Sine I informed public about Melquiades amplifiers I receive many emails from people asking me about the operation points of 6C33C. For whatever reasons people mistakably took me for some kind of 6C33C guru, which is am not. Moreover I could be hardly be a consultant to any electronic engendering subjects, probably with an exception of the very narrow fields where I personally happened did something and was able (mostly accidentally) to reach some conclusive results. This post is my sharing with public about my experience regarding the 6C33C’s operation where I today do feel comfortable with my findings and with my result.

I do use 6C33C for years and I experienced it in many amplifiers, I like this tube. However, I begun to actually drive it on my own since the Melquiades project started last year. There were countless experiments done during that time to optimize the 6C33 used and more or less optimum scenario was found. However, with the Super Melquiades (3 channels of Melquiades) the task was to push the envelop of Single Ended’s capacity. The Super Melquiades’ MF and HF use ½ of the 6C33 and the LF channel uses one full 6C33.

My primary attention of this post is to share my finding of how the LF channel of Super Melquiades was resolved as the solution turned out to be not that I was expecting.

Forget the paper’s operation data on the 6C33. The 125V and the current up to .5A perhaps goes a job in a voltage regulator of MIG-29 but it has no relation to sound. Certainly, in my case the 6C33 was used and tested with fixed bias, no other biasing methods considered civilized for this tube (sound-wise).

My initial sentiment was to replicate what Vladimir Lamm did with this tube - 175V/300A. He suggested that he used this tube for his designs sine 1971 and that he was the person who opened this tube to public. Obviously he knows what he does with 6C33, however, he uses 6C33 regulated but I do not, and do not practically like anode voltage regulation. Secondary Lamm is highly conditional in his visions and this restricts his output. So, when I made Melquiades I had in there 200mA transformer and it was no surprise that I had the very best bass result at 180mA/200V. Any further increase of current worsened sound whish obviously was die to the core saturation, or at least it was what I though… the ironic part was that Melquiades with it’s little transformer ad way better bass then 310mA Lamm ML2 drive at full throttle. I attributed it to many Melquiades’s specifics but not to the lower plate current.

Last few days I spent a LOT of time playing with the Super Melquiades. The Super Melquiades has 450mA 1/8 output transformer with inductance of primary of 15H.  It is non-sectional little-monster that specifically made for LF operation and that roles off at 630Hz. It powered by 1A power transformer with multiple secondary 275V, 300V and 330V. After a full wave rectifier the channel has input 10H/500mA chock, followed with 100uF, then 50R resistor and 350/10.000 capacitor, sitting directly next to the primary of the transformer.

After I set it up I spent quite a time to driving my LF line array and to listen the thing under the different operation, voltage and current. Then, armed with scope and function generator I begin to measure the thing trying to find out some correlations between the auditable and measurable. Dima Kereev (“deemon” within this site) pretty much ran the show with the “measurable” and he was the one who interpreted the results of all my measurements. We spent countless amount of time on telephone and now I learned how to measure with a oscilloscope the pH factor of my coffee. :-)

Anyhow, to my big surprise the fashionable believe that the 6C33 loves high voltage turn out to be incorrect. With increase of voltage and dropping current the tube certainly give up more power but bass become contrived and pretentious. The channel kind of demonstrates a willingness to push bass even if there is not need for it. With the specious recording there was bass but there was no Space – the primary evidence that the sub-audible bass was not there.  In ALL scenarios I find that 45-50W on plate is more suitable power dissipation for this tube. So, the measurement-wise at 275V and 170mA the channel pushed at full power down to 23Hz with 5.2V at input,  then with decrease of input voltage to 4.8V the channel was able to care this power down to 14Hz. The rise of current helped a little, however driven by continues 20Hz  at 5V input signal the tube (any tube that I tried) had a tendency to heat the plate.

At 250V the total out was also strong but the behavior was in some way identical with 275V on plate. The exception was that the lowest non-distorted max power was at 21Hz and with reduction of input voltage the channel begin to distort at the same 14Hz. The heating up of plate was way less aggressive and sonically the bass had way less nervousness and anxiety then when it was driver at 275V

I liked the most when I drove the 6C33 plate with 230V. I run all possible currents and… surprise, surprise I ended up at 150mA.  At this current I had in somewhat reduced total output power (34W on plate ~12W at aoutput) but the tubes measured very nicely. The most important it has a phenomenally interesting sound bass- very reach, very low, very articulates but at the same time very… VERY… “loaded”  At full power the channel ran down to 16.5Hz and with a reduction of input voltage from 5.2V to 4.8V the amp holds the perfect shape of wave down to ~7Hz. The anode at 20Hz and 5V of input hold the temperature very nicely.

Before the closing up the amp I thought to dump those finding on my site. I do not know how generic my result might be but I hope some of you would find it useful. At least it might be a collective reply to a dozen of guys who contacted me asking how I used the 6C33C, I never said to them anything defiantly before….

Rgs,
Romy the Cat

Romy the Cat wrote:

I liked the most when I drove the 6C33 plate with 230V. I run all possible currents and… surprise, surprise I ended up at 150mA.

The 6C33C was available with 1,2,3 anodes. The version with one anode known as 6C41C.  It kind of smaller tube with the rest practically identical characteristics with ½ 6C33C. Did anyone ever hear how this tube sounds. Bas Horneman  uses it in his amps but he does not have experience with 6C33C. I have seen before that the sampler packaging tubes did not sounded as good as this larger brothers, even if the where identical tubes but with the reduced glass volumes.

So, did anyone have an OK sounding amp with ½ 6C33C and who tried to replace it with 6C41C? If so then could you share your observations?

Rgs,
Romy the Cat

Romy the Cat wrote:

... did anyone have an OK sounding amp with ½ 6C33C ...

Rgs,
Romy the Cat

Romy the Cat wrote:

could I instead of trashing the tube to flip it to the left filament and continue to use the unused part of the cathode, plate and filament?

Romy the Cat wrote:

Theoretically it might be the case but the entire tube now should be filed up with the residual gas of right triode operation and the getter should be greatly eaten.

Guy,

could you, please, specify the workpoint for the tube and workload you use in your comments.

SY, Genn

Guy, but you measure harmonic distortions and get out of your measurements juts numbers. How those numbers relates to anything else you do not know. The harmonic distortions are not audible but only the mechanism the crates them; however your “numbers” do not describe the mechanic of the harmonic distortions creation. Not to mention that the auditable distortions are not the delta between the outputs and input but much more complicated creation.

It is possible that with higher current 6C33C measures better and even has lower plate impedance but it does not necessary serve any useful purpose to me.  I personally get correct sound with good electricity at 165mA or perhaps 185mA and with higher currents I need to use a transformer with different amount of the secondary turns as Sound become too sharp. Why would I need it?

I really do not know how to interpret the results of your measurements besides to accept then as “just numbers”. Probably if you need to make a good amplifier to satisfy your distortion analyzer then you needs to get a good SS amp from 80s that has .00005% distortions. Sorry, I do not need it. I have much more objective ways to assess the harmonic pattern of amplifiers.

Rgs,
Romy the Cat

guy sergeant wrote:

I've usually found valve rectifiers to sound preferable.

For HF channels only. Still with shaft switch diodes and properly implemented input choke filters there is not advantages of tube rectification, quite opposite, and particular for the LF.

guy sergeant wrote:

I am quite interested to build a pair of these amplifiers to see whether certain conclusions I have drawn about what has been good and bad in amplifiers I have made in the past are valid. I've usually found valve rectifiers to sound preferable. I don't normally like huge reservoir caps. The Melq has lots of features I'd shy away from usually which makes it intriguing.

Well, if you are wiling to build a pair of these amplifiers then you will be able to drive output tube with any current you wish. I still feel that at 320mA you necessary burn the tube. Also do not forget that this is a fixed base and the infamous 6C33C temperature stability. I personally feel that instead of driving the tube with huge currents it is better to get more sensitive speakers with higher transient capacity.

guy sergeant wrote:

I'm also a little concerned because I don't like the fat 'tubey' sound of the 834P one stage of which someone else pointed out was operating in less than ideal circumstances.

I’m not concern. Moreover I find that all the rest my phono-correctors are too fast, too “gipsy-sounding” and I am getting rid all of them leaving myself ONLY with ET2 connected to 834PT. Generally, if you care to add some “vigor” and some nervousness to Milq then it might be gone but boosting the plate current but I do not feel that it is necessary.

Rgs,
Romy

OK, first off all I have moved the latest posts to the “How to drive 6C33C” thread.

Regarding the operation point. My initial intention was to give Melquiades’ output as ML2 did: 175V, .310mA. What is important is to understand that for LF channel I would like to have much power as I could get, within the audible limits and reasons. I never dialed up power with LF channel but juts dumpling, which forced me to order two separate transformers, unfortunately. For the MF and Upper bass channels I really did not care about power and was only conserved about the relations about the harmonic correctness and “speed”. Another reasons why I did not care about the operation point in the beginning because with 8 sections of the Landhale transformers and 100-150mA of the spare gap size it is possible to configure any imaginable loading and play with different currents. 

The Milq’s design goes not really cares about the operation point of the output stage and when the amp was done it was spent quite a lot of time and efforts to fine what would be more suitable to drive that 6C33C. So, my operation points are juts what I ended up due to the experiments and listening and not something that I believe “should be”. BTW, the Super Melquiades plate transformers are 1A with triple secondary: 330V, 300V and 270V, minus the drooping in the rectification and LCRC: 10H -100uF - 50R – 10.000uF

Ronnie wrote:

The 230V 165mA point for the LF sections does not look so bad to me...

Ronnie, do not forget that the 130mA or 165mA for “bad electicity” is the operation for the MF channels. The MF channels are ½ of the 6C33C, not a whole 6C33C.

guy sergeant wrote:

I haven't measured anything I was just looking at the data sheet for the valve.

I do not know what you fantasy yourself. I have very little understanding it interest to looking at the distortions patter. They are juts numbers. Put the amps in bath and measure it’s water displacement and then claim that it does 63.473 of “something”. That “something” has as much to do with Sound as temperature in Sahara desert relates to appetite of polar bears in Antarctica. BTW, if it does means anything (not for me though) then I might tell that when the Melquiades and Super Melquiades were made Dima and I spent quite few affords/time to run different tests with generator and scope. We did not measure the distortions but Dima has a lot of experience to interpret those images and he told that the Milq at full power has around 5% of distortions, whatever it means

guy sergeant wrote:

When I get around to making these I shall make them as close as possible to your detailed design. I want them to be representative of what you are listening to so I can better understand the elements of playback you feel are important. There doesn't seem to be any point in changing anything.

Guy, all that I can tell you that you use completely WRONG motivations and if your want to have a replica of what I’m listening then do not west your time. It is completely idiotic intention to subscribe Romy the Cat’s believe in sound, some kind of Framer’s believe in sound of a believe in sound of some kind of Moron from AA. The only motivation that you should have would be your implicit nailing down what exactly is wrong with your current amplifier, what exactly you would like to change and in which direction. There was nothing more inelegant ever written about audio upgrades then this:

http://www.goodsoundclub.com/TreeItem.aspx?postID=432#432

Once again, my feelings of importance should not bother you. Develop your own and follow them. Read agene the release notes for the Milq. The Melquiades meant for the people who have objectives, not just the blind frustrations….

hagtech wrote:

After many years, Romy has come to enjoy this style of amplification.  If I read his words correctly, the sound of these machines brings him more emotion and connection with the music and musicians.  An admiral goal.  Who the heck cares about distortion numbers when you have this?

Actually, Jim, it has absolutely nothing to do with “emotion and connection with the music and musicians”. My feelings about the necessary to “appropriately saturate” reproduced sound with second harmonic do not derive from any “emotions” or even from music itself but rather from very simple reproduction of even the individual sounds. (There is even no need to go into the complexity of music). A single note has a pitch, or a centered frequency of maximum amplitude. However, the signal hit the pitch only instantaneously and a single note it s juts clams up to the pitch and then descends down. The pitch is always the same and even the “roundness of the pitch” (or a duration of long a single note stays at its amplitude summit) is always the same (for the same duration of the note). What is different would be the profile of the parabolas with which sound rolls to its pitch and roles out form the pitch’s summit.

I have no numbers. My RTA can work in a distortion analyzer mode and Dima bugs me to measure the Melq’s distortion pattern. I really have no motivations or needs to do it. There is something that I recognize as “THE correct acoustic distortion” and this is how I tune my playback. Let to visualize the distortions as the reflections. The 2nd distortions would be the initial reflection of acoustically produced sound, the 3rd would be the third and so on… In the real world we never register the direct sound, but always deal with reflections. The reverberation radius (the distance where we hear 50% of direct sound and 50% or reflected sound) is very small and if I remember correctly then for the large places like Carnegie or MET it is within a half dozen feet. The point is that our perception is always trained to perceive reflected sound and the dominating amount of information about the notes, tones and their expansiveness lives not in the instant hitting of the pitches but in the note’s harmonic.

Now, let pretend that you in the Carnegie Hall, filed with people (necessary for a proper reverberation patter), you have a single violinist is taking tones on stage and you are in a chair with a remote control and you might move your chair: from a few inches from that violinist to the end of the Hall. Moving back and force you moderate amount of distortions (primary and secondary reflections) and eventually you can find a place what you feel that the sharpness and evenness of that violin’s notes would be correct. Interesting that the “correct distance” for that violin will very-very little fluctuate with the volume of the notes and our hearing in natural acoustic space can accommodate our hearing attention (more about it later). However, the microphones can not do the intentional accommodation and they register everything alike and unconsciousness. In order to fight with it we need to inject into the recorded-reproduced process the “acoustic space-like behaving” superstructure that would overwrite the microphone’s vision and mimic for our awareness the acoustic environment – the space where out awareness is accustomed to operate and where all roles of phono-harmonies were born. The way to do it is to position between the absoluty correct signals that we get from he sources (let pretend that it is absoluty correct signal) and us, the listeners, a proxy layer. What this semi-transparent layer should do? I do not know but after this layer will be trespassed by Sound the Sound should have similar to the original acoustic structure patterns of the primary and secondary reflections. So, here is where the mechanism the phase randomanization kicks in, where is the none-linearity according to the “objective measurements” become “beneficial” and many other thing take place. Here is where my rules of “The Beach Effect” operate and here the humanity-like outcome of the electronic-barbaric methods becomes the key.

We should never forget that the “The Beach Effect” (or let it call it the “language of electronic transformations”) has also own negative consequences that always poison the core of purpose - Sound. For instance in that single violin on the Carnegie Hall example we never have any serious fluctuation of the “best harmonic distance” depends the volume of the notes but with a playback it is the Bitch and each mV of input signal on the input grid would require own behavioral patters of the above-mentioned “proxy layer”, or own algorithm of “The Beach Effect”. It is very-very complicated and as far as I know there is no one in audio who ever conquered this subject with a minor exception (minor because it was most likely accidental and because it was never made at the demanded level) of Lamm L1 preamp. There are many other “operation complexities” for the “proxy layer” but what it the point?

The point is that the definition of success of that “proxy layer” is not something that needs to be descried mathematically or measured by a distortion analyzer. Sound after passing the “proxy layer” should have acoustic-like signature. This acoustic-like signature is very much no subjective parameter and any person who has civilized and not corrupted reference points in music ends up with absolutely identical reference results.  Let pretend that a group of 1000 people have an imaginary ultimate 1000000-bands equalizer, the one that do nothing negative to sound besides juts EQ the amplitude. We take music, EQ if up and down randomly and then let those people to de-EQ it back. Anyone among with those 1000 people who are familiar with the way how life music impacts them will painlessly adjust that EQ and will end up with very much identically final outcome, means their Sounds will be fundamentally the same. Others will set up their own sounds to reflect their or cultural and musical corruptions - they will be somebody whom I call the Morons ™ and I always question what they do in audio.

So, the point is that no mater what an amplifiers or phono-corrector do they should output that very same “identically strategic final outcome” (an entire systems?) This is how I try to assess the distortion patterns - juts by listening and interpret my reaction to the heard Sound. It is import to note that I do not pay a lot of attention to “how it sounds” but rather “how Sound affect my own inner-me”, looking in myself as a reflection of the sound instead to the sound as a reflection of my imaginary model.

hagtech wrote:

Anyway, the trend here is the same as for the phonostage (corrector, whatever).  The EAR834 input tube has similar distortion characteristic at its bias point.

In the end a few words about the EAR-834PT. I do not use it raw but with Expressive Technologies transformer. The Expressive does with EAR-834PT very interesting thing; although I did not detect that it did such an interesting things with any other correctors that I used, at least not to the same degree with other correctors…

It is hard to explain what it is because it is unprecedented and would sound like a BS for anyone who did not experience it. Let for instance take any good transformer and place it in front EAR-834PT. It will do 20dB -30dB gain and if a transformer is really good then it will not loose any notable quality of the Sounds. I did it with a dozen transformers and let take the S&B/Bent step-up for a reference. It does not do anything wrong and produce perfectly OK Sound but when fan begins when the Expressive substitutes the S&B/Bent (or any other magnetic that I tried) The Expressive hugely INCREASES dynamic range of the Expressive-EAR-834PT tandem (or at least make it to appear this way). Also, it widen the extreme frequencies quiet dramatically but at the same time it combines all the sounds across the entire frequency range in some kind of very uniformed, very unswerving and very connected substances. It is very difficult to describe and it should be heard.

So, I do not know if my admiration of the Expressive-EAR-834PT tandem derives from to Expressive’s or from 834PT. I remember that when I use the raw 834PT with MM cartridges I still had not problems with 834PT sound, but what one can say in context of the MM 5mV Grado?. When I drive my 3mM SPU into the 834PT direct it has enough gain but I do not like sound and I use my SPU with the 65dB Expressive-EAR-834PT tandem.