John Broskie's Guide to Tube Circuit Analysis & Design

09 January 2012

Happy New Year

Wow, this will be my 13th year of posting the TCJ. That is a long time and a lot of words, schematics, and photos. Well, like any other reasonable fellow, I hope for the best, but I plan for the worst.

Since 1980, I have been reading about audio's impending demise. Yet, I am certain that tubes will still rule and are not about to be vanquished. On the other hand, I can imagine wire going the way of the dodo bird. As WiFi insinuates itself into more and more electrical devices, wire, cables, interconnects... become less needed. My laser printer only plugs into the wall socket; computer mice no longer sprout wires, nor do keyboards. How long will it be before the toaster, microwave oven, and the vacuum cleaner are WiFi enabled? And as class-D power amplifiers become even cheaper and more capable, it will make more sense for loudspeaker manufacturers to include them in their speakers, as three class-D amplifiers and an elaborate three-way, active crossover will prove cheaper than one high-quality inductor or film capacitor. Adding WiFi ability will also cost little. Besides, just about everyone hates cables as much as my dog hates his leash—and for the same reasons: they restrict too much and just get in the way.

So, what will high-end audio cable companies sell in a wireless future? My guess is that they will sell high-end audio ether, the all-pervading, infinitely elastic, massless medium formerly postulated as the medium of propagation of electromagnetic waves. (Hey, why not? If failed and outdated Keynesian economics can make a comeback, why not the ether.) Just imagine placing large void-filled containers in between your speakers and the signal source. Or, perhaps, high-end audio wave-guides (a system of material boundaries in the form of a solid dielectric rod or dielectric-filled tubular conductor capable of guiding high-frequency electromagnetic waves) might be the answer. In other words, these would look much like existing high-end audio cables, cost just as much as high-end audio cables, but they will not physically connect to anything. (Will UL approval be needed?) You may laugh, but just wait...

Okay, back to 2012. It started out a bit roughly for me, as I spent three days not able to ship through the website, as it would always hang at the last step, the printing of postage. Many cryptic error messages came up, but no one at knew what they meant; nor did Google reveal any links to an answer. Finally, in a chat with representative was the solution revealed: NEVER USE "Ø" CHARACTER IN AN ADDRESS LABEL. That was the problem that blew up Amazing. We have all heard of the dreaded divide by zero error; well, here a capital O with stroke error.

The year 2012 just might be the year that I finally get around to updating the structure of this website. It's just too big and ordering it by chronology make little sense. Ideally, I would like to subdivide it into many categories, such as phono stages, line stages, crossovers, headphone amplifiers, hybrid power amplifiers, balanced circuits, I-to-V converters... Then post much more often, but post much shorter entries, which would then move off the front webpage and appear in its appropriate category. (Presently, I only post a third of what I am actually working on and I would like to up the ratio substantially.) In other words, if a reader was only concerned with tube-based electrostatic headphone amplifiers, he could read the long webpage devoted only to tube-based electrostatic headphone amplifiers. We will see.



New 5687 Aikido All-in-One PCB

5687 Aikido All-in-One PCB
The 5687 Aikido PCB is back, but different. Now, all the tubes are 5687s and the PCB is an All-in-One effort that holds both the high-voltage B+ and heater power supplies, just like the octal and 9-pin Aikido All-in-One PCBs. I know that I am not alone in loving the 5687 and after I assembled the PCB shown above and listened to the first few notes, I felt as if I had reunited with dear friend, affection and devotion swelling within me.

Just like our friends, tubes have their own personalities. A 6SN7 doesn't sound like a 12B4 or a 5687. It certainly isn't so much an issue of being better, but of being different, as each tube has its strengths and weaknesses—just like our old friends. I started listening with Philips JAN 5687s from the 1980s. Amazingly good. I know that many tube lovers pay hundreds of dollars for NOS tubes, so the sonic splendor that these relatively cheap tubes produce is truly wonderful and cost effective. I then pulled out my stash of Pearl Audio's NOS CryoValve-treated Tungsol 5687s. Mercy, mercy, mercy. Adding Pearl tube coolers only increased the sonic glory. Rather than risk sounding like an audio reviewer, whose spillage of fanciful adjectives implies an intoxication more chemical than sonic, I'll just say that I loved the sound, every aspect of it. It was like meeting up with an old lost friend who shows up one day surprisingly rich, cultured and erudite, dressed like a British diplomat, accompanied by a famous babe actress, and driving a Ferrari. Who knew he had it in him?

Moreover, it had been a few years since I last seriously listened to the 5687 and I was eager to hear what double barrel 5687 sounded like. Double barrel? The previous 5687 Aikido PCB only used a 5687 as the output tube, with a 6CG7, 6DJ8, 12AU7, 12BH7... as the input tube. The new 5687 Aikido All-in-One PCB uses a 5687 for both input and output tubes. In other words, there is more 5687 flavor as a result, which is a good thing. A bad thing is that the heaters draw a total of 1.8A @12Vdc, which makes an all-12AU7 Aikido All-in-One's 0.6A seems so very puny in comparison, which indeed it is.

So what have we bought with this huge torrent of heater current? Big cathodes and high current. The 5687 offers a low plate resistance and a high transconductance, so it can deliver four times more current than the 12AU7 with the same cathode-to-plate voltage. Why is a high idle current important? We all know that low impedance loads require high current, but many do not realize that high-capacitance also requires high current, as it is current that charges and discharges capacitance quickly. If insufficient current is available, the result is a slew-limited sound that can only depress our ears, as the turgid, slow sound is gloomy in the extreme. Understand that a low output impedance is not enough. For example, a 12AX7-based cathode follower can present an output impedance of just 600 ohms, but its trivial 1mA of idle current will not be able to melt away the high-capacitance of a long length of fancy interconnect. In contrast, a cathode follower that presented a higher output impedance, say 1kohm, but a much higher idle current, say 10mA, can easily burn through the same cable capacitance. In other words, the faster the rate of change in voltage (the higher the frequency) and the greater the capacitance, the more current is needed.

5687 Plate Curves

Getting more current flow through a triode is easy: just increase the plate voltage. The higher the plate voltage, the more current a triode can conduct. An easy solution, but not practical one, as we quickly run into a triode's plate dissipation limit. Moreover, the higher the B+ voltage, the lower the capacitance in the power supply. Why? Money and real-estate, PCB real estate that is. High voltage capacitors cost much more than low-voltage capacitors; in addition, they are less volumetrically efficient. And as there is only so much room on a PCB, we cannot use a fatter capacitor than the board has allocated space for it. This is why a 5687-based, low-voltage, Aikido line-stage amplifier makes so much sense: although a 5687 can withstand a cathode-to-plate voltage of 330V, it can work beautifully with only 75Vdc of cathode-to-plate voltage in an Aikido line-stage amplifier with a relatively low B+ voltage of only 150Vdc. In other words, we can get away with a lower B+ voltage, because the 5687 is so well suited to low-voltage use. (Configured as a headphone amplifier, a lower B+ voltage, say 160Vdc, is the only way to go.)

The Aikido topology may be sneaky, but it is not magic; it works because its Aikido cathode-follower output stage knows exactly how much power-supply noise will leak from the input stage (imagine if you were given the answers before taking a test). There is a practical limit, however, to how large a power-supply noise signal can be nulled at the Aikido’s output. I have received a few, very few complaints from a those who have complained of too much power-supply noise leaving their Aikido line-stage amplifiers. Usually, the hum is 60Hz (or 50Hz) and the result of poor grounding practice.

On a few occasions, the hum came in at 120Hz (or 100Hz) and was power supply ripple noise. I asked for an AC reading of the power-supply noise and I just about fainted when I read the reply e-mail proclaiming 40V of peak noise, the result of an LC power supply oscillating like mad. To be honest, 4mV would surprise me; 40Vpk makes me wobble. Sorry, but 40Vpk of power-supply noise cannot be scrubbed away with a 6SN7's 3Vdc of cathode voltage, as the cathode voltage would have to be substantially greater than one half the power-supply noise for the Aikido cathode follower to work, as it is a single-ended affair that can only give up so much idle current before shutting off.

The lower the power-supply noise going into the Aikido circuit the better. Thus, we want the largest voltage-dropping RC resistor value possible, as it reduces the ripple appearing at the tubes’ power supply connection, but high current means that we would lose too much voltage across this resistor, so its value must remain low. Thus, we can only increase the power supply capacitor's value, so we must strive for the lowest raw B-plus voltage possible, as it will allow using a much larger valued reservoir capacitor (and limit the heater-to-cathode voltage to a safe voltage as well). In other words, by using the 5687 we can get away with a relatively low B+ voltage, as the 5687 can draw an amazing amount of current at low plate voltages and a lower B+ voltage buys us much more more power-supply capacitance.

The 5687 Aikido line-stage amplifier I put together uses power-supply capacitors rated at 200V and a 120Vac power transformer secondary voltage, yielding a raw power supply voltage of about 170Vdc before the first RC filter and a final B+ voltage of about 150Vdc at the tubes. With output-stage cathode resistors of 180 ohms, the idle current is 12mA in the Aikido-cathode-follower output stage; the input stage uses 390-ohm cathode resistors that result in about 7mA of current flow through the input stage. All of which, brings the total line-stage amplifier current flow to about 40mA.

As is my preference, I used two power transformers, one for the B+ and one for the heater regulator. They were industrial transformers that I had sitting about. If I had to buy two new transformers, I would get something like a Hammond 185D12 for the heaters and a Hammond 166F120 for the B+ voltage; or, if you prefer toroid transformers, the Triad Magnetics VPT230-110 for the B+ and the VPT12-4170 for the heaters would work well. On the other hand, if I were to buy a single power transformer, I would opt for the Edcor XPWR054-120 , which offers 120V-0-120V and 12V-0-12V secondaries. Its 2A heater winding may seem a tad weak for the job, but as the B+ only draws a fraction of the high-voltage winding's potential (40mA out of 100mA), we can squeeze quite a bit more from the heater winding, without stressing the transformer's core.

I have been recommending Edcor transformers for the past few years, but I didn't have much actual experience with them, so I bought a power transformer from them to test. Nice. Well made and wonderfully inexpensive. My only complaint is that the "Edcor" name is not painted on the transformer's bell cap; instead, it's on a clear sticker, which looks more than a tad tacky. Nonetheless, highly recommended.

The 5687 Aikido Stereo All-in-One PCB and kit and tubes are available at the GlassWare-Yahoo store. (I only have a very limited number of boards, alas.)


Single PS Circlotron
My renewed love affair with the 5687, got me thinking about building a balanced-output headphone amplifier for my Sennheiser HD650s using the 5687 in a modified circlotron circuit.

I agree; it does not look like a circlotron circuit. Nonetheless, it functions like a circlotron, exhibiting the same Zo, gain, and distortion. Lets replace the large electrolytic capacitors that cross-couple the triodes' cathodes and plates with 100V batteries.

The batteries allow the 5687s to draw 18mA each, while only 1mA flow through the 3k and 14k resistors. In other words, the 3k resistors help set the idle current, but do not flood the triodes with current. In contrast, the large-valued electrolytic capacitors in the previous example act like the batteries in many ways, but they cannot pass DC current, so the cathode and plate resistors much see the same current flow as the triodes do, as they are all in DC series. Since the triodes get all their current from the B+ connection and ground, the capacitor-based circlotron is restricted to class-A operation, whereas the battery-based version is free to break into class-AB. With 300-ohm headphones as the load, class-A is fine; indeed, it is preferred.

In order to get the same 18mA of current flow through each triode, much lower resistor values will be needed.

Of course, the 167-ohm cathode resistors could be replaced by 18mA constant-current sources, which leads to this fleshed out design.

Click on schematic to see close-up

In this circuit, the 5687 triodes run 10mA idle currents and the B+ voltage is 160Vdc. The input stage uses a differential amplifier that can accept either a balanced or unbalanced input signal. The output impedance equals rp/(mu + 2), just as it does in big circlotron power amplifiers. For many readers, this is a good place to stop reading, as they have a workable schematic. Bye. For the rest of us, this is where things get interesting. First, note that the input stage's differential amplifier offers a PSRR of zilch, nada, zip, as all the power-supply noise will be present on its two outputs. Terrible—or is it? The single-PS circlotron output stage also offers a PSRR of 0dB, because of the two constant-current sources. Thus, relative to ground, the balanced output will hold all the power-supply noise, but differentially no power-supply noise will appear. Nice. The headphone drivers really do not care anything about what is going on at ground; their only concern is the voltages across their leads.

Couldn't the output coupling capacitors be removed? Only if you like to live dangerously. The only certainty is that both triodes will share the same idle current; but their cathode voltages depend on the tube matching within the glass envelope. Couldn't a DC servo loop be used to force an equal cathode voltage on each triode? Yes, most certainly.

Click on schematic to see close-up

I have added a negative 12V power-supply rail and eliminated the output coupling capacitors. The DC servo keeps the outputs a 0Vdc, even if the 5687 tube is cold and not conducting or pulled from its socket. This feat of magic is the result of including the 1N4001 diodes. Under normal operation, the 5687 cathodes will always be at some voltage positive relative to its grid, so the diode falls out of the circuit, as it is reversed biased. But at startup, when the cathodes are cold, the DC servo's output voltage will swing up positively, as it tries to pull up the output voltage and the diode will turn on, conducting and connecting the DC servo's output directly to the constant-current source, pulling up the output voltage to 0Vdc. (Missing from the schematic is an extra 1N4001 diode that spans from ground to the 6922's shared cathode connection to their 10mA constant-current source, as at startup, we do not want the 6922 cathodes to be at -12Vdc, while the grids are at 0Vdc.) How do you get the +4Vdc power-supply rail? Easy. Just ground the output of a +12V regulator and use its positive input voltage (about +4Vdc) as the OpAmp's positive rail voltage.

Okay, what if all these solid-state constant-current sources just creep you out? What if you long for a pure-tube version of the above? The following circuit uses cathode resistors instead of constant-current sources and a split-load phase splitter instead of the differential amplifier; in addition, it runs a much higher B+ voltage.

Click on schematic to see close-up

The input grounded-cathode amplifier and split-load phase splitter present a PSRR of -6dB, which means that half of the power-supply noise will leak, in phase and in equal amplitude, from each phase output. The cathode-resistor based circlotron output stage also exhibits a PSRR of -6dB, so the balanced output will also present a PSRR of -6dB. The 2k plate resistors are matched by the two 1k cathode resistors in series, which define a 2k resistance, which, in turn, defines a 50% AC voltage divider with the 2k plate resistors. Once again, differentially, it will not matter to the headphones. Note how we matched kind to kind. Using the above circlotron output stage with a differential amplifier input stage would be a bad idea, just as using the above frontend with the CCS-based circlotron output stage would be a bad idea. Why? We do not want the power-supply noise to mix with (or modulate) the audio signal; we want the 5687 to be oblivious to the power-supply noise.

Also note how the "Garter-Belt" bias technique is used, wherein each triode biases the other triode, evening out their idle currents. The 1N4007 diode is there to protect split-load phase splitter triode at startup, when its cathode is at 0Vdc and its grid is at 150Vdc. Why are the output coupling capacitors there? I am just too nervous about burning up my beloved HD650s. (A single coupling capacitor could be used, if its 1M resistor terminated into the other output, not ground. Yes, 22.5Vdc would be present on both outputs.)

As a bonus homework problem, can you evaluate what the output impedance is from ground to just one output? Hint: the correct answer will probably amaze you; it's not anything close to rp/(mu + 2).

(Before anyone is tempted to e-mail me, asking which version I would build, the answer the the first with the differential amplifier and bipolar power supply and DC servo.)



Next Time
I hope to have good news about an RTIE capacitor delivery.



Kit User Guide PDFs
Click image to download

Download 5687 Aikido USer Guide PDF

BCF User Guide

Download PS-3 User Guide

SRPP+ User Guide

Aikido All in One user guide
Janus regulator user guide

Aikido Phon Preamp user guide
Aikido 9-Pin Stereo User Guide


E-mail from GlassWare Customers

Hi John,

I received the Aikido PCB today - thank you for the first rate shipping speed.
    Wanted to let you know that this is simply the best PCB I have had in my hands, bar none. The quality is fabulous, and your documentation is superb. I know you do this because you love audio, but I think your price of $39 is a bit of a giveaway! I'm sure you could charge double and still have happy customers.
     Looking forward to building the Aikido, will send some comments when I'm done!
   Thank you, regards


Mr Broskie,

I bought an Aikido stereo linestage kit from you some days ago, and I received it just this Monday. I have a few things to say about it. Firstly, I'm extremely impressed at the quality of what I've been sent. In fact, this is the highest quality kit I've seen anywhere, of anything. I have no idea how you managed to fit all this stuff in under what I paid for it. Second, your shipping was lightning-quick. Just more satisfaction in the bag, there. I wish everyone did business like you.

Sean H.

9-Pin & Octal PCBs

High-quality, double-sided, extra thick, 2-oz traces, plated-through holes, dual sets of resistor pads and pads for two coupling capacitors. Stereo and mono, octal and 9-pin printed circuit boards available.

Designed by John Broskie & Made in USA

Aikido PCBs for as little as $24


Support the Tube CAD Journal


get an extremely powerful push-pull tube-amplifier simulator for

Only $19

TCJ Push-Pull Calculator
Version 2

Click on images to see enlargements

TCJ PPC Version 2 Improvements

       Rebuilt simulation engine
       Create reports as PDFs*
       More Graphs 2D/3D*
       Help system added
       Target idle current feature
       Redesigned array creation
       Transformer primary & secondary
              RDC inclusion
       Save user-defined transformer     
       Enhanced result display
       Added array result grid

                                       *User definable

TCJ Push-Pull Calculator has but a single purpose: to evaluate tube-based output stages by simulating eight topologies’ (five OTL and three transformer-coupled) actual performance with a specified tube, power supply and bias voltage, and load impedance. The accuracy of the simulation depends on the accuracy of the tube models used and the tube math model is the same True Curves™ model used in GlassWare's SE Amp CAD and Live Curves programs, which is far more accurate than the usual SPICE tube model.

Download or CD ROM
Windows 95/98/Me/NT/2000/XP

For more information, please visit our Web site :

To purchase, please visit our Yahoo Store:           Copyright © 1999-2012 GlassWare           All Rights Reserved