Modulator with Class B Triodes or Class AB2 Beam Power Tube

[rsumperl]

Greetings everyone,
 So, if power output and power requirements were the same, would you rather go with push-pull triodes in Class B or push-pull beam power tubes in Class AB2?

Thanks,
 Ray

[N2DTS]

I would think there is no difference, both need driving power.
The triodes don't need bias or screen voltage, which should be regulated.

The advantage of  beam power tube in my book is running one in AB1 for no driving power and low distortion.

Triodes are hard to beat though, zero bias, or bias with diodes, no screen supply needed.
The downside of triodes is the driving power, and the fact they need power only on part of the waveform.
I just use a solid state amp into a driver transformer which does not care about the changing load as much as tube circuits...

Hi fi amplifiers were all AB1 screen grid tubes, or class A...

 
Brett

[w3jn]

Depends largely on the power required, and tubes available.

It isn't generally true that triodes do not require bias supplies.  811s are kind of the exception; most others do require bias - 805s, 810s, 833s, etc.  Nor is it always true that pentodes or tetrodes require a separate screen supply - for the 100W class EL34s and many other tubes can derive their screen voltage directly from the plate supply.

Tube broadcast transmitters used triodes more often than not.  If you're looking for that sort of power, don't bother reinventing the wheel.  However if you have a supply of 4-125s or 4-250s, take a hint from the T-368 modulator schematic.

If you're designing from scratch, it's better to select tubes that are inexpensive and readily available as opposed to scarce exotics, rather than worry about triodes vs tetrodes/pentodes.

[W3FJJ]

I would think there is no difference, both need driving power.
The triodes don't need bias or screen voltage, which should be regulated

Not True, I'm Running 304TL's  (triodes) AB1, no regulation on bias supply. Directly coupled 2a3 audio driver. No problem modulating 375 watt carrier
with 1800v on plates...

[WD5JKO]

I would just like to add that the Hi-Fi beam power tubes with > 10ma/v transconductance such as EL-34, 6550, 7868, 7591, 8417, etc. need a lot of idle current for low IMD. I once tried a pair of 8417's in a Viking I, set up class AB1 with the driver R-C coupled. I set the modulator idle current at 50ma (like the 807's were set). On the scope it did well. Listening to it revealed severe distortion. Boosting the idle current to 100ma made a huge difference with these tubes. Tubes such as the ones I mentioned like the idle current plate dissipation to be at least 2/3 of the tube rating. This puts a good part of the tube dynamic range in class A.

Jim
WD5JKO

[W3SLK]

To echo what Lamont, KK4Broadcast Operator would say, "Run triodes for the audio and tetrodes for the RF!"

[WA1HZK]

I did a lot of experimenting with 833's. The biggest improvement was reached in the modulator when I went with Chuck's advise and regulated the bias supply. The circuit is not too complex and it's posted on this site & my web site www.criticalradio.com Just scroll down the page to the WA!HZK Ham Stuff link. It's a fact, triodes will sound better with a regulated bias supply in the modulator.
Keith
WA1HZK

[WD5JKO]

It's a fact, triodes will sound better with a regulated bias supply in the modulator.

   Keith,

  I agree with you, to a point. If your triodes draw grid current (class A2, AB2, B), then a stiff bias supply is a darn good thing. If they don't draw grid current (class A, AB1), then the bias just needs to be well bypassed over the audio range, and have a bleed current of a few milliamps depending on the grid swing and miller capacitance of the big triodes.

  So do we want the bias supply to be 'stiff', i.e. low impedance but not voltage regulated, voltage regulated only, or to be both 'stiff' and voltage regulated? We can design for one, the other or both. This is important.

   Consider a modulator with low Mu triodes. Fixing the bias to an exact voltage has a consequence when the AC line voltage varies since the plate voltage will NOT be regulated. Therefore the tube idling current will vary a bunch when the AC line voltage varies. This requires you to have a handy knob to vary the bias day to day.

   Consider a bias supply that has good line voltage regulation, but is a series pass type. This can only source current, but cannot sink current. Since the tube grid current requires the bias regulator to sink current, and it cannot, the bias fluctuates with grid current. This approach is not a good choice.

    Consider a bias supply that has good line voltage regulation, but is a shunt regulator type. Now we can sink current which is what we need for a modulator that draws grid current. The problem here however is that the modulator idle current varies with the AC line voltage.
   
   Now consider a 'stiff' bias supply that is not voltage regulated. This supply will vary voltage proportional to the AC line voltage, and this will help maintain constant modulator idle current over the same swing of AC line voltage. The 'stiff' aspect of the supply would maintain the bias wherever it is at, to variations in grid current. This type of bias supply is a variation of the shunt regulator with a voltage divider setpoint instead of a voltage reference setpoint.

  The concern I just made is magnified with Beam Power tubes or Tetrodes. Here we will see massive swings in idle current when the G1 bias is VR fixed, and the screen and plate voltage is not. The high Gm HI-FI tubes are a real good example where this effect is extreme. I prefer with these tubes to VR the screen grids, and have a 'stiff' but NOT VR bias on the control grid. Then the idle current is pretty stable over a wide range of AC line voltage.

Jim
WD5JKO

   

[W2XR]

It's a fact, triodes will sound better with a regulated bias supply in the modulator.

   Keith,

  I agree with you, to a point. If your triodes draw grid current (class A2, AB2, B), then a stiff bias supply is a darn good thing. If they don't draw grid current (class A, AB1), then the bias just needs to be well bypassed over the audio range, and have a bleed current of a few milliamps depending on the grid swing and miller capacitance of the big triodes.

  So do we want the bias supply to be 'stiff', i.e. low impedance but not voltage regulated, voltage regulated only, or to be both 'stiff' and voltage regulated? We can design for one, the other or both. This is important.

   Consider a modulator with low Mu triodes. Fixing the bias to an exact voltage has a consequence when the AC line voltage varies since the plate voltage will NOT be regulated. Therefore the tube idling current will vary a bunch when the AC line voltage varies. This requires you to have a handy knob to vary the bias day to day.

   Consider a bias supply that has good line voltage regulation, but is a series pass type. This can only source current, but cannot sink current. Since the tube grid current requires the bias regulator to sink current, and it cannot, the bias fluctuates with grid current. This approach is not a good choice.

    Consider a bias supply that has good line voltage regulation, but is a shunt regulator type. Now we can sink current which is what we need for a modulator that draws grid current. The problem here however is that the modulator idle current varies with the AC line voltage.
  
   Now consider a 'stiff' bias supply that is not voltage regulated. This supply will vary voltage proportional to the AC line voltage, and this will help maintain constant modulator idle current over the same swing of AC line voltage. The 'stiff' aspect of the supply would maintain the bias wherever it is at, to variations in grid current. This type of bias supply is a variation of the shunt regulator with a voltage divider setpoint instead of a voltage reference setpoint.

  The concern I just made is magnified with Beam Power tubes or Tetrodes. Here we will see massive swings in idle current when the G1 bias is VR fixed, and the screen and plate voltage is not. The high Gm HI-FI tubes are a real good example where this effect is extreme. I prefer with these tubes to VR the screen grids, and have a 'stiff' but NOT VR bias on the control grid. Then the idle current is pretty stable over a wide range of AC line voltage.

Jim
WD5JKO

    

Hi Jim,

Like Keith, I am also using a shunt-regulated bias supply for my class B 833A modulator. It is a pair of 6080s, and being of the shunt-regulated topology, it obviously sinks the grid current developed in the modulator grids to ground as they are driven positive. Contrary to what many fellows may think, although the bias voltage does not track the line voltage with an active regulator, it is actually quite rare for me to tweak the bias adjust knob to maintain the normal 50 ma/tube idling current. Perhaps I also have very good line voltage regulation here as well.

Prior to the installation of this shunt regulator, I used a traditional unregulated bias supply, but with an extremely heavy bleed (I think the bleeder resistor was on the order of 1000 ohms). The low value bleeder served to not only improve the voltage regulation, but to minimize the voltage developed across the resistor as the modulator grids swung positive; the developed voltage would of course result in a bias voltage shift, which is undesirable.

Comparing these two different bias supply schemes I have used, unlike Keith I did not notice a significant improvement in distortion reduction when I went with the shunt-regulated supply. At this point, in my rig at least, I view the regulated supply as more "gilding the lily" than anything else.

The mileage of other users of class AB2 or B modulators may vary depending upon the type of grid bias supply utilized, and how hard the grids are driven positive.

73,

Bruce

[k4kyv]

I did not notice a significant improvement in distortion reduction when I went with the shunt-regulated supply. At this point, in my rig at least, I view the regulated supply as more "gilding the lily" than anything else.

I have always run class B triodes in the modulator.  This includes triode  connected 813's.  These run perfectly at zero bias. Other tubes I have used include 211's, 833A's and 810's. The latter all need bias.

I agree that the only advantage to tetrodes is that you can easily run them class AB1, with only minimal driving power to overcome losses in the circuit, and the bias supply doesn't need to be regulated.

With class B triodes, the bias and plate supplies both need to be well regulated. I started out using electronically regulated bias supplies, including VR tubes and DC amplifier type shunt regulators.  

I now use "brute force" biasing - that is, a DC supply capable of anywhere from several hundred mills to an amp or more, with a heavy bleeder, and the bias is tapped off the bleeder.

Why the change? Why did I abandon electronic bias regulation?  A problem I had noticed was that the modulator tubes static plate current was unstable.  This related to small changes in line voltage.  Even a few volts change, representing 1% or 2% change, was multiplied for the static plate current, so that I was getting more like 25%  changes with only a few per cent change in line voltage.  The problem is that with  electronic regulation, the grid voltage remains exactly the same while the plate voltage varies.  This relative change in bias voltage to plate voltage ratio is amplified by the tube just exactly as running fixed plate voltage and varying the grid voltage would be.  After all, this is what an amplifier tube is designed to do. It matters not whether the plate voltage is fixed and the grid bias varies, or if the grid voltage is fixed and the plate voltage varies, the result is the same; the change in plate current will be at a far greater percentage than the per cent change in grid or plate voltage.

I take the a.c. mains voltage that feeds the brute force bias supply right off the terminals that feed the modulator plate transformer, just before the T/R relay.  That way, any change in plate voltage due to line variations will be proportionally reflected as a change in bias voltage.  The plate voltage increases, so does the negative bias.  The result is that the static plate current now varies at the same percentage rate of change as the that of the line voltage - a couple of per cent at most.

A class B amplifier operates at minimum distortion at one and only one setting of bias voltage.  To maintain minimum distortion, any sag or boost in plate voltage must be reflected by an equal sag or boost (percentage-wise) to the bias voltage.

Ideally, the best source of bias voltage would be a DC amplifier type of shunt regulator that takes a sample of the modulator plate voltage, and uses this sample to precisely control the bias voltage.  This would take into account not only line voltage variations, but plate voltage sag due to drawing modulator plate current from the HV supply.

Next to less-than-ideal audio driver regulation, sagging plate voltage is probably the #2 most prominent cause of distortion in class B amplifiers (this includes slopbucket leen-yars as well as AM modulators).  Here, you have two factors working against each other.  As drive is applied to the modulator, the tubes draw more plate current, which causes the +HV to sag.  At the same time, the grids draw current which, due to resistances in the grid bias circuit, tends to cause the grid bias to increase, when you really need for it to decrease in proportion to the drop in plate voltage. The result is excessive distortion unless the grid bias voltage is well regulated.

The electronic regulator is unsatisfactory because of the above stated reasons, unless you have a source of rock solid a.c. mains voltage, or unless the regulator is designed to control the exact voltage by sampling the DC plate voltage.  The next best approach is a heavily bled brute force supply, in which the bleeder current is many times the maximum grid current, so that modulator grid current has a negligible effect on bias voltage.

Look at commercially built AM transmitters.  The vast majority of the all-tube plate-modulated broadcast transmitters, the Collins KW-1 ham transmitter and the Johnson Kilowatt all use an unregulated bias supply with at least a somewhat heavy bleeder.  This was done for a reason, especially considering that a simple DC regulator, maybe one using nothing more than a VR tube, would be considerably cheaper to build than a heavily bled, high current supply that inevitably requires heavier duty components.

[KM1H]

Ive no particular preference and have used both with equally good results.

To say that tetrodes are inferior is hogwash. There are way too many BCB rigs that ran 4-400's (4-250 also) and 4-1000A's to lend any credence to those beliefs.

Carl

[N2DTS]

In a qso today, Chuck pointed out many triodes can run quite a bit of power in ab1.
I never really thought about that before, but his 304 TL's are good for 600 watts in AB1 (1100 in ab2).

So I was wondering what triodes could put out what power in AB1.
If you run 811a's at 1500 volts, zero bias, what amount of power could you get out of them before they draw grid current?

I suppose 100TL's would make some power before going into grid current.
Could you run a pair of big triodes in ab1 to modulate a 100 watt rig cleanly?

Brett


 

[KE6DF]

"So I was wondering what triodes could put out what power in AB1.
If you run 811a's at 1500 volts, zero bias, what amount of power could you get out of them before they draw grid current?"

In AB1 the control grid never goes positive and therefore draws no current.

So, if you run with zero bias, you will always be drawing grid current on 1/2 of each cycle and therefore not be in AB1.

AB1 would require a negative bias.

[W3FJJ]

Brett, as iyh mention zero bias tubes such as 811, 3-500 can't be run AB1, but
here's a single ended class A audio amp using 833 triode, they claim they getting 150 watts out.

http://www.wavac-audio.jp/sh833_e.html

[KM1H]

The 3-500Z is designed for AB2 and can deliver -40dB 3rd order in SSB service with +5V bias at 2500V Ep and grounded grids. By raising the grids slightly above ground the bias will go more positive resulting in a degree of negative FB on voice peaks. This increases drive requirements around 20-25W.

I see no reason equivalent performance cant be met as modulators.

The 572B can be run as an 811 in AB1 with a bit of air but all AB1 operations are horribly inefficient; especially 304TL's considering the filament power needed.

Carl
KM1H

[KE6DF]

There is one of the medium power 1930's triodes designed to run in AB1 -- that's the 845.

It is sometimes used as a driver for larger class B modulator tubes.

If you look at the curves, it is in some ways the opposite of a zero bias tube.

The plate current at zero volts on the control grid is very high -- higher than most other similar sized tubes.

In AB1 the peak plate current you can draw is when the voltage on the control grid just touches zero.

[k4kyv]

About any low-mu triode can be run AB1.  In the 50's the cheap surplus 304TL (available for less than $1 each on Radio Row in NYC) was widely used in AB1 linear service.  Loads of filament power required, but the advantage is the low drive requirement, no need for regulated bias, no need for low internal driver resistance and plenty of peak cathode emission  capability. A pair would make a good modulator for a kilowatt or more DC input, without the need for an expensive class B driver transformer.

Because of its extremely low mu, the 845 is not recommended for rf service.  They were once popular for use as class A Heising modulators.  Because of their low plate resistance, they make excellent class B driver tubes.  A late 30's RCA 100-watt broadcast transmitter used a quad of 845's in class A pushpull-parallel as modulators.

Unfortunately, audiophools perceived some kind of magic in the tessitura of these tubes, and now widely use them as single-ended finals in their $80,000 stereo amplifiers, so the price of n.o.s. US-made 845's has been driven beyond what most hams could or would pay. More affordable ones are now being made in China, with the usual Chinese tube precautions to be observed.

[W2XR]

About any low-mu triode can be run AB1.  In the 50's the cheap surplus 304TL (available for less than $1 each on Radio Row in NYC) was widely used in AB1 linear service.  Loads of filament power required, but the advantage is the low drive requirement, no need for regulated bias, no need for low internal driver resistance and plenty of peak cathode emission  capability. A pair would make a good modulator for a kilowatt or more DC input, without the need for an expensive class B driver transformer.

Because of its extremely low mu, the 845 is not recommended for rf service.  They were once popular for use as class A Heising modulators.  Because of their low plate resistance, they make excellent class B driver tubes.  A late 30's RCA 100-watt broadcast transmitter used a quad of 845's in class A pushpull-parallel as modulators.

Unfortunately, audiophools perceived some kind of magic in the tessitura of these tubes, and now widely use them as single-ended finals in their $80,000 stereo amplifiers, so the price of n.o.s. US-made 845's has been driven beyond what most hams could or would pay. More affordable ones are now being made in China, with the usual Chinese tube precautions to be observed.

Completely agreed, Don.

As you and I had discussed on-air last night, I have had excellent results with the Chinese-made 845s in my class A audio driver, and I'd recommend them to anyone. They sound every bit as good as a pair of older U.S.-made 845s I have here, are reasonably priced, are built like a battleship, and have proven to be reliable in service.

73,

Bruce

[KM1H]

The bias shift circuit described by Orr in the 50's, and on here, was for the 304TL to reduce the ZSAC between syllables.

Carl

[N2DTS]

I wonder what sort of power a pair of 100TL's would be good for in AB1.
They are biased quite high, -185 volts...

I don't see many other common triodes with low amplification factors....
What about 812a's?

Brett

[k4kyv]

I have had excellent results with the Chinese-made 845s in my class A audio driver, and I'd recommend them to anyone. They sound every bit as good as a pair of older U.S.-made 845s I have here, are reasonably priced, are built like a battleship, and have proven to be reliable in service.

I saw some at a hamfest a couple of years ago and they looked to be constructed just like US made ones.  Hopefully, by now the Chinese have worked most of their quality control problems.  Everything else is made in China these days, so why not tubes?

[WA1HZK]

I run 845's as the drivers to the 833's also. I use the China ones and wrapped up the USA ones until the price goes to a zillion dollars then my notes will come from the south pacific. I had the thought of using triode connected 813's as drivers for the 833's but have not run that experiment yet. Due to 845's costing me cash and a box full of 813's I think that's a worthy experiment.
Keith