Class C bias....how far into class C/how much bias....what does it affect?

[K8DI]

So this came up out of a conversation/comparison of modified broadcast transmitters using 4-400A finals. 

RCA manuals for the BTA 1R series show -400v bias, and the indicated current and grid resistor support this (20mA on 20k), as well as actual measurement on a working example.
Bauer 707 manuals show identical currents and resistors.

Tube data sheets call for -220v bias in typical plate modulated class C operation.

Some well-respected builders/modifiers set them up for grid drive (vs. the original oscillator) and set the bias at the data sheet values, vs. the original transmitter design value.

I am going to assume RCA's engineers, and those at other manufacturers, had solid reasons to go much higher than the "typical" on the data sheet value with the bias.  What I want to find out is what those reasons might have been, and what happens/what's different with the final's operation between those extremes...efficiency? fidelity? tube life?

what do you think?

Ed

[M0VRF]

Class C doesn't use any DC bias!

[K8DI]

Class C doesn't use any DC bias!

Say what?

Please explain what on first glance is just plain wrong. The grids have DC bias.  Maybe you mean something else, or you’re trying to say it’s from self rectification?

My question isn’t about the source of the bias, but the amount. With a grid leak setup, the DC voltage is Ohm’s law dependent on the grid resistor and the DC grid current.

Ed

[W4AMV]

Hello Ed,

Yes, that value is about right.

The cutoff bias is ~ screen V/ amplification factor

The 600 V/ 5.1 gives 118 V and for class C a good rule is 3 x that or 350 V.

This should result in a grid conduction angle and subsequent plate conduction angle of 120 to 140 degrees. About right for class C.

You can confirm all by using the Eimac tube computer or spreadsheets.

73' Alan

[K8DI]

The cutoff bias is ~ screen V/ amplification factor

The 600 V/ 5.1 gives 118 V and for class C a good rule is 3 x that or 350 V.

This should result in a grid conduction angle and subsequent plate conduction angle of 120 to 140 degrees. About right for class C.

You can confirm all by using the Eimac tube computer or spreadsheets.

Alan, where does one find this Eimac tube computer? 

Ed

[W4AMV]

Eimac No5. Tube Computer.

See:

https://www.naic.edu/~phil/hardware/xmiterhf/doc/Eimac_AB5_Tube_Performance_Computer-2.pdf

And look for the image as well,

The document pdf provides the details/instructions.

[W9BHI]

I run a pair of grid driven 4-400's in class C, using grid leak bias.
24 ma.of grid current, @-220 volts for 2 tubes just as the Eimac spec sheet states.

[K1JJ]

Hi Ed,

I would set the operating parameters to what is recommended by Eimac for plate modulated class C.  Then play around with everything adjustable, looking for maximum audio peaks, best efficiency, cleanest IMD, etc.  Get the rig flat and clean from 30 to 9KHz or better using sine tones.

Generally I've found that recommended screen current, but a little more grid current than recommended, makes the best audio peaks.  Also, play with the tank L/C ratio and loading.  The highest plate voltage you can run (within reason) also makes tubes come to life.

Most of my experience in tube class C centers around the 4D32, 813, 4-400A and 4-1000A tubes.   I prefer tetrodes for final amplifiers, though have had great luck with triodes.

As far as grid bias, I usually use 50% fixed protective bias and 50% grid leak. (or just enough fixed bias to idle the tube near cutoff when the drive is cut.)  Play with the grid leak resistor value and drive level for best peaks and cleanliness.  Trying various parameters looking for the highest and cleanest audio peaks using a simple audio tone is the key to fine tuning a rig. Consider high level negative peak loading to protect the mod xfmr from damage.  (used for xfmr longevity, not for louder audio)

Consider audio NFB and a MOSFET solid state audio driver for your modulator.   The WA1GFZ board works great. I use three of them for various rigs.  (4X1 plate mod, pair of 813s plate mod, single 813 series mod rigs)   Strive for transformer-less audio except for the mod xfmr. (reduced phase shift, especially for NFB loops)

Tom, K1JJ

[K8DI]

Eimac No5. Tube Computer.

See:

https://www.naic.edu/~phil/hardware/xmiterhf/doc/Eimac_AB5_Tube_Performance_Computer-2.pdf

Thanks!

Ed

[DMOD]

As far as grid bias, I usually use 50% fixed protective bias and 50% grid leak. (or just enough fixed bias to idle the tube near cutoff when the drive is cut.)  Play with the grid leak resistor value and drive level for best peaks and cleanliness.  Trying various parameters looking for the highest and cleanest audio peaks using a simple audio tone is the key to fine tuning a rig. Consider high level negative peak loading to protect the mod xfmr from damage.  (used for xfmr longevity, not for louder audio)

Tom, K1JJ

This is good advice and has been my approach as well. Protective bias is a must in case RF drive goes south. But even after all the calcs have been done one still has to tweak.

Phil - AC0OB

[K8DI]

Tom, Phil, what is you opinion on using the RCA’s audio bias supply (~ 180v) for protective bias on the finals?

Ed

[K1JJ]

Tom, Phil, what is you opinion on using the RCA’s audio bias supply (~ 180v) for protective bias on the finals?

Ed

Ed,

180V fixed bias is a little high.  The GL bias works best when it is at least 50% or more of the total bias. Based on the 4-400A datasheet below I see that -85V of fixed bias will safely idle the 4-400As at about 140 mA. (like linear operation)  This is a good level in case of loss of drive.   In class C it requires  -220VDC. So you will need a grid leak resistor that will produce   -135 V.       (-85V + -135V = -220V)

I would use a simple 85V zener regulator on the output of the -180V bias supply.  You will then have  -85V regulated.    

I've had several rigs that needed a regulated fixed bias supply because they would slowly charge up the tube bias under operation, (RF rectification to the fixed filter cap)  so it's not a waste of effort to do so.

If you need help calculating the series pass resistor for the zener or the GL resistor values and power ratings, let me know. (GL uses 24 mA total bias current,  135V / .024 mA = about 5.6K @ 10 watts will do)

I have experimented with using higher values of GL  (more drive required) with great results. You can drop the class of operation DEEP into class C and see the plate efficiency increase.  I see no adverse effects in performance.  It does take a lot more drive as you approach the class C / D limits.  Start with the normal specs and then try the bigger grid leak values with more drive and slightly higher grid current to optimize performance.



Tom, K1JJ

4-400A datasheet:
https://frank.pocnet.net/sheets/088/4/4-400A.pdf

[K8DI]

Based on the 4-400A datasheet below I see that -85V of fixed bias will safely idle the 4-400As at about 140 mA. (like linear operation)  This is a good level in case of loss of drive.   In class C it requires  -220VDC. So you will need a grid leak resistor that will produce   -135 V.       (-85V + -135V = -220V)

I would use a simple 85V zener regulator on the output of the -180V bias supply.  You will then have  -85V regulated.   

I've had several rigs that needed a regulated fixed bias supply because they would slowly charge up the tube bias under operation, (RF rectification to the fixed filter cap)  so it's not a waste of effort to do so.

If you need help calculating the series pass resistor for the zener or the GL resistor values and power ratings, let me know. (GL uses 24 mA total bias current,  135V / .024 mA = about 5.6K @ 10 watts will do)

Tom, K1JJ

Tom, I didn’t figure to directly use the 180v supply, at least a pot or divider or something to set a voltage….I was just describing the supply..

I’m not clear on how to stack the bias supply and the GL voltage in a circuit, though. Feed the bottom of the GL resistor with bias and make sure the supply can sink the GL current? Or a different method?  Besides resistance and bypass caps do I need to more  to isolate the RF vs audio bias feeds? Finally, where do I meter the grid current in such a hookup?

Ed

[K1JJ]

Ed,

There's a lot of ways to do it.

Here are some samples showing GL and fixed bias from two of my old rigs.

This 813 rig uses both fixed and GL bias.   The fixed bias is always in series with the GL resistor to make them add.  There is neutralization included in the grid circuit.  I also use a C/L/C  (reverse pi network) grid input circuit quite often.  The meter goes before the RF choke in series with the DC fixed bias.  Be sure it is RF bypassed to ground after the meter.

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


50C5 rig using GL bias only.  If added, the fixed bias goes from ground to the grid circuit (in series) and RF bypassed.   This circuit uses a different RF driver grid matching.  The fixed bias is always before the grid RF choke to keep RF out of the fixed supply.  

http://www.amwindow.org/tech/htm/50c5/50c5.htm



And here's "Summer Breeze"  a 4D32 plate modulated rig I recently built, showing GL and fixed voltages:

http://amfone.net/Amforum/index.php?topic=45973.0

Schematic file below:


T

[DMOD]

So this came up out of a conversation/comparison of modified broadcast transmitters using 4-400A finals. 

RCA manuals for the BTA 1R series show -400v bias, and the indicated current and grid resistor support this (20mA on 20k), as well as actual measurement on a working example.
Bauer 707 manuals show identical currents and resistors.

Tube data sheets call for -220v bias in typical plate modulated class C operation.

Some well-respected builders/modifiers set them up for grid drive (vs. the original oscillator) and set the bias at the data sheet values, vs. the original transmitter design value.

I am going to assume RCA's engineers, and those at other manufacturers, had solid reasons to go much higher than the "typical" on the data sheet value with the bias.  What I want to find out is what those reasons might have been, and what happens/what's different with the final's operation between those extremes...efficiency? fidelity? tube life?

what do you think?

Ed

What is the actual voltage across IR607? This voltage/51ohms should give the actual grid current. Multiply this resultant grid current X 20k and that should be the average bias due to grid leak. Neither the RCA 1R1 nor the Bauer 707 use a protective bias per se.

BTW, the 1R1 schematic is incorrect. The 4-400 grid RF feedpoint at the 20k should split off to a junction of two 100 ohm current sharing resistors, but this is not what is shown.

Below is what it should be:

Phil - AC0OB

[K8DI]

What is the actual voltage across IR607? This voltage/51ohms should give the actual grid current. Multiply this resultant grid current X 20k and that should be the average bias due to grid leak. Neither the RCA 1R1 nor the Bauer 707 use a protective bias per se.

BTW, the 1R1 schematic is incorrect. The 4-400 grid RF feedpoint at the 20k should split off to a junction of two 100 ohm current sharing resistors, but this is not what is shown.

Below is what it should be:

Phil - AC0OB

When I first started going through the transmitter, I injected DC across every meter shunt (it was cold/unpowered of course). I used a series resistor and a series-connected known DMM current meter, and adjusted the DC supply to give the manual-correct current, then made note of the transmitter's multimeter reading at that setting (it is calibrated in percent). So I know that at the right current, the percent is X. I did this to compensate for any error in the shunt or meter. I also replaced a couple shunts because they were way off..

That said, before frequency conversion commenced, the grid current (converted from the percentage reading) was 21mA, across 20K -- 420v bias.

My first foray into grid drive from a modern radio was less successful than desired; I was trying a broadband/transformer setup. Pretty sure that isn't going to work out and that I'll be doing bandswitching in the grid network.

When I saw the schematic, I too questioned it. It seems wrong. But.... That is how it is wired. See pic attached below.  I asked a buddy who may know something and he suggested that it may be to decouple the grids from each other as well as the drive. I don't know if that was the plan at RCA or something else. The wire from IR503 (20K grid leak) goes to the 100 ohm and grid pin, then to a terminal, then the 100 ohm to the second grid pin.  (Ignore the loose screw on IC603; the IL601 lead below was disconnected as part of today's experimentation.)
Ed

[DMOD]

That said, before frequency conversion commenced, the grid current (converted from the percentage reading) was 21mA, across 20K -- 420v bias.

My first foray into grid drive from a modern radio was less successful than desired; I was trying a broadband/transformer setup. Pretty sure that isn't going to work out and that I'll be doing bandswitching in the grid network.

When I saw the schematic, I too questioned it. It seems wrong. But.... That is how it is wired. See pic attached below.  I asked a buddy who may know something and he suggested that it may be to decouple the grids from each other as well as the drive. I don't know if that was the plan at RCA or something else. The wire from IR503 (20K grid leak) goes to the 100 ohm and grid pin, then to a terminal, then the 100 ohm to the second grid pin.  (Ignore the loose screw on IC603; the IL601 lead below was disconnected as part of today's experimentation.)
Ed

Undoubtedly it is wired per the schematic but that is not the proper method for sharing grid drive/current. It doesn't make sense since the grid of IV302 only has only 100 ohms of isolation with respect to grid IV301. If the drive sharing was properly distributed, each grid would have 200 ohms of isolation from the other grid.

I still think this was a schematic error that was carried through to production.

Look at the screen grid bias. Each screen grid has 100 ohm current sharing resistors (IR324 and IR325) and it is split off equally, as it should be. The Bauer 707 design did the same thing.

The - 420V bias does not (according to the Eimac 4-400 specs and curves) exceed the -500V max rating, so the RF drive voltage must be such that the positive portion of the sinusoidal waveform is less than or equal to 120 degrees of positive pulse, which means they are really driving the finals hard from the 6146 stage.  

According to the Eimac 4-400 Constant Current Characteristics curves, it only takes -120V to completely cut off the plate current. This implies the driver stage is supplying at least 300V of positive peak RF.

I think the -220V spec was a protective bias spec in case you wanted to "design-in" some protective bias. Apparently the tube was rugged enough that neither RCA nor Bauer felt it needed protective bias, since they had fast acting over-current cutout relays.

Phil - AC0OB

[K8DI]

Undoubtedly it is wired per the schematic but that is not the proper method for sharing grid drive/current. It doesn't make sense since the grid of IV302 only has only 100 ohms of isolation with respect to grid IV301. If the drive sharing was properly distributed, each grid would have 200 ohms of isolation from the other grid.

I still think this was a schematic error that was carried through to production.

I agree it seems wrong.  It is easy to change, just swap the jumper and resistor on the right, and move the feed to the center terminal post. Since it is also easy to change back, I will change it...

The - 420V bias does not (according to the Eimac 4-400 specs and curves) exceed the -500V max rating, so the RF drive voltage must be such that the positive portion of the sinusoidal waveform is less than or equal to 120 degrees of positive pulse, which means they are really driving the finals hard from the 6146 stage.  

According to the Eimac 4-400 Constant Current Characteristics curves, it only takes -120V to completely cut off the plate current. This implies the driver stage is supplying at least 300V of positive peak RF.

I think the -220V spec was a protective bias spec in case you wanted to "design-in" some protective bias. Apparently the tube was rugged enough that neither RCA nor Bauer felt it needed protective bias, since they had fast acting over-current cutout relays.

Phil - AC0OB

The 6146 is loafing while driving the snot out of the finals.. plate current is about half the datasheet value, and even an old weak 6146 is adequate to get the 20mA grid drive, probably by design for long, reliable operation.

The overload relays are simple, rugged and reliable. I wish such things were still commonly available. There's HV, LV, plus separate modulator and final overloads, and a relay to disconnect the plates until/unless LV is present. I think it would be hard to actually catastrophically break the thing. This one lived on air in the Midwest for 40 years, and the only actually bad (vs. drifted/out of spec) parts were an open cathode resistor on a shorted tube, a dead spot in the low power audio bias reostat, and an open resistor that was not stock/part of a modification in the final screen circuit (that I reversed). If parts had been replaced in the past, they used visually identical parts and did it in the same manner as the original. Just looking in the back makes me smile.

Ed