The Super Melquiades Amplifier is essentially the 3 parallel running Melquiades amplifiers optimized for driving my Macondo Acoustic System, with line-level or inner-stage filters. This SET uses a full 6C33C at LF output and ½ of 6C33C for HF and MF channels. The key element – the unique driving stage in the Super Melquiades is identical to the regular Melquiades. Also, the Super Melquiades has an external, explicitly input-choke filtered, hugely overbuilt power supply(s).

The amp was designed/bult with no reservations to price or spent efforts, would it be design, inventory or assembling. The result of that labor of love sounds accordingly with the expectation. This image is just one left channel. The right channel is not done yet and will come to it’s duty sometimes in a couple month.

Rgs,
Romy the caT

Great! Please let me know when you are ready to unveil it. In the meantime, is there something I could start working on? Can you pass me transformer parameters? What about tubes and sockets? Caps?

Thank you.

Alex Yakovlev wrote:

Can you pass me transformer parameters? What about tubes and sockets? Caps?

It hard at this point to sat anything defiantly about the Super Melquiades sound. Positively to derive the speakers directly from amp with no speaker-level crossover does provide multiple and multiple benefits. So, Even if to subtract all other “differences” then what might be said about the Super Melquiades sound?

It is not far from where it was expected to be with, at this point, 3 clearly identifiable,  “different” moments.

1) The Super Melquiades has bass that way exceeded what was expected in term of its quality and weight. Perhaps this absolutely unheard by me before bass is responsible for the fact that the amp sounds identical at different volume and it does develop full bass even if it plays at insultingly low volume. In fact, I did noted that I tend to listen the Super Melquiades slightly softer then I usually listen my system.

2) The midrange channel is the exact replica of the regular Melquiades but in the midbass region Super Melquiades sounds larger, more fundamental and believe me or not but it subjectively pushes my upperbass horn lower.

3) The HF channel sound very different then Melquiades-Regula. The half of 6C33 do sound different then the whole tube and I still am trying to figure out how can I get the best out of it. The half of 6C33C is not instantaneously gratifying and it requires some thinking to learn and realize what it does….

Rgs,
The Cat

Chris, at this point it is hard to say anything defiantly. Subjectively it sounds like the Super Milq has slightly less HF extension and transient then the regular Milq, although the output transformer of the HF channel of Super Milq runs higher and flatter. From a different prospective the HF channel runs just a half-anode of the 6C33C. I remember that when I experimented with a half of the tube with the regular Milq I had in some way the similar result. Using half of this tube rely nicely removes this tube gilts and schmaltz not I have to lean to do it without loosing some transient bubbleness.

Well, I would not say that Super transient is HF challenged. It rather sound now at HF like any other better amp but it does not has the crassness, sense of instantaneousness and the phenomenal transient that regular Melquiades had. The Super Melquiades at its H sounds very musical and relax but not “revolutionary musical” as the Melquiades-regular did.

I still would like to stay with ½ of 6C33C and in worst care scenario I will put into the Vitavoxes S2 drivers the older aluminum surround diaphragms and they will push out the transient of the transients. Before I move to such a radical solution I think that there is two places in the Super Milq HF channel that have some room for improvement.

The HF channel has 13.5dB voltage devider at input that is made with RN60-55 resistors. They are wonderful resistors and I do love them but there are better. In the regular Milq  there were used S102 in the devider and I defiantly should replace them the RN60s.. Also it will be some resistors cancellation in input stage that would effectively remove one resistor from the signal path. I sincerely feel that it will have a positive effect.

The second thins is even more important. I consulted Dima with it and he explained that in the Super Milq and the regular Milq the between stages capacitors perform very different duties. In the Milq it was juts a coupling cap but in the Super Milq it is a filtering cap. According to him they work very differently and the demands for the cup in the Super Milq is order s of magnitude higher. In the Milq I use a wonderful 2 uF Electrocube 950 (I love this cap) that “sounded” fine. In the HF channel of Super Melquiades I shot to 3.700Hz that requires 410pF cap. None of the known to be good caps have such a small value, and to be over 400V at the same time. So, I put in the game Silver Dipped Mica. I really do not know how it sounds in this application. It is possible that the Mica cap does something with transient. I’m still on the market for a good pF-level coupling cap. Initially I meant to use in there air-caps (like in my phonocorrector) but it looks like it will not work out at 400V. I ordered for a test vacuum capacitors and will have them next week. Will see how then behave. Meanwhile if you have any suggestions for 400-2000pF caps the let me know. Or perhaps with time pass by the Mike will open up… Who knows….

Rgs,
The Cat

Allo,
Well, the divider may certainly be responsible for some softening of transients - in particualr, what values are you using there?

Also, the received wisdom is that silver micas are created far from equal; no experience with them myself but I believe Thorsten may have something to share in this area.

My concern is not so much the type of cap as to the value; air cap is probably the ultimate type, it should be quite easy to get some that are happy at 400V plus, but I really do wonder if the high output impedance of the 410pF/100k filter you are using is responsible.

So, it will be interesting to see how the vacuum caps compare (though from what I've seen they are probably as big as the amp). In the meantime, may I suggest trying the original big value coupling cap (the Electrocube 950) with a low impedance filter ahead of the amp?  I know this changes everything, but it should provide another interesting point of reference. If your preamp is happy driving it, maybe 4nF/10k or even 10nF/2.5k?

Oh yeah, if you are going to stay at an x/o of 3.7KHz, an aircore transformer becomes even easier...

Cheers
cv

PS Btw, I think I have mentioned this before, but for my planned aircore OPT amp, I will be using a step-down input transformer wound on a tiny toroidal nanocrystalline metglas core. This will

a) provide a 1st order high pass filter (fed via a series resistor to form an LR filter)
b) provide stepdown for gain matching and better drive of the input tube
c) achieve voltage isolation so that the drive can be directly coupled to the output tube (negative supply voltage to the first stage), also battery bias of 1st stage via the secondary

no caps anywhere; the toroid is wound (tediously by hand) with almost a single layer winding: 2/3 of the way round is the primary, secondary 1/3 and it goes all the way up into the MHz.

Result: the only signal path cap is the power supply resevoir to the output valve. input stage needs VR tube regulation - has to be very quiet as the dc coupling/split supply means all psu noise is presented to the opt

Chris,

My frustrations with HF of this amp reminds me the Agatha Christie novels….  however, like anything else it has an end, even a temporarily end….
After some experiments and still being dissatisfied with the HF I ended up consulting with Dima.  He spread me a theory according to which the HF was killed by my coupling Silver Mica cap. All my arguments that silver mica work OK he demised claming that it is known how mica works in parallel application but generally people do not use it is serial application. Well, it sounded like a reasonable argument. I personally did not believe him as I firmly stay that sound  (instead of The Sounds) does not come from sonic signatures of the parts but from the global topological implementations. However I decided to try. I placed between the stages my default 2uF Cube 950 (with speaker lever crossover) and it was like return back to home after a long journey. The amp immediately paid back with beautiful and VERY extended HF. In fact they are even more interesting and more extended then with a regular Melquiades!!! Do not forget that now the S2 driver does not see the upperbass’s coil and it makes it sound VERY clean!!! It was exactly one of the main objectives that I visualized for this project.

Now, I still do not know if he Dima was right or wrong about the Mica. It is possible that your proposal of high impedance does play a roll in there as well. Or perhaps Dima was correct and the only cap should be blamed for anything. Dima made me, contrary to my weak protests, to bult my own Cupper Foil with Teflon cap that I most likely will try soon in the HF channel. Perhaps air or vacuum will be next. HOWEVER it still will NOT help in there if the high impedance was the reasons of HF killing.

Frankly speaking I so love the Cube 950 that if I get the correct value I would not even experiment with anything else. The lower value I can get might be 1000pF (.001uf) that would require 40K bias resistors. I do not think I might go so low…

I really begin to hate all of thing! When will that all soldering stopped!!!?

Well, will see.
The caT

Romy the Cat wrote:

After some experiments and still being dissatisfied with the HF I ended up consulting with Dima.  He spread me a theory according to which the HF was killed by my coupling Silver Mica cap. All my arguments that silver mica work OK he demised claming that it is known how mica works in parallel application but generally people do not use it is serial application.

Romy the Cat wrote:

Dima made me, contrary to my weak protests, to bult my own Cupper Foil with Teflon cap that I most likely will try soon in the HF channel. Perhaps air or vacuum will be next.

Romy the Cat wrote:

I placed between the stages my default 2uF Cube 950 (with speaker lever crossover) and it was like return back to home after a long journey.

Romy the Cat wrote:

Frankly speaking I so love the Cube 950 that if I get the correct value I would not even experiment with anything else. The lower value I can get might be 1000pF (.001uf) that would require 40K bias resistors. I do not think I might go so low…

Romy the Cat wrote:

I really begin to hate all of thing! When will that all soldering stopped!!!?

Paul,

Thanks for pitching the Johnson capacitors. They do not look “large enough” for me. I used in past the following caps:

http://www.oselectronics.com/ose_p96.htm

They have very good size to value ratio. The ironic part that I just called to the manufacturer and they informed me that their caps are 600V rated. This turn out to be very good news!

Dima, keeps treating me with his worrying that at high voltage the air caps will act as a capacitance microphone and will pick up all acoustic and electromagnetic dirt form air. Perhaps he is correct but I remember how wonderful the air caps were in the corrector…

From a different prospective the result I got from the Cube 950 was so satisfying that it made me very satisfied. So, I ordered a party of Cube 950 1000pF (they turned out to be cheap like dirt) and if a series tandem of them will do OK in this high impedance application then I will try to un-wind some turns from the 950s making them smaller. Also, I can always drop the loading resistor a little… In anyways now I am in good shape, since I learned that the problem was the cap (filter) but not the transformer or any other “bigger” pain.

The problem is that I know that I will always be drooling when I look at the air capacitors… Probably I should burn this amp once in order to get the answers about the air caps. When I designed the chasses I actually made the provision for 2 pair of 1650pF air filters. I have space in there… I have the sealed enclosure for the capacitors… I have the capacitors themselves… I have a life insurance….

I have to do it….

Rgs,
The caT

Added in 3 hours:

I gave to the air cap some listening attention and frankly speaking all my enthusiasm very much evaporated. Yes, the air cap provided in SOME WAYS the effect that it did in phonocorrector, in term of clearing up the granularity and grain at HF. The effect is instantaneously notable and very pleasing. However, all together with the air cap the sound become very disgusting.

I kind of hare to convert myself into a Moron who audition different caps and write about it his conclusion, although it looks like I’m turning into this. I king of feel that I am wiling to get very specific Sound of Melquíades’ HF and I know that it is able to give… but there is this cap on my way. Well, at least if I sell this story to myself under this souse then it does not sound too depriving…

Anyhow, returning back to the sound of the air cap. It is, in THIS CASE, very mushy, very echoed, very foggy and with very strange harmonic context. It more sound like a room with heavy clapping echo. Interesting, that it nothing like the air-cap affect that I observed in phonostage when the caps did not care a lot of voltage.

I ma sorry, that my good air-cap hallucination turn out to be a hallucination. Again and again, after trying practically everything in cap world I’m returning to the peaceful and secure world of the Cube 950….

Rgs,
Romy the caT

Chris,

What I most concern is that my pursuing to fix the HF channel I make myself similar to zillion audio Morons who run across Audio like wounded in the ass animals and keep evaluation components, elements, cables, soldering and defectively of the wires. Although I perfectly understand that the validity of this actions but their method I found completely idiotic, furthermore - their objectives almost always condemned to be failed. In dominating majority of cases they do not have methodologically objective evolution methods and practically all of them (including me) get relative and severely conditioned results.

I try never (how about now? :-) to be engaged into voicing of sound by matching parts and even now doing my current crassness I am not happy with what I do. All those endless chasing of the results by selection of parts affects the Sounds but not Sound. Mostly people who play with those foolish games use initially very poor or even horrible, frequency unusable (dispute to be audiophile-approved) components. So, in thier cases the change between the “obvious garbage” and “almost-garbage” produce a “notable result”. I personally use proven, frequency commercial grade, NUTRAL elements that themselves do minimum or nothing to Sound and THEN I try to shape sound by topologic decision instated of the voicing chose. In fact the real valuable Sound does not come from a voicing anyhow and the really serious Topological Sound smashes through anyhow, even through the very crapy “sounding” components.

Well, it wounds like I “use” or I “shape sound”… In realty this exposure of mine to my believes that “I do something” is very temporarily and frankly speaking it started it being instigated by Dima and few other local folks. I hope when I finish the Super Milq project then my amusement with solder will be over and I will return to my quiet lagoon of listening music and beaching/teaching to audio Morons that thier sound sucks…

In the end a few words about the actual subject :-)

There are many different vision what and why going on in there including just an “absurd“ one that the line level filtering is fundamentally faulty because the gain stages do not operate within the full scale of harmonic envelop. I do not know what is going and I feel that the problem with the cap is topological – means the filtering demand to cap is very different then to the compiling cap + high impedance + high voltage. Probably not all caps can handle it. In a couple days I will have my Cube 950 that is my proven default  NUTRAL cap. I should had began with it and do not allow myself to go in o this crassness. I will think about the further reasoning and conclusions after I try the 950 in there.

Rgs,
The Cat

I hear you... I've never gone for the school of thought that says, let's correct the inadequacies of component/topology A by throwing in some "complementary" inadequacies in B.

Or, as a feline mathematician might put it, if each of the "sounds" is flawed:

Sound =   ( (sounds1 ^ k) + (sounds2 ^ k) + (sounds3 ^ k)...) ^ (1/k)

where k = musical demand coefficient

k = 1 for Patricia Barber Cafe Wotsit
k = 2 for classic multitracked rock
k = 4 for say, Rach 2
k = 8 for Verdi Requiem at full tilt

where the most extreme moron has failed to understand that k is even and is trapped in a circle of listening and tuning to k = 1 music...

???
cheers
cv

slowmotion wrote:

Voicing not BS, but......

Rgs,
The caT

Romy the Cat wrote:

So, voicing is a big totally artificial pile of BS and even the conversations and thinking about voicing is BS.

Guys,

Thanks, for your help with all my HF channel frustration.

The saga is not over. The latest update is the following:

The “secured” Cubes 950 (1000pF and 500pF) were placed between the Milq’s HF channel and the result was dissatisfying. The sound was very clean but FH were rolled off. Then 1000pF was placed in input of the amp against of 43K of HF channel (the preamp Rout is 7R). The HF returned back but sound picked HUGE amount of HF noise - anything at HF was resonating. I presume that the brand new Cubes of the low value should work in there for a while before they should be listened….They were juts too new....

I knew that I would have problems to filter HF at the Milq’s level :-(… In the end, I put back between the stages 2uF of broken-in Cube and placed at the speaker level 3uF of Vitamin Q (the configuration that I’ve been using for years). With this configuration the Super Milq HF channel gave out what I needed from it. I have to note that it has not rally anything to do with what I use to and habituated. It was very problem free sound, having which I would not peruse anything else. In fact it was way better sound then I use to as there was no upper bass coil into the upper FH tone.

Anyhow, then, I thought that in order do not overload the low primary inductance of HF channel I should go for lower coupling cup. I took Cube 950 .047uF and placed it there. The funny part that it, although sounded well, it still got some very remote resemblance of the HF noise that 1000pF had when it was placed at input.  I put it to burn for a couple days to see when it would lead me. So far, the 2uF inner-stage and the speaker-level filter performed absolutely and incontestably the best.

Rgs,
Romy

I really do not what is gong on there and why. Thorsten's comment that line-level crossovers do not act like speakers’ level crossover confirms what I feel and what some other people suggested me. Yes, the reactance of the crossover component, the inductive reactance of the driver coils, the lack of harmonic modulations in the amplifier stages when a stage operates within a limited frequency range… all of this certainly impact the final result. So what?

The experiments with the Super Melquiades HF channel are over. I found the optimum filtering configuration (or at least it what I feel presently) and closed the amp up. To my surprise the high-pass filter at the very input of the HF amp, before the first stage, did not sounded well. It had in a way too mechanical sound. I abandoned the frequency calculations and started to use mic-feed RTA on that HF dedicated channel in order to have the identical slopes with diferent filters. i did it  for the sake of more or less objective listening evaluations as evan with the very identical sloped the chanal where I had the filters at it's input sounded very thin and lean. When is puff it up with LF, moving the crossover point lower, the channel picked some “body” but still had sound very artificial and synthetic: It more sounded like “frequency pusher” instead of “normal narrow frequency sound”.

The filter between the Super Melquiades’ stages did not work due to necessity to use very low value caps against the high impedance and consequential HF roll off. However, the sound di not indicated any artificiality and was very natural, although HF channeled. Probably the first channel of the Super Milq should run full frequency. I do not know if that is the specific demand of that specific UHF rube that runs in the first stage or that is a genetic rule or perhaps something else. I know that it certainly turned out to be a rule in my case. Interpreting that in Super Melquiades’ LF channel I do have the lowpass first order filter of 65Hz sitting before the input stage… and I DO LOVE what the bass channel dose. Perhaps the problem that my frequency-limited first stage has is applicable only for HF? Or perhaps whatever the frequency-limited first stage dos at LF made me so love what the Super Melq’s LF channel does? Who knows!

Anyhow, the best, very much the best result that I was able to get form Super Melquiades’ HF channel was with the full range channel (2uF coupling cap) and the speaker level cap in series with the driver. Interestingly that when I began to lower the coupling cap I liked sound form HF channel less and less, although I it was way beyond the crossover point of driver.  I was hesitant to let the output stage to operate full range, due to low inductively of the HF output transformer primary (~1-1.5H). I left between the stages .047uF that makes ~ 35Hz high-pass filter and will let it to run as is for a few weeks to see what happen next. Still, when I go with the full-blown coupling cap of 1-2uF the HF become softer and more natural. Mystery, isn’t it?

Rgs,
The caT

Hm,

since I made my last post I was walking around my room, listing music and a feeling that something was very  still wrong did not live me. It always there was some sense of presence of HF in sound. It was not my typical full-Sound where it is imposable to say if it was enough "frequencies" but it was some kind of Sound + the "frequencies". I tested the  1st stage HF tube. It was fine but I still replaced it. I tested and juts in case replaced the output tube, even tryed thefew different brands of them but the problem did not go away.

Then I said: “OK screw all theory!” I opened the amps and replaced the coupling cap from .047uF  to 2uF. O wonder! It fixed the problem immediately. Then I trued few more caps in there and I learned that when the larger caps were placing between the stages I got the WAY better HF integration . Pretty much anything over .68uF worked very well. I would like to remind that the HF channel works from 3.700Hz electrically....

This all settles down a very important conclusion that for HF channel of multi-amping the presents of the modulative LF harmonics is incredibly important and apparently the HF channel amplifiers MUST run full range.

The 2uF-coupling cup was permanently placed to the place what it should be.

I hope the time I spend to learnt this make some of you to think then you do your multi-amping…

Rgs,
Romy

Morning,
I'm not disputing what you heard, but your conclusion that the LF must modulate the tweeter section can't possibly be right. 0.047uF is a -3db rolloff of 34Hz. If you're saying that over 0.68uF was fine, then by implication 0.47uF is not as good, that means that a cutoffs of 2.3 vs 3.4Hz on the tweeter channel make a difference!

Hence I'm thinking that LF modulation has nothing to do with it.

I think it has to be something else, to me it now sounds like the impedance issue is the most likely. Roughly 1k Zout with the 0.047uF; then again, given that the driver will have a Zout of at least that, can the 6c33 it be that sensitive? Maybe... who knows?

Anyway, you seem happy with it now, so I guess the case is closed and unsolved.

cheers