The AM Forum

This circuit has puzzled me for long enough, and while it has worked for decades, it just doesn’t sit well with me. In the attached RF bias circuit, both grid leak and adjustable bias are employed, but shouldn’t the current meter be placed in the circuit at point “A”? It would make sense if the adjustable bias were added later but it seems that the meter was never relocated.

Proper indicated grid current is 10 mA.

I believe it better belongs on the other side of that solder blob next to "A".  ( Or is that a Fahnestock clip?)  That is, between the top of the 7K resistor and the RFC.  There, it would measure grid current, and not a combination of grid current and the bleeder current of the bias supply down through the 7K resistor.

Argh. Yes, I meant to put “A” where you suggested. Initially I thought about eliminating the grid leak and the removal of the 7k resistor, but why make life difficult? The RF bypass cap would then connect between the meter side of the RFC and GND.

The bias is connected to the wrong point.  The neg bias should be connected in series with the grid resistor to ground.  Grid current meter can be left between the grid resistor and the bias supply.  Grid resistor should be adjusted to drop the remaining grid voltage needed after subtracting the bias voltage from the total grid voltage needed.

Your bias supply should be treated as a two terminal black box (+grounded) (- to the bottom of the grid resistor).  Grid meter in between the grid resistor and the neg terminal of the bias supply.  The meter will only read the grid current and not be affected by any bleeder currents.

The grid current creates a load on the bias supply just like any other PS only the voltage is negative.

Fred

I personally do not like ammeters as I prefer sampling voltages across low value resistors and using voltmeters calibrated in mA per mV.

However if this is a protective bias circuit with grid leak, the ammeter should be in series with the grid choke immediately to the left of the choke, with a RF shunt cap to ground just to the left of the grid choke and to the right of the ammeter.

The meter IS functioning but it is not correctly measuring the total grid current in the circuit.




Phil

This circuit should work just fine, although I would put the meter just to the left of the RF choke. Any current flowing in the circuit in either direction (toward the grid or toward the negative power supply) must flow through the 7K resistor. We like to think of current as the flow of electrons through a conductor, but by convention, current is defined as the direction of flow of POSITIVE charges. This is counterintuitive. Try to imagine positive charges flowing up out of the ground point. Another way to say this is that when grid current flows, the power supply is under LESS demand, not more.

The bias supply is providing protective bias only, and the operating bias is being provided by the grid resistor. If the grid current meter were located between the top of the 7K resistor and the RF choke, then with no RF excitation, it would read zero. However, with the meter located as shown in the schematic, there would be a small amount of current flowing from ground through the meter toward the bias supply. When RF excitation is applied, current (positive charges) flows from ground to the grid, and the bias supply has nothing to do. In either condition, all current is flowing through the 7K resistor. (Actually, in the circuit shown, there would also be some current through the variable resistor). There is no current flowing from grid to bias supply or vice versa. All current is going through the grid resistor either toward the grid (RF on) or toward the bias supply (RF off).

This did not make sense to me when I first read about it. It is discussed in detail in the old radio handbooks.

If you insist on envisioning current as flow of electrons through a conductor, it still makes sense. The bias supply is shoving electrons downstream toward the grid. With no RF excitation, and no grid current, those electrons have nowhere to go but down the 7K resistor to ground. When RF is applied, the electrons coming from the grid meet the electrons coming from the bias supply. Again, they have nowhere to go but down the 7K resistor.

I hope that this helps (and that I am right).   :D

Ron

As I look at your schematic again, another thought comes to mind that may help illustrate my point. As I said, if the meter is placed just to the left of the rf choke, then it would read zero if there is no rf excitation. Likewise, if it is placed at point “A” it would read zero, or close to it, with rf excitation. The current is flowing through the 7k grid resistor either toward the grid or toward the bias supply. Or, if you prefer, electrons are either flowing from grid to ground, or from bias supply to ground. There is no current from grid to bias supply or vice versa. (At least with the circuit shown.)

Ron

Probably a few ways it will work, the meter is in the wrong place in the drawing posted.  When you read the tube manuals it says you can use grid leak bias or a combination of fixed and grid leak bias.  The circuit shown is only using grid leak bias even though there is some fixed bias.  When the grid leak voltage exceeds the bias voltage, current can flow back through the power supply since it is actually in parallel with grid resistor.  The bias supply should be in series with the grid resistor.  The shunt resistance of the bias supply will cause the negative grid voltage to be lower for a given grid current.

When you read the tube specs it will tell you what the grid resistor should be.  That stated resistance with the correct grid current will produce the correct negative grid voltage for the class of operation.

When you use some fixed negative bias you must reduce the value of the grid resistor.  The new value of grid resistor with the same grid current will drop a voltage amount that when added to the fixed bias will be the same level of negative grid voltage as specs require for the class of operation.

When the bias supply is in series with the grid resistor it does get loaded to the amount of grid current.

There are other advantages to having the bias supply in series with the grid resistor.  It puts the supply further away from the RF end of the circuit.  Placing a bypass cap at the cold end of the RFC and then the grid resistor with another bypass cap, then the meter follow by the bias supply to ground.  This puts the meter and bias supply two filter sections away from the RF on the grid.

Also the grid meter only reads the grid current regardless of any currents in the bias supply bleeders.  If there is no grid drive the meter will read zero but the protective fixed bias will still remain on the grid.

Fred

This circuit is called the, "He couldn't make up his mind as to all grid leak bias or all protective bias, so he split it."  :D

Yes in the original circuit the meter will read something but without calibration it will not be reading Grid current only because of the circuit configuration.

In order for the original circuit to work properly one must have a variable potentiometer in series with the 7k resistor.

You would have to remove the tube and then "zero" the meter.

Otherwise, the meter is reading some Bias Supply current as well as grid current.

Since an ammeter is being used, a series current measuring device, one might as well use a totally series circuit as my "modified" circuit shows.

Phil

Thanks, guys. It's nice to hear thoughts from others before making modifications. The guy who built this transmitter in the 60s was quite capable, but since he's now a recent SK, I'm on my own figuring out the "why" behind the design.

The meter will be moved to the cold side of the RFC with a bypass cap added from that point to ground/chassis. As for the parallel bias supply/grid leak scheme, I believe it may be like that so the bias supply augments the grid leak bias (the bias pot has a very subtle effect on the current).

Should a diode be added between the supply and the grid leak part of the circuit so that the bias developed by RF isn't loaded down by the bias supply? In other words, current would only be drawn from the bias supply if the bias drops below a certain point.

FWIW, the bias is for four 6146s in parallel, so for Class C operation, this would be 14mA (3.5mA x 4) according to the Sylvania Technical Manual.

The problem with your circuit is that it is designed wrong.  The protective bias is not augmenting the grid leak bias even though it looks like it is.  Once the grid leak bias exceeds the fixed bias the bias supply now shunts the grid leak resistor. All of the grid voltage is grid leak.  Adjusting the pot only changes the value of grid leak resistance,  that's why it has little affect on the grid current.  Adding the diode will eliminate this problem.  Put the diode where you have the "X" in the drawing.

The way the circuit is now the meter will read some current even when there is no grid drive  Moving the meter to the RFC will eliminate this problem.

The better solution is to completely redesign the circuit but that will mean replacing the grid resistor with some other value.

Fred

I should add the one good point with this parallel fixed bias supply.  Once there is grid drive there is no need for the fixed bias.  Meaning that if for some reason the fixed bias wasn't working it would only be a crisis if there was no grid drive.

With my series fixed bias circuit, that I described in my posts, the fixed bias voltage is always needed as it is supplying part of the total grid voltage.

The grid leak resistor works out to be ~ 6750 ohms so a standard 6.2k or 6.8k resistor @ 3W and you're in the ballpark.

I would not put a diode in there since you want an average grid leak bias to be developed by both the positive and negative going excursions of the driving RF waveform.

So, in order to have complete control of biasing and in keeping with the KISS principle, here is another circuit for consideration:


Phil - AC0OB

Phil,

Your circuit is correct.  The fixed bias is in series with the grid resistor which is what I keep saying.  Only problem is the grid resistor would have to be reduced to take into account for the fixed bias voltage.  Whatever that fixed bias is, it is subtracted from the proper level of grid voltage (whatever that is suppose to be,  I have to check the 6146 specs).  The remaining voltage needed is used to calculate the grid resistor needed based on the correct grid current, (4x6146s) 4x3.5ma= 14ma.

Fred

BTW, I am assuming this is for a Plate Modulated Class-C circuit?

As drawn, and with the 10k wiper Pot. at ground, the voltage would be - 84V at 13.6 mA Ig with a 6.2k RGL and a plate voltage of ~ 600V.

The bias supply is going to supply any additional negative bias voltage, if needed. During RF grid drive, the bias supply voltage and the grid leak voltages add to increase the total negative bias voltage.

Looking at the 6146B/8298 spec sheet, the nominal grid voltage is -92V with a resistor is 27k.

However, you also have to also look at footnote "r" which says, "obtained from grid-No. 1 resistor OR from a combination of grid-No. 1 resistor WITH either fixed supply or cathode resistor."


Phil - AC0OB
  

Phil

Right, so with 4 6146s the grid resistor would be about 1/4 of 27K  which is what he has in the circuit (7K).  Problem is if you're going to use some fixed bias you have to correct for the fixed bias voltage which is being added to the grid leak voltage.  They don't tell you everything you need to know in the tube manuals.  This is true in the series fixed bias circuit (like the one you posted).  The circuit that N1BCG is using this is not true because his fixed bias is in parallel with the grid resistor.  Your circuit the fixed bias is in series with the grid leak voltage.

Fred

In my HB 6146 xmtr I remember using 24K for a grid resistor.  The reason was that I used a small cathode resistor to drop a voltage that I measure to indirectly measure the plate current. In my xfmr all the currents are measured indirectly from voltages.

The voltage developed by the cathode resistor was subtracted from the required grid voltage resulted in using less of a grid resistor.

Right Fred, and the 10k 5W variable resistor does that. It would allow you to go from -84V up to the bias supply voltage with the R2 Rgl resistor.

I find that with a series bias system you need a certain degree of freedom in order to allow grid leak voltage to develop.

In the case of my circuit and for power tubes, I use the factor 0.875. That is, I take the stated grid resistance and multiply it by 0.875 to determine the final grid leak resistance, so as to allow both a degree of freedom AND control.


Phil - AC0OB

FB on all the comments.

I've used the same circuit below on all my AM class C rigs, from 6146s, 4D32 PDM rig, 24-Gs, 813s up to 4-1000A finals. It is simple and includes everything needed in the grid circuit - input matching, metering, neutralization, fixed bias, grid leak bias and a shunt resistor (RF bypass) for more stability.

http://www.amwindow.org/tech/htm/813/813.htm

I think Elon borrowed the name for his car... :-)

T

So much good info!  Phil, do you use that bias circuit or did you create it for this discussion? The project is temporarily stalled with an upside-down RF deck while I locate a >25W iron (or bum one off my ham neighbor) but I'll post a follow-up.

Hi Clark,

I actually use that circuit in my homebrews and upgrades.


Phil - AC0OB

Phil,

The circuit you posted is designed for what tube??  Do you use the same 6.2K resistor for any of the tubes you use or do you use different value grid resistors?  The rest of the circuit would be the same for any tube

Fred

In post #16, the pdf file shows the grid resistance range for 4, 6146's in parallel.

Obviously, the grid leak resistance is going to be different for different tube types and the total amount of grid current required.


Phil - AC0OB

OK FB

That's what I thought but wasn't sure.  Ok on the grid resistor being for Clark's 4x6146s

Also take a look at Tom's K1JJ's circuit, left center, where he uses fixed bias.

An 11k was suggested as a starting Rgl for a single 813 from the specs for Plate Modulated Class-C.

For example, 11k/2 = 6.5k so he chose a 7K.

In my single 813 SG modulated transmitter (180W), I used a 9.1k Rgl but I had a variable bias supply. 0.875 X 11k = 9.675k.

But neither his nor my values will apply to any new designs.


Each new design must evaluated according to the tube specs and the circuit configuration.


Phil - AC0OB

The only issue I have is the value of the grid resistor.  Both your circuit and JJs is the series design which is what I use and what I keep suggesting to Clark to use.

Problem is neither you or Tom are taking into account the fixed bias voltage.  I know JJs 2x813 design.  I was just looking at the RCA tube manual.  They have a single 6146 xmtr in the back pages.  Usual grid leak resistor is about 27K for the 6146.  They have only a 12K grid resistor but the xmtr also has a fixed -45V bias supply.

This is the point I keep trying to make.  The fixed bias gets added to the grid leak voltage.  I keep saying that you should reduce the value of the usual grid resistor to a value to only drop the remaining voltage needed to reach the correct neg grid voltage for class C.

What do you think??

Fred

Re-reading your last post, I see that you are reducing the usual grid resistor by .875.  So, you're only reducing slightly and allowing the adjustable fixed bias to pick up the remaining voltage needed with a greater range of adjustment.  OK should be FB

Regarding the ratio of fixed vs: grid leak bias...   (Fixed bias + grid leak = recommended class C voltage at grid RF choke.)

I've found this ratio is not critical for high fidelity performance as long as the fixed bias "barely" cuts off the tube enough so that when the RF drive is gone, (grid leak voltage = 0) the finals do not go past maximum dissipation. I use a Variac on my fixed supply and simply set it so that the 813s idle at about class B. Pretend the amplifier is a linear amp and set the idle for this plate current with no drive. It is now protected against loss-of-drive meltdowns.

Then adjust the grid leak resistor value so that when the rig is running, the recommended class C voltage appears at the grid RF choke.  (fixed + grid leak combined)

Some hams feel that the most grid leak is best... I say use the minimum fixed as protection and then get the rest from the grid leak, whatever that works out to be.  I know some hams who run 100% grid leak bias and use no fixed - instead they use a failsafe shutdown protection circuit when a certain plate current is exceeded. (No drive condition)

The bottom line is to set the rig up right, you need to run IMD, THD and triangle tone tests while looking at the scope and SDR spec analyzer. By carefully playing with all parameters - the grid voltage ratio, plate loading, plate Q, screen current, screen voltage, grid current, RF drive, HOW FAR into Class C/D you can run it for max efficiency by increasing the grid leak resistor and increasing RF drive, increasing audio negative feedback, etc., will have the biggest effect on fidelity, PA cleanliness and PA efficiency. You will find how high and clean your audio peaks can go by experimenting with all parameters.  It does make a difference.

The grid leak and fixed bias is but a small portion of what needs to be adjusted to fully optimize a rig for class C AM.

T

I beg to differ as we ARE accounting for both voltages.


OK, what equations or models would help you do that?

And I said earlier and in the circuits I showed the bias supply voltages DO add to the grid leak voltages in a series circuit.

And the facts you are missing are these:

The peak RF drive to the grid is the primary determining factor for the value of the grid voltage in a series circuit.

Too low a value of grid leak resistance means the voltage developed across the grid is too low, may never bias the tube into class C, and requires a large amount of grid drive power because the grid impedance is too low.

Too high a value of grid leak resistance may allow the grid-to filament/cathode potential to go into arc-over territory and exceed the grid input power (heating) values.

One can control the plate current of a Triode Final in two ways: 1) by adjusting the RF drive, 2) by adjusting the control grid bias voltage.

In a Tetrode, the plate current is controlled by: 1) by adjusting the RF drive, 2) by adjusting the control grid bias voltage, and 3) adjusting the screen grid voltage.

In either case, what we want to do is to allow about 140 degrees of positive grid drive  to pulse the final for class C.

BTW, the tube manuals and the tube cookbooks are only starting points for tube stage design.

Just because you see a 7k value mentioned on a schematic doesn't mean it is suitable for yours or anyone else's  design.


Phil

OK Phil thanks for the added information.  I did note that you were taking into account for the fixed bias in my follow up comment.  Tom also posted a detailed explanation on the method he uses for adjusting both fixed and grid leak bias.

The number of xmtrs I've built I never copy anything from other xtmr designs.  I do all my own designing.  This way when it doesn't work I can't blame anyone else.

Fred

It is a small portion of the total design but is also an important part of the design.

Too many times people think tube books and cookbooks are God given and if a 7k works well here, it is going work elsewhere. That is technically not correct.

Many more people are now interested in AM and are interested in homebrewing their own rigs, but are also a bit more technically inclined than the previous cookbook generation and are looking for technical guidance in building each stage.

So I think we owe them models and technical explanations rather just saying, "here, it works for me so just follow my template and don't ask questions."

Questions are good as it means they are attempting to add to their understanding of the subject.


Phil - AC0OB

That's fantastic Fred because you learn so much more than the tube cookbooks can tell you.

BTW, the 0.875 factor comes from empirical (on the bench) data from the design of many transmitters, both plate modulated and Screen Grid modulated.

Phil - AC0OB

Yep, we gots it now.

Actually a universal design that would work with simple calculations for any class C AM rig:

For a given HV, in the spec sheets:  1) find the recommended grid bias voltage in linear service   2) find the recommended bias for class C service   3) find the recommended grid current for class C.

Take the class C bias voltage then subtract the (fixed) linear bias voltage to arrive at the required grid leak voltage.

Resistor grid leak = E/I   =     Required grid leak voltage (drop across resistor) / class C grid current.

Set the fixed bias to the linear spec voltage (Variac) and install the calculated grid leak resistor. (grid leak adds its voltage in series, of course)   With the right amount of RF drive, you should have the correct grid bias at the tube pin at the correct grid current.

Simple as that.

T

Another good discussion on this topic back in 2012. Some former heavyweights weigh in. I like Bill KD0HG's bias circuit using a 6BQ5 shunt regulator.

http://amfone.net/Amforum/index.php?topic=31784.msg247501#msg247501

Jim
Wd5JKO

I read the whole thread,  seems to be same debate that we have on this thread.

Again, a hearty thanks for the suggestions! The re-working of the bias circuit was a success with the RF deck working well, the meter reading grid I accurately, and protection against drive failure.

Stay tuned. I have another circuit redesign challenge...

Can you post your final schematic, pls...

It turned out to be the bias supply, ammeter, grid leak resistor, and RF choke in series with bypass caps to chassis. Basically what Phil posted...

http://amfone.net/Amforum/index.php?action=dlattach;topic=43814.0;attach=58196 (http://amfone.net/Amforum/index.php?action=dlattach;topic=43814.0;attach=58196)

In my case, R2 is 7k, C1 is 0.05, C2 is 0.01, C3 is 0.02, and Lfg is 2.5mH because of the original design requirements of the transmitter.

Sumptin like this Clark?  :)

Ah, good. Now I can toss my hand-drawn version. Testing the 6SN7 circuit now...