masses of metal so as to keep the Qul as high as possible. Frankly, I have seen it done following and NOT following that rule! Need to keep the tube socket grid connections as close to chassis ground as possible so as to minimize the lead length of the bypass cap and or the ground connection. I sub mounted my sockets about an inch or so below the chassis. So to solve that issue, I built a large Aluminum plate L bracket to nestle up to the socket and keep the inductance small. Final note, I did not see it in the article, but adding a T/R relay to bypass the amp may be useful. Although at higher power levels this might be more trouble than an aid. And as Tom added, a pi match at the cathode input is used. The pair of 813's have a published Zin for GG. Can't recall, perhaps 120 ohms or so. An easy match to 50. I placed my pi match in a small shielded enclosure above chassis. My 2- cents. Alan.  

Oh, I forgot to add... A glitch fuse on the HV supply line near the bypassed plate choke. I used a single strand of wire from 50 ohm coax shield. Mounted on a HB piece of pcb. Other's can chime in on what they have used.

I am providing an update on my 813 amp project, which was started almost three years ago!  Well, I was inspired to finally get it done and hopefully have it ready in time for the AM rally.  Again, this was intended to be a junk box amp where most components would be scrounged or made including the chokes and tank.  The chassis is a cake pan!  I very closely followed the amp and power supply design from a February 1969 and May 1965 issue, respectively (links are earlier in this thread).

All that is left is:
1) Secure a plate transformer.  Finding one has been impossible (probably many are in garages and sheds hoarded by old hams), so I am leaning toward the Antek 800 VA toroid).  Guess I will have to buy a new one! :(  
2) Install the front panel, drill for meters and bandswitch, and tap tank coil
3) Finish some wiring

I know it is crude looking, but this was my first amp build and it was built in the spirit of "use what I have."  There are several changes I would have made, but I will move forward and get it done!

Hi Chris,

Looking good. Is the HV transformer on another chassis? I see what appears to be
6-500 V stacked cap bank. So that would be 3kV and I assume you are looking at 2.5 kV. I see 3- of the voltage divider R's for equalization, assume the other 3 are hidden from view. However, what is their R value and wattage? They look like flame proof R's but their wattage seems small from their physically size. For the typical case I tend to go with 5x the bleeder power on that voltage divider. So I would end up with 20 W R's give or take. Just to keep the heat in check. Yours look a bit small.

Alan

P.S. I visited the PS reference. He goes with 2 W on 100 k (600 k bleeder). Chris, I would use 10 W. I have seen these equalization R's heat, shift values, no longer properly equalize the HV and BOOM. The HV bank caps go. My 2 cents.

Hi, Alan.  I appreciate your review and input!  Yes, you dispense good advice.  I had yet to solder and install all of the 100K 5w resistors.  I need to place a Mouser order, so I will go ahead and install 10 watts or more.  Thanks!

I am still moving forward with the amp project.  I am doing a little bit every evening after work and family time.  Wiring is almost done, a few more components need to be added, and a fan will be installed for cooling.  I picked up a real nice cabinet on ePay and it is at the powder coater--pics to follow.  I appreciate everyone's guidance.  There are likely a million things that the experts could nit pick about, but this has been a fun project.

Chris,

It looks like you are making good progress.  I have to throw in my two cents though.  I am uncomfortable with the appearance of the high voltage capacitor string.  It may be the photo's perspective, but it looks like the caps are touching the chassis.  The cap at the top of the string would have 2000 volts between the case of the cap and the chassis.  It seems unlikely the plastic skin of the cap would withstand this for long.  I mounted my caps on a 3/8 inch sheet of plexiglas for this reason.  Also, your equalizer resistors are underneath and close to the caps.  This will tend to cook them and may lead to short life.  Carry on and enjoy your good work.

73,
Fox

Thanks, Fox.  I appreciate your input.  I can lift the string with spacers and put a thin sheet of fiberglass PCB board under it.  In regard to the equalizing resistors, the original PS design from 73 had 2 watt resistors.  I figured the 10w resistors would stay relatively cool (they are not touching the caps).  I can snip the leads, extend, solder, and recover the leads with fiberglass sleeves.     

I fortunately found a sheet of phenolic board and will cut to size and drill holes to match the existing chassis holes for the terminal strips.  I can then slip the board under the HV cap string and use slightly longer screws to mount the strips.  I had originally planned to do the string turret board style but had so many long terminal strips laying around.   

Given the same resistance value, the 10W resistors dissipate the same amount of power as the 2W resistors do, generating the same amount of heat, albeit spread out over a slightly larger area. Spacing from the caps helps, as does forced air movement below the chassis.

What larger wattage resistors do is to provide a safety factor to reduce the possibility of their failure. However, if one of the resistors in the string were to fail "open", that capacitor will see a higher voltage that may cause the cap to fail - quite dramatically. Also, bleed-down is compromised. This is a serious safety concern.

My solution to this failure mode is to use multiple paralleled resistors across each cap. This way, if a single resistor was to fail, the voltage division of the string would only change slightly and bleed-down would still occur when switched off. As a bonus, lower wattage, less expensive resistors can be used. The voltage division of a failed resistor in the string can be easily calculated to stay within the voltage rating of each capacitor.

Electrolytic capacitor life is dramatically reduced by high operating temperatures. The life of a cap rated at 85 degrees C is very short when operated at 85 degrees C, perhaps just a matter of months. Granted, that's pretty darn hot, but the cooler the better.


Don

Looks really great. I love the pi coil.

Are you planning a tune/operate switch? For some reason the power supply author suggested a 250V one on the secondary. I hope a better scheme will be used if wanted.

Using 100K resistors for the capacitor equalization gives only a very low current (5mA@3KV). Should a capacitor get too leaky, the voltage across it will drop quite a bit because the resistor string can't give up enough current. The remaining voltage will appear across the others. It should not be a huge issue since everything is new but at some point there could be a surprise if they don't all age the same.

Looking forward to more and completing, testing, etc!

Very good Chris. As a rule of thumb and the experts can correct me on this, you need about a 1/4 inch separation/1 kV between the hot potential and the flash over point in air. HV insulation will certainly improve that number. In any case at 3 kV without any added insulation sheets, you need an inch to be on the safe side. I recently got zapped on a 1 kV supply I am working on for a small amplifier. Did not observe the rule and had several arc over points! Had to re drill out the card via holes a bit and added kapton tape as an extra measure.

Placing equalization R's in parallel is a good idea. Although in the 1-to- 1.5kV supplies I made and did not do that, but wish I did. On a 3 kV amplifier that I worked on for a ham colleague, HV caps went because of long term aging and heating of the equalization R's! Several of them failed OPEN. We replaced the caps, and soon afterward, the new caps blew again! Lesson learned, replace the HV caps AND the equalization R's.

Alan

Chris, some additional comments and again, the experts are here to add or correct.

On this 813 amplifier as published, there is no means of placing the tubes at cutoff or into standby mode. A grid negative supply V is not shown as well no means of monitoring grid current as well relative Pout. Tune up of a GG amplifier is facilitated by those measurements. Plate current alone is not sufficient and in looking at the paper again, is not monitored.

On the PS, he uses the secondary to provide reduced HV out for the voltage doubler. I think the concern here is the transformer insulation breakdown voltage where the CT is not at ZERO potential. Is that a risk? I've used the voltage doubler successfully but it was driven by necessity, did not have the desired size xmfr on hand. However, the center tap was not used and I think that raises a flag.

Alan

Thanks for the added information and comments.  I did not incorporate a tune/operate switch for reduced output as the author recommended.  I am not a fan of hot switching and, as you suggested, not having the CT at zero potential.  Yes, there is no grid meter as I will tune by monitoring plate and maximum output via a separate watt meter.  I plan to use the amp with my Lettine 240, which has an output of no more than 25-30 watts.  I can't imagine over driving it at these levels.  The transformer is an Antek toroid with the primary and secondary taps wired in parallel for 800V at 1 amp.  I am guessing I should see around 2100V to 2200V no load.  I would honestly be happy with just a 90-100w carrier.

I will be cutting the phenolic sheet today and, with spacers, raising the capacitor string up as high as possible.

I also used the doubler as it has been impossible to find heavy iron, which is why I settled on the Antek.  The good iron are either in land fills or in the basements/garages of elderly hams who have hoarded them.  I certainly searched and searched but simply wanted to move forward with the project.  

The continued suggestions are most helpful!  I wanted to make this a true junk drawer project and utilize as many resources as I already had on hand (e.g., the modified cake pan for a chassis).

Addressing Alan' concern, I custom cut a piece of pheonolic plate , drilled holes to match the screw pattern on the chassis, and added nylon spacers to lift the board.  Let me add that doing this without unsoldering the capacitor string was not easy!  It took me at least an hour to do it properly!  I still have to cut the resistor leads and add some lead to keep them further away from the caps.  The dangling grey wire in the photo is for the blower, which had been mounted but not connected.

I am almost done and waiting on the cabinet and front panel to be powder coated.

I had the front panel and cabinet sand blasted and powder coated (hammertone gray/silver).  The most challenging part of the project, so far, was measuring and drilling out the front panel!  It took hours!  I used step bits of various diameters for the drill and the holes for the meters required a fairly large step bit.

I am waiting on a variac I ordered from eBay and will begin testing soon!

I'll make one more update as there doesn't seem to be much interest.  I have been searching for some nice 4" plus skirted dials for the plate/load caps.  Otherwise, all has been completed.  The bandswitch and meters were installed.  I will be applying power in a few days and have strict instructions from my wife to only do so when she's home!  ::)     

I have used Carolina Laser to make phenolic labels for past projects.  They do great work and their online creation tool is easy to use.

I appreciate all the past assistance/guidance. 

That's a very nice looking job!

I'll be interested to see how it runs and how the power supply design works out.

Greg

Looks nice!

Where are you (and lots of other builders whose rigs show up here) sourcing the shaft couplings, extensions, flex joints, bushings, etc.?  When I look to typical suppliers I don't find much, and what I do find seems overpriced...

Ed

I'm interested and watching your progress, Keep us updated...now the real fun starts......PowerUp time...
It looks very cool, nice job and great idea for a project..
I'm interested!!!
AG5UM

I don't let the Warden know when I do 'things'.

Life can be a dream when she's not around for the sh-boom, sh-boom.

klc

Two schools of thought on working alone....
1. Alone means alone....no help-on-the-way....no-one to call for help....
2. It's often dangerous to have someone distracting you, asking questions, telling jokes...
    or worse-yet.. startling you!! , make you jump...
    and of course ..if you grab onto 7000 volts, 600 amps... are you really going to need any help??
     or just your burial plot paid up???
3.  maybe your wife could get one of those "Baby monitors", and watch from the house?? just kidding..
Anyway.....
Great project, great work, and like was said... do you really want her to know Everything your doing???
If this is not funny, sorry, it was meant to be, just kidding...
AG5UM

I dunno.  I had 6kv at 2a ccs xformer driven, 180uF cap bank hit me.

Obviously I'm still here lol.

--Shane
KD6VXI

She's very wise!

When I fire up high voltage projects I have my wife standing by with the phone to dial 911 just in case.

Excellent workmanship! I envy your abilities.

Jon

Hello!  Just wanted to bring closure and let everyone know that I finished the amp yesterday.  It took far longer than I expected as life and work got in the way.  Nevertheless, I devoted yesterday to wrapping it up.  I simply have to have some phenolic labels made and that should be it.  

Looks nice.

Kl

Thanks.  I will be doing some more testing tonight.  The coil was tapped with the help of a grid dip. Hopefully, I was close.

Nice looking. I like it. 

Question: will the meters have actual plate voltage on them?

Thanks.  Yes, the plate current meter does have HV on it.  In retrospect, I should have placed the current meter in the negative lead of the plate supply.  I will be making this change in the very near future!   

Great progress!

Curious, you say you used the GDO to set the taps.
Exactly what process did you follow?

Verification of the loaded Q of the plate tank for the 813's is important and I must confess, I use a vector impedance meter to do this. Are you using the GDO to find the resonate point for each band and if so, how do you find the loaded Q? Do you use a frequency counter with the GDO to verify the 3 dB bandwidth?

I guess you might have followed the inductance values required for the tank L to create the proper Pi match for each band from the 73 article. Then resonated the taps with some fixed C to get the tap position for that L correctly set.

Yep, ok I guess that is what you followed, Chris. After I returned to the paper.
It is just a little bit of work and it is just ME... But I would go through the calculations for the Pi network to validate what the 73 article has presented. Just to be sure. Easy to calculate the inductance for those taps from the solenoid equation. And then make sure the C_plate and the C_load all pass the sanity check for the required plate load Z for the 813's. Again... just me.

Alan

Hi, Alan.  That's correct.  So far the taps appear close based on my calculations, but the real testing will occur this weekend.  You and I are on the same page. There will no doubt be some adjusting!

Hi, Alan.

Well, the 20 and 40 meter taps were way off--80 meters was fine.  Twenty and 40 meters are tapped on the second and sixth turn, respectively.  On 40 meters, I am seeing 1800VDC and 230ma under load.  My RF ammeter suggests around 95 watts, but I am driving it with maybe 15 watts.  I need an accurate watt meter!  At idle, the meters read 2.2KV and 55ma.  I'm going to do some more testing on 20/80.

Now to get it on the air!

It is looking good!

A MFJ (or other) "antenna analyzer" works well for checking the RF output tank.  Calculate the expected operating load for your tubes based upon the class of service and connect a small non-inductive resistor from one of the plate caps to a nearby ground point to simulate the operating conditions.  Leave the tubes in place with the plate caps attached to provide their usual capacitance.

Now connect the analyzer to the RF output jack and adjust the final tank controls for as close to 1:1 SWR reading as you can get on the analyzer and this should be very close to the correct control setting when the amplifier is used.

Do NOT forget to remove the little resistor before applying power or it will turn into a small hand grenade :)

This trick is very useful when troubleshooting tank components because it allows you to test connections and switch contacts "live" without danger and you can also temporarily substitute a low voltage/low power capacitor for a suspected bad unit to see if that is the fault.

I first used this when I built my triple 4CX800 contest duty amp many years ago and the settings for the roller inductor, vacuum plate, and load capacitor were all very close to my calculated values.

The grid dip meter is still a very handy tool.  Several years ago I worked on a Viking 500 for a friend, it would not tune up properly on 40 meters but worked well everywhere else.  I suspected a switch contact but all were fine.  The analyzer showed that it wouldn't match properly on 40 meters but it did appear somewhat OK at the low end of 40 and below the band.  The grid dip meter then confirmed my suspicion, the 40 meter coil tap was one turn off.  I had forgotten that the Viking 500 was available in kit form so this one never worked properly on 40 meters although it sort of worked at the low end of 40 with reduced efficiency.  

Rodger WQ9E

I used that trick when working on a Drake L-75 Amp that had all kinds of loading issues. I found some bad contacts on the band switch and away it went!

Hi Chris,

What Rodger outlined is spot on. If you have a swr bridge or some way to read a match that would help. I might add, not reverse engineering the 73 paper, I estimate for class B your plate RL value at 2.5 k for two tubes. Each tube has about 14 pF Cpk plus the stray. So your C_tune cap on 40 meters can be preset and a typical value is chosen for a loaded tank Q of 10-to-12. From Q=Rp/X_tune=12, I get about 107 pF on 40 meters and 50 pF on 20 meters. Application of the approach Rodger outlines, you can set and forget the C_tune cap. Now the Pi match reduces to just an L match and it becomes far easier to tweak the L tap. I would set the C_load to mid range to start and with the 2.5 k in place and tubes in socket (POWER OFF!) poke some RF into the antenna port and monitor on a swr bridge looking for 1:1.2 or better swr as you tap the tank L on 40 and then 20 meters.

I did go through the Pi net calculations for this case and I get 5.2 uH on 40 meters and so 2.6 uH on 20 meters. And on 80 meters, you can confirm, I see about 10.5 uH (that would be the entire L) and a C_tune of 210 pF.

See if that makes sense.

Looking at the photograph of the band switch on post 41, the interconnecting wires look rather long, and of small gauge.  Short wires, with minimum possible inductance and resistance are desirable here.  It is preferable to keep most of the inductance of the tank local to the coil, and not the interconnecting wiring.

A suggestion to improve the efficiency and stability of the pi tank would be to rotate the entire switch assembly clockwise 90 degrees (looking at it from the back).  Doing this would allow you to shorten all the wires, especially the common wire to the loading capacitor.  Perhaps also use a larger gauge wire, or copper strap, to reduce losses.  The wires need not be insulated, as long as they do not touch each other.  This improvement has the most benefit on the higher frequency bands, where the switch wiring is a significant portion of the total inductance.

I happened to use the same switch on an antenna selector this weekend. I use the short lead approach. Turned it over to put the knob on and now it points south. So be aware. Will have to file a flat on the shaft to correct the problem. I love those switches, though.

I have used a bunch of those BC-375 tap switches and they are nice.  Hallicrafters used one for the band switch in their HT-45.

The spring on these switches is robust to put it mildly so the turning force is pretty high and a lighter force spring could be used while still providing positive selection.  For an amplifier project I needed to gang two of them with one mounted 90 degrees to the other.  I liberated a 90 degree gear drive from an old WRL Galaxy transceiver parts unit to couple the two switches and I removed the spring from the switch driven by the 90 degree offset gear set so that one switch provided the detent indexing for both.  Turning two ganged BC-375 switches with their stock springs intact would be a handful!

Rodger WQ9E