Rerurn to Romy the Cat's Site

Melquiades Amplifier
Topic: The single-stage Milq and power Supplies.

Page 1 of 1 (11 items)


Posted by Romy the Cat on 05-03-2007

I am very much do not like the current state of my mind about the single-stage Milq. Let me to explain….

In 2001-2002 I spend some time trying to assess the sonic influence of PS to sound. After making numerous tests and prototypes I concluded that SS soft-fast-switching rectification with following high DCR and slow-core LCRC filter does fine. What was important that it did fine in full range context making sound in a way full-range-wise acceptably balanced.

Now, I am about to convert my Milq into a singe stage amp that will not work full range and will have very limited power (still, in class A) and the amp will most likely would require very different type of power supply. I remember that during my former sonic assessments I was particularly concern about the minute nuances of bass structure, the bass WHILE the amp will handled the full range. It will not be any bass in all 4 single-stage Melquiades and therefore my former experience about power supplies are absolutely irrelevant….

I spoke with a number of peoples about the power supplies trying to understand if any common knowledge exists abut the subject but it confused me even more. Everyone use own solutions that they claim work find works for them. All of them use low sensitively loads. Furthermore all of their conclusions contradict each other. To insult the injury I will say that no one from whom I heard looked in the direction of sonic attributes analyses for  power supply that feed speakers with dedicated or (limited) frequency range…

So, where it brings me? Pretty much to nowhere. I know that my former LCRC with enormous size of the last capacitor sitting right the tube cathode (ground in my case) might not work and it makes HF slightly hard and thinner with increase of dynamic, affecting the “space” quality of sound:

http://www.GoodSoundClub.com/TreeItem.aspx?PostID=4321

So what to do now? The possibilities are really endless. To stay with high impedance supply using the inductance filtering or the high capacitance filtering? To go for a simplify CRC filtration? To go for LCLCLC flirtation with minimal non-electrolytic capacitance? To go for tube rectification (many types) with different combination of filters? To go for higher AC frequency power supply? To go for transformerless regeneration? To go for a voltage regulation? There are many other opportunities and all of then very simple to implement. Still, I really am not wiling to try all of them not to mention that it is VERY difficult (if any possible) to evaluate a solution unit the 4 channels will be built and all channel are connected… I do not know… perhaps I need to make gain stage, mount in into Macondo (the PS will be separate anyway) and then juts change the type of the PS for B+? (The bias supplies are no controversial and bias supplies will remain as is)

It would be interesting to hear from anyone who made experiments with power supplies for narrow frequency range and high sensitively but still…. will this experience applicable for my given case (the type of the tube I use and some other specifics of my Milq)? I really do not have answers and… I am very glad that the Super Melquiades is up and running perfectly… letting me DO NOT DO ANYTHING… Such a great feeling….

Anyhow, if you have any thoughts then feel free to share them as I am kind of at lost here…

Rgs,
Romy the caT

Posted by Paul S on 05-03-2007
Romy, I have futzed around with caps more than I should have, mostly because they so obviously affect the sound, no matter where they are in the circuit.  In general, it seems like the best bet is to keep the values as small as possible, and in some case it sems like it's best to keep the physical size down as small as possible, too.  I have heard a lot about using poly caps instead of electrolytics, but it seems to me that electrolytics have their place where large values make poly caps not only impractical but audibly slower.  I am wondering if it might be a good thing to design ahead of a cap just in order to keep its value/size as small as practically possible in order to get better sound.  I happen to have all SS rectification right now, and I can't say at this point that I miss tube rectification.  But I can say that everywhere I have lowered the capacitor values to a minimum I get better sound.  It's probably too harsh to say that capacitors are a necessary evil, but...
 
Best regards,
Paul S

Posted by Romy the Cat on 05-03-2007


Posted by Romy the Cat on 05-03-2007
The major question in the circuit above would be what the minimum values of the last capacitors (currently they are market as 1000uF) would be necessary in order the LC chain “do not see” the loaded output transformer and the plate of the tube. The presumption that the "electrons" (or whatever it is) will go out of plate and if the amp has a sufficient capacitance in the last cap then the the "electrons" go to ground right there and will not roll back to the LC… Now, how the hell can I figure out the minimal values of the last cap before the transformer…

Posted by drdna on 05-04-2007

This is sort of a sideways answer to this question, but having used power supplies with vacuum and soild state rectification, I also played with Allen Wright's SuperReg circuit and I really liked the way it sounded. A more stable and grainless feeling to the music.  So when you are looking at the power supplies, consider this topology as a possibility.

Adrian


Posted by Romy the Cat on 05-04-2007

 drdna wrote:

This is sort of a sideways answer to this question, but having used power supplies with vacuum and soild state rectification, I also played with Allen Wright's SuperReg circuit and I really liked the way it sounded. A more stable and grainless feeling to the music.  So when you are looking at the power supplies, consider this topology as a possibility.
Adrian,

Yes, thanks I will consider this and many other options. Mostly likely I will make the amp itself, mountable inside the Macondo and the PS will be separate. I think to do it as I usually do (picture above) and then eventually to bring in home different ready to go supplied of 175V (tube, SS regulated etc) I do not see myself building them…. because it is too boring. It is interesting to listen them and to here what all those topologies and implementations do - not to build them.

Interesting that you mentioned regulation…. For low voltage SS devises it is given. The regular cheap 3-pin regulators are garbage and a long time ago I discovered that replacing them with better discreet regulators was a very serious step forward. However, I never understood regulation for tube plates.  5-6 years ago I was harassing everyone asking expanding me why it should mater as any plate regulator I head did some improvement and at the same time affected sound negatively in some ways (dynamics for off all). Eventually in 2002 I asks a guy to design for me a regulator with specific SONIC requirements, He did and I built it eventually. It used ultra fast 7788 pentode as amplifier and 6C19P as a regulator. It was driven by LCRC filter and it was kind of good… Here is the picture of what is left form it as now:

The purpose of that regulator was to investigate how voltage regulation of pentode’s screen affects sound in a first stage of phonocorrector. Also I drove the plates by the regulators to see what happen. With the screen result were extremely positive (where do you think the idea of the Melquiades input stage came from?). However, I would not say the same about the plates. Regulation of plates had very-very-very minor effect in “smoothing” LF but always slightly restricted dynamics. I have to tell that that regulator of mine was modified in all imaginable ways (at least 20 times) in order to be transparent on plate. Even in it’s the most dynamic-transparent state I asked myself why do I need this regulator? Eventually I decided that I do not need regulation of plate as I never understood what would be wrong if the anode voltage will fluctuate a little. I think the voltmeter voltage over a few hours is less important them minor lost of dynamics…

Well, sure it might very arrogant and perhabs stupid to make conclusion about the plate regulation judging only in context of one circuit and one regulator. However, I do have other regulators and I did try regulating the Melquiades plate, those I used off the shelf tube regulators. I do not know about the Allen Wright regulators and perhabs about many other good regulators and off the shelf power supplies. I do not mind to tried them when everything will be ready and when to try then will be literally to switch two cable form one supplely to other (like the Border Patrol PS). What I would like to avoid is to make alternative supplies myself, spending time analyzing all those mercury rectifiers, TV damping diodes, regulations methods and the rest DIY crap….

Rgs,
Romy the caT

Posted by Paul S on 05-04-2007
Romy, do you plan to float the current, too, or do you plan to keep that constant?

Is it correct to say that since all the tubes will see different loads, it may be that they will behave/react/sound differently on the nether end of the same source, or, as you have said, there remains the possibility of variable "blow-back".

But I'm guessing that you already came up with the 1,000 uF value in the supply's out links by listening?

So I'd hate to be the one to say that the lowest cap value would be the one that yielded the maximum acceptable noise under dynamic loading, and anyway it does not appear just that simple.

Why is it hard for me to picture unregulated supplies?  I am used to seeing more isolation per tube, and/or per tube tailoring, depending on OPT/load.

I am thinking of an interesting PS link I will try to locate.  It is for a multi-staged amp, but is an interesting "tank", with diodes, no caps in line, with a "stacked" ground plane.

Best regards,
Paul

Posted by Romy the Cat on 05-04-2007

 Paul S wrote:
Is it correct to say that since all the tubes will see different loads, it may be that they will behave/react/sound differently on the nether end of the same source, or, as you have said, there remains the possibility of variable "blow-back".

The tubes will be loaded differently slight. The Fundamentals channel will be 3dB idler then MF channel and tweeter will be even 6dB more idle. However I do not see it as a problem, in fact I consider even to drive the HF channels with up to 1W less plate dissipations…

 Paul S wrote:
But I'm guessing that you already came up with the 1,000 uF value in the supply's out links by listening?

Nope I did not. This is a purely blind guess in order to let choke do not “see” the plate and transformer.

 Paul S wrote:
Why is it hard for me to picture unregulated supplies?  I am used to seeing more isolation per tube, and/or per tube tailoring, depending on OPT/load.

Setting the plate loading will be done by transformers. I do not see needs to preserve any specials power souses isolation between the tubes. Do not forget: they are the stages of the same right or left channels. Also not of them will be handling any stressing LF of power and they will be in very “gentile” mode, band-passed at line level. Well, it is not the I’m trying to educate you – I rather am trying to figure out for myself what the hell I will be doing….

 Paul S wrote:
I am thinking of an interesting PS link I will try to locate.  It is for a multi-staged amp, but is an interesting "tank", with diodes, no caps in line, with a "stacked" ground plane.

Please find it. At this point I opined for any idea. I even think to drive it with some kind symmetric regenerator since it will be juts HF… Who knows….

The Cat

Posted by Paul S on 05-04-2007
OK, Romy, take a deep breath:

http://home.att.net/~chimeraone/axiom300bschematic.html

Sorry to say, it's clever as Hell.

If it makes you feel any better it was designed by some rocket scientist, or something, not you-know-who.

This is in fact a pretty cool self-balancing tank that "looks" plenty stiff to the tubes and OPT without conventional regulation.  It about has to be "fast", that's for sure.

The diodes at the power tubes appear to me to also serve as a sort of super-short local quasi-feedback arrangement.

It looks to float voltage with constant current, if that interests you.
 
Note that all caps are quite small, even with higher voltages, and all are perched "out of loop" at ground gates, just like your PS drawing, except here there can be no "blowback".

That has to be one hell of a main PS tranny; probably costs about the same as a car.

There's more than you want to read on the subject salted around the site.

I was all set to build a pair of these before I found my ML2s and took the easy way out ;>}

I'd like to hear from anyone who actually knows enough to comment about this.  In other words, I just pretty much shot my wad on this subject.

Best regards,
Paul S

Posted by Romy the Cat on 05-04-2007
It is done with own vision and very consistant with own vision. However, I’m not position to assess how it might sound. This guy drives Lowthers and believes that he is a belly-button of universe. I do not case about nether 300B sound not Lowthers so, it would be difficult to me to judge his judgment. There are ways to implement PS with no caps at all but one again – we do not make PS but Sound. How this “clever as Hell” sound? Do you know? Do you think that “rocket scientist” knows? Well, it might be interesting to head that axiom amps but not behind the burka of that guy ego.

Posted by Paul S on 05-05-2007

Well, I figured you'd react like that, at least at first ;>}.

Referring only to the PS, ignoring the tinfoil hats, etc., taking the whole thing out of context if you need to, it just looks like an interesting way to way to keep the PS tank "full" and clean under diverse loading, with minimal influence from "parts", especially caps, or backwash.  A sort of "ground zero" PS, and probably possible to do a simpler version with sand.
 
Meanwhile, I hear that pure silver foil works better than tinfoil...

Best regards,
Paul S





Page 1 of 1 (11 items)