After many hours and much efforts I have connected the new 6-channels Super Milq to my 6-channels Macondo, with omission of tweeters for now.  I made some measurements and initial adjustments and then was trying to listen it briefly. I did not listen it too critical as it would require more calibration on everything but I was interested about my initial raw impressions. Here are some instantaneous noticeable observations, without order, priority and without any attempts to make any generalizations.

1) The attenuator at Injection Channel turned out to be a total success: very-very-very transparent!

2) I will keep the Injection Channel without any filters and will filter my Tannoys at speaker level. I feel that it is nice to have in house a channel in my Super Milq that will be a full-range Melquiades, able to drive anything else…

3) The volume from single stage amps (Fundamental and MF channels, loaded 15:1) is the identical to all the rest channels – absolutely perfect. The 30K-10K voltage devised at upperbass was all that was necessary to balance all system. (look at the circuit above at provisos page:  http://www.GoodSoundClub.com/TreeItem.aspx?PostID=5207  )

4) I need to research more how I can capitalize on the upperbass channel, since it does not see the Fundamental’s channel drivers anymore.

5) The sound of a single-stage Milq is not the same to the sound of a two stage Milq - they are NOT identical amplifiers sonically. I can’t not say which is “better”, as I hear the single-stage Milqs only in context of narrow-filtered drivers.

6) Since the single-stage Milqs have own sonic signature I feel that I shell review how Macondo should be organized in terms to capitalize the opportunities that those single-stage channels offer.

7) Is any huge difference between Macondo driven by 3-channels Super Milq and 6-channels Super Milq? I would not say so. There are some differences, including some very positive but still are not as huge as I would like to think. From a different perspective the 3-channels Super Milq was perfectly calibrated to work with Macondo, the 6-channels Super Milq is not at this point. I might only presume that when the 6-channels Super Milq and Macondo will be properly integrated then it might be different.

8) It looks that the tube maintenance with the 6-channels Super Milq will be an ugly ceremony….

9) Screw APS PurePower people! During all those upcoming finalizing procedures I would need a stabilized voltage and I would desperately need constant quality sounding electricity. I still have none of them. It begins to frustrate me….

10) The very prominent changes in sound that I’m getting now are from MF and partially Fundamental Channels. With Fundamental Channels is slightly more interesting then it use to be but it mostly because it is now flexible in operation and setting up.

11) The new sound from MF channel is the punch-line of the entire project. Direct coupled to single-stage Milq the MF horn with S2 drives do sound different. The new MF channel has way less granularity and has much cleaner and much smoother sound flow. It is in way very similar to the sound that I was getting from electromagnetic drivers but in contrary to the field coil type of sound the new MF channel has dynamics and it is capable to become very swift and zippy if music is called upon. Still even it’s zippiness a different then it used to be…. is it very smooth but vivid at the same time. I like a lot that new MF sound, but at the same time I have some reservations. My concern is that with introduction of that super-lucrative and “palatable cleanness” the new MF channels lost some bubbleness on sound. I would question if it is good but thanks God I do not need to question it as I have a tool that really sets me free – the Injection Channel. Without the Injection Channel I would most likely have problem with the super clean MF driver but with the Injection Channel I might inject any structure I want into MF sound and it make me really review what I getting from the new MF channel. I was plying a little with MF + Injection and was able to get very-very interesting results – actually capitalizing on the fact of the MF’s new  purity and hipper-hygiene…

12) I would need to roll off the upper knee of my Fundamental Channel slightly faster…

13) The self noise from the 109dB sensitive speakers is = nothing short of phenomenal. All 5 channels emit no noise of any kind, even if I stick my head into the horns. There is no physical evidence that the system is running, beside the lights from the amp – very nice. The 3-channels Super Milq was very quiet but the 6-channels Super Milq took it even further.

14) The 6-channels Super Milq, even it has juts 3 more 6E5P for whatever reason it radiates more heat and warms room much stronger then it older brother.  Perhaps it is because I have more powerful forced cooling….

15) I was genius that I implement in the new Super Milq an ability to turn the amp on and off via a pedal. It is so cool do not bend my fat and lazy body and just to hit a pedal… I need the identical functionally for opening my refrigerator door…

jessie.dazzle wrote:

Upon reading this, I of course immediately want to know your reaction when running the S2 MF channel with the original metal suspended diaphragms reinstalled...

 A very reasonable question. I would like myself to know the answer and am planning to take the S2 with older metal suspension cone and to see how if it behave with single stage amp different. However, I will do it later on and it is not at my immediate radars as it would be satellites and I have a bigger fish to fry at this point (like founding the right operation points for single stages and to make it possible in context of those tubes, resolve some biasing conflicts….. etc…

jessie.dazzle wrote:

This is a really great idea, which BTW should also allow Koshka power up the new amps!

Hm, I never thought that I would make the amp top vulnerable to my Cat to turn the amp on. This is a reasonable concern. Uselessly Koshka do not care about anything unless it covered with fresh tuna let hope that she will not do it and in the worst case I will use a very strong pedal.

Eventually, I had a chance to spend last night with up and running new Milq (no tweeters was involved at this point). As I said before, the major fight was around the MF channel and I feel I did found a good configuration where the MF channel takes advantage of the single-stage amp but do not exhibit the improper volume/harmonic behavior. It is very interesting to see how S2 responses to single-stage amp. I have many observations and I will make them later, when the information will age in my mind…

Meanwhile I decided last nigh to pick a mono recording and just listen some real music, since the Milq/Macondo are already properly balanced.  I took the La Scala’s 1955 performance of Giordano’s “Andrea Chenier” with shining Maria Callas and Mario del Monaco. It played well, though I question if it makes sense to use that caliber music for audio evaluations. 

It was very fascinating to see the MF channel’s reaction to Sound. It has more of that “snow taste” and in a way sounds slightly less “Vitavox-like”. I am not quite sure at this point if it was positive quality and I need to play with settings around the MF channel more in order to get maximum-interesting sound out of the MF channel. What is very positive moment in all of it is that I might work explicitly with MF channel, without affecting anything else. If nothing else and only because of this moment then the entire idea of the 6 channels Super Milq is well-justified. Thankfully, there is a lot of else…

RonyWeissman wrote:

Is it the new Tradition OPT for the MF channel, sorry I haven't had time to follow threads lately? If so I am curious as I need to order an OPT for my full-range melquiades project soon.  (I found someone to build it for me finally!!!) .

Romy the Cat wrote:

If you read the very important post: http://www.GoodSoundClub.com/TreeItem.aspx?PostID=5255 then might understand that it lead to a painful research on the subject. I have ton of data now and many experiments done. If I have time I will upload some notes.

Romy the Cat wrote:

I did not expected that the main impediments in the super Milq project will be the organization (sound) of the filters, particularly the high-pass filters.

Paul S wrote:

Romy, on Fundamentals channel did you leave the 20uF cap at speaker level or did you re-configure that band pass at line level?

Nope, it is hoe it depicted the 20uF cap at speaker level. I see no problems with it is very narrow bandpass that cares no HF. However the interdiction of a high-pass filter at line level, right next to the low pass will create problems. The OPT at that channel is 60H and can care a full range signal.

Paul S wrote:

Are the RC feeds you mention pre-driver or interstage for A, B and C?

Paul S wrote:

Is everything still 1st order?

Paul S wrote:

Can you say already if you have enough drive with 5H on the MF?

Paul S wrote:

No need to adjust - bias on A, B, C?

Nope, in the two stages channels the driver cares 17mA that is very far from critical 35-40…

Paul S wrote:

I imagine "clarity" is astounding.

Nothing extraordinary. There are certain areas where I can report some improvement but nothing revolutionary. It will take time for me to learns how to capitalize on the advancements that I abstained now and how to convert it into more interesting sound….

This weekend the new 6-chenal Milq is not finished yet design-vise but functionally is already “there”. The amp was good enough to play with other 3-chenal Milq. Firstly for the last month I was able to listen full stereo from Macondo…. I spend some time with specific “targeted listening” regarding the nuisances that I was interested in. I made quit a number of exciting observations that will share as time goes by and then I decided to relax from audio and to listen some music. The electricity was not bad and the entire payback was responding not badly. Then, at 2PM, the WGBH kicked in….

In their “SymphonyCast” program the WGBH broadcasted Gergiev leading London Symphony with the 2007 Proms program. Gergiev is not my favorite guy and the Proms seldomly have interesting play but this time it was surprise. Gergiev brought all-Russian program:

Tchaikovsky: Romeo & Juliet Overture
Prokofiev: Piano Concerto No. 2 by Alexander Toradze
Tchaikovsky: Hamlet
Prokofiev: Symphony No. 7

The concert took place in August and WGBH broadcasted live-to-tape BBC recording. The sound was wonderful and as usually SymphonyCast has no abusive compression. The Prokofiev’s symphony was confusing as the Seven should be. The Hamlet was ornamented but boring – I heard more interesting play of this peace. The Second Piano Concerto was extremely appealing - it is seldomly played work and it is incredible complex and difficult. I need to hear it more and I tend to like what I heard; though there were many moments where Alexander Toradze was kind of “lost”. And then the very first piece that LSO and Gergiev played – it was the Romeo & Juliet Overture. That was nothing short of amazing – bold, meaningful and with great respect of own play – the phenomenal performance!

I have to tell that I have heard some new thing in the new Super Milq/Macondo that I liked very much and that I did not experience before. I hope it was not juts an “another good electricity day”…

OK, it took for a while but at this point I have nailed down all design aspects of the new 6-channels Super Milq, and particularly its input stage. I have listen and evaluated all solutions that will be implemented in there; selected, appraised and ordered all ingredients and now I am at a straight finish line with this amp. At this point I know how it will be finished, organized on context of my acoustic system and how it will sound when it will be done.

Here is the semi-final diagram of the 6-channels Super Milq input stage. I said “semi-final” because the values or 2 “red” resistors in the C and F channels will be slightly changed later on, but it will be very minor and will be no more then 20% from where it now. I will make the adjustments on that this image as it is the only drawing that I have and use for the assembly.

A few words about the Super Milq’s final solution.

Lower bass “A” channel: There is no change from what 3-channel. The Super Milq was the same: first order at 65Hz, two stages and 53W dissipation on 6C33C plate at 200V. The coupling cap I dropped to 4uF, did not go for electrolytic and still stay with film cap. I have deseeded do not do inductor as the filter and to say with RC chain. If I found 15H-30H indictor with VERY low DCR (20R-50R) and relatively small dimensions then I might go for it. So far the only inductor that I accept in there sonically was 11” tall and had weight of 25 pounds - I can not fit it in there.

Midbass “B” channel: Everything is straight forward in there.

The Injection “C” channel: The 30K 24-steps L-pad with RN60 resistors and Alma switch is nicely transparent. I do not high-pass this channel at line-level and as a result I have nice full-range Milq amp built-in with which I might drive anything I wish- very convent and I do not need to keep a separate full range Milq around to play other speaker’s drivers that I get from time to time. I can even to connect the second filament on the 6C33C and use more power from the channel. (The OPT has 250mA gap.) I found it very good to have this channel built-in nit to mention that it does very goods job to drive the Reds of the Injection channel. Perhaps now I might let my full-range stand alone Milq to go….

The Fundamentals “D” channel: The first singles stage channel. I use 5E6P in there. To mess-up two closed operating line-level filters was not good idea. Since the channel is low passed I did not detect any negative influence of the caps on the secondary. The channel has also attenuation at speaker level and is used ~ 6db below the MF channel.

The Mid Range “E” channel: This is a purely a shine of the whole project. I use 6E6P-DR as a driver  and  RL filter. Absolutely nothing in capacitive world (mica, teflon, polypropylene, polystyrene) was able to give me the result I got from the RL filter.  Air cap was the best among caps but RL filter was superior to air caps. It does apparently serves the purpose does not put anything in series… So, the RULE is: no capacitors in high-pass if top is wide-open. I shell write more about the selection of the filter for MF in future as it was very interesting journey.  It was exciting to learn that D, E and F channels do not need as lot current as I initially thought and I use 30mA at 200V, which dissipate moderate 6W on plate. The coke I use is Hammond 622ZA in a Co-Netick can. I have also another “alternative” very interesting output transformer for this channel to try….

The HF Range “F” channel: The 6E6P-DR at 30mA connected at tetrode. The filter is second order with Sowter 9858 Pultec MEQ-5 EQ Inductor 420/277/145/108/61/34 mH) and variable air cap. The variable inductance and capacitance allow me to write any prefer Bessel curve against any frequency I wish and I have now much more gain on in tweeter then I need. The driver is tetrode strapped to defeat the insertion loss of the 175:1 transformer and to have more then necessary gain to run the “Water Drop” Tweeter on transition slope, crossed at 30kHz-40kHz with second order. The shortcomings of the tetrodes (loosing of damping, increase of output impedance and so on…) are absolutely not applicable in my case as the ribbon tweeter care less about the damping. In fact if I have a tube with more transconductance then 6E5P-6E6P (30-35ma/V or 30.000mS in Western scale) that has >3-4W on plate and more then 2-3V of bias I would stick it in. I might use something like 7788 pentode only with more power and more bias. If you know any of them then please pitch it to me as to have too much gain on tweeter do not hurt. Do not forget that the chennal will not do “sound” but rather the “space noise”. Still, even what I have now is ~ 12dB more then I need and it is already is very good.

In the end: the layout of the things is really perfect and I do like very much how it fits. I practically (air cap is an exception) do not use any extra wire at all for signals and bias – just the natural leads of the parts. The bias attenuators, after much evaluations, I ended up to use Bourns 83 series that are a perfect balance between the geometry, wirewound multi-turn precision/reliability and sound quality.

http://www.bourns.com/pdfs/8384.pdf

It is pretty much it and not I need to render it in the final implementation….

One of the most interesting outcomes of the Milq conversion into 6 channels is the impact it had to it MF channel.  The tweeter is a separate subject, I still work on it, I know how it will be done but I did not hear the final result. I would like to report something special about fundamental channel but there is nothing special about it – it sounds as it was before. I have with the dedicated amplifier some flexibility to play with it but basically it the same. The upper bass not has own dedicate amp with 14dB superfluous gain and I can theoretically to put in there any driver I wish, not necessarily as crazy sensitive as Fane was. I might do it in future if I find more interesting sounding driver then 8M. However, the sound of MF did change quite interestingly.

If you read above then you know that my primary concern was the inappropriate harmonic changes with volume when a channel was high-passed and with opened top. After many experiments I summarized that it was related somehow to the filtering capacitor.  Not necessarily a capacitor as a “better or worst sounding capacitor” but rather a capacitor as a topological entity. A serial filtering  capacitor is not coupling of power supply capacitor and it is not charged by DC, therefore it’s dialectic is not polarized and it gets re-polarized with every electron of AC signal. It usually impacts sound statically, injecting permanent  coloration, and I never seen before that it had dynamic impacts fluctuating with volume. The Air cap was better solution it had no re-polarizing dialectic. However, the RL fitter was juts way out there. It immediately did not just answer all questions but make the question do not need to be asked.

The S2 driver with the RL filter sounds exactly how it sound without any filter at all, in fact it sounds slightly better then without any filter. Perhaps it has to do with unloading bass from inductance-challenged transformer but I did not hear any distortions what I drove the 4H MF transformer full-range. So what is so special with the sound of the new S2 channel? Here are a few factors:

1) Less “bad granularity” without loosing of the “good granularity”.

2) Way cleaner upper region, though without the “bad granularity” it might feel in beginning that it has “less HF extension”

3) The awareness of the driver about the full range or something that I call the “neighborhood watch effect”.

This last entry should be explained because it was EXACTLY what I was looking to get from my MF driver. When I listen my stand alone S2 now it reproduces above 1K but it has awareness about the whole spectra of music. It is not the bottom knew response and but rather some very long acoustic-like harmonics that make the driver to reacts upon the events that are outside of it reproducible range. If I have a bass note that shell not be auditable via the MF channel then this note of course is not auditable but the MF channel still creates a “perceptional event” that indicates that there is a hole in the music that should be filed with sound or with mental extrapolation.  The channel with capacitive filter does not do it – it masks out events from the non-reproduced regions. The driver in the way acts as a pointer that shows off what the others channels shell be and how they shell sound. Of course it is necessary do not subtract from it the S2 it’s speed, the phenomenal transient capacity, the “absolute tone” and the vintage ability “to go boiled” with tone when it necessary.  Hey, if the second channel will do the same then it looks like I will be able to play the brutal London Symphony’s valve trumpets and soprano trombones as they should be played…

I am becoming a more satisfied Pussy with my 6-chennal DSET project…

Paul S wrote:

If you have the Mercury SR2-9013 35mm master Dorati/LSO Nutcracker, that might be a treat

Hate that performance.  Dorati did not play Tchaikovsky appropriately. Each and single Dorati’s Tchaikovsky piece sound to me like a shopping in IKEA…

Paul S wrote:

Now, how to get 180 phase and 12 dB slope without a cap...

I do not think it is necessary with tweeters as the job of the tweeter is to inject nose not sound, unless the tweeter crossed too low, and in such case the selection of a ribbon tweeter was a mistake most likely.

Paul S wrote:

Who knows how many times the signal can be "re-constituted" via inductors?

I do not think that any re-constitution takes place. In a cap the signal that is in the filter zone of transparency flows still on the capacitance of the devise – it has no physical contact between input and output and the cap acts s a frequency depended series regulator. An inductor, however, acts as a shunt regulator, shorting the frequencies and the band-path of the RL filter flows directly hardwired to the load.

Last nigh I spend a couple hours to listening this thing. Since the amp is done in design phase and the final parts are still in order I can not finish it and close up – it kind of frustrating… Still, it gives opportunities to listen it knowing that it’ll be “like that” when it will be done.

I have right channel that is absolutely untouched old 3-chennal Melquiades with “old” Macondo. As the left channel I have the new 6-chennal Melq coupled with the new Macondo. Of course I feed them with the same mono signal and it gives a very good illustration where I have gone in term of Sound with the 6-chennal Melq revision.

As I said before the flexibility to write any sound I wish with 6 active channels is very appealing. Also, as I said before the top 2 channels on the new Milq/Macondo are the most interesting.

To best describe what the new MF channel does would be to imagine paintings that were painted by brushes of 8mm and 1mm. The narrower brash allow going into greater amount of details, if there is a need for it. Sonically it works very interesting as the channel goes into details not always but only when music calls to it. When it necessary it still paints the picture of Sound with the sweeping stoke of the large brash. Ah, the wonder of the 6E6P!!! What is very important to understand is that this new higher level of “resolution” has nothing to do with the idiotic illusion of hyper-resolution, escorted with harmonic castration, as the most of the nowadays manufacturers do (Kharma/Avalon/Magico and many other). Quite in contrary - the new Milq/Macondo MF channel has more dominant low order harmonic tail then the old MF channel, the tail that stays with volume. At the very same time the driver despite of the long harmonic tail runs across the dynamic range even faster then before.  Quite a positive result! What is also is very important to understand that the new higher level of “resolution” has also absolutely no Spectral-like HF dithering where the upper range hyper-structuralism is perceived by audio-Morons™ as  “quality”. The new MF channel goes up very forcefully but very pliable with grace and elegance. When the channel hits it’s limit (dynamic or response) it does not dive in convulsions but just crumbles with modishness and without calling to itself. It is in way reminds me the sonic fading of my channel armed with electromagnetic S2 driver but the new MF channel fades from the Sonic heights that the electromagnetic never ever approached.

The HF channel with the new Milq is completely different story and they are a totally new page of HF evolution compare to what I had before. I do not think I even heard that type of HF. It will be a separate set of the observations in the “Water Drop” thread when I feel I am ready.

I was playing Bach’s harpsichord concertos last nigh and I come to a very interesting observation that my Injection channel works quite different with new Super Milq then it was with old Super Milq. I was driving the old Macondo from the amp of the new Super Milq and confirmed that it was not because the differences in amplification. The new Milq/Macondo has subjectively higher dynamic range and height level of tonal saturation. That makes the colorizing-dithering effects of Injection less beneficial with the new Milq/Macondo. I still see the Injection benefits but I apply it now much more conservative then before – a very positive thing from my point of view. The good thing also is that the new Milq/Macondo’s in it’s new level of tonal saturation do not dive into the “Van Gogh colors” as my Injection Channel with Reds tends to do…

It looks like the new Super Milq/Macondo idea turns into something very interesting. I really would like see it all done and the amp is closed up and permanently installed. In reality I certainly do not know what is responsible for the positive upper range advancements in the new Milq/Macondo tandem. It might be the elimination of high capacitive 6C33C, it might be the use of the single stage amp, it might be the use of different output transformers, it might be the use the small topological changes that were made in there, it might be the use of different type of filters, it might be the use of different rectifiers and some other changes in PS section. If I were in audio business then I would catalog the changes and try to capitalize on them but since the success in audio for me is Sound in my listening room then I less care about the reasons of the result if the results are satisfying.

jessie.dazzle wrote:

While reading of your progress on the 6 channel amp, I was thinking ahead, and how I might approach the building of such amps as separate projects for each frequency band (separated physically as well). In my particular case, this would be in the interest of keeping things simple, but also to leave the ML2s to run the rest of the system while building/evolving the new amps. The idea being to remove the ML2s one frequency band at a time, and compare one addition (one frequency band) at a time against the ML2s.

Well, I understand what you are trying to do but I see some problems with this approach. I the way how you approach it you will not have optimized paths for input stage and very much might end up with 6 input cables. Alternately it might be done as 6 plug-in sections, using high quietly Tektronix knife-connectors… If so then the PS and everything else should be packaged with plug-in objectives.

If I have an aim to build an unknown DSET for myself then, knowing what I know now, I would chose for following approach:

1) Build one prototype of full-range SET channel, powering it from any old good lab power supply. If you do not have it then get on eBay for $200-$300

2) Deeside for yourself if the amplifier is worth pursue and delivers the results that I would like to see (sound, power, gain, measurements, stability, operation, parts etc…)

3) Built a power supply for whole DSET, calculating the power for 6 channels (with 30%-50% spare power) and use multiple taps (100%, 110%, 120%, 130%, and 140% on primary for plate transformers, enabling me to add channels to this supply. Do not under-estimate the thoughts about the PS to amp interface – it is VERY important decision.

4) Convert the built full-range channel into a first DSET for upperbass/lover-Mid channel powering it from the DSET power supply.

5) Order the output transformers for the rest channels and having thier dimensions begin to think how the whole DSET will be package chassis/chassiss – wise, fitters-wise, layout-wise etc…