[00940]

Disclaimer: someone asked me how difficult it would be to build an amplifier similar to the Zana 2. My first reaction was that it is just plain crazy to use 6C33C for headphones. Still, the concept has been bugging me and I had to try and see how this thing could even work in theory. This will be all sims, I really have no interest in pouring hundreds of € into something with high chances of failure.
 
Still, playing with theoretical amps is definitely part of the hobby for me
 
What I gathered from reviews is that the 6C33C in the Zana are run at 8W, with a 90r plate load resistor. The amp is dc-coupled in-between the stages and has no input cap. Driver is a 6sl7. Now, there is a lot in that which doesn't make sense to me.
 
- 8W ? No matter how I run the sims, lowish B+/high current or the other way around, 8W is too low to get the best performances of the 6C33C. It seems ok at 17W though, still a long, long way from its max power.
- dc-coupled ? Why ? The amp has by definition a massive output cap. Would a small interstage cap of the highest quality make such a big difference on top of that one ? And if the amp is dc-coupled, it means that the 6C33C's cathode has to be way above ground (to give breathing room to the input stage). Which means higher B+ than necessary and plenty of heat dissipated in the cathode resistor.
- 6SL7 driver ? Ugh, everyone on the web seems to be pushing for a decently low impedance driver for the 6C33C, to get an adequate hf response. So that means the driver has to get some help. µ-follower here we are.
 
Starting from there, I had my parameters :
 
- input stage must be a high-µ tube, to get decent gain to apply feedback and linearize the not very linear 6C33C.
- if the input stage is also the driver, we can use a solid state ccs to also get a low impedance output. Most high-µ tubes won't like to be run with too much current, so let's divert part of the current through another ccs, from tube anode to gnd.
- interstage ac-coupling. We can now use supplies adapted to each stage.
- 6C33C in fixed-bias. No need to worry anymore about cathode resistor decoupling nor the dissipation of the cathode resistor. A small negative supply is a small price to pay for that.
 
It all leads naturally to the following schematic. Sims are pretty good, even at 32R (sims promise 0.2% thd at 1Vrms into 32r). Distortion spectrum is nicely decaying. I really have a hard time to understand why someone would want to build such a monster though.
 

 
As it is an intellectual exercise, I'd be very interested in reading your comments and suggestions on this.
 
Btw, I found a prior attempt at this game: http://www.head-fi.org/t/340309/6c33c-b-se-otl 

 

[blubliss]

Here is something my friend built recently, needs to be cased and PS will be separate. Ignore the fans, the heatsinks used are scrap and not up to the task. I will get my friend to post the schematic.
The amp sounds really good with the HD800, find myself listening to this amp a lot recently even though I have the DiyT2 and new dynahi on hand.
It is incredibly musical and engaging with the right tubes
We may bring it to the upcoming meet in southern CA.

 

[stv1756]

.

 

[00940]

Cool thing.

  I can see how dissipation would require decent heatsinks.
 
Did you try it also in single ended ?
 
Pretty much the opposite of the zana in philosophy though: cathode follower output but no gnfb.

 

[FrankCooter]

I like to play with "off-the-wall"  theoretical designs as well. And occasionally I'll even build one. The first headphone amp I ever built was a single GM70 driving a 5K:300 transformer. Absurd, but not bad at all with Sennheisers.
 
Despite it's extraordinarily low rp, the 6C33, like most regulator tubes,is not the most linear  tube for traditional  output  duty. Works better as a cathode follower.  Add a ton of global negative feedback and and a very non-linear  12AX7 as the first stage, some sort of solid-state regulated power supply with lots of big electrolytic capacitors, and you're more than halfway to a solid-state amp. But perhaps this is your goal.
 
How about more linear tubes to start with and a consequent reduction of both raw gain and required GNFB?
 
If you really need a lot of front end gain you might want to look at a triode-connected D3A. An interesting substitute for the 6C33 might be a 13E1. Both are European types with excellent reputations. Try to run both stages off a single power supply. Using a plate choke for the D3A ought to put both stages on approximately the same B+ level.
 
If I were to actually build this I'd want to go with a low impedance Mercury rectifier based power supply with chokes and oil caps instead of regulators.
 
Keep up the good work. If nothing else it's a fun mental exercise, and if it turns into something I like, I just might build it.

 

[nikongod]

Any thoughts on Pentodes in pentode mode for the gain tube? 
You can get stupid-high gain out of pentodes (obviously) which buys you a lot of feedback. 
The LOOOOOOW miller capacitance of a pentode is also nice. 
 
I think the original amp was DC coupled to simplify the biasing process. I'm not sure I trust many people to bias a 6c33c with a pot... I can see things getting out of hand QUICK. Even quicker when you consider the realities of someone saying that the amp sounds so much "more better" when idled at 300mA instead of 100mA... In the case of  an amp like this owned by a "skilled owner" I'm still not sure AC coupling is the way to go. The 6c33c has some nasty grid current leakage which may complicate biasing with a 100Kohm grid-leak resistor. A smaller grid-leak resistor may work, but then you need a bigger cap and that is no fun either. 

 

[stv1756]

It was originally conceived as a balanced amp, so no idea about how it would sound single ended. I did try running the B+ at 250v, but had problems with the cathode/filament voltage. Normally one would lift the filament voltage, but we couldn't because we are using a switched filament PS. 

 

[00940]

  Despite it's extraordinarily low rp, the 6C33, like most regulator tubes,is not the most linear  tube for traditional  output  duty. Works better as a cathode follower.  Add a ton of global negative feedback and and a very non-linear  12AX7 as the first stage, some sort of solid-state regulated power supply with lots of big electrolytic capacitors, and you're more than halfway to a solid-state amp. But perhaps this is your goal.

 
The "goal" was just to see if I could come with something similar to the zana in ltspice, to get an understanding of the amp. Really nothing more. The designer seems to make a big deal of the fact he's not using the 6c33c in cathode follower mode, avoiding local feedback.
 
Indeed, the thing is using a whole lot of gnfb. Output impedance isn't that good otherwise. I completely agree with you, it probably would sound quite like a ss amp. On a side note, the 12ax7 isn't that bad as long as the plate load is very high and as long as you give it enough voltage. But, to be true, I used it in the sims since I didn't have a model of the 6sl7 used in the original zana (and didn't feel like searching).
 

 If you really need a lot of front end gain you might want to look at a triode-connected D3A. An interesting substitute for the 6C33 might be a 13E1. Both are European types with excellent reputations. Try to run both stages off a single power supply. Using a plate choke for the D3A ought to put both stages on approximately the same B+ level.

 
I knew about the D3A but not about the 13E1. That's quite a magnificent tube ! Quite difficult to find on the other hand. 

I can only find a pair of webshops selling it.
 
 

 Originally Posted by nikongod
 
Any thoughts on Pentodes in pentode mode for the gain tube? 
You can get stupid-high gain out of pentodes (obviously) which buys you a lot of feedback. 
The LOOOOOOW miller capacitance of a pentode is also nice. 
 
I think the original amp was DC coupled to simplify the biasing process. I'm not sure I trust many people to bias a 6c33c with a pot... ..
 
The 6c33c has some nasty grid current leakage which may complicate biasing with a 100Kohm grid-leak resistor. A smaller grid-leak resistor may work, but then you need a bigger cap and that is no fun either. 

 
To be true, I'm not too familiar with pentode gain stage. I'd have to look into it. Not sure more gain is actually needed though. The 6sl7 or ecc83 are already giving a lot.
 
I hadn't thought of dc-coupling as "fool-proofing". It makes sense.
 
Yes, youre right, 100K is probably high. From what I've gathered, it's the maximum value that could be accepted there.
 
 

 Originally Posted by svt1756  
It was originally conceived as a balanced amp, so no idea about how it would sound single ended

 
Oh, ok. Thanks 
 
 
Just for fun, what I'd consider a "sane" overkill amp, with 4p1l. The kind of thing I could see myself building one day (currently busy with a simpler 6e5p in parafeed).