4 x 833A solid state driver

[AMLOVER]

After reading few topics about grid driving triode B classs AF amps I decided to try the WA5BXO solid state sircuit with some changes.
My questions are :
a) does anybody tried it succesfully with 833As?
b) is it possible to drive it to the 8 ohm primary with a solid state hifi or tube amp?
c) what is the approximate driving power needed?
d) is it enough 30-50w for 4 x 833As parallel push pull?

I attach a schematic to describe my case better.
 
Stefano

[AMLOVER]

Sorry this is the much better schematic.

Stefano

[K1JJ]

Hi Stefano,

I have not tried that circuit.  However, if you want to get rid of the transformer and external 50 watt amplifier, consider this WA1GFZ-designed, class A, all  MOSFET audio driver. Transformers = phase shift distortion and negative feedback limitations.

I use this MOSFET audio driver into my 4-1000A modulator grids.  One volt drives the 4X1's to full output. Notice how simple it is in the pictures.

The audio is superb using this driver. It uses negative feedback too.  The Greek AM stations over there would probably like to try it too.

73,

Tom, K1JJ


Audio Driver Thread:  (The schematic is in the third post)

http://amfone.net/Amforum/index.php?topic=23632.0

[WD5JKO]

Stephano,

   The John Coleman, WA5BXO circuit was intended to use a tube P-P driver where we just use the primary of the output transformer as a center tapped choke. He does it again on the modulator tube input side using a center tapped grid choke.

   The way he did it, the audio driver provides no power, and instead just provides the voltage swing needed. On the positive excursions where the modulator tubes draw grid current, the emitter followers are there to provide high power gain with almost unity voltage gain.

   The circuit as you drew it will work, but puts all the voltage gain requirement upon the driver transformer. Here you will need a quality transformer or performance will suffer. Your SS 50 watt amp, if rated at 8 ohms is about 20 v rms, or 28v peak. So you might need a 1:10 turns ratio (to each secondary side) to get the grid swing to 280v peak. The higher the turns ratio, the more the ideal transformer model degrades.

  I suggest if you are going to use John's circuit, that you reconsider and use a Vacuum tube driver. Here the transformer evils are minimized, and the ratio is unity, or 1:1.

  As Tom pointed out, the WA1GFZG circuit has merit and is applicable as well. Here you have more circuitry, but no transformers to deal with.

  Let me also put another idea on the table, and that is the "Power Drive Circuit" by George Anderson from Tubelab. Here is the link:

http://www.tubelab.com/PDcookbook.htm

  In this case, George uses tubes along with FETS as applicable. The "Power Drive Circuit" is similar to the WA5BXO circuit in that both use either emitter followers (John's), or Source followers (George) to provide low impedance drive to a vacuum tube grid.

 Whichever method you choose, look at George's design example. Any of the three circuits mentioned need to be designed custom to your specific application.

Jim
WD5JKO


 

[K1DEU]

Many hams used this SS Driver 7+ years ago. Designed by Steve KA1SI / WA1QIX

Many have seen and used the Gates audio driver boards with 4 807's  where a voltage gain 807 is coupled to your Class B audio modulator grid through a 807 Cathode follower

Dave Snyder supplied many PC Boards


   http://www.classeradio.com/tube_driver.htm


Many are using Steve's 3 diode limiter also

   http://www.classeradio.com/tech.htm

[k4kyv]

The most obvious problem I  see  with Stefano's schematic is the pair of 1k  resistors wired in parallel, inserted in series with each 833A grid.  That defeats the whole purpose of the class-B driver circuit, which is to supply the audio voltage to the grids of the tubes from a low internal resistance source.

The average grid impedance to a class-B grid as seen by the driver is something on the order of 500 ohms.  Check out the impedances and turns ratios of broadcast quality driver transformers; you will find the secondary impedance to each grid (total primary to 1/2 secondary) to be about 500 ohms.  Inserting 500Ω of resistance between the driver and each grid would wreak havoc with the operation of the driver stage, even if the emitter follower stage itself were capable of driving the tubes to full peak grid voltage with low distortion.

[WBear2GCR]

another issue is that the secondary of the step up transformer has no load on it. In my experience a load is necessary for the transformer to work as planned and not exhibit nasty high frequency response. Sometimes an RC network or a straight C across the half or full secondary is indicated.

                 _-_-bear

[AMLOVER]

Tom,
your recommendation is very complicated for my solid state knowledge and seems to be perfect for zero bias or AB class. I need 600v peak to peak grid to grid and B class grid driven triodes. I'll be unhappy not to have that low distortion but I must go step by step to the top.

Jim,
I like very much John's sircuit for its simplicity and I'll stack on that as far it is the only sircuit I can construct and test. I have some questions but if I'll get my answers in the forum then I'll go for that.

K1DEU,
Gates 4x807 is my second coice in case I can't manage it with the solids.

Don,
I changed my schematic after your notices and I also added a tube predriver that I have available and I can use it there. The reason I am afraid to drive the transistors with cap coupling is that this amp (4xEL34) uses 720V on the plates and John suggests to connect the input caps in an amp with less than 300v plate voltage. If this can be overcome then I'll connect with pleasure the tube amp with cap coupling to the transistors basis otherwise I need to use the back to back coupling with all the disadvantages Tom refered.
As you can see my tubes are the 5868 that need 560v peak to peak grid to grid AF voltage, I hope I can go that far with this sircuit.
I have also got big zeners to get 100w regulation and so I deleted the power transistor (ECG198).
Is it right to use single ended transformers in order to have the ability to control the bias individually for each leg of the push pull output or a simple pushpull is enough?

Bear,
the transformer sircuit is exactly how John has suggested it, if adding some load is critical then let me know aproximate RC rates.

I attach the new schematic as I improved it after all your help, recommendations are greatly appreciated.

Stefano

[w5omr]

Coming from one who is currently using the WA5BXO circuitry (version 2, I might add), let me add these words, Stefano...

I'm using a Bogen Public Address amplifier using (4) 7868's.  It's probably close to 100w, thereabouts.

The basic design is the same, but instead of using one side of an output transformer (say, for example a push-pull 6AQ5 output transformer)) John decided instead of blocking capacitors on the plate output leads of the audio amp, instead used a 1:1 transformer for voltage isolation.

The reason we used ECG164's for the voltage amplifiers is because they had a high voltage potential capability with a butt-load of gain (Beta).  With 300Vcc, and ~50ohms of current limiting gives around 600Vp-p (minus the voltage drop) is closer to 580Vp-p.  That and it was enough (and a little bit more) to drive the grids of a pair of 250TH's, which take up to 460Vp-p grid drive in Class B.  That driving requirement is roughly the same for a pair of RCA 833A's.  If you want to drive (4) 833A's in push-pull parallel, you are going to need roughly 2x the driving voltage out of the WA5BXO audio driver circuit.

Now, with THAT being said, the transistor John used in his (4) 813's in push-pull parallel modulator, was an ECG198.  MUCH lower Vcc requirement, a bit higher gain, and more than enough to satisfy the driving requirements of (4) 813's in push-pull parallel.

Not that I'm judging you, Stefano, and I'm rather ignorant of the rules and laws of Italy, but a pair of 833A's in Class B, with 3000VDC, -70v on the grids and 400Vp-p driving power, will produce 1650w into about a 9500 ohm load.  That's an awful lot of audio for a kilowatt transmitter!

[w5omr]

I have also got big zeners to get 100w regulation and so I deleted the power transistor (ECG198).
Is it right to use single ended transformers in order to have the ability to control the bias individually for each leg of the push pull output or a simple pushpull is enough?

that circuit, with the ECG164's, Stefano, has a very stiff and well regulated bias supply, using the ECG198 as the regulator, and has a voltage swing from 0V to -180V. 

[W2PFY]

Not that I'm judging you, Stefano, and I'm rather ignorant of the rules and laws of Italy, but a pair of 833A's in Class B, with 3000VDC, -70v on the grids and 400Vp-p driving power, will produce 1650w into about a 9500 ohm load.  That's an awful lot of audio for a kilowatt transmitter!

He is using this on the broadcast band from 1.600 to 1.700 something. It's kind of like our CB band here as I understand it except they try to build high quality sounding stations. It sounds like they have a lot of fun over there. Stefano explains all this way back in another thread.

[w5omr]

He is using this on the broadcast band from 1.600 to 1.700 something. It's kind of like our CB band here as I understand it except they try to build high quality sounding stations. It sounds like they have a lot of fun over there. Stefano explains all this way back in another thread.

Ah.  explains a lot ;-)

[WA1GFZ]

My circuit will drive 833s. it is a simple adjustment of two bias pots to set the 833 bias to any negative voltage down to a maximum of almost the negative supply. The 833A wants -80 to -90 volts. The positive swing is almost 150 volts positive above the bias point. The circuit could be converted to IGBTs and run at an even higher votage. The circuit as it stands will make almost 280 volts peak to peak each output

John, Ask Tom how  many hours it took to blow up the ijy circuit. Then there was the INR rig that landed in Africa which also blew up.

It is stupid to drive a solid state device with a tube, sorry

[K1DEU]

Frank The IJY attempt was poor, No transorbs No parasitic snubers. Steve SI / QIX did not proof or approve IJY's article. I believe most know about this. Well perhaps not. However a lot have the IJY PCB and have put SI's  design / improvements on that PCB to have a non bug appearance. No one would contemplate using 807's to drive 833's familiar with the Gates maintenance issues. Some even know what happens to Rp as K emission drops. Dave Snyder was not a mean person and did not have much experience with triodes with out heaters.

There were at least 14 people using The IJY article and PCB 's. In those days Frank you could not eavesdrop on your daytime coffee breaks.

I think Steve's article has an excellent learning / study curve.

[WA1GFZ]

John, Agree, Dave is a good guy. I buy parts from him every flea market.
But there was problems with that design. I don't know of anyone using it.

[Gito]

I don't understand,why We need this circuits,
 
When We have a solid state power amplifier ,And we used A inter stage transformer ( output transformer)

Why don't we drive the grids of the 833's directly from the secondary winding  of the inter stage transformer,connect  the fix bias (-80 or -90 VDC)to the CT
connect the primary winding (8 Ohm or 16 Ohm} to the output of the solid state transformer.

I used a 50 watt to 100 watt solid state power amplifier to drive 4  813 (triode connected) modulator that way

And also  used the same design to drive two 833 in push pull.

[w5omr]

I don't understand,why We need this circuits,
 
When We have a solid state power amplifier ,And we used A inter stage transformer ( output transformer)

Why don't we drive the grids of the 833's directly from the secondary winding  of the inter stage transformer,connect  the fix bias (-80 or -90 VDC)to the CT
connect the primary winding (8 Ohm or 16 Ohm} to the output of the solid state transformer.

Audio through a transformer meets certain hysteresis loss.  Removing transformers removes loss and enhances frequency response, particularly low frequencies.

[k4kyv]

Audio through a transformer meets certain hysteresis loss.  Removing transformers removes loss and enhances frequency response, particularly low frequencies.

Depends on the audio transformer.  Some of the broadcast quality ones with interlaced primary and secondary windings come pretty close to the ideal transformer, although there is no such thing as perfect. For example, some of those old RCA broadcast transmitters have excellent frequency response down to 30~ or below, and the initial design came from the early to mid 1930s using a pair of 845s to drive a pair of 849s (later replaced by 833As).

Solid state drivers should be capable of much lower internal resistance than most practical tube types.  The Gates 807 driver is far from ideal.  I once built up a homebrew version, copying the Gates schematic verbatim, but I could not get the audio peaks out of the modulator that I could when driving with a quad of 2A3s and a proper driver transformer.  I tried an experiment, simulating the class B grid by running the homebrew cathode follower with a sine wave tone and monitoring the output with a scope, connecting and disconnecting a 500Ω resistor across one of the 807 cathode outputs.  The resistor caused about a 15% drop in signal level measured at that point.

Another problem I see with the 807 driver is that the peak audio output voltage far exceeds the maximum cathode-to-heater voltage. You need to use a separate dedicated filament transformer for each 807 (I believe the Gates does that, but I'd have to check my schematic to be sure) and leave the secondary floating to avoid heater to cathode break-down, and I could still predict the voltage peak exceeded at frequencies in the 10K and higher range due to internal capacitances. Nevertheless, the Gates circuit meets proof at less than 2% distortion and I have never experienced cathode-to-heater failure in an 807 either in my transmitter or the one at the broadcast station where I worked in the mid 1960s.

I have the complete 845 transformer coupled driver module removed intact from a Raytheon RA-1000. I have contemplated temporarily trying it out to replace the 807 driver in my BC1-T, and if it turns out to give superior performance, making the substitution permanent. The Gates 807 driver was more a cost cutting measure than a technical achievement.  The much heavier duty BC1-F that preceded the 1-T used a transformer coupled 845 driver with a small amount of negative feedback.

[K1DEU]

Dear Gitmo and many friends  (Why use more than a transformer for Class B Grids.)

A few references; A good transformer would have a mutual coupling index of 1:1  if we had a non loaded primary and secondary of 50 henrys each. If we place our inductance bridge on either winding and short the opposite our inductance would drop to zero henrys. These transformers are difficult to find and or wind. I would often carry my Extech 380193 LCR bridge, Radio Shack clip leads and a square root calculator to measure audio transformers efficiency.

In this world our enemy is hysteresis.  Alien UFO's can overcome these losses overcoming gravity and time/space but our understanding of levitation / magnetics is primitive. A little from The Australian Radio Handbook  

  http://www.hamelectronics.com/k1deu/pages/ham/transmitters/am/images/CHAPTR05_32_0001.jpg

A single class B grid is a difficult load varying between a short and a open. When its a high impedance, no sweat. But when its close to a short there is the challenge to what it is connected to. Only Cathode / Filament or source followers can properly drive a class B grid and not through a transformer.  Many hams inc. me on 75 meter AM would give a demo of direct connected VS Transformer connected for listeners.
  I and others would also  give a demo of a Modulation Transformer with and with out a polarizing (non-shifting) DC Current on its secondary. Yep with audio we do not want a strong polarized magnet field on or near our audio magnetics.
     Look at the one Class B grid like its a half wave rectifier. When its drawing great grid current conducting the Driver Transformer is going to be magnetically unhappy.....  Yes you can hear the difference even with voice only.

Don ; when we would do a proof twice a year I was never impressed with reading of 2% with recent PMEL> I could hear crappy audio.  My fav tune for my own distortion and not over processed proof was Elvis Presley; Jail house Rock instead of the National anthem at sunrise to see if the Long Haired Classical music only Station Owners had hang overs or were crabby! Yes good kids get fired and re-hired a lot.....

Regards a fellow curious tinkerer  John D.   Shaw,  K1DEU   god it's early

[K1DEU]

I am waking up and will contradict myself.

In my widdle (Red Skelton) Johnson Valiant my driver for Class B 811a's 30 Ma. Ip was simply (yes I still have it) a 6C4 with over 600 plate volts driven by a AU7 . The 6C4 Voltage gain stage running only 1.3 watts plate dissipation drove a single-ended 6550  +600 eP    whose cathode was coupled through a primary UTC 600 ohm to 600 ohm 10 watt line xfmf. For cathode bias I used a 73 volt zenier diode on the other primary lead to ground. Zero bias on 811a's.  I used a separate filament xfmr due to the excessive heater cathode voltage.  Twas very clean, Distortion wise !

http://www.hamelectronics.com/k1deu/pages/ham/transmitters/am/pages/modulation_power_mods.htm

So if your driver is verrrry low impeadence with lotsa negative feedback and a good tight coupled, extra core weight driver xfmr may sound like clean strings !!!

73  John, K1DEU  its still too early in the AM; may the rain arrive and wash all the damm snow and mud elsewhere all the way to next fall

[WA1GFZ]

The 11N90 FET has a 1 ohm RDS on and will deliver 11 amps all day long.
you will not get that out of a tube or transformer. The phase shift through the circuit is under 30 degrees out to 10K. It is easy to extend the frequency response and get even lower phase shift if required. Simulation files agree with test results.
I bet the component count is lower than the quad 807. That is where I scaled my circuit from.

[K1DEU]

   Egads quite the triode Frank.

Have seen many shot Gates driver boards due to lack of maintaince. Frank besides the 845 driver boards did you see Terry's PFY driver schematic for his widdle Westinghouse ? perhaps he has a copy with him...

It is actually Snowing  Grrrrrrr

[W2XR]

I know this has been discussed here before, and no doubt has been beaten to death, but if a transformer-coupled audio driver for a class B modulator is designed correctly, the performance can still be excellent.

The problem is two-fold. First, to drive a pair (or a quad) of 833As, the driver tubes must be capable of providing lots of audio power, significantly more than the grids of the class B tubes actually consume, and must provide superb regulation into the class B load. This is typically achieved by the use of class A triodes, and a good example is the 845: Gates used a pair of them in the BC-1E and the BC-1F, with 1400 VDC on the plates. At that operating point, a pair of 845s are good for close to 100 watts of essentially distortionless class A push-pull audio. Assume the ratio of distortion-free power the driver is capable of providing, vs. the power consumed in the class B grids to be at least 5:1. Suffice it to say, that is a lot of power the driver must be able to deliver. And class A audio power can be very expensive, and very inefficient.

Second, the driver transformer must meet the criteria as defined by Don (and others). And it has to handle the audio power the driver is capable of delivering without saturating. Not an easy task, and that is a big transformer with a lot of core material.

The real issue nowadays, is the cost and the availability of finding a suitable class B driver transformer, and the cost of building a suitable class A driver amplifier. When you figure in the cost of the 1400 VDC power supply, the cost of the 845s (although the Chinese ones are of excellent quality and reliability), the decision clearly leans in the direction of the 807 cathode-coupled circuit, the WA1GFZ FET driver, or another proven design approach that is certainly more practical for the task at hand, such as a tetrode or pentode-based amplifier with lots of NFB to improve it's regulation.

I use push-pull class A 845s with a driver xfmr from a Gates BC-1F to swing the grids in my push-pull 833A modulator. I was lucky to come across these parts from a BC-1F that was on it's way to the dumpster; otherwise, I would have been forced to use a different circuit. Prior to the incorporation of the push-pull 845 driver in my rig, I swung the grids of the 833As with a backwards-connected Dynaco Stereo 70 audio output xfmr, and used a triode-connected Altec-Lansing 1568A amplifier as the driver. It worked well, and I used this approach for years.

The push-pull 845s, with 6 dB of negative feedback from the plates of the 833As, to the cathodes of the 2nd voltage amplifier stage, literally blows that previous audio driver away. No comparision for fidelity/reduced distortion, audio bandwidth, or amplitude response. And it meets my criteria for a truly old-buzzard audio line-up from the 1940s. The audio takes a trip through five xfmrs before it hits the plates of the 2x 4-400As in the final. When I occasionally patch in the RCA BA-6A tube limiter, in place of the Inovonics 222, it goes through 7 audio transformers.

That's the way it was done back then. Certainly not the current state-of-the art, but it's what I wanted. As the ancient Greeks would have said, there is simply no accounting for taste. Everyone's radio and audio culture and religion are different.

And yes, once the PCB for the WA1GFZ audio driver hits the streets, I'm building one up. I'd like to hear the improvement over the existing transformer-coupled 845 audio driver I'm presently using. I know I'll be favorably impressed.

73,

Bruce

[Steve - K4HX]

I used a 50 watt to 100 watt solid state power amplifier to drive 4  813 (triode connected) modulator that way

Exactly. Many others have used similar to drive 833A and larger tubes with excellent results. To an amp and a transformer and then add transistors following seems an over complication for most applications (maybe all amateur radio applications). I'm sure the more complicated circuit is superior, but if I'm going to build a solid state driver for Class B modulators, there will be no transformers involved. YMMV.

[K1JJ]

Yes, a good solid state power amp driving a Linear Standard (or equivalent) backwards-connected audio transformer does very well as an audio driver. Many of the big  guns have done this for years and they sound excellent. I've used the same technique for many rigs here too.

The only problem is the phase shift - I could see the subtle effects, especially in the deep lows. It showed up when trying to determine the proper audio phase for my voice. In addition, I noticed the extreme highs were not perfect as well as the cross over point.  FB, but not perfect.  On a noisy band no one willl notice. But on a quiet afternoon, at S9+60 over, listening to an SDR sync detector into a big stereo amp system - you will hear the difference.

This was all fixed once I went with the GFZ transformerless MOSFET audio driver. Even though I have a 1KW RCA broadcash mod transformer in the circuit, the 4-1000A modulators produce a cleaner result. The proper voice phase is now easy to determine. I just use the same polarity as the class E rig. The Class E rig is as transparent as they get, so it's a no-brainer to use it as a reference. Beforehand, I could not use the class E-rig phase polarity. The low end could not handle it.

Now the highs sweep cleaner at the extremes and the lows go down to the limit of the mod transformer which is around 25hz or so.  Since adding the GFZ audio driver, I've had guys say they thought the 4X1 rig sounded better than the class E rig...  Technically it is not cleaner, but maybe a better match for my particular voice.

I think the ultimate tube rig I've ever built was a 4-1000A PDM modulated by a 4-1000A. No transformers at all. That rig swept like a balanced modulator, DC to 20KC perfectly. My point is, it does take extra effort to eliminate transformers from the system. Most guys sound FB with them. But go that extra mile to have a transformerless system (at least in the audio driver) and on a quiet day you will hear and see the subtle difference in audio transparency.

T