How can I accurately measure conduction angle of an RF amplifier?

Assuming I don't have a tube manual or set up operating parameters according to the typical conditions provided, and I wanted to juggle the grid bias and drive, how could I accurately tell what class of operation a PA is running?

Measuring plate efficiency might get me close, but other plate circuit losses might throw off the measurement.

I want to make sure the amplifier is pure class C.

 

The article I have attached will explain it for you.

 

So it doesn't require an expensive Conduction-Angle-Ometer?

 

You only use those when the conduction angle you are measuring is > 360 degrees.

 

I certainly appreciate the response with the article except it seems to be related to solid state devices rather than tubes. I didn't find it to explain how to measure conduction angle of a vacuum tube PA, or did I miss it?

 

if you apply plate voltage, and the tube does not pull any current, then it is cut off and class C. Seems like I remember for a plate modulated class c final you want to increase bias voltage so that it is twice what is required to cause no resting current to flow. This info could be completely wrong, and I hope that anyone will fill in the blanks and correct me where wrong. That was jus some things that I picked up during a few low power transmitter builds and was untechnical enough for me to understand.

 

Here's how I have experimentally set up a triode tube for class-C operation, without the data chart. You will need variable DC voltage sources for the grid and preferably for the plate, and a variable source of RF driving power. It is assumed you already know the correct filament voltage and the approximate plate dissipation of the tube.

Assuming you know close to what plate voltage and current you will be running at the final, apply normal filament voltage but no rf grid drive, and apply fixed negative DC voltage to the grid and then turn on the DC plate voltage, adjusting grid voltage until plate current drops to zero. Preferably, you would have some means of increasing the plate voltage in several steps up to the normal operating voltage, to reduce the likelihood of exceeding the plate dissipation of the tube while arriving at the initial adjustment. When you reach full plate voltage and the plate current is just barely reduced to zero with negative grid voltage, that indicates that you have found the cut-off bias voltage for the tube for that plate voltage. For class-C, the bias voltage should be set to at least twice the cut-off voltage; for best linearity and efficiency in plate modulated AM service it should be more like 3 times cut-off. Now, apply a small amount of rf grid drive, just enough to cause the plate to draw some measurable current. Dip the plate current or adjust the tank circuit for maximum RF output. Now, increase the grid drive until you reach maximum plate current, while maintaining the RF loading adjustment for the desired value of plate current. While maintaining the determined cut-off bias voltage, keep shuffling grid drive and plate loading until further increase in grid current results in no further increase in plate current, keeping the DC plate current approximately at its rated value. This indicates that the tube is being driven by the RF to saturation. Now increase the grid drive 25-30% more, until you are well past the saturation point where plate current stopped increasing with further increase in grid current. It should now be running class C. At this point, measure the DC grid current and bias voltage. Using Ohm's law, calculate the grid leak resistance and power that would be dissipated in the resistor at those parameters, and insert a suitable resistor in the grid circuit. Re-measure the DC grid voltage and current to make sure you still have the correct bias and grid drive on the tube. It should now be operating fully in class-C service. If the amplifier is to be used for CW, or you don't have fool-proof overload relay protection, use a fixed bias of approximately the DC cut-off voltage, and develop the remainder with the grid leak resistor. If you are plate modulating in AM service, this should give good modulation linearity, as indicated by very little change in plate current as modulation is applied and increased up to 100%. Increasing the grid leak resistance requires an increase in RF driving power to maintain the original value of grid current, so you should have available a certain amount of reserve RF driving power.

 

Don,

That was a well thought out and well-explained response. Thank you for taking the time to go through that. The only thing missing is a means of observing the actual waveform. I didn't ask that initially but I had it in mind.

Textbooks sometimes illustrate what the partial (quarter cycle or 90 degree) class C waveform would look like when observed on a scope. I wonder, if in a laboratory style test setup, if it's possible to observe the 90 degree waveform on a fast oscilloscope and observe the changes as parameters are changed? I'm thinking that the pi-network would have to be disconnected from the plate and replaced with a purely resistive load simulating the tank circuit so that the flywheel effect would not take place, and the RF loosely coupled to the scope. Under those conditions could a demonstration similar to the textbook drawings actually be observed? I've never seen such a live demonstration, only drawings or sketches. Then again, my library is not all-inclusive. If possible, couldn't the demonstration be scaled down to something manageable, where a commonly available resistive load be used and also be somewhat safer than at a high power level? Has this type of test already been demonstrated? Am I expecting more than is possible with simple equipment?

Seeing it actually taking place and observing it on instrumentation always makes for a good solid learning experience. What we learn and accept about radio transmission is invisible to the eye in most cases.

Thanks in advance for any follow ups to my question.

Eric - WB2CAU

 

You would want to display the plate current, not plate voltage. Maybe a low value carbon composition resistor (Glo-bar?) in the cathode circuit and let one channel of a dual trace oscilloscope monitor the voltage drop across the resistor, while the other channel monitors the signal voltage fed to the grid. The resistor would have to go into the circuit where it would register plate current only, not combined plate and grid currents. This means all grids returned directly to the cathode, not the common B minus/ground, and be careful of capacitive by-passing effects. I would think the fly-wheel effect of the tank circuit would have to be preserved in order to maintain unaltered class-C operating parameters.

 

Generally, bias at 3X cutoff gives class C. (Old school rule of thumb)