[Maxx134]

We can discuss it first because we might not agree on everything though lol !

Naaa lol I am very open to new stuff.
Remember in this thread when we had alternative approaches which led to @SonicTrance making the new design amps?
Well I also found out that the WCF design compromises were also resolved by member @coinmaster in a design collaboration called "leamski follower".
So see this thread led to alot of diverging directions.

 

[coinmaster]

The leamskie follower does solve the design flaws of the WCF but it's not a mod that is easy to do or that I would recommend someone try to do. It would be easier to design it from scratch at that point.
I have 8 6c33c tubes and custom wooden chassis for balanced leamskie follower that I haven't really bother to assemble yet because it would essentially be a novelty project and I've designed amps that are vastly better than the MK6 so I wouldn't really use it, and I'm working on new headphone tech right now among other things.

 

[baronbeehive]

The leamskie follower does solve the design flaws of the WCF but it's not a mod that is easy to do or that I would recommend someone try to do. It would be easier to design it from scratch at that point.
I have 8 6c33c tubes and custom wooden chassis for balanced leamskie follower that I haven't really bother to assemble yet because it would essentially be a novelty project and I've designed amps that are vastly better than the MK6 so I wouldn't really use it, and I'm working on new headphone tech right now among other things.

Yeah, absolutely. No point flogging a dead horse! I have to say that the amp atm is sounding very sweet, not saying that there couldn't be better amps though.. or better headphones come to that.

Edit: Those wood blocks might be a possible design issue lol . Saying that I don't mean single to you out... those HiFiMan headphones for example look pretty bonkers!

 

[Maxx134]

I'm working on new headphone tech right now among other things

Not to sidetrack, but I was fortunate to hear his prototypes, and it is in another league from anything I have heard, so hopefully the future looks ripe with possibilities.

Also, I will be posting updates to first page in coming days, with @baronbeehive help.

 

[Maxx134]

I am currently researching into the designer's choice of "non-polar" on cathode section, for instance any "esoteric" clues/reasons for his choice.

 

[baronbeehive]

I am currently researching into the designer's choice of "non-polar" on cathode section, for instance any "esoteric" clues/reasons for his choice.

The cathode caps do appear to be more significant than I originally thought, although I know from my other amp how relevant they are to SQ.

 

[CopperFox]

This is something we never experimented with enough, so thank you for leading the way here.
Normally one would not expect differences to happen in this location, but the realization is that we are not dealing with a regular circuit, but a WCF circuit, so this has not been explored much.

This is good to know! I plan to change my cathode caps anyways, so will take this into consideration.

We do know that the both the ESR and speed is benefitted from paralleled caps, but I was curious to know what was there a reason why the designer chose a non-polarized cap in that position, and I never dug deeper as to why.
I am starting to think it was for speed.

So currently, my issue is that I do not have enough capacitance there, and have seen the meter dip on bass hits, when using hungry planars.

One thing I noticed after installing the 2*330uf 100V caps was that the vu meters still fluctuate with some bass hits, but they move upwards instead of down. I don't remember them ever doing that before, not even with the 2200uf. Whether or not this change comes from these caps still needs further confirmation though as I'm sure some of the improvements I'm hearing must come from the new pcb and new components on it. Don't know how much though. I did test the new amp before installing any mods on it but didn't really pay attention to the vu meters' movements then.

I now have 2*220uf 100V Nichicon KZ's installed as cathode bypass (the resistors I ordered for the impedance mod have been stuck in customs for a week). I'll refrain from commenting on how they sound until they've been there for about two weeks.

 

[CopperFox]

Sonic has PM'd me with the Cavalli paper which has some quite complicated maths in it to show the WCF function and calculating the load for headphones. The Ka and Kr are assumed constants so they stay the same. As far as I can gather the calculations are not an exact science.

IF you would like a copy of the paper and can understand the maths lol, let me know.

I found this which seems to be a compilation of articles about the WCF that includes the Cavalli paper?

 

[baronbeehive]

I found this which seems to be a compilation of articles about the WCF that includes the Cavalli paper?

Thanks for that! This might explain why, at least in the WCF circuit, there is not much difference in sound, at least to my ears, regardless of what the Ra is in the Cavalli equation to calculate headphone load. It seems there are difficulties in this calculation before Cavalli finally arrived at his value, and possibly even after that. Or maybe there actually isn't much difference in sound as I originally thought. I have a speaker amp with adjustable outputs, 4, 6 and 8 ohms and I found the same there, not much difference if any. So I will be interested in your findings, especially as you have a very low impedance headphone which the paper you found mentions the difficulties with such difficult to drive headphones in the WCF, and the small current available within the WCF.

 

[Maxx134]

I found this which seems to be a compilation of articles about the WCF that includes the Cavalli paper?

This is excellent for understanding the WCF design. Thanks! I will look up what info I have on it as well.

The impedance mod, is basically what is already done in the MK8, to be set for higher impedance Headphones.

 

[Maxx134]

For instance, on the HD800, it actually made the headphones more open in the midrange. So no modding was needed for the HD800 to sound great.
It was optimal for these higher impedance Headphones.

Yet I changed my MK8 to be set for a 50ohm load instead of the 300ohm of the Sennheiser headphones.
This made the HD800 revert back to it's "brighter" nature, with less open mids.

So the differences heard were from extreme differences in the impedance mod change.

From my limited testing on planars, I do have notes in which I still observed some tunning differences on the output tubes. It made a slight difference for me so I will find them.

 

[CopperFox]

Ok I have now tried 2*220uf 100V and 330uf 100V + 220uf 100V in addition to the 2*330uf 100v (and single 470uf, 1000uf, 2200uf and 3300uf).

The 2*220uf had much in common with the single 470uf in that it had great dynamics and also similarly suffered from excess attack / edge on certain kinds of passages. It did have more detail than the single 470uf though and it also sounded brighter in comparison.

However, the best overall is the 330uf 100V + 220uf 100V (550uf in total). This combination has great dynamics and detail and there is neither too much edge/attack or smoothness. In my use of low gain and very low impedance planars it is just about right in many respects.

This combination also seems to have the overall widest range of usable different power tubes. I'm now using a combination of Tung-Sol 6080WB and Shuguang 6N5PJ... and those Tung-Sols hadn't really sounded right to me before, and the chinese power tubes didn't really combine well with western tubes before.

If I had to rank the cathode bypass capacitors/capacitances it would be like this (first five Nichicon KZs, 6-7 are Nichicon KW) :
(Note that I don't even have high impedance headphones so I don't know how they would work with those. I also use the low gain setting.)


1. 330uf 100V + 220uf 100V (does most things (everything?) right, overall the best sound I have heard)
2. Single 470uf 50V (good dynamics and crunchy mids <--- those sound fun to me. Also excess attack and relatively low detail/resolution.)
3. 2*330uf 100V (good but overly smooth, not best dynamics)
4. 2*220uf 100V (good dynamics but excess attack and obnoxious slight brightness)
5. Single 1000uf 50V (sound is... ok)
6. Single 2200uf 100V (very flat, not usable, long burn in)
7. Single 3300uf 100V (fails to burn in)

(EDIT: 1000uf+ needs high gain to burn in and function properly. Many prefer that but I prefer low gain so that does affect my ranking. My impression is that high gain needs higher capacitance than low gain. High gain with the 470uf had the worst excess attack I've heard.)

At some point with the single 470uf I was considering to stop experimenting as it was a sound that I liked but I did also want to find something to fix some of its flaws. Although it took a long time I'm now happy that I did.

 

[baronbeehive]

If I had to rank the cathode bypass capacitors/capacitances it would be like this (first five Nichicon KZs, 6-7 are Nichicon KW) :
(Note that I don't even have high impedance headphones so I don't know how they would work with those.)


1. 330uf 100V + 220uf 100V (does most things (everything?) right, overall the best sound I have ever heard)
2. Single 470uf 50V (good dynamics and crunchy mids <--- those sound fun to me. Also excess attack and relatively low detail/resolution.)
3. 2*330uf 100V (good but overly smooth, not best dynamics)
4. 2*220uf 100V (good dynamics but excess attack and obnoxious slight brightness)
5. Single 1000uf 50V (sound is... ok)
6. Single 2200uf 100V (very flat, not usable, long burn in)
7. Single 3300uf 100V (fails to burn in)

At some point with the single 470uf I was considering to stop experimenting as it was a sound that I liked but I did also want to find something to fix some of its flaws. Although it took a long time I'm now happy that I did.

Thanks to your work on this we are getting a comprehensive picture on the amp's function. I will discuss this with Maxx when we update page 1.

We have ideal values for Coupling, WCF and now cathode bypass caps. It looks like you have found the sweet spot. I would suggest though that your finding of 330uF + 220uF would be replicated if there was a 570uF value caps, which there isn't. So it looks like we are looking at a value of 570uF achieved by bypassing the 2 caps you mention.

Your results concur with much of our findings and look pretty consistent. The larger values being rather dull sounding while the smaller values being brighter, to generalize.

The reason for all this is that the cathode resistor and the cathode bypass cap form a high pass filter, increasing the capacitance lowers the cutoff frequency which increases gain at low frequencies. A cathode bypass cap is used to maximize gain which would be lower due to the negative feedback loop from the cathode resistor. Negative feedback decreases gain and promotes linearity. Because the cathode bypass cap eliminates the negative feedback loop, the amp will have more gain with the cap than without it. The capacitance of the bypass cap determines which frequencies are boosted. If a smaller capacitor is used, only higher frequencies will be bypassed. A large arbitrary value cathode cap would ensure proper cathode decoupling well below audible frequencies so that there would be no hiss, or other noise as this would be cutoff by the cutoff frequency. The higher the capacitance would increase the bass response by lowering the cutoff frequency of the filter and therefore increases gain at low frequencies. That may or may not make sense!

Edit: another problem with large value electrolytics is the increase of ESR that can be helped by using smaller value bypasses to the cathode bypass which we have done.

 

[baronbeehive]

This combination also seems to have the overall widest range of usable different power tubes. I'm now using a combination of Tung-Sol 6080WB and clear top Shuguang 6N5PJ... and those Tung-Sols hadn't really sounded right to me before, and the chinese power tubes didn't really combine well with western tubes before.

I never did manage to get hold of those Tungsol power tubes because they were very difficult to get hold of now apparently!

So did you manage to evaluate the impedance mod yet? I was interested to see if you could hear any difference.

 

[Maxx134]

Because the cathode bypass cap eliminates the negative feedback loop, the amp will have more gain with the cap than without it. The capacitance of the bypass cap determines which frequencies are boosted

This is why some tube circuit do not have any cathode caps, in order to not have any added gain, so as to be more linear with less THD.
So in this position (cathode cap) we can effect and produce the tube harmonics and "thickness".
There are manufacturers that intentionally use this area to increase their tube harmonics, by choosing their choice of cap. Not talking about size/value, but cap type. They would purposely put in a low or non-audio grade cap in that position.

At some point with the single 470uf I was considering to stop experimenting as it was a sound that I liked but I did also want to find something to fix some of its flaws. Although it took a long time I'm now happy that I did.

I am glad you did as well. This gives us great insight into the variables to making a preference choice.
I will be using your posts for updating first page soon

This adds complexity on what choice I will make. I tend to think you found the optimal value/values, and proportion, so now we can focus on cap brand sound trait effects on the tubes..

I have already researched a few different electrolytic cap brands and will post some general consensus of thier character type soon.