Transformerless push-pull tube circuit??

[Learner]

[quote author="CJ"]
Learner, did you ever build that LA2?
:?[/quote]

Well, I am planning to redo my PCB layout again on Protel so it would be another "little" while I guess............. Besides, I would like to be more confident in building all sorts of pre amps first in order to gain a lot more experience to understand the concept of circuit analysis thoroughly by applying it in practice. Since it would be very helpful when comes down to trouble shoot a circuit because things dont usually work right the first time, so still need a lot of practicing and learning from making errors while building a good variety of prototypes.

There has been so many occasions where I had to rip the whole circuit apart to rebuilt, always seems to find a "better" way of doing things just when I have finished wiring but never before or during wiring..... I don't want to be doing that to clones but I know I will be, so I might as well do it on the prototypes....

Also, I want to add some IC filters to the sidechain in the LA2a and stick a handfull of notch filters in parallel between the pre and the output stage. I am too busy with programing interrrupts in uControllers at the moment not to mention arrays and pointers in Borland C, so I'll have to leave the analog wiring exercise during the holidays......

 

[PRR]

> is this a reasonable prediction by SPICE??

Yes. SPICE says this plan sucks.

Look at it. You start with +0.5V peak. You split that to U1, making 1.5V peak, and U2 making -1.0V peak. Push-pull, but not balanced. 2.5V peak between grids.

Then you feed a couple 12AU7, which might have a gain of 10. What do you end up with? 1.6V peak.

And you'd get the same result with JUST the U1 and R10=22K.

Why are U2 and V1A V1B not adding any gain?

You are driving the tubes' inputs push-pull, but taking the output push-push. The two tubes are FIGHTING each other! (If you re-bias to class C, you get a frequency doubler....)

If all else were equal, output would be zero. But you have an apparent error in the U1 U2 stage: 0.5V unbalance. The 12AU7's intrinsic gain of 10 or 15 is badly reduced by misconnection, so apparently it runs with gain around 3, a not unreasonable answer and not one we like.

You need to recombine the two tubes' outputs OUT of phase. A center-tapped transformer is THE way to do it.

If you can't stand transformers: the White Cathode Follower is THE simple quasi-push-pull tube stage. It isn't true push-pull and is limited to class A operation, but is dead-simple.

Next come the several Futterman variants. Even if you don't abuse them (ask them to drive loudspeakers), they are pretty tricky beasts.

For loads of 10K and several volt signals, if power is not precious, a simple resistor-coupled emitter follower is usually your best bet. A single 12AU7 can pump several milliAmps to a load, 6V6 or 6BQ5 several dozen mA, with utter simplicity.

In your plan, what is the idea behind R6? R3 R4?

 

[Learner]

[quote author="PRR"]

You are driving the tubes' inputs push-pull, but taking the output push-push. The two tubes are FIGHTING each other! (If you re-bias to class C, you get a frequency doubler....) If all else were equal, output would be zero. But you have an apparent error in the U1 U2 stage: 0.5V unbalance. [/quote]

That's what I got after I balanced out the input voltage to the grid of both tubes a frequency doubler, by changing the Rf of the inverting amp to 30k.

The circuit is now balanced(pretty close) and has out of phase output on their plate of 3.6V rms, I could probably use this as a balanced output?

You need to recombine the two tubes' outputs OUT of phase. A center-tapped transformer is THE way to do it.

I have added an op amp at the output of the plate but now the output on both plates are in phase and has 7V rms from V1A and 5.2V rms for V1B, can this be used instead of a transformer?? Is the simulation correct??? or what I meant to say is would this work in real life as predicted in SPICE????

If you can't stand transformers: the White Cathode Follower is THE simple quasi-push-pull tube stage. It isn't true push-pull and is limited to class A operation, but is dead-simple.

Sounds like a good place for me to start, simple is GREAT!!!!

I don't mind transformers if they weren't so costly and that's not even considering the factor of weight and extra space for isolation.

In your plan, what is the idea behind R6? R3 R4?

Well, R3 and R4 is to bias the plate current for each tube. R6 I am not quite sure what it is like some sort of feedback...... BUT it has to be there otherwise the simulation will not work.... same with the feedback caps C7 and C8.

 

[analag]

Ok here's what I would do if I'm building an OTL amp. Go to www.alphacore.com and buy an O-core toroidal core and wind it by hand. I actually did a few including a 1300VA power trannie for my solidstate reference amp, it's easy and if you need help I'll be glad to assist, you can customize it any way you want, dual polarity, dual secondary for better imagery if you build a stereo unit. The circlotron is more symmetrical in design than a Futterman, your choice.

Analag

 

[Learner]

[quote author="analag"]Ok here's what I would do if I'm building a OTL amp. Go to www.alphacore.com and buy O core toroidal core and wind it by hand. I actually did a few including a 1300VA power trannie for my solidstate reference amp, it's easy and if you need help I'll be glad to assist, you can customize it any way you want, dual polarity, dual secondary for better imagery if you build a stereo unit. The circlotron is more symetrical than a Futterman design, your choice.

Analag[/quote]

Winding transformer, mua!!???? :shock: :shock: :shock: :?

Man, that is way out of my league. I can cope with mathematics for electronic calculations and enjoy building prototype circuits using ICs, transistors and tubes but winding transformers is just way too much for me........ Mainly because I don't have a proper set uip for it, and trying to build things without a proper set up/tools is just a real pain.

 

[PRR]

Argh. Curse me for a novice. C5 C7 C8 C6 short the opamps' outputs to each other.

A 47uFd cap is not a perfect short. And your Ideal opamps will pump infinite current into a low-low-low-impedance load. Try that with any real opamp and it will distort like crazy. Put a current probe on the opamp output (or on a 47uFd cap) and at 1KHz you will see a large fraction of an Ampere of AC current.

> R3 and R4 is to bias the plate current for each tube.

No. They serve no fuction in your first version. (If you take the output differentially, you need R3 R4 but you don't need or want R5.)

> R6 I am not quite sure what it is like some sort of feedback...... BUT it has to be there otherwise the simulation will not work....

Can't be feedback, or not any significant amount. It drives the low-low-low impedance of the Ideal opamp and the 47uFd caps. Also: what is it feeding back, power supply ripple? That's all it can feed-back from!

> same with the feedback caps C7 and C8.

As for the simulator blowing up: if you have C7 and C8 (why?) but nothing between them to leak-off the DC, most SPICEs will object because they can't compute the DC condition at the junction of C7 C8 (even if we don't care).

> the output on both plates are in phase and has 7V rms from V1A and 5.2V rms for V1B,

No, they are OUT of phase. Does your simulator have a dual-trace o'scope? If not, use a differential voltage probe. If they were in-phase, the voltage between them would be smaller than the voltage from one output to ground.

> would this work in real life as predicted in SPICE????

Unfortunately, SPICE is pretty correct here (though I'm not sure you are reading what it is saying).

What are the DC voltages around U3? Can you buy an opamp that will stand that voltage?

And look at the Big Picture. 1V in, 17V out, from three opamps (all Ideal, one hi-voltage) and a bottle? Unlike your first attempt, this amplifier DOES amplify, but that sure is a complicated many-parts way to do what it does.

If you were in my classroom, I'd take away your SPICE computer and give you pencil and paper. You won't learn anything by throwing parts randomly at a SPICE board. ANY audio tube problem can be solved on a napkin.

> can this be used instead of a transformer??

To do what? What is your goal here? The thread says "push-pull": do you mean push-pull tubes or a "balanced" output? You have found one way to get a balanced output, but you could get the same result with one 12AU7 or 12AT7, 3 resistors, 2 caps, no sand. All that other stuff is just messing you up.

 

[Learner]

[quote author="PRR"]

To do what? What is your goal here? [/quote]

I am looking for a tube output circuit that can drive 600 Ohm load, also can be used for balanced and unbalanced output. I might try and build a variant of this circuit since I dont have 12AU7 tubes just to see if it'll work as predicted in SPICE, since I can get unbalanced output by coupling the node of R3, R4 and R5 or balanced output by coupling the plates and run inverted signal into the second tube.

 

[analag]

The good old circuitmaker download LTSpice and I'll send you some real tight models to work with. Don't build that circuit, you'll be disappointed, you have opamps surrounding the tube, so what's the tube there for. What do you want, balanced in and unbalaced out, or the other way around. Any way use the tube at the input and have it drive an SSM2141 or SSM2142 depending on what your needs are.

Analag

 

[bcarso]

Learner, the latest schematic now has the tubes in parallel---not push-pull. The new resistors are preventing the a.c. near-short that PRR noted before, but are otherwise doing essentially nothing. If the aim is to apply negative feedback from plate to grid, it is effectively shorted out by the low op amp output Z and 47uF coupling C.

 

[Learner]

[quote author="analag"]The good old circuitmaker download LTSpice and I'll send you some real tight models to work with.[/quote]

Hey, please do. I would be looking forward in seeing it! :thumb: :grin:

you have opamps surrounding the tube, so what's the tube there for.

Oh that, those are just thrown in to get some voltage gain. I am only interested in building the tube stage, so I can combine/mix match various configuration and stages eg. differential or singleton to WCF or SRPP...... to see how it works/sound.

What do you want, balanced in and unbalaced out, or the other way around. Any way use the tube at the input and have it drive an SSM2141 or SSM2142 depending on what your needs are.
Analag

I am looking for flexibility, bal or unbal in plus bal or unbal out all switchable on the front panel.

 

[Learner]

[quote author="bcarso"]Learner, the latest schematic now has the tubes in parallel---not push-pull. [/quote]

Thanks for pointing that out, I do realize that but am no longer pursuing down that path. Originally I was looking to build a class B push pull output stage because of its efficiency and power to drive lo-Z load, but at the same time I do not want to pay for the output transformer since it costs more than the whole circuit it self!!! (like at least 5 times???) :shock:

If I want to have a switchable balanced or unbalanced output by adding another transformer on board, the total cost of 2 transformers is more than 10 times the circuit itself!!! (not to mention the weight and space it needs inside the rack)

I will still build the WCF and the SRPP stage to learn how it works and it'll make good comparison to other layouts, hopefully 1 day in the near future I'll build a a true class B push pull transformer circuit since it has been such a classic layout for so long.(it's a must have to my tube prototype circuit collection) :grin:

The new resistors are preventing the a.c. near-short that PRR noted before, but are otherwise doing essentially nothing. If the aim is to apply negative feedback from plate to grid, it is effectively shorted out by the low op amp output Z and 47uF coupling C.

Those resistors also effects the gain when you couple the output to both plates for balanced output, you need to invert the signal going into the second tube of course. The higher the value of those resistors, you'll get more gain but currently their value of 500K are pretty much close to optimum, likewise the value of R3, R4 and R5 also influence the output level of both bal and unbalance output. If you increase R5 a little bit more eg. to 30K ohms, you can sqeeze a little bit more out of the unbalanced output.

 

[analag]

As soon as you broke it down the circuit popped into my head, I wish I could post a schematic but alas I have no web site yet, coming soon. Ok let me describe it as best as I can. Differential input, with the cathodes tied together, you will need a dual polarity supply, current source the cathode for better CMRR, if you use a solidstate current source then the negative supply rail can be as low as -12V with great results. Grounding one grid is all you need to go unbalanced at the input. The SSM2141 is a balanced input unbalanced out chip, so use the tube to drive that, both plates through 2uF or so caps, 100k resistors going to ground feel free to experiment, with the value(s). The output from the SSM2141 can be your unbalanced out, for balanced out, you could take it from the tube or use the SSM2141 to drive an SSM2142 unbalanced in to balanced out transformer imitating chip. Switch them in and out to suit your taste.Voltage gain should not be a problem with a 12AU7 or 5965 in a line level device. By the way I build my tube models off of the data sheet then fine tune from the ample supply I have in my arsenal for real reliable results (so I can keep my projects cost effective)and I go way beyond the 12axx series.
Drop me an email I'll hook you up. If you wish.

Analag

 

[gyraf]

Hi Analag,

We have a common upload resource for GroupDIY:

http://groupdiy.twin-x.com/index.php

With direct linking possible and all. Provided for free by one of the regulars here :thumb:

Just sign up for an account, and you're flying..

Jakob E.

 

[analag]

Did I forget to tell you, you just made my day?

Analag

 

[analag]

Had a problem posting the schematic

Analag

 

[gyraf]

That dosen't work. What I do is open the image, right-click, wiev properties, and then copy+paste image location.

In this case:

http://groupdiy.twin-x.com/albums/userpics/10049/demo.JPG

and with "image" clamps:

Jakob E.

 

[gyraf]

:grin:

Control gate of the FET with a CV, and you have a hybrid VCA..

I suspect this is what's inside the "black-box" in e.g. the Drawmer 1960 "tube" compressor

Jakob E.

 

[analag]

Excellent, I did a practice image upload and it worked, thank you. I think I will fire up the simulator and mess around with the FET and see what happens. Thinking about an OTL reference amp.

Analag

 

[PRR]

> Control gate of the FET with a CV, and you have a hybrid VCA..

You'd think so, and you "do", but a hi-Z long-tail under the cathodes in limiter duty distorts much worse than a low-Z source (zero up to 1/Gm).

Basically a hard-worked limiter has a long useful range of Class AB operation. If fed constant-current, it can't get out of Class A.

 

[PRR]

> you have opamps surrounding the tube, .... those are just thrown in to get some voltage gain.

Your V3 voltage source can be set for any voltage. 300mV through an opamp gain of 3 is same-as setting the voltage source to 900V and omitting the opamps.

If you omit all the pointless parts:

Even if you need a phase-splitter: the SPICE voltage source (unlike many real ones) does not have to have one terminal grounded.

> I am looking for a tube output circuit that can drive 600 Ohm load, ... flexibility, bal or unbal in plus bal or unbal out all switchable on the front panel.

Are there any real 600 ohm loads left?

And when people throw "600Ω" around, they usually have in mind levels like +4dB VU, which is like +20dBm or 11V peak. Which means 11V/600Ω= 18mA peak. This is HARD WORK for a small vacuum tube.

Like your plan with the 20K plate resistor: even if I did not know what a 12AU7 was, or had some very good tube to use, the absolute maximum current into the load is 250V/20K= 12.5mA, well short of our 18mA goal (if we ever find that historic 600Ω load). Knowing a little about 12AU7, I doubt you have even 8mA flowing in the plate resistor. If we turn the 12AU7 hard-off, most of that 8mA diverts into a 600Ω load, but that's only 8mA*600Ω= 4.8V pk, 3.4V sine, not even +13dBm. Taking 16dB headroom, we can't crank the VU meter past -3dBm, or the -7VU mark on a meter calibrated for the customary +4dBm at 0 VU.

> I don't mind transformers if they weren't so costly

There ARE good autio interfaces that don't cost a pound of iron. TASCAM -10dB 10K unbal works very well in small studios. Hell, it is how 99% of HiFi playback systems use.