Grid input tuning for 813 rig

[KK7UV]

This is the transmitter design known also as K1JJ, Tesla 500,...  a pair modulating a pair.

I am attempting to get the grid input network somewhat optimized on the bench with an antenna analyzer, realizing I may have to do some tweaking when it's installed in the transmitter.  One of the attached schematics depicts the input network with tubes and grid bias circuit shown.  The other schematic depicts the bench testing setup, which is where my question lies:

For bench tuning (out of the transmitter circuit) with an antenna analyzer, how do I connect the simulated grid load resistance and the grid input capacitance? Do the R and C go across the tuning cap (stator to rotor)? or do they go to the common ground (which would be the chassis in the transmitter)?

I am using a 33pF mica cap to simulate the grid-to-cathode C of a pair of 813's according to the RCA datasheet (16pF per tube)
I am using a 5830 ohm resistance to simulate the grid input Z according to the ARRL handbook formula (drive power (W) / grid current (mA) squared x 622,000.   For my application this is 5.4w drive, 24mA dc grid current.

[KC2ZFA]

look here too: https://forums.qrz.com/index.php?threads/the-k1jcl-memorial-transmitter-resurrection.873671/page-3

[KL7OF]

I built a Tesla 360 and built the all band grid tuning unit from the Calif Handbook. 2 coils and a breadslicer if I remember right.  14th Edition....... Good Luck.  Steve

[K1JJ]

This is the transmitter design known also as K1JJ, Tesla 500,...  a pair modulating a pair.

I am attempting to get the grid input network somewhat optimized on the bench with an antenna analyzer, realizing I may have to do some tweaking when it's installed in the transmitter.  One of the attached schematics depicts the input network with tubes and grid bias circuit shown.  The other schematic depicts the bench testing setup, which is where my question lies:

For bench tuning (out of the transmitter circuit) with an antenna analyzer, how do I connect the simulated grid load resistance and the grid input capacitance? Do the R and C go across the tuning cap (stator to rotor)? or do they go to the common ground (which would be the chassis in the transmitter)?

I am using a 33pF mica cap to simulate the grid-to-cathode C of a pair of 813's according to the RCA datasheet (16pF per tube)
I am using a 5830 ohm resistance to simulate the grid input Z according to the ARRL handbook formula (drive power (W) / grid current (mA) squared x 622,000.   For my application this is 5.4w drive, 24mA dc grid current.

Hi Steve,

I would go across the grid to cathode pins of the tube for the simulation.  If it gets squirrelly, then try the other way. Though, the test should tell you basically what you need without the added loads.

That said, most of my modern class C or linear builds now use a reverse pi-network for the input.  You can get it to tune and work well from 10-160M easily.  Use a variable cap to ground, a physically small roller inductor in series and a smaller variable cap to ground into the grid.   The old link input was handy to allow the grid to float using the neutralization  circuit.  But generally, well designed tetrode rigs do well without neutralization in GG linear or grid driven service.

Good luck with the project!

Tom, K1JJ

[KD6VXI]

If your tetrode isn't behaving, ground the screen to the rf deck chassis, put the B+ to the screen at chassis and - of that power supply to cathode.

Bam

Stable.

My 4cx5k ran that way for a couple decades.  Stable, better than 10 dB gain.

--Shane
WP2ASS / ex KD6VXI

[w9jsw]

Here is my grid tuning for my K1JJ plate modulated 813 rig.

C? .01 are 1kv disc caps. L is an airdux coil with a shorting switch that is tapped for each band. 300pf an air variable cap that is mounted on the front panel (small one from RF Parts, I can give you the PN). No neutralization used on 160/80/40. I am unstable on 20M but never go there anyway. Seldom see AM there. I tuned this with an mfj in circuit using the grid swamp R only, no added components. Grid swamp is a pair of Caddock TO220 non-inductive 1K resistors in series, bolted to the chassis directly between the 813s. All the grid components are tightly mounted close to the 813 pins in the center area between the sockets.

The pic is the same circuit as implemented in my new 4-400 deck. It is not operational yet. For that one, I used globar non-inductive R for the grid swamp but may go back to the Caddock ones as they are cleaner/simpler.

When I tune up, I preset my turns counters to the appropriate remembered numbers, preset the grid input var cap for example on 80M it goes to 10:30 on the dial. I then key the transmitter and fine tune the grid current using the cap to 30ma. I also dip the plate and then adjust the loading for around 375ma of plate current. I run a fixed 2KV. Easy-peasy.

I also dropped the tesla screen PS and choke approach. Using 25K screen dropping resistor. Lots simpler. Works great!

[K1JJ]

Here is my grid tuning for my K1JJ plate modulated 813 rig.

C? .01 are 1kv disc caps. L is an airdux coil with a shorting switch that is tapped for each band. 300pf an air variable cap that is mounted on the front panel (small one from Nebraska Surplus, I can give you the PN). No neutralization used on 160/80/40. I am unstable on 20M but never go there anyway. Seldom see AM there. I tuned this with an mfj in circuit using the grid swamp R only, no added components. Grid swamp is a pair of Caddock TO220 non-inductive 1K resistors in series, bolted to the chassis directly between the 813s. All the grid components are tightly mounted close to the 813 pins in the center area between the sockets.

The pic is the same circuit as implemented in my new 4-400 deck. It is not operational yet. For that one, I used globar non-inductive R for the grid swamp but may go back to the Caddock ones as they are cleaner/simpler.

When I tune up, I preset my turns counters to the appropriate remembered numbers, preset the grid input var cap for example on 80M it goes to 10:30 on the dial. I then key the transmitter and fine tune the grid current using the cap to 30ma. I also dip the plate and then adjust the loading for around 375ma of plate current. I run a fixed 2KV. Easy-peasy.

I also dropped the tesla screen PS and choke approach. Using 25K screen dropping resistor. Lots simpler. Works great!

Yep, I use the exact same RF grid drive ciruit on my 4D32 rig and on Commando, the pair of 6146s. I would call it an L-C  or L input network.  Very simple. The reverse pi-network using an extra input variable cap will give us an exact 1:1 swr, but the L-network with <1.5:1 in most cases is adequate.   Also, the .001 cap in series with a 2K resistor to ground across the grid is an excellent way to calm down any rig.

On the screen DC feed....  For the 813 series modulated rig I use a screen dropping resistor, simply cuz it's a floating circuit anyway. But the disadvantage of using the big dropping resistor vs: the Variac controlled screen supply with a choke is the large heat waste. Also the screen current is not adjustable on the fly; ie, if you change your loading or plate voltage, the screen current is not optimum; ie, stuck to where you initially set the resistor value.  The advantage is probably a slightly better audio frequency response (at certain freqs) using the resistor vs: the choke.

But there's many ways to skin a rat...

T

[KC2ZFA]

John, what’s the airdux number ? or what’s the length, turns over inch, coil diameter, and wire gauge ?

thanks and looks gud!

[w9jsw]

I will measure it and let you know. I know that it is around 20uh, and about 1.5in in diameter. Tom needs to show what he used. I think he wound his on a form. May be cheaper.

[K1JJ]

Any reasonable small mini-ductor will do.  My own 813s X 813s rig uses a 40T Minidux  with 1.5" diameter.  This cover 160M.   I adjust each band tap by using the MFJ-259B experimentally, going into the grid 50 ohm RF input..  No big deal to find a near 1:1 tap when the variable cap is tuned.  Fine tune it later when the rig is running to find ~ 1:1.

The only problem is sometimes choosing a variable cap that has too high a minimum capacitance. It will be unmeshed on the higher bands, like 15, 10M, etc.  Though I have antennas for the higher bands, I am mostly a 160-40M guy anyway

T

[w9jsw]

Here is the air variable cap I use on 2 of my rigs. On the new one, I have it resonant from 160M to 10M. Not sure it will actually do 10M as I am still bringing it up, but it is a goal.

https://www.rfparts.com/20apl300de.html

My airdux is like Tom described. Had also thought about using a set of wound toroids, but this was easier.

John

[KK7UV]

Thanks for the replies everyone.   I had considered an L-network early on, but with matching a 50 ohm SS exciter to 5830 ohm grid, the Q formula works out to 10.75.   I had thought this was too low after reading the 50s-70s handbooks which said the input network is recommended to be between 12 and 15 for Q.  But, perhaps that was specific to the link-coupled parallel LC network?

Anyhow, if I can't get my current iteration working, I will try the L-network.  I am currently using link-coupled parallel LC with T-106-2 toroids (one for each band 40-75-160) as shown in my original post - simplified for one band.  With the simulated load resistor (5830) and grid-cathode capacitor (32pf), I can adjust the primary link winding and get very low SWRs on each band on the workbench with the network extracted out of the transmitter.   But firing it up in the transmitter results in very high SWR on the exciter and tiny drive.   As I looked futher into the RCA 813 datasheet, I see that the grid C can vary between 13 and 19.6pF per tube, and therefore up to nearly 40pF for a pair.  If I measure the pair mounted in the chassis sockets from grid to ground (cold, no RF) I get 43pF.  I am now wondering if I should try 43pF for a simulated load value and report back.   

[K8DI]

What is the actual circuit you are using (not the published 813 type circuit, but how you have it wired)?

I think there's possibly an error with the one at the top of this thread.  If you think about it, the LC will resonate, but can't deliver drive between the grid and cathode.  (Speaking for RF, NOT DC here) The cathode is the filament is grounded through the capacitors on the filaments to ground. The network completely floats at the cold end -- it is isolated from RF ground by the choke. If you were to move the .001 (or add one) from the bias end of the choke to the cold side of the resonant LC/other end of the choke, you'd have an RF path (through the ground) that completes the circuit by providing a return path for the RF back to the transformer..

If your bench setup is like the second of your two variations, it will test well because your simulated load is across the LC. If you wire your actual TX more like the first (necessary as the bottom of the LC cannot be directly connected to the filament) it won't pass drive to the tube....

Ed

[W1RKW]

on my 813 JJ rig, I wound my own on 1 1/2" PVC. I don't recall how many turns I used but I used a similar method K1JJ described in his last post to ball park tune it. I didn't worry about taps for other bands since it is single band.  I fine tuned it for near centering of the variable cap once integrated into the system.

[KC2ZFA]

But firing it up in the transmitter results in very high SWR on the exciter and tiny drive.  

before you redo the input circuit put a receiver type variable air cap in series with the link to ground and see if you can tune for least SWR looking into the amp. Use a 3-section job, like 500pf per section, and experiment with how many sections you’ll need per band.

Peter

ps. the observation above regarding the .001 before the bias rfc is correct. the cap should be on the rf-hot side of the choke. If you decide to neutralize use a feedthrough cap to connect to the neutralization strip or cap on the tube side of the chassis.

[w9jsw]

I drive my rig with a SS amp that uses a pair of MF150 in push pull. Does fine. Needs about 30W of drive. SWR is low. I keep a dual needle meter on the drive input to insure good swr. Always low when I tune it up. I have it set that the optimum drive level results in 30ma of drive at the peak point of resonance of the LC circuit.

[KK7UV]

Ed - thanks for seeing my error.  See the attachment for the complete grid circuit with your suggested capacitor added in red.  I will try this out and report back.

John - My SS exciter is a DDS-VFO and buffer stage driving a QRP Labs P.A. (p-p IRF510 MOSFETs) and gives me adjustable 5-12w output into 50 ohms.  It is followed by relay-selected lowpass filters for each band.

[w9jsw]

K1JJ/others need to opine.

As I understand it, the amount of swamp I have is proportional to how hard I have to drive it to get the proper grid current. I think I am generally around 20-30W depending on band. If you reduce/eliminate the swamp, the drive lessens. For my rig, the swamp R stabilizes the input. Some guys don't use swamping R because it requires excessive drive.

Tom, do I have this correct?

[KL7OF]

K1JJ/others need to opine.

As I understand it, the amount of swamp I have is proportional to how hard I have to drive it to get the proper grid current. I think I am generally around 20-30W depending on band. If you reduce/eliminate the swamp, the drive lessens. For my rig, the swamp R stabilizes the input. Some guys don't use swamping R because it requires excessive drive.

Tom, do I have this correct?

My tesla only needs 15 ma grid drive for full output.  I don't swamp the grid tuner, Just tune it to 15ma.  Steve

[w9jsw]

K7LOF - do you have a schematic that you can share of your grid input as implemented in your rig? Want to see how you tied it into the circuit along with how you did the grid leak bias.

[K1JJ]

K1JJ/others need to opine.

As I understand it, the amount of swamp I have is proportional to how hard I have to drive it to get the proper grid current. I think I am generally around 20-30W depending on band. If you reduce/eliminate the swamp, the drive lessens. For my rig, the swamp R stabilizes the input. Some guys don't use swamping R because it requires excessive drive.

Tom, do I have this correct?

John,

Yes, you have the right idea. Extra drive needed is the drawback of swamping.  

All rigs are different. Use of swamping and  neutralization may vary with bands.  A sloppy layout on 10-15M might require both neutralization and swamping. These add-ons will depend much on layout and parts chosen too.  RF chokes, certain good quality bypass caps - and lead diameter and length will have an effect on stability. Whereas, a sloppily built single band 160M rig will probably require no swamping or neutralization despite its shortcomings.

A well built tetrode rig on 10M with excellent layout (and well-bypassed screen) might get away with just swamping.

I usually prefer to do both just as good practice.   Neutralization will also depend on whether you are using a triode (mandatory neut)  or tetrode (optional neut).  Though with swamping, even a light 5K/ .001 load across the grid to ground, can help, despite the little extra drive needed.  On 160-40M rigs, I sometimes don't use neutralization on tetrodes, just swamping with 2K.

I like to use the non-inductive "chip" resistors (25-100W) that are heat sunk to the chassis. They are perfect for the job.  I usually start at 2K and can add some padders if there are any instability troubles.

It's always good to have a few tricks up your sleeve to tame rigs; even ones that are laid out well may need some help on certain bands sometimes.  Good results are directly proportional to how much time and effort was put into the designing and building process.... and how much is then put into the troubleshooting shakedown.

I like to beat up a new rig to its MAX limit, looking for bad parts and bugs.  Once it runs well, then back the power down to maybe 50-75% of max and it will usually be cleaner and last a long time without trouble.

T

[KL7OF]

K7LOF - do you have a schematic that you can share of your grid input as implemented in your rig? Want to see how you tied it into the circuit along with how you did the grid leak bias.

I will look for the schematic. It was 20 some yrs ago.... Those tubes run zero bias...

[KL7OF]

K7LOF - do you have a schematic that you can share of your grid input as implemented in your rig? Want to see how you tied it into the circuit along with how you did the grid leak bias.

Hi John..  Here are some cheesy cell phone pictures of my notes from 2001 and Page 572 Editors and engineers handbook #14.  Over the life of my Tesla I have made many  changes.  I have used a bias supply with grid leak but I think now at 2400V on the plates, They run zero bias.  I may have a leak resistor in there or some diodes now, I can't remember.
   I added a switched in 400pf mica padder for 160M in the grid tuner .   Steve     


Added:  I used to use an 807 for an RF driver.  Now I use a 2E26 (collins 310B)  driver.  The rig doesn't take much drive on 160-40, about 10-12 ma.  and about 18-20 ma on the other bands. I get plenty of output.

[w9jsw]

Thanks, Steve. This is very helpful.