Bias supply current requirements

[K8DI]

So, for the sake of discussion, I’ve got a triode, in a typical power amp circuit. If I drive the amp with rf, grid current flows. If it’s idling, it doesn’t.  Bias is derived from the drive, some would say it’s self rectified, others would call it grid leak, not sure that either is really accurate, but whatever.  For the sake of discussion, say it’s a 100mA when driven.

Now say I want to add static bias to the amplifier. I want to push it from AB to B to C operation, or I just want less plate current when not driving it. Does my bias supply need to deliver 100mA all the time, or only at idle, or less, because it’s only bias and not drive, or more, because there’s no drive, or what?

I’m trying to get a feel for how much dissipation a bias regulator is going to need. If I undersize it, it will shut down when it overheats, and the bias will disappear, leading to the expected consequences of losing bias..

Ed

[w8khk]

That is an interesting question, Ed.  The bias supply needs to deliver very little, if any, current to the grid circuit when there is no drive, and even less current when the RF drive is producing self-bias via the cathode-grid rectification of the RF on positive peaks of the drive signal.

The two sources of bias (fixed-bias supply and self-bias) are additive, in series.  The bias regulator is responsible for making sure the bias voltage does not soar above the desired value when self-bias is added to the fixed-bias supply.  There are two ways of accomplishing this goal, and each presents a different demand upon the bias supply regulator.  And in each case, a different demand is also placed upon the bias supply proper.

One simple type of bias supply is the brute-force, or resistive bleeder supply, which requires little to no regulation, but places more current demand upon the bias supply (transformer, rectifier, etc.)  The idea here is bleeder, or load resistor, is sized such that enough current flows such that the bias voltage does not increase appreciably when the final is grid driven, producing additional bias voltage, and rising above the desired value.  This is undesirable, because if the bias continues to rise with drive, then even more drive is required in order to make positive peaks during modulation, then the self bias rises, making more drive necessary, etc, etc.  So with a brute-force bias supply, a lower resistance load (bleeded) resistor is required to minimize the voltage rise with drive, consequently requiring more current from the transformer and rectifier to provide the additional power dissipated in the load resistor.  In this type of bias supply, the builder may choose to add a series regulator to provide a constant voltage across the load resistor.  Then as the final tube is driven, and the self-bias ADDS to the fixed bias, the regulator produces less current from the bias supply, maintaining the voltage across the load resistor.  So in this case, the demands upon the bias supply and regulator are maximum when there is no drive, and they are reduced when drive is applied.

The alternative is to provide a shunt regulator, instead of the series regulator.  This type of supply requires the minimum current from the transformer and rectifier in the fixed supply, but the regulator must handle more current, and thus more dissipation, when drive is applied.  In this supply, a tube or solid state device is place in parallel with the supply output, and performs the same function as the load resistor, but it varies its resistance to maintain the same output voltage when the self bias adds to the overall bias during periods when the final receives drive.  The self bias rectifier and filter simply maintain a voltage source to the shunt regulator, which conducts lightly (high resistance) to hold the bias voltage down to the desired level when drive is not present.  Then when the drive is applied, the load on the rectifier and filter is miniscule, as the shunt regulator continues to hold down the additional voltage generated by self bias.

To plan the current, or power requirements of the self bias supply, it is necessary to know how much  current and voltage are generated by self bias, and how much total bias is desired.  Then, if a simple series regulator is used, the load resistor must be sized to dissipate the overall bias voltage during drive, and the rectifier filter (and the regulator, if present) need to be sized to handle the power required by the load resistor when drive is not present.

For the shunt regulator, size the shunt elements (tubes or solid state devices) such that the overall bias may be dissipated during drive, then size the transformer and rectifier to provide enough voltage and current to maintain the desired bias voltage when drive and self bias are not present.  When no drive is present, there is little to no effect on the bias supply, so this condition may be calculated looking at the bias supply alone with no RF amplifier attached.

The earlier handbooks have examples of typical shunt bias supplies using one or more 6AS7 or 6080 dual triode tubes in parallel.  These tubes have extremely low plate resistance and can very effectively regulate the bias current, if you choose to use hollow-state devices for a vintage appearance.   These articles explain clearly the design requirements for a bias regulator.

[Opcom]

if it's worth it, a relatively simple circuit like this could be used. It's a DC coupled amplifier based on the old simple STK0050 power amp IC. The STK-0050 is online and a well known simple 50 Watt amplifier.

The diagram as-is was for thinking about a very solid source-sink bias supply for a grounded grid amp, but at higher voltages, enough for the tube in question, it should work the same. By changing transistors or resistors as needed, it can source or sink any amount of current necessary within reason to keep a steady voltage on the output terminal, at least in LTspice, where I used the same diagram for a deflection yoke driver at +/- 300V and 3A.
A pot can be used to set an input voltage to the amp, so as to set the bias.

This avoids the penalty of a shunt regulator and heavy bleeds, and the output voltage, set from a nice low power regulated source, will follow the input voltage whichever way the current goes.
It could be done with vacuum tubes as mentioned. Use the DC coupled transformerless amp circuit.


I did not find this kind of regulation necessary, but I am using a choke-input filter type bias supply with a 200mA rating and a 120mA bleeder. In the Old Books, 3x grid current was suggested as the 'heavy bleed' and I use more since the 200mA supply was already there, the high current and big transformer having been chosen by the original builder of the transmitter. The grid current is on the 4-1000A is 25mA average.

To try and reduce deviation of the bias voltage at the grid by the action of the tube being driven and grid rectification, a low resistance to the tube grid is used. I have just the RF choke and a 250 Ohm WW resistor and the drive has very very little effect on bias. This has been entirely satisfactory.

I guess the choice depends upon how close to perfect things must be.

[K8DI]

Thanks for the insights, guys.

It seems, I should build a bias supply whose regulator is barely able to keep up with the idle bias current and bleeder resistor, so it just maintains regulation while idle, and then the grid current from drive can also flow thru the bleeder, and the bias supply will deliver less and less current as the drive increases. As long as the bleeder can eat the current the bias won't rise. As long as the bleeder load on the bias supply doesn't overheat the regulator at idle, the regulation will be maintained, and the with that, bias will be constant.  Further circuit additions like Opcom suggested where the supply eats the current instead of the bleeder enhance the ability to keep everything stable.

I'm wanting to use a TL783 high voltage series regulator to control bias, mostly because I have a couple already built on little PC boards so they're easy to use. I mostly need to balance regulator dropping voltage vs current draw (e.g. device power dissipation). These devices work well, but if you drop a bunch of volts and draw some current, they get hot and shut down...

Ed

[w8khk]

It seems, I should build a bias supply whose regulator is barely able to keep up with the idle bias current and bleeder resistor, so it just maintains regulation while idle, and then the grid current from drive can also flow thru the bleeder, and the bias supply will deliver less and less current as the drive increases. As long as the bleeder can eat the current the bias won't rise. As long as the bleeder load on the bias supply doesn't overheat the regulator at idle, the regulation will be maintained, and the with that, bias will be constant.  

Ed, I agree with your assumptions, so long as the current through the bleeder in standby periods (no grid drive) is at least as great, and preferably slightly greater than, the current produced by the self-bias during periods when the grid of the final is at maximum drive current.  

If the fixed bias does not create enough voltage across, and current through, the bleeder resistor, then the fixed bias voltage will rise somewhat with drive.

I assume you will be running this rig as a class-C modulated final, or a CW final, since you are contemplating a combination of fixed and self-bias.  If you wish to run it as a linear amplifier, then it is best to use a well-regulated, fixed bias supply only, with no self-bias.  For a class-C modulated final, it may not even be necessary to regulate the fixed bias supply; simply run enough current through the bleeder such that you have sufficient safety bias if drive is lost, and that the increase in bleeder current with drive does not allow the fixed bias to increase significantly.

Since you are planning on series, instead of shunt regulation, keep in mind that the demands upon the regulator are minimal.  You will be seeing less demand upon the bias supply during transmit periods, and maximum demand on the regulator when in standby.  Since the load current does not increase on transmit, then you need almost no headroom in the regulation circuit when transmitting.  Thus your design should plan for minimum voltage drop across the regulator, and maximum drop across the bleeder resistor, resulting in minimum power dissipation in the regulator, which is further reduced by the drop in regulator current during transmit periods.  

It is suggested that you determine the desired self-bias voltage, and grid current during maximum grid drive, then size the bleeder resistance and wattage accordingly.  Then provide a bias power transformer and rectifier that can produce slightly larger voltage at a current equal to, or slightly greater than the calculated bleeder current (based upon regulator voltage drop specifications).  Now the regulator will have minimum voltage drop during standby, nominal dissipation, with the dissipation decreasing during transmit periods.

[DMOD]

So, for the sake of discussion, I’ve got a triode, in a typical power amp circuit. If I drive the amp with rf, grid current flows. If it’s idling, it doesn’t.  Bias is derived from the drive, some would say it’s self rectified, others would call it grid leak, not sure that either is really accurate, but whatever.  For the sake of discussion, say it’s a 100mA when driven.

Now say I want to add static bias to the amplifier. I want to push it from AB to B to C operation, or I just want less plate current when not driving it. Does my bias supply need to deliver 100mA all the time, or only at idle, or less, because it’s only bias and not drive, or more, because there’s no drive, or what?

I’m trying to get a feel for how much dissipation a bias regulator is going to need. If I undersize it, it will shut down when it overheats, and the bias will disappear, leading to the expected consequences of losing bias..

Ed

What tube or tubes are we talking about, and in which one class are you're going to operate?

It is difficult to give advice when generalities are given.

For high power tubes, I prefer an adjustable grid bias (protective bias) WITH grid-leak bias in order to test for and to set the operational sweet spot.

Below is a circuit I used on the SG modulated 813. It is operating in Class C.

As a rule of thumb, I use a grid-leak resistor value which has a value of about 75% of the suggested grid leak resistor value from the tube data. This gives the Adj. bias a certain amount of control while also allowing grid-leak bias to develop as well.  

I used a grid current of ~ 16 mA for calculations.


Phil - AC0OB

[K8DI]

Thanks for the discussion, folks.

 Like I said, I wanted to get a feel for the current requirements, so as to drive my designing and planning.

Situation:  getting big parts for RF or power supplies is difficult or expensive or both. Stuff is too far away to pick up or too heavy to ship. Hamfests remain out, and when they start up, I'll stay away until Covid is contained and I have been vaccinated (I am in a high risk group, so's my wife). But I have SOME stuff.

I have a 250w Thordarson multi-tap type mod transformer on a chassis with triodes. (type 19M17)  I can plate modulate something.

I have an REI/Hallicrafters loudenboomer amp, with a power supply I built that is good for about 2400v at 400mA. I'd prefer the voltage a bit higher, but it's what I have.

When I use that as an amp, I can run 150ish watts carrier, after that, modulation gets clipped. The 3-400z that it was designed for, and the easily substituted 3-500z, can be run class C and plate modulated, at a higher power level. Spec sheet says 500+ watts carrier.  All I gotta do is bias it into C, modulate the power supply, and hope the effective output impedance is still within the tuning network's range, right?  This is where my thoughts came from...

The REI version of this amplifier varies somewhat from the Hallicrafters HT45 schematics I've found. Specifically (and conveniently for this idea) it does not directly DC ground the grid! It is RF grounded by three mica caps, but DC grounded through the grid current meter. Thus adding bias shouldn't require major surgery (although I am not at all worried about keeping it original, I bought it to enjoy, not preserve).

I've also read of K1JJ's playing with biasing linears into class C for AM operation, so besides plate modulating it, I'd like to try that too.

All of this presumes a bias supply exists, which it does. When I built the PSU for the amp, I included a small transformer/rectifier/filter setup. It has about 160v output with a 100K bleeder, the iron is 50 watt sized, so I should be able to pull 250mA or more from it. I want to put the bias regulation in the amp chassis, where I can put a pot to adjust it that I can easily reach.

Now that I have some feel for the current requirements, what's to stop me from experimenting, other than eventually blowing some stuff up?

Anyway, call me nuts, or give me advice, I am happy to listen.

Ed


 

[KD6VXI]

Biasing that amp to class C can be accomplished with a center tap diode string.

50 6A10 diodes will put it into class C with ease, for modulating.

Less than that for what Tom discussed.  You really need a scope for running amps this way, as the modulation will be changed.  IE, 100 pct modulated fed into a slight class C linear will come out overmodulated.


--Shane
KD6VXI