Advice for final tank circuit

[K9ACT]

All I am suggesting is that one can not flatly accept everything we we read.  Too frequently, someone makes a statement which may very well be true in his case and it gets repeated and passed on for generations like the Ten Commandments.  This is surely such a case.

A 6146 is not a high power triode. Producing .6 W vs .2 W with a tube is trivial and costs nothing so perhaps not a bad idea.

Producing 75W vs 25W is not trivial and is a major design issue.

So, perhaps it is time to re-evaluate that rule of thumb.  Let's find out what is really needed and not force people to over complicate a project because someone said so 50 years ago.

My experience with power triodes is that the grid drive requirements are just about what the tube data suggests.

This includes 811, 812, 8000 and 810.  The only rule of thumb I could come up with would be to follow the tube data.

js

[KM1H]

Yep, JS pretty much said it.

Those of us that have spend years or decades actually building things tend to look warily at old published info. The ARRL repeated 30's designs almost into the 60's and never bothered to revaluate them with actual test equipment.

These days it is so easy for even the individual to have test equipment that was affordable only by commercial labs in 1960.

Ive been assuming that Rons project is all on one chassis, compact, and with minimial lead lengths. If not then my earlier statement of 2 sets of coils with links and up to 4 variable caps will hold. It never hurts to design the driver for its maximum efficiency as its easier to reduce drive than create it out of the air.

The comment about the 833A is interesting also. If a Viking II with a pair of 6146's can drive a pair of 833A's to full power than one MUST assume that a single 6146 can drive a tiny little 254 almost out of its socket. Just dont expect as a high efficiency at 1400V compared to 3000V.

Carl
KM1H

[W8ACR]

Hello all,

I guess it's time to describe my project in more detail. Years ago, I acquired a GK500 that was in very poor shape. The PS and MOD decks were shot (they had actually been under water!!). The RF deck was mostly intact but a few key parts were missing. I was able to build usable PS and MOD decks with parts on hand, but they were significantly different from the original design. I used a UTC S22 mod transformer which did not fit on the original MOD deck. I therefore had to add a fourth deck - one just for the S22. It was all a quick and dirty hack job, and quite ugly, but the rig actually worked half decent, and It was used on the air for the past ten years or so without any major problems. I have always wanted to do a better restoration, and in particular, I wanted to get the mod transformer back on the MOD deck. I recently acquired a 125W mod transformer that will fit on the MOD deck, so it was time to do the project. I tore down the RF deck to nothing but sockets and transformers and rebuilt the OSC and BUF stages according to the original GK500B schematic. I decided I wanted to redo the final stage with a tube more suited to the new lower power level. Obviously, a 4-125, 4-65 or 813 would have been more suited to the original circuit, but  I didn't have any of these on hand, so I chose the 254W since it was about the right size for the power level and it presented the challenge of designing a new circuit.

The RF deck has been rebuilt, and the OSC and BUF stages seem to be working properly. For simplicity, I left out the bandswitching ciruitry and decided to make it a single band 160 meter rig. Plate and grid voltages on the 6146 are correct, but it dips down to 10 mA at resonance and I can't get it to load up any more than this. If I tune it off resonance, say up to about 25 mA, then the 254W grid current will rise to about 20 mA, but that's as far as it will go. I was hoping to use the simple capacitance coupling of the original design, but no go. It seems to be acting like an impedance mismatch, and that is why I thought adding a tuned circuit at the 254W grid might solve the problem. Once I solve the 6146-254W coupling problem, then I'll tackle the 254W plate tank issue which should not be difficult.

I have at my disposal a military voltmeter capable of reading up to 5000VDC, an O-scope, a frequecy counter, a signal generator, and an HP VTVM. I could draw and post a schematic, but think of it this way, ahead of the 254W it is a GK500B in the OSC and BUF stages, and from there on it is a plate neutralized triode amplifier with conventional plate tank and variable link output to the antenna.

Hope this helps you to understand what I am trying to do. Thanks for you comments and suggestions.

73, Ron W8ACR

[N3DRB The Derb]

Ron,

try some link coupling. I always had more success at doing that. It's inherently more flexible.

[K9ACT]

I could draw and post a schematic, but think of it this way, ahead of the 254W it is a GK500B in the OSC and BUF stages, and from there on it is a plate neutralized triode amplifier with conventional plate tank and variable link output to the antenna.

First of all, a tuned grid on your PA is not an option.  Not even ARRL handbooks build triode amps that way.

Secondly, words are not a substitute for a schematic.

I for one, am getting tired of words.

How can you, (let alone us at the other end of your keyboard), possibly understand what you are doing without making/seeing a schematic?

js   

[W8ACR]

Jack,

Sorry for being lazy, I'll post a schematic later tonight. Thank you for your input. I do have several schematics that I am working from, it's not just in my head.

I think I used the wrong term when I spoke of a tuned grid. I simply meant that I needed to have a grid tank, just as you have in your 8000 rig. The Globe King 500 has no grid tank, it feeds the grid of the 4-400 directly from the plate tank of the 6146 via a coupling capacitor. I was hoping to do the same with the triode, but I guess it won't work. I will post the schematic of the GK500 RF deck as well as the circuit that I was trying to use. If you want to look up the GK500 schematic it is on the BAMA site at edebris.com.

73, Ron W8ACR

[Gito]

Hi

I think the RSGB and ARRL hand bpook is a "profetional" hand book and accepted world wide and the article it wrote must come from theory and practise,they must have experiments and building transmitter to find what they wrote.It's not the made of one man,but of many man experience.
So I believed it.
Just one thing that  we must miss, The power dissipated in the grid leak resistor.
If we used an automatic bias ,than the power turn as heat is R X I (grid current) and if i'm not wrong the data book does'not include it on the driving power needed.

For me designing a transmitter first find the tube with  the power output We need,find the input impedance of the tube,if it possibe find a driving tube that's capable with 3 times yhe power needed to drive the Final tube,find a tube with enough HV so when loaded gets a output impedance like the input impedance of the final tube,so we need only one resonance tune circuit at the Grid circuit of the final Amp.
I've made several transmitter,one is a 833 final drive with two 807  at 500 v,using a pi section to match the 833,I must load it (807) to it's limit to drive the 833,
It' all home made maybe a bad design?
But it's running for 3 years without any complain.

Gito
 

[Gito]

Hi

another transmitter I build three 833 .3000 V HV loaded to 900 ma that's an input of 2700 watt,with a 813 driver 2000 v on plate.

All homemade and it.s been running for three and a half year.
As it said before maybe not a good design.

Gito

[K9ACT]

Kudos to Gito.

Great looking stuff.  If it works it's a good design in my book.

Thanks for sharing these pics,

Jack

[K9ACT]

One other "detail".... I never made the connection between GK500 and a Globe King 500 and assumed it was some exotic old rig with some weird tube in it.

My Elmer K9WEK/Dan replaced the RF deck in his GK500 with a pair of 8000's pretty much like mine and can be of a lot more help on your project.  I have asked him to join in this discussion.

Jack

[W8ACR]

OK Jack,

Thanks, that would be great if Dan could join in on the discussion.

I am attaching three schematics with this post. First is the Globe King 500 schematic. Next is a schematic from the 1952 ARRL handbook showing a triode final amplifier fed with simple capacitative coupling to the driver. Next is a hand drawn schematic of the circuit I built and which did not work.

Ron

[W8ACR]

Here are three more hand drawn schematics of proposed fixes to my drive problem.

Ron

[KX5JT]

It was easy to turn this monitor on it's side here at work, it's a flat panel LCD but working the mouse when the cursor moves in different directions is another story!

[K9ACT]

OK Jack,

Thanks, that would be great if Dan could join in on the discussion.

In the mean time, famine to feast.  My knee jerk reaction was to suggest that it could be a triode/tetrode  thing but after finally getting that schematic right side up, all I can do is again wonder if the Handbooks are invincible.

I looked through a number of old ones and could not find a triode without a tuned grid.

I am curious about RFC5.  Unless it's a parasitic choke, there would be no RF on the grid.

The GK500 is a tetrode and plays by different rules as it obviously works.

At this point I will defer to Dan.

Jack

[k9wek]

Hi Ron,

Hi Ron,

I think you have gotten some pretty good advice from the guys on this forum.

Start with the tube data sheet as has been suggested.  I'm not familiar with the 254W tube, but I'll make some general suggestions that should apply to any tube amplifier.

Since you are running the 254W at reduced plate voltage, if you reduce the plate current proportionally you know it will work because this is what happens during modulation.  However, the power input (and output) may be reduced more than you'd like.  You will not need as much grid current, or power, to drive the tube under these conditions.  Decide on what level you want to operate the tube and assume a grid current and driving power to begin your calculations.  In general, the grid current should be kept proportional to the plate current given in the tube data sheet.  If several operating conditions are given you can usually interpolate between them.

Now you need to design the grid circuit.  Calculate the input impedance (resistance) from the formula in all the ARRL Handbooks.  Input impedance (ohms) = driving power (watts) x 622,000/(grid current)^2.  Now, for a Q of 12 calculate (or use the chart) for the tuned circuit values.  For a simple tuned circuit (single ended, not balanced) the reactance of L and C will be 1/12 of this value.  There is some confusion on how to go from the single-ended values from the chart to values for a push-pull (balanced) circuit.  The reason you can use half of the C value in each section is because the grid load is across only half of the tuned circuit, and therefore an impedance of 4 times that value is reflected across the whole tuned circuit.  This makes L be 4 times as big and C be 1/4 as big.  This accounts for why C of each half can be half the value for the single ended circuit. 

This factor of 4 also applies to the L and C values for the plate circuit when you go from single ended to balanced.  Remember that the load impedance (resistance) that you should use to calculate the plate circuit values should be 1/2 the DC plate resistance.  The charts already have taken this into account.

I would throw out option 1 and option 2 as they don't have any advantage over the original circuit.  The original circuit (balanced plate circuit) is the one recommended by all the old timers, and should be the simplest to neutralize.  In order to use a pi network output circuit you need to use a balance grid circuit in order to get out of phase voltage for neutralization.  Option 3 is what Jack is running.  You can also bring the bias to the center tap of the inductor with a bypass capacitor to ground there instead of the coupling capacitor and RF choke.  I prefer grounding the center tap of the capacitor (the degenerative circuit that takes more drive, but the drive apparently shows up in the output) because it has proven to work and be more stable for me.  In any case, remember to use an RF choke if there is RF at that point in the circuit.  This is not shown on your schematics.

Also, only ground (for RF) one center tap of either the coil or capacitor on either the plate or grid tuned circuit.

I won't take sides on the debate about how much extra power your driver should be capable of.  Once you have your amplifier working, you can determine what voltage and current you will have to operate the driver at.  Link coupling between the driver and final has advantages and disadvantages.  With 2 tuned circuits there is likely to be more loss and more adjustments to make to change frequency.  In some cases you can't get practical L and C values for a Q of 12 in the grid circuit, so you will use more capacitance and higher Q which will waste driver power.

You have to get the amplifier working and then design the driver.  It would be nice to have lots of extra driving power so that you can experiment.  You may want to link couple a separate transmitter (one or two 6146's?) to the grid of the final during development.

I hope all this is relavent and gives you some ideas.

Dan K9WEK

[W8ACR]

Hello Dan,

Thank you for your input.

Here are the parameters for the grid driving circuit of the 254W tube as taken from the tube data sheet. I am extrapolating somewhat for a plate voltage of 1400V.

Plate Voltage 1400V
Plate current 180mA
Grid Voltage -330V
Grid current 45mA
Driving Power 24Watt
Platew power input 252Watt

Grid input resistance is therefore 24/2025X622000=7371 Ohm

If I use a single section capacitor, the proper values for a Q of 12 would be 140pF and 50uH.

If I use a split stator capacitor of 70 pF each side, that is a total capacitance of 35pF. This would require an inductance of 200uH for resonance. Is that correct? That seems like a pretty large value of inductance.

I do want to use the split stator capacitor. Do my calculations sound right to you?

Thanks, Ron W8ACR

[k9wek]

Ron,

Yes, your calculations are right.

You can use a smaller coil (meaning lower L) and bigger cap, but that would mean a higher circuit Q than 12.  This especially happens on the higher bands where you can't get the cap small enough because of tube capacitance.

Shooting for those calculated values gives a Q high enough for good grid voltage swing flywheel action) into the grid current region.  This is necessary for best efficiency as you can see from the tube curves.  If the grid waveform flattens out on peaks, the tube won't conduct as well.  A higher Q wastes more driver power heating up the coil.  Its a compromise.

The Q of the coil (maybe 200) needs to be much greater than the Q of the circuit (12 or more) or its losses become more significant.  Therefore use a physically large coil for best Q.

Dan K9WEK

[N2DTS]

I think I would try option one, that looks like it would work well to me.
I don't think the Q is critical in the grid stage, you can likely fudge around with that if the coil gets too big.

Triodes are a LOT different in the drive catagory from tetrodes!
My 2X813 rig takes very little drive for full power out, and I run it way into class C by increasing the grid leak resistor, and it still takes much less drive power then the pp812a rig did.
The 812a rig took twice the drive I think, for half the power out, and the 812a is a wimpy triode.

The 833 is a monster triode, so maybe using an 813 to drive it is not unreasonable if you want long life.
Still, the 813 is good for 300 watts of carrier, which seems excessive to me....300 watts to drive an 833 or even two seems excessive, even for ccs broadcast ratings....



Brett
 

[K9ACT]

I am sort of running Option 3 but with a Pi output tank.

Perhaps it's just familiarity but it certainly looks simpler and cleaner than the other two AND as mentioned, it only requires about half the drive specified in the tube date because of the regeneration effects.

Having said all that, it has always puzzled me why one needs two tank circuits on top of each other so to speak.  Either one will tune grid and following plate.

Discussing this with Dan yesterday, it seems to have more to do with physical location than any absolute technical need.

If the driver is on the same chassis, inches away, it presents an entirely different scenario than if they are far apart connected with coax.

The Globe King in one example that works with a single tank circuit.

The problem seems to be impedance matching and transmitting high voltages over a coax.

I am just paraphrasing Dan and perhaps oversimplifying  things but it made sense when he suggested this.

Perhaps he has given more thought to this or someone else has some ideas on the subject.

Jack

[k4kyv]

...it has always puzzled me why one needs two tank circuits on top of each other so to speak.  Either one will tune grid and following plate.

Discussing this with Dan yesterday, it seems to have more to do with physical location than any absolute technical need.

If the driver is on the same chassis, inches away, it presents an entirely different scenario than if they are far apart connected with coax.

The Globe King in one example that works with a single tank circuit.

The problem seems to be impedance matching and transmitting high voltages over a coax.

One of the advantages of using two separate tank circuits with link coupling between them is that each stage can have a common grounding point for all the components in that stage, allowing each stage to "float" relative to other stages in the transmitter.  There can be differences in rf potential from one point on the chassis to another just a few inches apart, not to mention what can exist between two separate chassis bases in a rack mounted rig.  These rf potential differences can induce regeneration if every component in each stage is not grounded to a single point.  Link coupling isolates each stage from the others.  

With capacitive coupling, the rf potential differences between points on the chassis cannot be isolated, since the rf ground return from the capacitive coupling is via the chassis or ground bus.  So with one tank circuit and capacitive coupling, the stages are more likely to be flaky.

I have had that exact problem with the BC1-T on 160m.  Gates just grounded everything to the nearest convenient point on the chassis or cabinet frame, and I noticed some flakiness in tuning the original configuration on 1230 kHz into a dummy  load.  There was even more after the 160m conversion.  I re-routed the grounds in the rf driver and final stages to use single grounding points as much as possible, but with components mounted on three levels, at the driver stages, 833A's, and the upper deck with PA tank components, it was not practical to find one single grounding point for the entire driver-PA circuitry without a spider web of ground wires.  Re-routing the grounds helped, but there is still some flakiness in the driver tuning.  For example, if I don't dip the 807 driver plate current  exactly, the grid drive to the 833s may increase according to the grid current meter, but the 833As run orange hot and rf output drops off.  This tells me that I still have a parasitic oscillation somewhere, and that the extra grid drive is at some frequency other than the operating frequency.  I need to check the waveform of the rf drive to the grids and see if I can find exactly what is going on. I suspect it has something to do with rf potential differences between the grounding points in the driver stage and the final.

Another advantage of two separate link-coupled stages is that one of the links can be made variable, which makes it very easy to adjust the grid drive to the following stage.

Of course, link coupling leaves you with the disadvantage of having to adjust two separate tuned circuits between the stages.

One way around that problem is to use "unity coupling" between the stages.  The plate coil is a tuned circuit in the conventional sense, but the grid coil is wound on the same form as the plate coil, preferably interleaved between turns of the latter.  It is not tuned, but the tight inductive coupling between coils allows it to function just like a single tapped coil, and yet still provide isolation between stages that would not be possible with capacitive coupling.

[KM1H]

The 1952 HB amp is OK but Id either shunt RFC-4 with about 3K or replace it entirely with a resistor. C-11 should be at least a .02. RFC-5 is fine, it is simply a parasitic choke resonant way above 160M. A low value carbon resistor does the same as well as may eliminate the need for neutralizing; Ive mentioned these things before.

Lets skip Circuit 1, there are several problems with it.


As for Options 1, etc you certainly like to eliminate bypass caps!

Since it has been established that its all on one chassis I see only one need for multiple circuits and link coupling. That would be for stagger tuned broadbanding since you have power to burn in a properly functioning 6146 driver. The same can be done with two coupled coils as mentioned a few times earlier. In either case a Q lower than 12 is desirable.

What has not been answered is the condition of your tubes as well as the amount of drive to the 6146 grid. Until that has been answered we are laboring with too many unknowns.

Carl
KM1H

[W8ACR]

Thanks again for all the comments and suggestions. It is painfully obvious that I have a lot to learn here.

The 6146 is a good tube, and the 6AG7 seems to be working just fine with a nice dip to about 10 mA at resonance. Plate, grid and screen voltages are correct on all tubes as well.

If I use inductive coupling to drive the 254W, should I use twin lead, coax, or a single wire with the other side to chassis for the coupling lead? Also, should I also use a Q of 12 to design the plate tank of the 6146? The tuning cap of the 6146 is grounded on the rotor in the original design, so I'll have to redo the plate tank on the 6146 if I want to use inductive coupling.

Thanks, Ron

[K9ACT]

>The tuning cap of the 6146 is grounded on the rotor in the original design, so I'll have to redo the plate tank on the 6146 if I want to use inductive coupling.

Another interesting situation here.

We think of inductive coupling as a ground  isolation trick but in most cases, the link goes to a coax connector mounted on the chassis.  Then a coax to the other chassis and another connector and piece of coax to the next link.  There are all sorts of theories about which end of which coax to ground, adding even more confusion to the subject.

If you use a link to twisted pair and a receiving link, there is no ground connection at all and no need to even think about how the driver is tuned.

To confuse things further, I am using a Pi tank in the driver to a link in the power amp grid with coax between.  Again, where to ground the coax?  Does it really matter?

js

[WU2D]

Have you decided on Rice (grid) as shown in the original schematic or Hazeltine (plate) neutralization? You can use the Rice on either a link or Pi output and it has many advantages in implementation and power rating of components, but the Hazeltine is a little better system according to many handbooks and most BIG RIGS use that. 

Don't forget to think about the driver - it must be neutralized especially if you go with a bottle rocket like the 6146.

Mike WU2D

[Gito]

Hi Bret.

why using a 813,because after the 807, we got 6146 and to jump to 813,
so as I used the 3 to 1 formula driver wattage  ,if  i used a 807 or 6146 ,I need 2 807 aor 2 6146, when using two tubes in parallel ,sometimes one of the tubes goes weaker,so I must replaced /find another tube to have the same power,it happens to the two 807 in my 833 transmitter ,which I regret later.
So to over come it I used the only option I have a single 813,of course not with it full rating,you can used a 1500 v supply load it to 120 ma giving a  180 watt input or 126 watt  output, that's about 3 times the drive power needed for two 833.(40watt) . so follow the leader (ARRL) of course any body can have different opinion ,but I only follow the Leader.
 
Sometimes,we over looked the parameter data of a tube.
Like the 254W a "small" tube and small output power comparing a "monster" tube like 833 tube.

The input drive of 254W tube needs is    25 watt  drive power .
The input drive of 833    tube is             20 watt  drive power.

So, You can clearly see which tube needs a higher drive ,even 833 gives a higher output power compare to 254W ,25 watt is 25 watt and 20 watt is 20 watt right.

Ron attached a schematic diagram of  a 175 watt transmitter on 160 meter band from the ARRL hand book it used a 6L6 as a driver that is capable of 20 t0 30 watt output (Class C) an a 812 final that needs only 6 watt drive power.So the driver is more than 3 times the 812 needs

Ron You can used this circuit but with a pi-section coupling between the driver and the final,but of course with different values that;s  needed
Because with pi-section ,you can match and load easily the output impedance of the driver and the input impedance of the final and the power needed.


I re attached the circuit of the 175 watt transmitter in upright position.


Gito