John Broskie's Guide to Tube Circuit Analysis & Design
 

The Experiment (12-01-2003)

I wanted to devise an experiment to see if the Tube CAD Journal could support itself, so I could drop another ball and publish once again a full journal each and every month, complete with readers' email, articles, and even a construction project per issue. For those who do not understand why putting out a journal once a month is such a big deal, consider the following. Aside from the job of picking the topics, writing the articles, deriving the formulas, drawing schematics, modeling circuits in SPICE, actually building circuits and testing them, creating two versions of the journal (one for HTML and one for PDF), and posting the journal, the real work lies ahead. The PDF on grounded-cathode amplifiers, for example, has been downloaded about 20,000 times since it was first posted. Now, let's say that 20% of its readers may have some questions or problems understanding some portion of the article, but only 1% actually get around to writing an email. Well, that means I will receive 200 emails. So what? How much work can answering 200 emails be over one year? First, some replies take several pages and require new schematics and graphs and added explanation. Second — and more importantly — that's not a total of 200 emails a year, but for just one article; and I have written over 50 articles already. If I start to publish the Tube CAD Journal every month, those 200 emails will soon explode into thousands. Do the math. The writing of an article is only "the anode cap of a very large sweep tube" of work (I could not bring myself to mention icebergs).

What's the experiment? In the past, to support the Journal I have offered an ongoing metaphorical "bake sale" (the TCJ software); this has not generated enough income to allow me to break away from an existing commitment to bring out a monthly journal. Now, rather than continue to have the bake sale during the intermission of this concert (as it were), I would like to put the horse back in front of the cart by first offering the bake sale, to make certain the event can actually be funded. In other words, I have a new TCJ tube-related program for sale and if the sales are big enough to justify canceling some of my present commitments, I will put out a regularly scheduled Tube CAD Journal. If the sales aren't up to the task, I will continue to write the occasional blog entry and long article as my schedule permits.

But what sort of program to offer? As I have received hundreds of emails asking for a push-pull software program, it seems that push-pull amplifiers are no longer shunned. So here it is: the GlassWare Push-Pull Calculator. With this program, you can quickly find out if your OTL or transformer-coupled amplifier project will produce your expected results. Eight different push-pull topologies are covered:

      Cathode-Follower Totem Pole Amplifier

      Grounded-Cathode Totem Pole Amplifier

      Mid-Referenced Totem Pole Amplifier

      Circlotron Amplifier

      Right-Grounded Circlotron Amplifier

      Grounded-Cathode Transformer-Coupled (fixed bias) Amplifier

      Grounded-Cathode Transformer (cathode bias) Amplifier

      Cathode-Follower Transformer-Coupled

 

Twenty-four calculations (output wattage and impedance, distortion, dissipation, idle current…) are just a button-push away. Seven graphs display the details behind the tabulated results, including the output wave form and the combined plate curves (the Live Curves™ mathematical model powers the simulations).

"Sounds great! How much is it and how can I buy it?" you ask. The price is only $29 USD.  If you want a CD (to give friends as stocking-stuffers, say), those will be $3 more.  It's available at the GlassWare Yahoo! store.

 
-  John R. Broskie

 

New Article and Blog Entry
Yes, it’s been a while since the last post. But I believe the wait was worth it, if you interested in, puzzled by, or even bothered by the circlotron amplifier… most tube practitioners are all of the above. Like an octopus hiding behind a cloud of black ink, the circlotron’s functioning hides behind the unfamiliarity of its topology. In the attempt to disperse the cloud of misinformation, we discover a new circuit topology that functions identically with the circlotron, but looks nothing like it.

The Blog entry below should please the hybrid and simple amplifier enthusiasts.

Miscellany
My ISP viewed my mailing 800 email new-journal-posting notifications as a form of Spam. So it looked as if the “Please notify me when a new article is posted” portion of this site had to go. Fortunately, Yahoo! Groups provides a way out. Thus, if you wish to be emailed a notice when a new article has been posted or an error has been caught, join the Yahoo! Group “tcj_readers” by emailing to tcj_readers-subscribe@yahoogroups.com (Note: if you choose “special notices only,” you will not get the updates sent to you.) And, as always, be sure to give the Tube CAD Journal companion programs a good look, as I think you might find them quite useful. (Besides, they are dirt cheap and they help justify keeping this journal going, as they are -- for now -- the only income-producing portion of this site.) Please send in your comments and suggestions.

 

 21 October 2003

Unusual tube circuit

Reader, Donald, sent in the following link.
http://www.diyaudio.com/forums/showthread.php?s=014c2cc0e4d4f1c1e090b4f87346a828&threadid=20686 which then leads to the following link: http://www.ne.jp/asahi/evo/amp/el86/report.htm

Interesting indeed—or is it? (The Japanese website held much more interesting schematics.) A generic amplifier (the original post only specified a gain block, which could be pure tube or solid-state) with a difference; namely, its feedback resistors pair is a triode: in other words, the triode’s mu (actually, mu + 1) is used as a voltage divider (a plate amplifier, see email_2001 ).

If used with a solid-state amplifier, then this circuit probably holds to the tubes-make-nice-distortion school of audio design, using a tube’s euphonic distortion to make a nice-sounding hybrid, a school of thought I adamantly oppose. I do not like tubes for their distortion, but in spite of their distortion.

Here's an analogy: imagine that fine wine is being siphoned from the oak barrel via a twenty-foot length of rubber hose, a hose that lends a slight garden-hose-on-a-hot-day flavor to the wine. People complain. Thus, over time, the rubber hose is replaced by a steel-alloy pipe, but which lends a harsh, metallic flavor to the wine that sends convulsions through some tasters, like aluminum foil against a tooth filling. Not surprisingly, many prefer the taste of the old rubber to the new metal, in spite of the technological advance that the steel pipe represents. Idiots! The many-advanced-degreed enologist explains that careful testing shows that the metal pipe only contributes a fraction of the suspended solids that the old rubber hose did and thus wine must taste better because of it. Well, while the argument runs on, someone purposes a remedy: the wine will be pumped through the steel pipe first, then through the rubber hose. Now everyone will be happy… Or will they?

It seems to me that the solid-state amplifier would still be doing too much, that the same 25mV-range-of-linearity sound that the transistor-based differential input stage exhibits would still be in charge of the input signal. How much can we expect the triode to do, considering that it does not amplify the signal in any way? And what happens at start up, when the tube is cold and not conducting? Or when the tube is wiggled in its socket? Does the solid-state amplifier slam to one rail, while the woofer smolders until the triode conducts or reconnects?

But before dismissing this design altogether, let have some fun. Couldn't a vacuum tube signal diode be used in place of the triode, as it too has an rp?

Or, what if we use a triode as the bottom resistor of the two-resistor voltage divider that sets the negative feedback ratio of an amplifier? Both resistors influence the sound out of the amplifier. The schematic above right makes the point. The triode’s cathode resistor is being shunted by the triode’s rp/mu (the inverse of its gm) and the amplifier is safe in terms of DC offset, even at start up. In fact, the triode could be pulled from its socket, while the amplifier was in use, with the only result that there would be a momentary popping noise and the gain would fall off a bit.

(By the why, when you see any bipolar power supply, imagine that it implies an additional power supply rail voltage three times the single rail voltage, as a voltage tripler can easily be added to the bipolar power supply. In the circuit below, we see a conventional bipolar power supply with four extra diodes and three more caps. This voltage tripler is a full-wave type that won't need too much filtering.)


Still, what bothers me is the idea that the throwing the ball to the solid-state amplifier first. Is this a good idea? If the solid-state amplifier fumbles, we have no hope of winning. So, how do we reorder the amplifier so that the tube gets first crack at the input signal? In my posting on simple amplifiers (
http://www.tubecad.com/page12.html) I displayed a topology that might prove to be a real sleeper, shown below.

The idea here is that the tube and solid-state amplifier form a simple two-stage amplifier, with the tube in charge of the input signal and the global negative feedback. Yet, the loudspeaker is never at risk, as the solid-state amplifier’s feedback is never left open. Both the tube and the solid-state provide voltage gain, so a low-mu triode, such as the 5687 could be used. If more second harmonic coloration is needed, then the schematic below might be the way to go. Once again the triode is charge of both the input and the feedback; but as the triode undergoes greater current swings than the triode in the circuit to the left, it will produce more distortion.

The circuit at the right uses the triode’s impedance at its cathode added to its cathode resistor to define the input resistor of an inverting amplifier. The solid-state amplifier inverts the signal and is in charge of the negative feedback loop this time. Either a resistor or a constant-current source can be used to provide a current path for the triode. Ideally, a servo-loop could be implemented that would allow removing the 10µF coupling capacitor. (A FET and IC and few resistors and capacitors is all that would be needed.)

If using a complete solid-state amplifier seems daunting, or if you prefer single ended amplifiers over push pull amplifiers, then any of the following three amplifiers would prove interesting.

The amplifier above is single ended and uses a choke to load the P-channel MOSFET. The DCR of the choke against the idle current defines the DC offset. If the DCR is low enough, the coupling capacitors could be jettisoned. If the amplifier were to be used in a bi-ampped system, with the it powering the higher frequencies, then the crossover capacitor could do double duty as the output coupling capacitor.


This next amplifier uses the same basic idea, but replaces the inductor with MOSFET based constant-current source. The B+ had to be raised to allow more voltage swing, as the constant-current source cannot swing below ground the way the inductor can. (“Constant-current source,” like so many terms in electronics, is a misnomer, as a true source would be able to provide its own voltage and current. A better name for this type of circuit is “constant-current regulator,” as it only regulates the amount of current flow. A 1Meg resistor placed across a true 1A constant-current source would see 1,000,000 volts develop across its leads.)

The amplifier above uses N-channel MOSFETs (which I prefer) and will more reliably self-center. (Note that all three of these amplifiers will exhibit a good deal of second-harmonic-distortion, as the triode will turn on more readily than it will turn off and the same holds true for the MOSFET, thus the two device's non-linearity are going in the same direction, which means that curvature will compound, not subtract. Still, there is a negative feedback loop to lower the output impedance and distortion. Furthermore, all three of these simple amplifiers require strict class-A operation, which means heat and massive power transformers.)

In contrast, the circuit above can be run in class-A, B, AB. Two current mirrors relay the tube-based differential amplifier’s balanced current swings to two resistors. One resistor terminates into the output and the second terminates into the negative power supply rail. Thus each output MOSFET sees the same magnitude of drive voltage, but in opposing phase. With the feedback loop closed, the triodes are in charge. In the absence of a negative feedback loop, this amplifier would have a near infinite output impedance and excessive gain. The current regulator that loads the triodes’ cathodes also sets the idle current of the output stage.

As it stands, this amplifier close to being truly functional. What’s missing is output stage Zorbel-type networks and a means by which the DC offset could be adjusted. The obvious solution, a potentiometer that spans both cathodes is a poor choice, as potentiometers can break contact at times or become scratchy. A better idea is to use a pair of parallel and series resistors to reduce the potentiometer’s possible malfunctioning, as shown above.

Of course a DC servo loop could replace the potentiometer altogether. As for output devices, other MOSFETS or, even, NPN transistors could readily be used. In fact, vacuum tubes could be used with the addition of coupling capacitors and negative bias power supplies… or something like what you see to the right. Here the two current mirrors connect directly to the output tubes’ grids. Both triodes function as cathode followers, which means that the global feedback ratio would be greatly reduced. The extra PNP whose base is grounded creates a cascode-like structure that prevents the PNP transistors from seeing too much voltage (300 volts is pretty much the voltage limit for PNP transistors).

 

//JRB

 

     

Tube CAD does the hard math for you.

This program covers 13 types of tube circuits, each one divided into four variations: 52 circuits in all. Tube CAD calculates the noteworthy results, such as gain, phase, output impedance, low frequency cutoff, PSRR, bias voltage, plate and load resistor heat dissipations. Which tube gives the most gain? Tube CAD's scenario comparison feature shows which tube wins.

Windows 95/98/Me/NT/2000/XP

For more information, please visit our Web site : 

www.glass-ware.com

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

www.tubecad.com           Copyright © 1999-2004 GlassWare           All Rights Reserved