Happy Halloween
Fall is going on strong where I live in Colorado, with an abrupt switch from warm to cold. In fact, the leaves are putting on a better display of colors this year, no doubt due to all the global warming caused by the increased prevalence of tube-audio gear here.

Single-Tube, Two-Triode Magic
Where to begin? Creating complexity is easy, but seldom satisfying; making simple but effective is difficult, but ever so gratifying. Imagine if mousetraps didn't exist, but we still needed them. An enterprising company, such as Apple, might then come out with the iMouse, which would require frequent USB recharging and WiFi access, along with holding a video camera for its AI mouse recognition (surely, you would want to be alerted when a mouse was no more) and a syringe to deliver a painless, but lethal injection and small oven to incinerate the unhappy rodent; a technological wonder for only $700, two for only $1,000 USD.

My last post held many single-tube, two-triode designs, which amazed me with their ability to deliver so much with so little. In contrast, an audio IC can easily contain a hundred parts, resistors, capacitors, diodes, transistors… Here is an example of a simple two-triode design from Post 626:

The only cheat was the constant-current source, but it can be replaced by a single LM317 and one resistor; failing that, we can replace the constant-current source with a large-valued cathode resistor that terminates into a negative power-supply rail. The triode on the left provides all the signal gain; the one on the right; current delivery. The external load impedance of 300 ohms represents a headphone driver.

While going through old email, I found a line-stage amplifier circuit that I had sent a friend. The circuit was also a two-triode affair that delivered less gain and a higher but still low output impedance.

This line-amplifier circuit had three goals: only 12dB of gain; some harmonic enrichment, where the triode's glory would shine; and a good PSRR figure. Several unsought features fortuitously appear, such as the same triode cathode voltage, which sidesteps the CCDA's heater issue of having a large voltage differential between cathodes.

If the B+ voltage were 300Vdc, the cathode-voltage difference between triodes would be 150Vdc, which would absolutely require referencing the heater power supply to 75Vdc. By the way, the CCDA's PSRR is only -6dB.

Another feature is that the cascade of a grounded-cathode amplifier into an anode-follower (aka plate follower) results in no phase inversion at the output. By the way, if we replace the 24k plate resistors with 5mA constant-current sources, we get stellar PSRR and the ability to use a much lower B+ voltage.

(The unspecified negative feedback resistor must be calculated to compensate for the high gain from the first stage with the constant-current source loading.)

Returning to the original 12dB line amplifier, the one potential problem is the getting the right capacitor values for the Aikido Mojo to work well. Electrolytic capacitors come in very coarse capacitance increments based on the following base values 1, 1.1, 1.2, 1.3, 1.5, 1.6, 1.8, 2, 2.2, 2.4, 2.7, 3, 3.3, 3.6, 3.9, 4.3, 4.7, 5.1, 5.6, 6.2, 6.8, 7.5, 8.2, 9.1; this is the E24 spread of base values, which hold 24 values; E48, 48 values. In addition, electrolytic capacitor tolerance is usually a sad 20%. In contrast, we can get resistor values in much finer gradations and with much tighter tolerance (E96 and E192), as little as 0.01%, even 0.001%. But before we abandon the two-capacitor Aikido Mojo technique, we should see how it works.

Without the second cathode capacitor, the one that bridges the cathode to the B+ voltage, the power-supply noise appears at the output in anti-phase to the ripple. This means that the triode on the right has over amplified the amount of ripple appearing at its grid, so we must introduce a small sampling of the ripple to its cathode to create a power-supply noise null at the output. In SPICE simulations, the PSRR at 100Hz was -53dB, using the 0.33µF, 100µF, and 1.2kµF capacitor values.

I must point out that this is only the raw PSRR, which does not include any of the usual PSRR-enhancement techniques usually applied to power supplies, such as chokes, RC filters, and regulation. For example, by increasing the B+ voltage a tad and inserting a 3k series power resistor, we get a vastly greater PSRR, as electrolytic capacitors and added resistor form an RC filter, whose -3dB frequency is below one Hertz (0.57Hz).

Here is the resulting final PSRR with the RC resistor:

This is close to one hundred times better.

2nd Aikido Mojo Approach
Let's move on to a different Aikido Mojo technique, the two-cathode-resistors alternative, which takes advantage of the wider range of resistor values. We start with a stock 12dB cascading line amplifier.

The PSRR without any Aikido-Mojo magic is a paltry -19dB; with Aikido Mojo, -78dB, an almost 60dB improvement, which translates to a 1,000 fold improvement. In other words, 1V of ripple is reduced to 1mV; 1mV to 1nV. What is required are an additional resistor and capacitor.

Note that the internal coupling capacitor value has been increased to 1µF from 0.1µF. Both this capacitor and the 100µF electrolytic capacitor can be increased in value, but shouldn't be reduced in value. The principle behind this Aikido-Mojo variation is the same as before, but uses a two-resistor voltage divider rather than a two-capacitor AC signal voltage divider. (The 100µF capacitor terminating into the B+ voltage is effectively, in AC terms, a dead short.)

Since the 6SN7 offers a lower amplification factor (mu) than the 6DJ8, the anode follower's negative feedback resistor ratio differs from the 6DJ8-based design example. Here is the SPICE-generated PSRR graph.

If we rather use a 6DJ8, we must alter the part values; and we can lower the B+ voltage.

The PSRR for this circuit is -72dB at 100Hz. Not bad, but we can do better. Let's start with a core circuit and then add the Aikido Mojo.

3rd Aikido Mojo Approach
Unlike the previous design, the first stage gets an RC power-supply filter.

Also note the anode follower's cathode resistor is now bypassed with a large-valued capacitor, which will result in a lower output impedance. The addition of the B+ voltage RC filter delivers a PSRR of -44dB. If we add one resistor, however, we get a vastly improved PSRR.

"Cynosure" means an object that serves as a focal point of attention and admiration. The word derives from the Ancient Greek word for dog tail, kunosoura, which became in Latin, cynosure. Well, this dangly resistor reminds me of a dog's tail. (See Posts 426 and 536 for more cynosure designs.) While I am at it, I will point out that Aikido is the Japanese martial art that uses one's opponent's aggression against him, while the Aikido Mojo uses the power-supply noise against itself. (A friend once said, "I like hearing John talk, as he is not only entertaining, but you also earn college credits at the same time.")

We no longer have to inject some power-supply noise into the anode-follower's cathode to force a power-supply-noise null, as we can limit the amount leaking from the input stage to create the desired power-supply noise null. We do this by undoing some of the RC filter's filtering. Yes, by doing less we get more.

If you are wondering what the distortion is like leaving this circuit, here is the Fourier graph of harmonics based on an output signal of 1Vpk at 1kHz.

The THD is below 0.1%. Note the strong 2nd harmonic and the greatly reduced higher harmonics. Next, we see the PSRR graph.

That's some PSRR!

Another Aikido Mojo technique is to feed the anode follower's negative feedback loop directly with a resistor.

The assumption here is that, due to the RC filter, the input stage does not leak much power-supply noise at all.

The resulting improvement isn't as great as the previous variation due to resistors in the 15M range no longer are offered in tiny resistance increments, but the coarse E24 spread of values.

While this PSRR is not as good as the previous variation, it's exemplary.

Artificial Intelligence Image Creation
The newly inserted AI panel in my Firefox browser kept pestering me to let it show me what it could do. I relented. I asked it to create an image of a tall, dipole, loudspeaker, with two midrange drivers and one tweeter. Here is what Google Gemini produced:

A ridiculous design, but nonetheless effing amazing. I often bemoan my inability to 3D model, as it would come in so handy in illustrating my new ideas. Hell, if AI is this good, do I really need to learn how to use a 3D modeling program?

Back to the Past
I stopped making interconnects about 15 years ago—far too long a pause. What's changed is a friend got hold of a spool of silver-plated copper stranded twisted-pair wire with a shield and Teflon sheathing. The fellow who sells the bulk cable has strict policy of not selling to audiophiles, only to electrical engineers and scientists and technicians. (I so understand his opposition to audiophiles. Imagine an audiophile customer calling to complain of the wire's overly caramelized lower midrange or overly brusque highs.) Well, soon enough the soldering iron was hot and interconnect designs filled my mind.

Designs, what designs? It's just wire after all, so what design choices can exist?

If we are only dealing with two wires, we have at least the choice between twisted or untwisted. (Of course, we have left out coaxial cable with a single wire surrounded with shield, which cannot be twisted.) In addition, we can add ferrite beads or magnets to the two-wire interconnect. (Do not laugh. They are added and the cables sell for hundreds of dollars.) Even if we are left only with twisted or untwisted, we can opt either for a constant twist or a twist that spins in the opposite direction somewhere along the length. Since stereo long ago beat mono, we usually have two interconnects, not one. Thus, we can get fancy by braiding to pairs of wires.

I actually just made this arrangement, but have yet to listen to it. Now let's add a shield to our two wires and things get interesting.

The usual setup grounds the shield at both ends.

This is how most non-audiophile-grade interconnects are made. Note that it makes no sense to treat this interconnect as being directional. It's considered bad form, however, to allow current to flow through a shield, so the following modification is made.

The shield is not only grounded at one end, so no current can flow through the shield. The new problem is that the unterminated shield acts like an antenna; not good. The workaround is to add a tiny-valued capacitor.

No DC current can flow through the shield and only a trickle of ultra-high-frequency signal can jump across such a small capacitance, as a 10pF capacitor exhibits an impedance of 800k at 20kHz. But this is enough to stifle the antenna-like properties of the shield.

Another addition we can make is to go shotgun (i.e. double-barrel shotgun), the use of two shielded cables with any number of internal wires. The bulk Teflon wire I got held a two-wire twisted lead inside the shield. I treated the two wires as a single wire.

If nothing else, we have halved the DCR of the wire, as both the hot and ground got two twisted wires, rather than one wire.

Moreover, the dirty little secret of stereo practice is that an audio signal, in just the right channel, will provoke equal return current flow through both the right and left channel interconnects with a stereo power amplifier, which using a dual-mono line-stage amplifier will not change. Why not? The current flow does not know that it must belong to only the right channel and allow current flow only I the right channel's interconnect, as it simply takes the path of least resistance; in this case, the two ground paths offered by both interconnects present equal resistance, so the current flow down both. By the way, the RCA plug and jack was a huge blunder; instead we should have used something like a three-pin XLR connector that would hold right and left pins and one ground pin. See my Post 434 for more details.

Well, with the shotgun arrangement, we effectively have isolated the right channel's return currents from the left channel's hot cable (and have created duplicate ground-return paths). I believe this explains why the double-shotgun often sounds better than the single barrel cable.

My first attempt at making a fancier double-shotgun was the following:

The hot cable uses a driven shield, i.e. a shield that sees the same signal as the hot wire inside. How did it sound? Different certainly, but not to my taste, as the high-frequencies seemed a tad too forward, but your mileage may vary. If we add antenna-nulling capacitors, the interconnect will look like this:

My second attempt made two changes—both applying to where the shields attach.

Okay, this new interconnect arrangement won my heart and ears, as it delivered the biggest stereo image. So, are we done? No. I decided to resurrect an interconnect design I had created in the mid-1980s, the Broskie Faux-Balanced interconnect.

Dang. It beat my last variation on the double-shotgun. I value stereo imaging greatly, and this 40-year-old design delivers it. How does it work? Here is what I said in post 240:

Stop and think about the voltage relationships within an XLR cable: two anti-phase signal wires, one ground wire, and shield, which attaches one chassis to another. In theory, no current flows through either the ground wire or the shield, as the two anti-phase signal wires complete a circuit within themselves. If the positive signal lead sees an instantaneous voltage of +1V, the negative signal lead must see an instantaneous voltage of -1V. If it doesn't, then it is not a balanced signal.

Ideally, there would be no voltage differential between either the two grounds or the two chassis, so there could be no current flow between the grounds or the chassis. This is not what happens with RCA-terminated unbalanced interconnects, as the shield conducts current, even if a separate ground wire is included.

Note how the shield must carry current and how it connects the signal ground to the chassis ground. Also note how in the balanced XLR cable, the shield is neutral to both signal wires; or if you prefer, equally drags down both signal conductors via its capacitance to these wires. I wanted to make an RCA-terminated unbalanced interconnect that behaved more like a balanced XLR interconnect. Here was my solution: the shield no longer attached to both chassis; instead, it connected to a two-resistor voltage divider network that split the AC signal on the positive conductor and delivered it to the shield.

Thus was born my Broskie cable, which sounded quite fine. I used the Mogami cable and I was able to squeeze the two 1/8W resistors in the RCA plug's housing. A later development was the addition of two small-valued capacitors at the output end of the cable, which helped to prevent the shield from acting as an antenna.

I made this cable for friends and the most common complaint was that it cost too little (about $15 for a meter length) and everyone knew that the more you paid, the better it sounded. I also shared this design with a few cable manufacturers and it has been sold commercially, with neither profit nor credit to me.

Well, our old friend Thomas Sterns Eliot put it best:

We shall not cease from exploration
And the end of all our exploring
Will be to arrive where we started
And know the place for the first time.

"Little Gidding," Four Quartets

Here's a crash course in interconnect designs:

blog0240.htm,
blog0241.htm,
blog0423.htm,
blog0434.htm,
blog0459.htm

Speaker cables:

blog0242.htm

Music Recommendation: Audiophile Hi-Res System Test
As a subscriber to the Presto Music streaming service, you get access to high-res (or higher-res) albums not available at Amazon Music, Qobuz, or Tidal. Presto Music offers many 24-bit, 192kHz albums not available at other music services. (Sadly, none of the music streaming services offers decent database searching and filtering options. For example, you cannot limit the vast catalog to only 24-bit, 192kHz albums.) One 24-bit, 192kHz find was Audiophile Hi-Res System Test. This is a test album on steroids. What I appreciated was the fine-resolution spacing of frequency tones, starting 10Hz and ending at 96kHz, which even professional audio reviewers have a hard time hearing. Here are the tracks:

01 Presentation

02 Jazz Group Live

03 Twelve Guitars, One Double Bass

04 Percussion 1

05 Voice And Jazz Group

06 Harp

07 Jazz Bass & Piano

08 Piano

09 Bassoon

10 Violin Concerto

11 Baroque Music

12 Children Choir

13 Organ

14 Piano

15 Harpsichord

16 Percussion 2 Djembe Drum

17 Percussion #3 (To Be Played At Medium Level)

18 Xylophone

19 Glockenspiel

20 Piano, Vibraphone

21 Percussion 4

22 Timpani

23 Test Presentation Channel Identification

24 Left Channel

25 Right Channel

26 Phasing Test, In Phase

27 Phasing Test, Out Of Phase

28 Presentation, Sound Perspective

29 Bells, Center At 6'

30 Bells, Left At At 6'

31 Bells Right At 6'

32 Bells Left At 18'

33 Bells Center At 18'

34 Bells Right At 18'

35 Tone Presentation

36 Tone 10Hz

37 Tone 20Hz

38 Tone 25Hz

39 Tone 31.5Hz

40 Tone 40Hz

41 Tone 50Hz

42 Tone 63Hz

43 Tone 80Hz

44 Tone 100Hz

45 Tone 125Hz

46 Tone 160Hz

47 Tone 200Hz

48 Tone 250Hz

49 Tone 315Hz

50 Tone 400Hz

51 Tone 500Hz

52 Tone 630Hz

53 Tone 800Hz

54 Tone 1000Hz

55 Tone 1250Hz

56 Tone 1600Hz

57 Tone 2000Hz

58 Tone 2500Hz

59 Tone 3150Hz

60 Tone 4000Hz

61 Tone 5000Hz

62 Tone 6300Hz

63 Tone 8000Hz

64 Tone 10000Hz

65 Tone 12500Hz

66 Tone 16000Hz

67 Tone 20000Hz

68 Tone 25000Hz

69 Tone 32000Hz

70 Tone 40000Hz

71 Tone 50000Hz

72 Tone 64000Hz

73 Tone 80000Hz

74 Tone 96000Hz

75 Burn In Track Presentation

76 Burn In Track

//JRB

AI Summary
Adobe's AI's report on this post, which interestingly enough ignores my section on Google's Gemini AI (jealously perhaps?) :

The paper discusses the design and performance of a simple yet effective two-triode line-stage amplifier, focusing on its power supply rejection ratio (PSRR) and various circuit configurations.

Single-Tube, Two-Triode Amplifier Designs

The text discusses the simplicity and effectiveness of single-tube, two-triode amplifier designs compared to complex audio ICs. ​

• The author contrasts simple designs with complex audio ICs, highlighting the satisfaction of simplicity. ​

• A simple two-triode design is presented, emphasizing its ability to deliver significant performance with minimal components.

• The design includes a constant-current source, which can be replaced with an LM317 or a large-valued cathode resistor. ​

• The post includes various line-stage amplifier circuits, including a 12dB line-stage amplifier with Aikido Mojo, showcasing its features and performance metrics. ​

Aikido Mojo Techniques in Amplifiers

The text elaborates on the Aikido Mojo techniques used to enhance amplifier performance, particularly in power-supply rejection. ​

• The Aikido Mojo technique aims to improve PSRR (Power Supply Rejection Ratio) and harmonic enrichment in amplifiers. ​

• The 12dB line-stage amplifier with Aikido Mojo achieves a PSRR of -53dB at 100Hz, which can be improved further with additional components. ​

• The use of a cynosure resistor significantly enhances PSRR, achieving -78dB, translating to a 1,000-fold improvement.

• Various configurations and modifications are discussed, including the use of different triodes and resistor values to optimize performance.

Interconnect Design Innovations

The text explores the author's journey in designing interconnects, focusing on various configurations and their sonic implications.

• The author revisits interconnect design after a long hiatus, experimenting with silver-plated copper stranded twisted-pair wire.

• Different configurations are tested, including twisted vs. untwisted wires and the use of shields.

• The "Broskie Faux-Balanced" interconnect design is highlighted for its ability to mimic balanced XLR interconnects, improving stereo imaging. ​

• The author emphasizes the importance of design choices in interconnects, noting that even simple wire arrangements can yield significant sonic differences.

Music Recommendation for Audiophiles

The text recommends a high-resolution music album that serves as a test for audio systems, highlighting its unique features. ​

• The album "Audiophile Hi-Res System Test" is noted for its extensive range of frequency tones, from 10Hz to 96kHz. ​

• It includes various tracks designed to test different aspects of audio systems, such as channel identification and phasing tests.

• The album is available on the Presto Music streaming service, which offers high-resolution albums not found on other platforms. ​

Did you enjoy my post? Do you want to see me make it to post 1,000? If so, think about supporting me at Patreon.

User Guides for GlassWare Software
Just click on any of the above images to download a PDF of the user guides.