Class B, zero bias - satisfactory circuit arrangement?

[WBear2GCR]

That is the Altec 1570B.
I have two examples of that amp.
Very nice, especially considering how small the output iron is...

A good way to drive the output tubes, fwiw. Standard connection for "triode strapped" 6W6 tubes which are like 6V6 but lower power, iirc.

Of course it does not have enough gain to be a speech amp, it takes standard "line level" input, you'd need to add another stage in front to use it as a modulator design.

As far as feedback, I'd take feedback before what would be mod iron, IF you needed it for some reason. The circuit is pretty good already, and if you use high quality mod iron it will be pretty good without feedback, imo. There are ways to use resistive dividers to get feedback from the secondary without having to find iron with a separate "feedback winding".

                  _-_-bear

[Opcom]

1. shows how to use neg feedback on a tetrode driver for class B modulators. In the push pull circuit, the method is similar to that used in some BC rigs, taking the end plate signal and sending it back to an early stage. These also avoid pitfalls of tying G2 to plate which usually results in low power output. In the examples the G2 is run at the same potential as the plate but that is not a rule. The point is that G2 is run from a DC source and has no signal on it. The resistors shown in the single ended amp therefore connect as a simple voltage divider across the plate winding and have nothing to do really with G2.

2. is a graph comparing tolerance to varying loads of the amp architectures in attachment 1 compared to a 2A3 push pull amp, and also a straight amp with no feedback.

[Opcom]

3. is a complete driver for class B load

4. is the high level class B modulator. It shows 813's but the KT88, 807, almost any tetrode or beam tube will do. If the tube is a pentode, the suppressor may be connected to G2 or to ground (probably best although there is a diagram using an 803 otherwise), and if it is a beam tube, the beam electrodes should be connected to ground. It also shows a different way of driving the class B tubes, using a cathode follower instead of by the plate and delivering lower power than if the circuit were taking power from the plate, but either way will work as long as the driver makes enough power. The circuit calls for 20W dissipation on the Class-A 807! Notice the drive is to the 813 is only 65v RMS, 74V peak. That is because it does not take much to get high signal current when both the screen and control grids are driven positive.

My advice to the OP is to use a 6L6 or better to drive the class B stage, and not have to run full tilt dissipation on it. Or if possible change over to a push pull circuit with teo 6L6's. But only some experimenting will tell.

[The Slab Bacon]

Oh - driving BOTH the screen and grid tied together?

Triode connection is usually the screen tied to the Plate via a resistor.

If you tie the screen and grid together it is more or less screen drive... but ok.
I doubt that is terribly linear, as screen driven tubes are very non-linear... but it would be interesting to read about it in more detail.

Edit: wonder what the gain looks like with this connection??

              _-_-bear

Bear,
       Tying G2 to plate is usually used for low power class A Operation. Whth this configuration you have pretty much full Ep on the screen. You usually need some kind of biass to keep the tube from melting down.

G1 and G2 tied together and fed, usually creates a class B type situation. With little or no positive biass on the screen the tube usually sets there quietly waiting for something to do. This arrangement just about always needs to have the grids pushed into grid current to produce full current flow through the tube. (henceforth a big triode). Therefore this is usually used for higher powered class B operation. (little or no negative biass, a lot of grid current, and a much lower Z input)

Just think of a grounded grid leanyour with a tetrode and both grids tied to ground. With that circuit arrangement, you have basically created a class B triode amplifier.

[k4kyv]

Tying the screen and control grids together forms a high mu triode.  Tying the screen to plate forms a low mu triode.

The low mu connection with a 6L6 makes a usable substitute for the 6B4G, although the plate resistance is not as low as with the latter.

The high mu connection (with some tubes with a resistor in series with the  control grid) makes a good zero-bias class-B triode.

[WBear2GCR]

Lovely!

Of course I am wondering why, if this works as advertised it is not very popular with audio applications. Eg. making a 6L6 (etc.) into a 2A3 would be a neat trick, even if it tried to sit in Class B (positive bias is always an option?).

               _-_-bear

[k4kyv]

Would you positive-bias a 2A3/6A3/6B4G in audio applications?

[w4bfs]

Would you positive-bias a 2A3/6A3/6B4G in audio applications?

no reason to unless low b+ is all that is available .... in class a2 ....grid current

[WD5JKO]

The low mu connection with a 6L6 makes a usable substitute for the 6B4G, although the plate resistance is not as low as with the latter.

  Patrick Turner of Turner Audio thinks the 6CM5/EL36 or 6FW5 (no plate cap) makes a superior triode when connected as such to better the 2A3/6B4 series tubes:

http://www.turneraudio.com.au/2323-triode-integrated-6cm5.htm

"6CM5 in triode gives µ = 4, Ra = 600 ohms, gm  =  6.6mA/V.
snip
 learnt one of the best kept secrets about the 6CM5 :-
Although its a beam tetrode it will sound very good as a triode.
When I first built it and compared it to a pair of Quad II
amps I had repaired, the 6CM5 seemed to sound better. The 6CM5 was meant for line output in tv sets, and capable of as much cathode current as a KT88. But anode dissipation is limited to around 18 watts in a class A or AB situation.  It could even be used as a single ended triode with about 4k anode load for about 5 watts with Ea = 375V and Ia = 50mA, and methinks it will out perform 2A3!"

Another thread about 6CM5 that shows grid 2 drive, cathode drive, etc. Also talks about 807 G-G in audio:

http://home.alphalink.com.au/~cambie/EL36.htm

Jim
WD5JKO

[KI4YAN]

It's true, many of the older sweep beam tetrodes make very good high-mu triodes when you tie the screen to grid via a 20K or so resistor, and drive the screen. Some, such as selected variant of the 6BQ6GA/6CU6, which are rated as 14W, will handle 28-30W plate dissipation EACH, with NO problems, and when strapped as a low-mu triode, the 6CU6 is a VERY close match to a 300B!

[w1vtp]

That Turner guy is interesting reading. 

http://www.turneraudio.com.au/about-turner-audio.html

His personal life just keeps going on and on - I couldn't stop.  He has an interesting opinion about our lady-friends

[WBear2GCR]

Patrick is a good fellow.
I've been online with him in various groups for a very long time now.
First class guy. Sorry to read of his medical problems... just sent him an email
after reading that.

                     _-_-bear

[steve_qix]

Wow !  Ok, well - I've been solving the same problem over here for the past week or so.... driving, in this case, a pair of 807s in triode connected mode, and I have FINALLY and successfully done it - with high fidelity, and have learned quite a bit in the process.  Here are the results of this work (and it was a lot of work!!!):

1) The 807 (and probably many other beam power tubes or tetrodes) CANNOT be connected in a straight triode configuration.  Period.  You *must* run the grids at different potentials.  The screen voltage must be higher.  DO NOT use a resistor to accomplish this.  Doing so is folly.  The grid is a non-linear load, and introducing a linear element (a resistor) in series with the control grid (a non-linear load) will result in serious distortion.

I operate, with the 807s, using a 60 volt difference between the screen grid (at 55 volts) and the control grid (at -5 volts).  This difference is accomplished using 3 20 volt zeners in series - with a 100 ohm resistor in series with the zener string to protect it - all shunted by a 220uF capacitor (this is a non-critical value - 100uF will probably work just as well).  The driver is connected directly to the screen, and the control grid is connected to the screen and driver through the zener-resistor-capacitor network described above.

2) Do everything you can to GET THE TRANSFORMERS OUT OF THE DRIVER.  This is relatively straight-forward to accomplish.  Use anything other than a transformer - use a good cathode follower (with a triode); use a MOSFET (which is what I do).  Just don't put a transformer in there.  Bad!

Transformers are linear resistive elements, albiet not large resistances. They are imperfect in their ability to transer energy.  So, when the grid(s) go from an infinite impedance (negative voltage) to some finite resistance (being driven positively), there is a significant change in the load as seen by the transformer & driver, and this will result in a slope change on the driving signal (due to voltage drop) and this means DISTORTION.  Negative feedback will help, but nothing will compare to a source or cathode follower driver.

Also, transformers introduce phase shift - and this in turn limits your ability to put negative feedback around the whole modulator (and include the modulation transformer in the loop).

3) Put negative feedback around the driver, and put an overall loop of feedback around the whole modulator.

4) Results:  The 807 modulator is part of a Viking II.  Using the STOCK VIKING II MODULATION IRON and the 807s connected in the QIX triode configuration (using the zener-resistor-capacitor network, detailed above), I am able to achieve the so-called "class E modulation pattern" (otherwise known as "shark fins") with a transformer coupled modulator.  

No, it is not as good as a class E rig, but it's pretty darn good.  The mod iron limits the positive peaks to less than about 125%, and of course the low frequencies start to come apart below 40 cycles, but still, that's not bad - considering it's a transformer and not a very good one at that.

Hope this information is useful!

Regards,

Steve

[G8VOQ]

http://homepage.ntlworld.com/richards_internet/RADIO/zerobiasmodulator.html

The link shows a modulator circuit that drives a pair of KT88's. I think originally the manufacturers employed 807's. In fact, I'm minded to use 807's myself.  The circuit is wired for class B, zero bias. The modulator is to modulate an 813 running at 150W input.

Given that the PSU is probably designed for class B, zero bias (PSU needs decent regulation I believe) it seems that one could do worse than try to keep the original class of operation and zero bias arrangement.

However, even though the circuit as is will provide satisfactory performance, I'm completely rebuilding, so could afford to make improvements.

Improvements mentioned revolve around the drive arrangement to the finals, that is around the 6L6 driver. Specifically, arranging for some feedback. Other improvements seem to be ditch the driver transformer. Even employing push-pull 6L6's, but perhaps that is not necessary.

Anyhow, the circuit shown is my starting position, but no doubt improvements can be made to the drive part of the circuit.

I don't know what the value of C13 was.

I think I have got C1 and C4 value correct.

I've got 300V to play with for the speech amplifier and 6L6 driver. Apparantly 750V for the finals.

[steve_qix]

http://homepage.ntlworld.com/richards_internet/RADIO/zerobiasmodulator.html

The link shows a modulator circuit that drives a pair of KT88's. I think originally the manufacturers employed 807's. In fact, I'm minded to use 807's myself.  The circuit is wired for class B, zero bias. The modulator is to modulate an 813 running at 150W input.

Wow, that circuit has enough stuff wrong to write a book about it    Will it produce audio?  No doubt!  Will it sound good?  Well, beauty is in the ears of the beholder, but I don't think so    If you're going to use tubes, just use a cathode follower.  In the modulator itself, that resistor between the screen and the grid has to go.  Use the zener-resistor-capacitor network I talked about in an earlier post.  Drive it with something low impedance, and it will sound great!

Another thing - a pair of 807s is not really sufficient to modulate an 813, unless the 813 is running at really low power (about 120 watts input or less).  You did mention the 813 is running at 150 watts input - even that is pushing it a bit with 807s, but it might work.

If you run the 813 at ICAS ratings (2000 volts at 200mA), you will need something a whole lot bigger than 807s.  There are a number of triodes that can be safely used at 2000V which would make good modulators for an 813.  I'm assuming and suggesting a common power supply for the modulator and final - always the best way to go!

Regards,

Steve

[G8VOQ]

What I'm trying to do is employ the transformers, so I'll be limited by them. I'm not that bothered about getting the maximum power out from the 813. There was a "kind of" variac arrangement giving an 813 plate voltage of 750V /1000V /1250V /1500V. It was suggested that plate voltage be set to 1000V at 185mA, which gave the legal limit of 150W input to the PA (150mA plate - 35mA screen).

So, the 813 was supposed to be run at 150W input to the plate. And of course the PSU is a function of that as well as the modulator arrangement. As I mention the set had KT88's installed, not 807's. But, I was wondering whether I could revert to 807's. They are cheaply available as NOS.

I was given a circuit with the set, and when I dismantled the set, it came out the same schematic, so I think I've drawn the circuit as was.

R17 & R18 are quite sizeable resistors.

R19 & R20 (there but not indicated as R19 & R20) are the two 15K resistors connecting g1 and g2 on the KT88's.

[The Slab Bacon]

Interesting circuit ?? I would also make many changes to it. by the component values (r-c coupling of lower stages) it looks line it will produce some very thin "ear bleeding" audio. 100 Pf coupling caps are just too small to produce any real audio fidelity. It looks to have a little rudimentary feedback wrapped around the 6L6. With a little component value tweaking, It might be able to sound somewhat ok.

The 6SJ7 input / mic amp is prolly too much gain and may cause a hum pickup problem. But with 100pF coupling caps the other stages prolly wont pass it. Ib triode connected class B service with around 750v you can beat the snot out of a pair of 807s for about 120w of audio. Which would be enough for 200-240w DC input to the final. (look at any basic 807 / 6146 boat anchor) But............
Instead of beating the hell out of a pair of 807s, a pair of 809s or even better 811As would be loafing along at that power level.

If you are using tubes designed to run AB, why not run them that way?? If you want to run class B, why not use tubes designed to run that way?? 809s are getting a little scarce, but chinese 811As are cheap and easy to get.

A pair of 807s in AB1 driven by a phase splitter will be capable of making 100w without much problems. 811As will work just fine at 750v, they'll be just loafing along. Why reinvent the wheel when someone has allready done it for you?    

[k4kyv]

http://homepage.ntlworld.com/richards_internet/RADIO/zerobiasmodulator.html

The link shows a modulator circuit that drives a pair of KT88's. I think originally the manufacturers employed 807's. In fact, I'm minded to use 807's myself.  The circuit is wired for class B, zero bias. The modulator is to modulate an 813 running at 150W input.

Wow, that circuit has enough stuff wrong to write a book about it    Will it produce audio?  No doubt!  Will it sound good?  Well, beauty is in the ears of the beholder, but I don't think so    If you're going to use tubes, just use a cathode follower.  In the modulator itself, that resistor between the screen and the grid has to go.  Use the zener-resistor-capacitor network I talked about in an earlier post.  Drive it with something low impedance, and it will sound great!

I used a very similar circuit with my very first attempt at a class B modulator.  The best I could ever do was space shuttle audio.  A transformer might work if you can find a good one designed for class B driver service, but the only ones I have ever seen were made for a push-pull driver.  The "ham radio quality" units don't make the  grade.  I use a UTC LS-series driver transformer in one of my homebrew rigs, and an RCA broadcast transmitter driver in the other. Acceptable quality class B driver transformers are difficult to design and build, and therefore very expensive.

The zener network sounds like the way to go.  Back when I built mine zeners weren't available yet, or were speciality items, so I used the resistor (20k IIRC, from data in the ARRL handbook tube  charts). 813s will work with the grids tied directly together.

Many slopbucket linears have been described over the years using tetrodes and pentodes with all grids grounded directly, which means the control and screen grids are tied together and operate at the same voltage. One that comes to mind is described in the 1957 Handbook, using 4 1625s in parallel.  You had to modify the tubes by going into the base and moving the lead to the suppressor grid/deflection plate from the cathode pin.  A commercial version of the 1625 amp was put on the market using 837 pentodes, which bring out the suppressor grid to a separate pin, instead of 1625s. I wonder if many G-G amps using screen grid tubes wouldn't be a lot  cleaner if the zener network were incorporated in those as well.  But apparently some tubes operate well with both grids at equal potential while others require less voltage on the control grid. The problem is how to determine which tube requires which, other than trial-and-error, since data for zero-bias triode operation is usually not given in the manufacturers' tube charts.

Steve, where does the negative control grid voltage come from?  You mentioned in the previous post using the network with 60volts worth of zeners, so that 55 volts on the screen results in -5 volts on the control grid.  But isn't some kind of DC bias supply needed as a "keep-alive" circuit to maintain the 60v across the zeners, so that the control grid follows the screen? Another thing, I had always assumed that the purpose of the resistor was to reduce the amplitude of the audio signal applied to the control grid, compared to the screen, and that a tapped secondary similar to the tapped primary on the "ultra linear" circuit would work better, although I never tried it.  But with the zener, the control grid voltage follows the screen exactly with the same amplitude audio voltage, but with the control grid DC biased 60v negative relative to the screen.  So you must have used a DC bias supply somewhere in the circuit to maintain this potential difference and to cause a constant -60 volts to remain across the zeners. Zener action requires some minimum DC current to flow before the reverse potential appears across the diode.   Maybe a schematic would clear things up.

Another thing - a pair of 807s is not really sufficient to modulate an 813, unless the 813 is running at really low power (about 120 watts input or less).  You did mention the 813 is running at 150 watts input - even that is pushing it a bit with 807s, but it might work.

According the Handbook  chart, the triode connected 807 circuit is supposed to be capable of delivering 120 watts. Allowing for transformer losses you should still be able to get more than 100 watts of audio, which would be plenty to modulate 150 watts input, with a little positive peak head room to boot.  IIRC, the RCA manual lists the maximum ICAS audio output from a pair of 807s in class AB2 as 120 watts.

[G8VOQ]

I have a question: 400W PEP is my limit and that is power measured out at the antenna terminals. What is 400W PEP in terms of average DC power input to the plate?

(Could be measuring between apples and oranges here). Not terribly important issue.

But anyway, the old set was made to run 150W into the plate, and that measure was the legal limit then.  So, I've basically got a transformer to supply 150W to the plate.  And for some reason an 813 was chosen as the PA.

The actual transformer is 1100V 0V 1100V @ 150mA.

[G8VOQ]

Instead of beating the hell out of a pair of 807s, a pair of 809s or even better 811As would be loafing along at that power level.

A pair of 807s in AB1 driven by a phase splitter will be capable of making 100w without much problems. 811As will work just fine at 750v, they'll be just loafing along. Why reinvent the wheel when someone has already done it for you?    

Well, I've been wondering whether to use the regular connection (tetrode) with 807's.  I think at some point the thought was, with a little adjustment in the driver (feedback) or whatever, class B, zero bias was going to be okay. I was also thinking that perhaps my modulator PU is configured / designed for the class B setup, and I'm keeping and employing the tubes in the modulator.

It looks like my modulator circuit is non too brilliant, so should one do a wholesale rebuild using "proven" circuitry. I guess there could be an argument for that.

Probably the tubes I have are okay, it's just the circuit as is, is non too good.

What to do, seems still up in the air.

[WA1GFZ]

Steve,
Cool idea with the zener offset makes sense as the control grid pulls current. Then you just need a pulldown to -5 volts.
Any current through the isolation resistor would change gain. The zener forces a constant voltage offset. Tom and I found you have to hold the phase shift through the driver as low as possible. We had to increase the coupling cap values and change resistor values to hole a low phase shift low over a broad frequency range. I think we ended up around 30 total degree phase shift out to around 10 KHz. Then we could apply feedback from the mod transformer secondary and live with the phase shift introduced by the iron. I think Tom is running around 10 dB feedback. Many driver circuits I have seen have a nasty phase shift over frequency which may prevent shark fin audio because the high end has a drastic phase shift.
When Tom got it right he was able to reverse the audio phase for better positive peaks.

[k4kyv]

What exactly is a "shark fin" waveform?

[Steve - WB3HUZ]

Something the Class E crowd came up with to compete with the smugness of Flex Radio owners.

What exactly is a "shark fin" waveform?

[WA1GFZ]

Tom Vu came up with shark fin audio in 1991 as we tested his first PDM rig.
Back then he thought is was some weird distortion

[steve_qix]

I can't say when the term "shark fin" actually came about.  I can say that we used to call it "Christmas Trees pointing to the (left or right)".

All low distortion, low phase shift, wide frequency response transmitters are capable of this pattern, and it is not an easy pattern to reproduce.  I first noticed it during the early '70s when I built my first PWM transmitter when I was in college.

Anyway, that's what it is.  It's just one of many measures of a good modulator.  Many modified rice boxes, flex radios, class E rigs, tube PWM rigs, tube grid modulated rigs, etc. can reproduce this pattern, which is typical of the male voice.