Liked my S.E.X. without C4S Load

[sammyk]

Just built my S.E.X. 3.0 recently and I'm in love with it. I've entered the next stage of sonic nirvana, so just wanted to extend a personal thank you to everyone at Bottlehead who makes these kits possible.

Now it turns out I fit into the category of "You might even be one of the people who prefers the stock unit."

After installing the C4S, right off the bat the differences were obvious. Low end was definitely improved, and possibly greater level of imaging, at least individual instruments "stuck out" more. The trade-off however seemed to be a bit less liveliness, and to my ears, some additional harshness. I also could not tell a huge difference on noise floor.

I just want to understand why I do not prefer the "upgrade" (i.e. what's wrong with me). If the C4S load causes the triode to operate more linearly with less distortion, am I losing a substantial amount of the pleasant even order harmonic distortion that makes up the SET sound in the first place? Is that pretty much the primary explanation here?

[Paul Birkeland]

The driver stage in the SEX amp does not generate a whole lot of distortion, as the plate loading resistor used is already a really high value.  My suggestion is that you use the C4S upgrade for a few weeks, then put the 150K resistors back in place (you can just disconnect the white/red wires going from the T strips to the PC board at the terminal strip end, then coil them up and move them out of the way). 

When you go back to the plate load resistors, you'll gain more insight into what you're hearing. 

[sammyk]

Will report back in a month. Perhaps some more break-in time is needed for the film caps.

[Tom-s]

And? How did the switch back go? 

[sammyk]

Disclaimer before you read the rest of my post:

Inevitably some people are going to read this to aid their decision on whether or not to continue with the C4S load themselves based upon what I say, so I first want to preface that my (and others) experiences are largely subjective, dependent not just on the S.E.X. amp itself but the entire audio setup. If you think about the multitude of variables involved in the process from: AC power quality, power supplies, source material, DAC, AMP, shape of your ears, psychological influences, wires that are used, capacitors in the circuits, speaker drivers, crossovers, etc, etc.  I believe every little bit can influence in some regard so Y.M.M.V... Please A/B for yourself to know what you like best and what works best for your system. —sammyk

I've been doing some heavy daily listening since my last post. Previous complaints with C4S:

• Increased Harshness
• Perceptible Noise floor
• Lack of Musicality/Liveliness

1. Increased Harshness: Straight out of the gate, there was an overall harshness to the sound that was unpleasant to me. I theorized perhaps it may be that the C4S was making the system more transparent and I could hear the relatively fresh film caps better (I am firm believer in film cap break in, especially in boutique output and crossover caps). This did resolve to some extent, maybe in the range of ~20-30% better after 15 days of listening. Perhaps this would have gotten better after more time, but something else did make a surprising and significant difference here. I had two 4 pole Mundorf MLytic HV+ 470uF 550VDC power supply capacitors leftover from another tube amp I was taking apart, and I used them to replace the two 100uF 450V power supply caps in the S.E.X. while rewiring to make the 4 pole caps location in the circuit in between the diode bridge and chokes as recommended by Jensen Caps http://jensencapacitors.com/CustomerData/Files/Folders/4-pdf/13_4-pole-capacitor-application-note.pdf.

This change made the C4S induced harshness go away pretty much entirely. It also increased overall SQ, lowered noise floor, and notably increased response in the low frequency giving a more holographic sound stage. The theory is that C4S removes the power supply capacitors from the signal, but as I've seen and others in the past have reported, there still seems to be some interactions here that affect SQ.(See Paul's comments on the two stages of the S.E.X.)

2. Perceptible Noise floor: With this one I wasn't sure if the C4S was making a difference here. I can confirm it does, however, it is only noticeable if you are using lower impedance headphones. Without C4S I can barely hear something on the 25ohm headphones, but with 600ohm, dead silent. With the C4S, noise floor is even less perceptible on 25ohm headphones, maybe if you are trying really hard. Likewise with the C4S, 600ohm headphones are dead silent. And with the power supply upgrade that I previously mentioned, the difference became even smaller. (See Doc's comments on how noise floor reduction is accomplished and why the filter cap upgrade works)

3. Lack of Musicality/Liveliness: This one is really what made me decide to question whether or not I should stick with the C4S. I love tube amps (esp. single ended triodes) because of their unique musicality. Even with upgrades as mentioned, and additional time for break-in, the musicality never came back. To make a simplistic analogy the C4S made an impact that made the amp sound somewhat more "solid-state" in that the sound was tighter, colder, accurate, the bass hits harder, but it is not particularly "interesting, or warmly inviting." Not relaxed. Serious. A bit sterile. According to Morgan Jones in his Valve Amplifiers book, Cascodes behave more similarly to Pentodes than Triodes, without any of the negative features of Pentodes. I assume that because the Cascode should be so low in distortion this means the high odd order distortion of the the Pentode's characteristics shouldn't matter. I also assume that means if there was any pleasant even order distortion from a triode configuration, this would not exist with a Cascode. Ultimately, I do not know if the theory applies in this situation. (See both PB/Doc's comments). This week I tried some USSR film caps for the output and interstage, but it didn't get me back to the non-C4S sound that I knew was there without the C4S.

Conclusion: I think the C4S has a lot of benefits in: decreasing noise floor, increasing frequency response, deepening parts of the sound stage, etc. However, it also changes the stock S.E.X. sound that I really liked, enough for me to say I'm going to run no-C4S for the time being. Perhaps in the future when some major component changes come in (say new DAC or speakers). Or I'll probably just build another S.E.X. and have that be perma-C4S. In any case, compared to other amps I've listened to, the S.E.X. in both configurations blows most of them out of the water.

[Paul Birkeland]

This change made the C4S induced harshness go away pretty much entirely. It also increased overall SQ, lowered noise floor, and notably increased response in the low frequency giving a more holographic sound stage. The theory is that C4S removes the power supply capacitors from the signal, but as I've seen and others in the past have reported, there still seems to be some interactions here that affect SQ.

The SEX is a 2 stage amp, the C4S works on the first stage, and the 100uF power supply caps are the power supply node for the second stage.

According to Morgan Jones in his Valve Amplifiers book, Cascodes behave more similarly to Pentodes than Triodes, without any of the negative features of Pentodes. I assume that because the Cascode should be so low in distortion this means the high odd order distortion of the the Pentode's characteristics shouldn't matter. I also assume that means if there was any pleasant even order distortion from a triode configuration, this would not exist with a Cascode. Ultimately, I do not know if the theory applies in this situation.

It doesn't.  Think of the C4S as being able to use a 10 mega ohm plate load resistor without the associated 15,000V power supply that you'd need in the SEX amp.

[Doc B.]

The cascode in this circuit is in the constant current source loading the triode. The Morgan Jones reference is to a cascode signal amplifier. Not the same thing.

I think you have a good description of the sonic differences, but the thing that is happening in terms of electronics theory is the stock resistor loaded input/driver circuit has more dominant second order distortion - often referred to as a more musical sound - at a given output level than that circuit does with the C4S installed, up to some higher signal (i.e. volume) level point where the C4S loaded triode leaves its more linear behavior and goes into distortion, seemingly harder. But that is only because it the transition from linear behavior into distortion is quicker at that higher signal level. The stock tube setup goes into its more "graceful" gradual distortion sooner ( i.e., at a lower signal level). Thus the C4S loaded tube is actually more faithful to the original signal than the stock tube up to that point where distortion comes in for the C4S tube. This transition point is typically at a pretty high listening level. This is not really the same as the effect of high negative global feedback in an SS amp, nor higher odd order distortion in pentode circuits.

I also don't understand the sentence

With the C4S, impedance is even less perceptible on 25ohm headphones

I'm guessing you mean noise floor instead of impedance. What is happening is that the gain of the input circuit increases a few dB with the C4S, thus the noise floor should be amplified a bit relative to the stock circuit, as is the music signal. The lower the headphone impedance usually means higher sensitivity and thus the noise floor will often be more perceptible with low impedance cans. But at the same time the power supply noise is rejected better by the C4S and thus goes down even though the gain improves.

And the filter cap changes could be heard regardless of the C4S addition, because only the first stage is loaded with the C4S. The output stage is not under the influence of the C4S.

This is only meant as constructive criticism. I think it is worthwhile to understand the circuit a little better before trying to apply theory to what I think are very good listening observations.

[Deke609]

This is all very fascinating stuff.

Sammyk - thanks for posting your detailed observations. I don't have a S.E.X., but I like many others have had similar experiences with what I'll call the "cleaner but leaner" trade-off. 

... [T]he thing that is happening in terms of electronics theory is the stock resistor loaded input/driver circuit has more dominant second order distortion - often referred to as a more musical sound - at a given output level than that circuit does with the C4S installed, up to some higher signal (i.e. volume) level point where the C4S loaded triode leaves its more linear behavior and goes into distortion, seemingly harder. But that is only because it the transition from linear behavior into distortion is quicker at that higher signal level. The stock tube setup goes into its more "graceful" gradual distortion sooner ( i.e., at a lower signal level). Thus the C4S loaded tube is actually more faithful to the original signal than the stock tube up to that point where distortion comes in for the C4S tube. ...

@ Doc B - this is really interesting info. As I'm sure many others on this forum would too, I'd love to learn more from you, PJ and PB about how various design choices impact the sound signature.  But writing takes forever, and there is only some much time in a day, week, month, year.  Any interest in offering online seminars on selected topics? I'd be willing to shell out some bucks to learn from a source that I trust.

cheers,

Derek

[Doc B.]

I get asked this by more advanced builders every once in a while. It is a cool idea, but in practice what I have found is every student wants a different lesson or wants the info at a different level of technical complexity. We have given amp design/build classes for up to 15 students in the past, and at the end there are always a few who say it was not nearly technical enough for them (though I would say it probably was, considering their practical experience level) and a few who say they were lost three minutes into every lecture. Unfortunately I don't claim to be that guy who can start at the most basic level and finish the lecture at the most technical level for those who want to stay with it.

If someone was to approach me with the idea that they would determine the specific - let me say that again - specific content to provide, and shoot and edit the videos I would be willing to discuss putting something together. As it is I'm so spread out with projects these days that I would never get something like that finished myself.

"How to design a tube amp" is what I usually hear for the topic. Cripes, that could take ten years to teach, not five 90 minute lectures. I'm still trying to figure it out myself after 25 years at it. But if the content was limited to, say, how do active loads work, or what is parallel feed and how does one determine component values, or what are the considerations in chassis layout, or how to design the most basic SE amp possible, that could be a manageable topic.

[Deke609]

Great! So there's potential willingness at BH. Sometime in the next week I'll start a new thread to see if there is enough serious interest to justify trying to get this off the ground.

[sammyk]

It doesn't.  Think of the C4S as being able to use a 10 mega ohm plate load resistor without the associated 15,000V power supply that you'd need in the SEX amp.

The cascode in this circuit is in the constant current source loading the triode. The Morgan Jones reference is to a cascode signal amplifier. Not the same thing.

Thank you PB. & Doc on the corrections to my layman electrical theory assumptions - I totally do not have a background in electrical engineering or tube amp engineering, just a hobbyist level of interest so I'm thankful for the proper information. And I think it will be helpful for those who may be curious about it in the future. I will edit the main post to point out the areas that I spoke about that are misleading.

The SEX is a 2 stage amp, the C4S works on the first stage, and the 100uF power supply caps are the power supply node for the second stage.

This clarification makes sense to me now. My motivation to replace the last cap was discussion in this older thread https://bottlehead.com/smf/index.php?topic=3929.0 on the potential effect of the last power supply filtering cap upgrade and for the benefit of the four pole topology. The experience that it somehow removed harshness associated with C4S upgrade was unanticipated. I was confusing the driver stage and power stage when talking about the impact of power supply caps on C4S, I was considering what PJ said regarding the possibility of hearing a difference https://bottlehead.com/smf/index.php?topic=303.msg1649#msg1649

The cascode in this circuit is in the constant current source loading the triode. The Morgan Jones reference is to a cascode signal amplifier. Not the same thing.

I must have conflated some things in my late night reasoning. Under Morgan's section on "Semiconductor constant current sinks" he briefly mentions that "A transistor cascode is broadly similar to a pentode..." so maybe that's where I got off on.

I'm guessing you mean noise floor instead of impedance.

Yes, meant noise floor not impedance, sorry about the typo (late night/early morning writing).

...electronics theory is the stock resistor loaded input/driver circuit has more dominant second order distortion - often referred to as a more musical sound - at a given output level than that circuit does with the C4S installed, up to some higher signal (i.e. volume) level point where the C4S loaded triode leaves its more linear behavior and goes into distortion, seemingly harder. But that is only because it the transition from linear behavior into distortion is quicker at that higher signal level. The stock tube setup goes into its more "graceful" gradual distortion sooner ( i.e., at a lower signal level). Thus the C4S loaded tube is actually more faithful to the original signal than the stock tube up to that point where distortion comes in for the C4S tube. This transition point is typically at a pretty high listening level.

This totally clears it up for me.

This is only meant as constructive criticism. I think it is worthwhile to understand the circuit a little better before trying to apply theory to what I think are very good listening observations.

It's very helpful. Thank you again for the clarifications. I also second Derek's ideas for doing some sort of digital classes or teachings, even perhaps a YouTube series similar to what Paul McGowan does with PS Audio, but with a focus on tube audio. It would communicate Bottlehead's brand with larger audiences and develop a greater interest in the products offered (business/marketing major).

[Doc B.]

I must have conflated some things in my late night reasoning. Under Morgan's section on "Semiconductor constant current sinks" he briefly mentions that "A transistor cascode is broadly similar to a pentode..." so maybe that's where I got off on.

They do both make decent constant current sources. The advantage of the Cascode transistor circuit is very high impedance and PSRR. And one less heater to power.

[Natural Sound]

I think you have a good description of the sonic differences, but the thing that is happening in terms of electronics theory is the stock resistor loaded input/driver circuit has more dominant second order distortion - often referred to as a more musical sound - at a given output level than that circuit does with the C4S installed, up to some higher signal (i.e. volume) level point where the C4S loaded triode leaves its more linear behavior and goes into distortion, seemingly harder. But that is only because it the transition from linear behavior into distortion is quicker at that higher signal level. The stock tube setup goes into its more "graceful" gradual distortion sooner ( i.e., at a lower signal level). Thus the C4S loaded tube is actually more faithful to the original signal than the stock tube up to that point where distortion comes in for the C4S tube. This transition point is typically at a pretty high listening level.

Fascinating.  I have experienced this "transition" you speak of. This post explains in theory what I'm hearing in reality. Over the years my ears have told me where the sweet spot resides and I adjust the volume accordingly. Once you know what you are listening for it comes naturally. Where things become complicated is with highly compressed (loud) digital recordings that have reduced dynamic range. Then it becomes difficult to determine where the distortions of the tube (C4s or not) come into play vs the the saturation of the recorded source. This is also where the other components in the audio chain and source materials have a big effect.

FWIW I've gone back and forth with plate resistor and C4S loading listening sessions in the past. Each have their pluses and minuses. That said, in all instances except one I return to Bottlehead C4s constant current loading every time. That exception is in my Crack amp. But my Crack amp is configured for a 6SN7 in lieu of the stock 12AU7 voltage amplifier. And I'm only using 1/2 of the speedball, The 6080 tube so its not a straight apples to apples comparison over a stock unit. 

Thanks for this comprehensive explanation, Doc B. I found it very enlightening.