The AM Forum

I am in the process of designing a 304TL RF stage and working on the PI network.

Why 304TL's and 813's? Because they are burning a hole on my power tube shelf and are asking why they are being ignored?  ;D

I am trying to locate some load data for the 304TL's for either the "Resonant Load Impedance," or "Effective Load Resistance."

Bands will be 40, 80, and 160m.

Thanks.

If you can't find anything, four of the internal triodes in parallel would be a place to start for RL @ Ebb.
What, say four 100tL's ?

Now THAT'S what I'm screamin'! I'm right there with 'ya man; been thinking about it on and off for several years now since I probably have all the stuff to bolt one together. Let us all know how it's coming along! BIG STRAP!

73, Phil

I was planing the same tube line-up for a xmtr.  I think the 304s are four 75s in parallel.  I was thinking 75m, probably 160m and maybe 40m.

Fred

Here is some data from Franks Tubes.

That is an odd one!
The TL/TH tubes look very cool, and really impressive when glowing almost white.
High voltage tubes though, they do not work very well at low voltages.

I never tried 813's as modulators, and never used TH or TL tubes in RF service.
When I had the 30K1 I collected 100TH and 100TL tubes to replace the 75TH tubes that rig used as modulators, and built up a PPP 100TH mod deck which works well but takes a lot of drive power.
800 watts of audio at 2500 volts, 900 at 3000 volts...my bias supply runs out at about 2500 volts.
If you use the TH/TL tubes, you MUST put them behind glass, they will freak people out.
http://n2dts.smugmug.com/Ham-radio/i-z345zrf/X2

Triode or tetrode setup on the 813's?
Is one cleaner then the other?

Get a hold of ER magazine issue #10 from Feb. 1990.

An article by Bill KD0HG sk on the Kilowatt class C RF deck using a single 304 TL with 2500v on the plate and PI network tank circuit 160-20m.

Al VE3AJM

This will be a triode pair of 813's with the system HV running at about 2200 VDC.

Four 75TLs.

Yeah, Fred DZT mentioned that too. I should have guessed better given Eimac's nomenclature starting with 3.

I have one but only two of the triodes light up. Use it as a display.

I believe it can still be used at half it's rating; i.e. 152 TL?

Chuck, W3FJJ built a final with a single 304TL.

Yes, why not a 152 TL.  Good idea.  Use the dead half as pp neutralization. 15 watts in, 350'watts class C out as per 1/2 of ARRL handbook specs.

I used to run a pair of 304TLs modulated by a pair of 833As.  It worked great!   I did run 3kV on everything (common power supply).  Used an 813 to drive the 304TLs.

A single 304TL could be modulated by a pair of triode connected 813s I suppose.  If you want to run a lower voltage, something like 2000V at 400mA may be practical.

I built  a pair of 304TL's in parallel, modulated by a pair of 304TL's
for 160-80 meters, I was hoping it would do 40m too, but
trying to neutralize eight  75t's in parallel is a bear!

A single 304tl in the final would of been plenty enough, but I was
looking for symmetry.

I like the 304TL'S more as modulators, than RF service.
I ran the modulators , AB1,  direct coupled with a pair of push pull 2a3 speech amp.

High Voltage Power Supply, was common capable of 2500v, but  I usually
ran it at 1800 volts,  300 mils, and got 400 Watts out..
No color on the finals and slight blush on Modulators
Just loafing...

Exciter was a DX-100 running 75 watts into the grids..

The rig got dismantled in the QTH move, but all the parts are still here, and
I like to resurrect it.. I am seriously considering changing the finals to a single 3-500z
though, and keeping the 304TL Modulator as is...

Chuck- W3FJJ

Thanks Chuck,

Do you recall the load impedance you used for the pi network?

Thanks

Phil

@w3fjj
i have read somewhere that a high µ triode like the 3-500z is not the best to be plate modulated in class c, tell me is that true?
73's from John

I did and Ray O. said the issue was no longer available but he does have the paper article and will send it to me

Ray is way cool in my book.

[/quote]

Thanks Chuck,

Do you recall the load impedance you used for the pi network?

Thanks

Phil
[/quote]

Hi Phil,
I can't find my notes, but I think I used around
3000 ohms with  (Plate voltage/2*plate Current)  Q of 12 as starting point.
 but tweaked it some, since I had roller inductor for L1.
I found I didn't need as much inductance for L1 as handbook
suggested.. So extra inductor in series with roller was
removed..


My Pi-net  was 500pf , 22uh , 2500pf . I used full 22uh
on 160 and about half that on 75meters..

GL   Chuck

Really nice work Chuck!
Matching meters, and plenty of them!

I ran some calcs.

R = (2,200^2)/(2X550) = approx. 4400 Ohms.

R = V/I = 2200/0.545 = 4037.

And from the 304TL spec sheet it looks like the plate-to-plate resistance for AB2 is 5,000 ohms so for single plate the load should be 2500 ohms.

So I think I'll run the Pi-network numbers for 3,450 ohms and see what I get and compare.

Grid bias selected is -380V@70mA (according to tube curves) with 40 Watt drive from an 807. The -450V bias supply is adjustable.

For Class C, the plate load (Rp) is approximated by Ep/2Ip.  So, for your parameters ,

Rp = 2200/(2 * 0.545) = 2018

Thanks Steve.



If the tube has listed an "Effective RF Load Resistance," as is shown for the 6146 for linear service,
http://frank.pocnet.net/sheets/049/6/6146B.pdf, this would make things a lot easier. :D

I know from personal experience the RF load data for the 6146 is right on.

I guess back then the assumption was the circuit designer was going to bring the tube into his lab and run impedance, load, and power experiments to suit the  intended design.

That formula has been around for a long time. I'll have to check Terman's book to see if it dates to the 1930s.

Anyway, if you have a reasonable range of adjustment on the variable components in your pi-network, calculating the "exact" load impedance is not critical. You'll tune up the transmitter anyway, and at max power out, the load impedance will be correct.

I know this comes from:

R = V^2/2P for class C,

and since Power = Iplate*Vplate

then

R = V^2/2(IV) = Vplate/2*I,

but does it take into account the Rp, the plate resistance of the tube?

The pi circuit will have a bandswitched L and extra C needed, and of course the two variable caps.

The plate load of the tube is determined by the operating parameters and, as such, is a static value. Plate resistance is a dynamic parameter and can be determined from two other dynamic parameters, the quotient of the amplification factor (µ) and transconductance (g_m) of the tube. Knowing that

µ = ∆E_p / ∆E_g , and

g_m = ∆I_p / ∆E_g , then

Plate resistance is

r_p = ∆E_p / ∆I_g .

To answer your question, when comparing the above formula to the one in my previous post for plate load,

R_p = E_p / 2I_p ,

I'd say yes. The factor of 2 is there to account for the much less than 180 degrees of current flow in Class C operation.

Triode-connected 813s are exceptionally clean.  Drive the grid and screens tied together.  I have used them in at least seven plate modulated rigs both as modulators and finals and the THD and IMD testing always came out FB.

I have run up to 3KV on plate modulated 813s. The modulators will need some forward conducting diodes or zeners in the cathode for bias to idle properly when above 2KV or so.

813s CAN certainly be run as tetrodes, as per Eimac specs in AB1, and with well regulated screens, grids and a stiff plate voltage they will perform very well too.

Some of us have used lantern chimneys for 813 air cooling and they will probably do at least 200 watts of dissipation with lots of air.  The lantern glass can be found on the web. The 813 filaments have plenty of emission to support bigger power output.

The 813, like the 4-1000A, is a very rugged and versatile power tube.

T

Tom,

Forward conducting diodes when run above 2KV.  Do you mean that the 813s need some negative bias when running as class B triode connected modulators??  Like 811s need some negative bias when running above 1300 volts.

I'm confused by your words "forward conducting diodes".

Fred

Hi Fred,

Yes, the triode-connected 813s will require some bias when run over 2KV or so. Depends on where you wish to idle them, in class B, class A or whatever.

Use regular 1 A  rectifier diodes  (or a 50W zener) and use a rotary switch to short them. This will select the proper bias and idling current.

The diodes are placed in the fil xfmr center tap with arrows (cathodes) pointing towards  the fil xmfr so that they become forward bised on key-up. This will give about 0.6 V per diode. It will require a bunch to get what bias you need.   You know the drill.


*   If someone wants the ultimate clean modulator system, drive the 813s with a MOSFET solid state driver.  No driver transformers. Put NFB around the modulators back to the input of the SS driver. Don't get any better than that.


WA1GFZ SS driver:  (I still use it today on the 4-1000A rig)

http://amfone.net/Amforum/index.php?topic=23632.0

T

What is the power output of the triode connected 813's?
Say at 2000 volts.
And do they run at 1200 or 1300 volts?


813,s seem like great tubes, I have about ten of them, and have had the same two in my big rig (rf) for 35 years I think.


I have some Russian 811a's in one of my modulators and they make me sound slightly Slavic, some all American gangsta tubes from the 40's would be better...Sam Spade type 813 sound...

No, that would be James Cagney type 813 sound.

No, I think he mostly got filled with lead, Sam Spade got beat up sometimes but always worked.

Power output would depend on how hard they are driven and the cooling - and how clean we want them to be.  But a pair modulating one or two 813 finals would be no problem.

I do remember a triode-connected pair modulating a pair in class C at 3KV making an 800 watt carrier output with a sine tone showing 130% modulation. This was at extreme full tilt with chimneys.  I was amazed at what those 813s could do - and the plates were still gray. (graphite plates)  With no air they did show color, however, which is not good.

If we run 813 modulators at 1200 volts, then they would need some POSITIVE bias to idle them properly. I'll bet the transfer curves would look OK at that HV.   The way to find out is to run some tones thru  for an THD test once the rig is up and running.  (RF output sampled)

T

I have an RF deck with four 813's in it, used to be my flex radio amplifier.
1200 watt carrier at ccs ratings in class C!
And about 100 watts with screen modulation!

I have no use for that kind of power (1200 watts), nor do I see any reason to push tubes hard, the extra 100 or 200 watts out you get does almost nothing.

Positive bias as modulators! That is interesting.
So about 1700 to 2000 volts would be the zero bias point?

Hooking up the 813's screen as low or high mu triode may affect bias set point for minimum distortion.

Has anyone ever done that?
Use various grids?
On the G76 they used 6dq6's as zero bias triode modulators using the screen as the grid I think, and grounding the grid?
Something goofy like that, and who figures that stuff out?

How would you measure the distortion?
 

Brett,

   Well one way is to use an SDR receiver. Distortion products down -40 db from a tone is 1% distortion.

My G76 can just get there when at 50% modulation. I have a negative peak limiter that has a soft threshold, and the distortion rises rapidly above this level. Fortunately I don't talk with a sine wave.  ;)

Jim
Wd5JKO

I have done it that way also, but the RF deck is added to the mix.

   This is like rear wheel horsepower instead of advertised Gross Engine horsepower. Rear wheel is where the rubber meets the road.

Measuring the audio distortion directly means you must properly scale the voltage with a compensated voltage divider (like 100:1 or 1000:1), and then feed that into an audio distortion analyzer for analysis. Problem here is the RF stage modulation linearity might be crappy, when the audio modulator is almost perfect. The result is still crappy signal.  :P

Jim
Wd5JKO

I did some calculations on the PI network using an effective load resistance of 2000 ohms with a Q of 12, and came up with nearly the same inductance values as found in Bill Kleronomos' (KD0HG) article (A Kilowatt Class RF Deck, ER #110, Feb. 1990) ) after doing some reverse engineering on his PI circuit.

I will be incorporating some of his ideas into my modified circuit.

The P-P 813 mod deck will be modeled after K1JJ's schematic.

Now to find someone who can cut and fashion the RF deck. I have too many other projects at this time.

Phil

Update;

Here is a draft of the RF Deck, Exciter, Power Supplies, Control, etc.

I am only going to use 3 of Bills (KD0HG) grid tuning coils, namely 160m, 80m, and 40 meters. I especially like his design because there are no DC voltages in the grid tuning circuit.

I also have to look through my stash of meters to find out what I have so the metering resistors may have different values, and I may include one more meter so not so many functions are monitored by one meter.

The 304TL grid tuning circuit is switched by a 4 pole bandswitch. I may modify that portion of the schematic to look less busy and to look more like Bills schematic.

Regards,

Phil

I am not sure that is going to work.
I have never seen a class C plate modulated RF deck without a tuned input to the grid.
You need to drive the grid with a lot of voltage and it makes current so there is a load on the driver...

I have seen it done in an amp using a tetrode where the tube runs in AB1, no grid current, a load resistor was used as the load....

Did we not carefully look at at the entire pdf attachment, and READ Phils last post?? No tuned grid circuit?? What?? Please look at all of it again...Sheesh...Looks good Phil. I worked Bill quite a few times with his 304 tx.

Ever heard of the Johnson Thunderbolt amp that used a passive untuned grid circuit as one choice for class C CW?? Theres no reason why a passive grid input circuit using non-inductive resisitors at the grid wouldn't work for a plate modulated RF deck.

Yet, that's not what Phil is doing with his grid circuit!? It is tuned...

Al VE3AJM

I only looked at the first part, DUH!

Below is an updated schematic which hopefully is more straightforward and less busy.

V1 is the oscillator and VFO amp, V2 is the Buffer/Driver amp, and V3 is the Final.

Basically, the output of the oscillator V1 feeds the 807 Buffer/Driver (another bunch of tubes I had).

The Buffer/Driver drives one of three, switchable, series resonant circuits. Each coil of the series resonant circuit is overwound on a companion coil.

The companion coil forms a parallel resonant circuit for peaking the grid voltage.

V2 has a screen drive adjustment for setting power levels.

All resonant circuits are isolated from DC by blocking caps. This keeps any LV or HV off the coils and since no DC current flows through them, the only heating is from pure RF.

Phil

Does that grid drive setup really work?
I have never seen anything driven that way.
And why worry about coil heating, because of the high power drive needed?
(out of an 807???)

I just have a hard time thinking you are going to drive a grid way into class C with loads of power and current with no direct connection to the grid, no design I ever looked at (in handbooks and so on) is like that.

The 807 acts as a wide band (1.8 to 7.3 MHZ) RF amplifier and should supply anywhere from 5 to 40 watts of RF power.

According to the 304TL data sheet, the drive power needed for Class-C operation at 410 Watts output is about 15 Watts@2200 volts Vp, so there should be enough drive from the 807 to do that. (Of course, there will be some circuit losses).

On the schematic there is a 0.01 uF cap from the high end of the companion (RF transformer secondary)  coil set (via the bandswitch) to the 304TL grid resistor. The reactance of this capacitor at 1.9MHz is 8.4 ohms, plenty low enough to transfer most of the RF energy from the top of the parallel resonant circuit to the grid. BTW, the "top" end of the parallel resonant circuit is toward the bottom of the schematic. Neutralization "N" RF comes in at the bottom end of the parallel resonant circuit which is at the Top of the schematic. The grid bias will be set to approx. - 350 Volts.

Bill inferred from his article that RF circulating currents in the PI circuit did heat that coil and recommended larger diameter coils or coils wound on ceramic or Mycalex forms.

It's been said here http://amfone.net/Amforum/index.php?topic=13025.5;wap2 :

For Class A operation: Z(load) = V(DC) / I(DC)
For Class B operation: Z(load) = V(DC) / [1.6 x I(DC)]
For Class C operation: Z(load) ~ V(DC) / [2 x I(DC)]

But this simplification bothers me. Does the following have merit?
When trying to calculate the plate load impedance, picking a 'class' seems to cover a very wide set of conditions. In a given class, because bias, drive, etc., are all different from rig to rig, the conduction angle is not a fixed value (240 deg., 180 deg., 60 deg., etc.therefore is not leading to a fixed Ep/Ip -vs- RL ratio for each class, I tried to base the ratio on the number of degrees of conduction.

OK:
class A conduction angle 360 degrees. Z(load) = V(DC) / I(DC)
class AB conduction angle from 359 ~ 181 degrees. Z(load) = V(DC) / [1 ~ <1.6 x I(DC)]
class B conduction angle 180 degrees. Z(load) = V(DC) / [1.6 x I(DC)]
class C conduction angle from >0 ~ 179 degrees.   {Z(load) ~ V(DC) / [2 x I(DC)] is in there somewhere..}

I don't recall the original calculations but it involved an exponent, pi, and some other stuff to figure peak cutrrent and actual plate swing.
I had written a program for it but the source code is lost.

Doesn't it make sense to base the RL or  Z(load) on a calculation taking into account the degrees of conduction? Or am I wrong about it?

Terman suggests anywhere between 120 degrees to 160 degrees as the best compromise in terms of conduction angle and efficiency.

R = Vp/2Ip is a short hand way or rule of thumb method often used when no resonant load impedance or when no effective load resistance data is available from the tube info.

If you want to be more exact, you can use the equations from the references below.

The paper below is a good read regarding Tank Circuits.