Class A Amplifier

[W2INR]

I have an linear, a pairs of 3-500z's in grounded grid. The are currently run in class B.

I would like to be able to run the same amplifer in class A for certain operating conditions.

So my thought is a relay and two bias supplies so I can switch amplifier class with a switch. Can it be done with the amplifer left in the GG config and just playing with the bias?

What is the best way to make this happen?

How would you go about it?

G

[Jim, W5JO]

G some of the old Class C transmitters used a series resistor string and a switch to go from Class C to Class A.

[Steve - WB3HUZ]

Remember, the grid only has to be grounded for RF, not DC. So, I would think you could put some bias on the grid, probably positive voltage in this case. I suppose a negative voltage in the filament circuit (where the Zener is located, if your amp has one) would do the same thing.

You might want to download this program and do some playing.

http://www.smeter.net/software/triode1.exe

[K1JJ]

G and I have talked about this and we've discussed two possible methods to go from grounded grid class A to grounded grid class B.

First, with the grids strapped to ground for DC, and a small zenor or diodes in the cathode will give the required class B.

But for class A, would it be better to put a 1 amp ~20VDC Negative supply (or whatever voltage is required to idle it in class A)  in the cathode?  (A negative supply in the cathode would make the grid look more positive) Or would it be better to bypass the grids for RF real well and put a positive voltage on the grids and idle it where it needs to be for class A?

Either way sounds like it would work, with the advantage being that the grids are strapped to ground with copper sheet for stability with the cathode bias.

But maybe there is some reason to lean one way or the other... or a better way to do it entirely. BTW, this supply will have to be stiff and regulated. In this particular case a string of resistors in the cathode will not work simply cuz it produces a positive voltage and a negative voltage is needed there to produce a positive grid voltage at the grid to ground - which is needed in class A. 

The unique problem here is we are dealing with zero bias triodes.

T

[Steve - WB3HUZ]

BTW, this supply will have to be stiff and regulated. A string of resistors will not work simply cuz a positive voltage is needed in class A.

I don't see why this would be so. If you are running Class A, the grids would not be drawing any current, thus no load on the supply. Or am I missing something?

[K1JJ]

BTW, this supply will have to be stiff and regulated. A string of resistors will not work simply cuz a positive voltage is needed in class A.

I don't see why this would be so. If you are running Class A, the grids would not be drawing any current, thus no load on the supply. Or am I missing something?

You are talking about resistors in the cathode, like a standard low level tube class A audio amp, right?  I was talking about a stiff neg supply in the cathode. (20V at 1A, etc)

No matter what was done, the problem is that even with the grid to ground and the cathode to ground, 3-500Z's will only idle in class B, depending on the plate voltage. To get true class A where the tubes run 2/3's or more disipation at idle, would require a positive grid. Any resistors in the cathode would just add neg grid bias and make them idle lower.

But maybe you were talking about shorting out a grid leak in class C to go back to linear. Yes, there would be no grid current in classic class A. But, since the grid is positive relative to ground, wouldn't it draw grid current in this case?  A GG amp usually draws plenty of  grid current in class B. So maybe the best it could be is class A2 not A1. This zero bias GG class A case is unusual. I starting to wonder if the 3-500Z is not a good choice for class A at all, even though it is perfect for GG class B.....  Gary needs an amp that can do both. What ya think?

T

[Steve - WB3HUZ]

Let me get the data sheet and take a look. I don't think it matters if the grid is positive or negative. Class A is Class A, and that means no grid current.

[K1JJ]

Yes, class A just means it dissipates lots of heat and runs close to 360 degrees of the conduction cycle.  Class A is usually no grid current (A1)... but if the grid signal swings positive relative to gnd, then it will be class A2.

But the more I think about it, I'll bet with 4kv on the plates of the 3-500Z's and the cathodes tied to gnd w/o a zenor, it will be idling near max dis. Take a look at the curves for zero bias and see when it hits 350-400W idle per tube. Might work as is... even if he goes to 4.5kv.. :-)

The other alternative is to run a tetrode with reg screen, like a 4-400. It will not draw grid current for sure in class A. I'll bet in true class A it will be as clean as the GG 3-500Z. Gary could even use a class A pre-driver and put some RF Neg feedback around it like I'm doing with the new Mr Ugly here.  That's good for -45-50db 3rd order. The GG 3-500Z's would not be quite that good, even idling high.

Let's see what you find from the tube curves, OM.

T

[WBear2GCR]

Negative supply on the cathode will make the grounded grid look positive...

However, at least with audio applications, taking a class B zero bias tube into class A means that you generally have to drop the B+ considerably to keep the plate within the power rating.

Rp also looks different in Class A vs Class B.

A good example is the 845/211 vs. the 805/838 - they're almost identical plates, if not identical.

Dunno if this helps or not... 

            _-_-WBear2GCR

     

[The Slab Bacon]

G, Tom,
            Like Bear says, you will have to dramatically reduce the Ep to keep the tubes within their dissapation ratings if attempting to run class A. I guess you are looking for some kind of ultra linear, low distortion amplifier. Keep in mind that changing to class A operation will DRAMATICALLY reduce the power outpoot of the amp. If you are still looking to get high power outpoot it may be a much better choice to design and build an amp using a MUCH larger toob (s) that can loaf along at the heavier dissapation requirements of running class A linear.    (Just my take on things)

                                                                                     The Slab Bacon

[Steve - WB3HUZ]

Modeling shows the following. These numbers don't quite seem right. Will try to compare to spec sheet.

3-500Z Class A

Plate V: 2500
Grid V: +22 (or -22 on the cathode)

Drive to get

320mA Plate Current
42 mA Grid Current

Input Power: 800 Watts
Output Power: 324 Watts

[Bacon, WA3WDR]

The 3-500Z in class A would pretty much have to be class A2 (grid current drawn), because otherwise the tube would be pretty much cut off.  However, if the grid would be drawing current over the entire drive waveform, the problem is simpler, because in that case the grid won't be too different from a resistive load - and I think that's best for low distortion operation, which would make sense if that's why you want to use class A to begin with.

The grid could be bypassed to ground and tied to a positive bias source.  A finite resistance can be inserted in series betwen the supply and the grid; the right value of resistance might reduce distortion.  My guess is that the optimum resistance value for this purpose would be around 20 to 500 ohms.  For simple tests, forget the resistor.

In class A, bias that puppy up positive for 500 watts no-signal dissipation per tube, and bias it negative to cut it off when not transmitting, or kill the plate supply.  Dissipation will be lower when transmitting, because some of the power will be output power.

You speak of bias in class B, so I guess you are running about 3KV plate.  I think you would want to lower B+ quite a bit for best performance in class A.

At 3KV, a positive bias of about +3 volts would produce 167 mA of plate current, or 500 watts input.  That's the bias point to use.  For maximum output, you would want to go from there to 333 mA peak and 0 mA negative, but that would drive your grid negative, and linearity would suffer.  (Ironic - here for best linearity, you want the grid to draw current all the time.)  Taking the cathode-grid voltage to +6 peak, and 0V trough, would be better for linearity.  I don't think you'll get much output that way, though, because your current would only vary from about 140 mA to 200 mA.  You would need a high-Z load, but you'd run out of voltage swing.  However, there might be an advantage in terms of linearity, I don't know.

At 1000V plate, you could run +48 V of bias and 500mA of plate current.  Then the cathode drive would cause the cathode-grid voltage grid to vary from 0 to 96V, plate current would vary from maybe 20mA to about 1 amp, and I think you'd get a lot more power out of the amplifier.  I think that the optimum plate load Z would be about 500 ohms for a single tube at 1KV plate voltage in this service.  Also I think that the RF input impedance would be lower than usual.

I'm not sure where the ultimate optimum operating point would be for this service.  But I think you probably want to switch grid bias and plate voltage.

Alternatively you could try class A1 with negative bias, low drive and a very high plate voltage.  5KV plate with -13V bias would give you 100 mA plate current, and varying the grid from -26 to 0 would swing the plate current from about 0 to over 200 mA.  Here I think the optimum plate impedance would be about 5K to 10K, I could be wrong.  But again, you need to change the grid bias and the plate voltage.  Also, I think the RF input impedance would be higher than usual.

<edit> Current and impedance figures are for one tube.  I could be way off on the A1 plate Z, as Steve indicates.

[Steve - WB3HUZ]

Anode Load Resistance

Plate V: 2500
Grid V: +22 (or -22 on the cathode)

Drive to get

320mA Plate Current
42 mA Grid Current

Input Power: 800 Watts
Output Power: 324 Watts

Anode Load Resistance: 6490


From Bacon's Numbers

Plate V: 2500
Grid V: +22 (or -22 on the cathode)

Anode Load Resistance: 17.79k

[WA1GFZ]

Most amps float the minus side of the power supply and put the grid meter between the fil CT and ground. The positive bias would go in series with the meter.
The high plate Z if true would mean the pi network Q would be quite high. You could go the the next tap on the band switch if lucky.
I use to run my 4CX3000A class a by just shorting the zener string. It would do about 1600 watts DC input. When I talked a little louder the changed from class A to a more strapping mode though. When I put 100 watts of carrier into it the input power would sit at about 2600 watts input.

[Bacon, WA3WDR]

I see Bear and Slab already mentioned the plate voltage back there - I skimmed the posts too fast.

[wavebourn]

What about class A modulator? 

http://www.diyaudio.com/forums/attachment.php?s=&postid=1033281&stamp=1161198777