Selecting the PA screen grid resistor in an AM tetrode design

[wa3dsp]

I am working on a Ranger but this question applies to the many tetrode 807, 6146, etc designs out there. It seems that there are many different approaches to this. When I was restoring an AF67 and later an A54 the screen was derived from the modulated plate voltage through a 25K resistor period.

On the Ranger there are many variations in mods but you often see a 30K from the plate supply with a 20K from screen to ground. The stock design used a clamp tube which becomes a high impedance with proper drive so essentially the 30K was the only thing there.

If you take the data from the tube tables for a 6146 in class C plate modulated with a plate voltage of 600 volts the screen should be ~175 volts at about 9 ma.  So 600-175 = 425 / .009 =  47K screen dropping resistor

So the Multi Elmac's ran the screen fairly hard? or is there something I am missing here?

Also is there a reason why you would use a voltage divider here? This throws away some amount of modulation audio. I know floating the screen is a bad thing and possibly this is some amount of protection should the series screen resistor open but that seems like  a minimal risk for the expense of wasted power.

And lastly in the old handbooks and some designs you see bypassing of the series resistor for audio frequencies. You rarely see this in amateur commercial designs though. Is this something that is necessary or desired?

73 Doug

[Steve - K4HX]

The article at the link below may be helpful. It's for a pair of 6146s, but the concepts apply to a single.

http://amfone.net/Amforum/index.php?topic=32893.0

[wa3dsp]

OK, so let me see if I understand this correctly for the 6146. From the tube tables it shows 8-9 ma screen current at 175 volts or so depending on the version of 6146.

If I use those numbers and my plate voltage is 650 volts then in your example R1 = 88K and R2 = 132K

Are these numbers anything like you used? Please give real numbers including capacitance in your final circuit.


73 Doug

[Steve - K4HX]

You'd have to ask Dean what he did on his DX-100. I did something similar on a Viking II. Don't remember the numbers (that was almost 20 years ago) but I used adjustable wire wound resistors and set them for the best scope pattern.

[wa3dsp]

QRZ shows Dean as being in Tuscon AZ but no email.

[KA2DZT]

If you are modulating the screen through a B+ dropping resistor off the modulated plate voltage you don't use any voltage dividers.  There should not be any resistor to ground.

For one 6146 the screen resistor should be about 50K with 600-650 volts plate.  The screen by-pass cap should not exceed .002ufd.

Connect you screen resistor from the cold end of the plate choke directly to the screen.  Use a .001ufd 1KV disc ceramic to by-pass the screen.  This cap should be located at the tube socket.

Fred

[wa3dsp]

Fred,

   Yes that is what I would have expected BUT there are so many different circuits out there that it is hard to determine what is science and what is just "Lets try this."

The Ranger article in the July 2011 ER has a 30K to plate with a 20K screen to ground.

The original Ranger had a 30K to plate.

The Multi-Elmac's uses a 25K to plate.

All these rigs have similar voltages and use 6146's

Then there is K1KNX's work using a mixture of modulated and unmodulated plate voltage with a bypass around the resistor to modulated plate voltage -  link above.

Rather confusing.

[VE3LYX]

I built my 6146B linear(Actually a 6293 but I have had a 6146B in there too). I used a best guess like your calculation then fired it up and measured the actual voltge under load (carrier only because I was being careful) I found I had to change it some . I forget what it is but it took more then I thought it would to get the voltage in spec. Some tets dont care but I think the 6146 series does. it is an unusual tube in that it does not take a lot of drive to make it work. My theory, anyway, is that most things powerful are also capable of self destruction.  I used just a down leg from the plate supply(B+ but before the RFC for mine. Some might find it crude even but it idles well with no residual Rf and pops in smoothly when a signal is introduced. I have no schematic. I just built it establishng first the DC circuits and values for proper tube operation per 6146B tube data sheet and then wired in the RF stuff afterwards. Not much to it. Tuned input  (with input link), cathode backed up 4 ohms to keep things in check. Typical grid resistor I think 4.7 K but twas a while ago. 150 pf in to the grid from the input tuned circuit. Tuned output  (LC parallel tank) outut from a 4 turn link over the output tank coil (centred. )I originaly built it shunt fed with Pi but have come to prefer the hot tank with a link out. A homemade RFC and typical ham style parasitic suppressor on plate lead. I was not sorry I established the screen resistor that way because it works well and I KNOW it is right cause I measured the voltage. before and after. I have had it on a 700 volt supply but because of the fact I have only 1/4 watt AM HB exciter to drive it it seems pointless to run it full throttle when it isnt being hardly even pushed. Normally I run it at 1/2 that and it does what I want with this exctier. This one tube, even at the same voltage, does more powerwise and behaves better then a two stage HB linear does with parallel 6l6s out.  I also discovered these tubes are senstivie to proper heater voltage. If it is a bit low power falls dramatically. At least in the 6293 tube. A bit low means less than 6v at the minimum. Probably the 6146B is no different.
Don Ve3LYX
I took the lid off and stood on my toes to see. I cant see the screen resistor well and I am also colour blind severely so I would have to take the cables all off ,take it upstairs and ask my wife "what colours is this resistor?" Then there would be the "What?" "I dont know what you want." "Which one" etc etc  and I decided it would be quicker (and was) just to use your calculation first(which you have)  and verify and adjust as necessary. 

[Steve - K4HX]

Trust me, the article Dean wrote is legitimate. Dean was on the air for many years and always had an excellent sounding signal. The values were derived from actual measurements.

Fred,

   Yes that is what I would have expected BUT there are so many different circuits out there that it is hard to determine what is science and what is just "Lets try this."

The Ranger article in the July 2011 ER has a 30K to plate with a 20K screen to ground.

The original Ranger had a 30K to plate.

The Multi-Elmac's uses a 25K to plate.

All these rigs have similar voltages and use 6146's

Then there is K1KNX's work using a mixture of modulated and unmodulated plate voltage with a bypass around the resistor to modulated plate voltage -  link above.

Rather confusing.

[W2PFY]

Dean is in the NE  area. Send Tim, HLR a PM for his details.

[Steve - K4HX]

Another thing to consider is that if you haven't removed other sources of distortion (mod transformer, modulators, etc) doing the screen dropping resistor mod is not of much benefit.

[KM1H]

The screen resistor to ground provides a constant load and keeps screen current from going negative.
This is a common procedure used in high power tetrodes where negative screen current is normal under some tuning conditions and resulting in SSB distortion; it is even mentioned in some Eimac spec sheets.
The G3SEK Tetrode Boards introduced in the 90's take care of that, and other tetrode IMD problems.  That circuitry added a 5-6dB IMD improvement to my 6M NCL-2006.  http://www.ifwtech.co.uk/g3sek/boards/tetrode/tetrode-1.htm

As far as AM I must have missed that post and will be retrofitting to already restored rigs and adding to new ones for myself and customers rigs after first confirming results. It appears that the series resistor is fairly critical....does it hold with tube changes?

Carl

[wa3dsp]

The story behind the resistor to ground stems from, among others, Johnson's use of the screen clamp tube to protect the final if no drive was present. In the Ranger they used a 6AQ5 in combination with a 30K screen resistor. When there is full drive the plate of the 6AQ5 is at high impedance so essentially there is no screen to ground resistance. In one Ranger article in ER (July 2011) the author removed the clamp tube and used fixed negative bias on the PA grid. At the same time he inserted a 20K resistor from screen to ground leaving the 30K screen to Plate. Perhaps this was to compensate for the clamp tube not being there but I am not sure why as it would be at a very high impedance in operation. I could be wrong but I don't think the negative screen problem was an issue with the 6146 but rather other higher power tubes.

Generally the formula    (Plate voltage - screen voltage) / screen current is used.  So for a 6146 operating at 600 volts this would be  -   (600-170) / .008  = 54K   for a single resistor  or divided in the .4 / .6  ratios as described in WA1KNX's article.

Why Johnson uses 30K or Multi Elmac used 25K in their designs I do not know and at this point there is no one to ask.  Other manufacturers used values closer to the formula. It would be a lot easier on the clamp tube and much less wasteful of power to use the higher values. Perhaps they were trying to squeeze every last watt out of the PA. One thing is for sure in the Johnson equipment they were certainly not trying to conserve energy.

73 Doug

[VE3AJM]

Dean wrote an article in the AM Press Exchange on all this. This may help you out until you can get in touch with him directly.

Its issue #71 from May 1989, and its available here online.

Al VE3AJM

[VE3AJM]

Its the same article that Steve K4HX had posted a link to earlier in this thread.

Al

[K4IMZ]

The Ranger article in the July 2011 ER has a 30K to plate with a 20K screen to ground.

The original Ranger had a 30K to plate.

Anybody have any source for the Ranger R15 30k 25W chassis mounted wire-wound Resistor ??

I have not located a source.

73 , Randy

[WQ9E]

Randy,

They are a stock item at Mouser:  http://www.mouser.com/ProductDetail/Ohmite/L25J30K/?qs=sGAEpiMZZMtlubZbdhIBIK0OZQc57YR%252bMO6M%2fHrbngA%3d

[ke7trp]

Theory, math and schematics are in any of the older hand books on this subject.  Get a mid 50's book.  My favorite is the 1954.

Most of my rigs here use modulated screen voltage using a dropper, My 2x4, my GK500, ranger and valiant.  No resistor to ground. Not needed here. Many of the bypass caps are way to tight at 005 for the WRL products. It really cleans up the output and makes a real nice clear sound once the mod is done.

The biggest difference was on the king. They used a choke to "self" modulate the screen.  With that unhooked and a dropper installed, there was a very different clear tone from the rig. Modulating the screen with the plate in phase is a no brainer.. DO IT.

[K4IMZ]

Randy,

They are a stock item at Mouser:  http://www.mouser.com/ProductDetail/Ohmite/L25J30K/?qs=sGAEpiMZZMtlubZbdhIBIK0OZQc57YR%252bMO6M%2fHrbngA%3d

THANK YOU !! 

73 '  Randy

[WD5JKO]

In reading through this thread, a couple of things kept running through my mind.

For a class C amplifier, consider the screen circuit for the three conditions, key down, key up, and AM. I have some issues with some of the suggestions in that they do not consider the CW key UP condition.

One thing about a series dropping screen resistor is that with no excitation the screen voltage will rise to the plate voltage value. Sure some rigs use a clamp tube, but some do not. Most tubes have a maximum screen voltage rating, and exceeding that value by a large factor is not a great idea.

Always ask yourself when making a mod here is what happens when running CW? Does the safety bias keep the tube safely at a low plate current when key up? Think of a rig with a 4-400 that had a screen choke originally to a 500v supply, and now you have a series dropping resistor from a 2-4 KV plate supply? What happens when the plate supply is ON, and the key is UP? At the minimum you should beef up the screen bypass capacitor voltage rating to at least the plate supply value.

Many tubes when plate modulated need a reduced modulation percentage at the screen. The 6146 is one of these tubes that often over modulate the screen such that the screen goes too low while the plate is still up in positive territory. The result is a sudden and premature "white lining" when looking at the transmitter modulated envelope. Ever see a rig that can hit 100% modulation negative when the upward modulation is only about 85%? This is one reason why. A resistor divider between the B+ and modulated B+ can address this issue where the parallel combination of the two is equal to that of a single series dropping resistor, but the ratio of the two can set a lower modulation percentage for the screen grid.

The case of a series dropping resistor, and a screen to ground resistor will work similar to what I just described since the b+ should be an AC common, but will also act as a DC voltage divider to limit the maximum screen voltage when there is no drive, and no clamp tube present.

My Gonset G-76 is interesting in that  there is a screen dropping resistor with a VR tube in series with it, and a screen to ground resistor. There is also a 12AQ5 clamp tube (junction of screen dropping resistor and the VR tube). This rig has no fixed bias supply. When in CW mode "key-up" the VR tube extinguishes as the clamp tube clamps, the plate current is almost zero. On AM the modulation is excellent. This is a very clever circuit using a sweep tube final.

Configuring the rig to display a trapezoid pattern on the scope will reveal issues with modulation linearity.


Edit: here is what RCA says about the 813 screen supply when in CW mode, see image below.
Edit 2: Looking at Eimac data, the maximum screen voltages for the following tubes is:

4-250 = 600V (RF amp)
4-400 = 600v (RF amp), and 800V Audio amp
4-1000 = 1000V

So running high plate B+ on one of these tubes with a screen dropping resistor (no voltage divider or clamp tube) could be a big problem if the amplifier is keyed without excitation, like CW key up. A clamp tube for one of these tubes would have to withstand the full B+, so the usual 6AQ5 is out...maybe a 6BQ6/6DQ6 with a separate G2 supply?

Jim
WD5JKO