[carlo]

Lots of people to thank before I continue (note, I didn’t get any discounts or anything from the vendors listed, they were simply helpful as all hell):

zzz, skippy101, and CaptainBubba – for being patient with me as I asked questions about everything from rectifiers to transformers to headphone jacks.

Kyle from Reference Audio Mods – took the time to give me a lot of advice on what to try in the project. Huge help in terms of technical information.

Kevin from DIY Cable – Took the time to talk caps and volume controls and answer all of my questions with the utmost courtesy.

Bob Crump, John Curl, chiggy, and all the others who’ve posted a wealth of information in Audioasylum’s archives.

The good people at http://www.triodeel.com/ for their kick ass tube data sheets.

Vertigo-1, kelly, Zanth, tanfenton, Hirsch, violetta88, and everyone else at #headphones for listening to me ramble about this **** for months.

Welborne Labs, Partsconnexion, and Allied for prompt and hassle free service.



So anyway, I picked up a practically completely populated Melos Gold Reference/Maestro board (it’s the same thing) from geom_tol and decided it was my opportunity to experiment with boutique parts and get a reference level headphone amp in the process. I started by getting the amp functional, all caps were given ample time to break in (this project has been going on for a couple of months). then I experimented with parts starting at the first resistor the signal sees through the last component, then went back and a/b’d what was close all over again (the holco/tant competition was fierce).

“Ultimate Melos” is already taken, and I think “Maestro” is a stupid name for a headphone amp (I feel that way about a lot of hifi products), and so this amp is The Maestrobator (imagine thunderclap).

kelly type disclaimer: Most of the parts were switched and a/b’d, my impressions of which are below, but it must be remembered that my opinions are just what I’ve heard with the parts in this board (feasable that the “sound” of a part would be different with another design). If you don’t think resistors and caps make a difference fine; I didn’t think resistors would until I tried it, but should you feel tempted to post something stating this whole thing is a crock, don't - lets just agree to disagree. My goal is to get the most out of the circuit as possible, not to necessarily “beat” other amps (I have no idea if it does or not). I’m doing a write up of the amp that’ll be more formal and go into deeper detail (it’ll be available on the net free for all), but for now I’m just giddy and sharing

On to the amp:

Modifications Parts List:

Off Board:
Cardas GRFA (gold/rhodium) RCA jacks
Neutrik ¼” locking headphone jack (pictured above a custom acoustically damped holder disguised as a cigarette box)
DACT CT-2 stepped attenuator
Hammond 160VA toroidal transformer
Corcom EMI Filter (note: disconnected in pic since that one went screwy, now using another one)
Corcom isolation transformer
Silver plated fuse and holder
Generic IEC socket (whatever Welborne Labs sells)
All power wiring is Harmonic Tech 14ga
All signal wiring is Cardas 22ga stranded (also tried Tara Labs RSC)
Added warm-up switch (keeping voltage running to the tube plates and output FETs but not the tube grid).
Omitted: mute relay, balance (not needed, didn’t want it in the signal path).

Close up of amp turned on sorry, couldn’t disable the flash for the full effect of tube glow. those are unused tants at the top of the pic by the way):

Power Supply
Low Voltage supply:
Blackgate NX Series (the red caps at the bottom of the board) 1000uf 25v replace the stock 470uf 16v Xicons.
Two Solens are used as shunt caps (didn’t like them in the signal path, more on that later).
I used Lynk for high wattage resistors (the green things to the bottom left) since I didn’t know how ungodly large they are and the local Radioshack didn’t have the values I needed.
Holco resistors (note: in the pics there’s two Rodersteins by the Blackgates, I was experimenting to see if their magnetic end caps added noise).
All diodes are Harris FREDs.

High Voltage Supply:
Blackgate Standard replaces stock Jamicon of same value (100uf/160v).
Harris ultra-fast soft recovery diodes.
Resistors from diodes to transistor gates are Vishay/Dale.
ICW replaces stock Dayton cap.
Parallel resistors to the tube plates are Rikens.

Signal Path (note all comparisons were made against each other in a/b mode, but not double blind):

Left Channel Mod: among the many convoluted traces on the board the left input travels through the power supply section. Should you have noise issues in that channel (I did) jumping the trace and rerouting to a less noisy part of the board eliminates the problem.

Caps (incidentally, you don't need the blocking cap unless your source has dc issues. I removed mine):
Auricaps – I found that they offer plenty of detail, separation, and low level information without harshness or frequency anomalies. Decay like a dream and are big and open, with them in place I get the closest to a soundstage with headphones as I’ve had in my setup.
Also tried: Hovlands Coupling Caps – way too hard in upper frequency with the amp, but plenty of bass and (maybe exaggerated) microdynamics. Lots of slam and speed (Sennheiser owners take note), but didn’t sound balanced enough through the spectrum for my tastes. Currently experimenting with it in my preamp (where I love it).
Solens FastCap– Mushy and veiled. Fastcaps my ass. Currently used in PS.
ICW – nice and smooth, but loses a lot of inner detail compared to the Auricap. Excellent cap for the money (couple of bucks each). Currently in high voltage supply.

Resistors (note: I didn’t find resistors made a dramatic difference unless its leading into or out of an Auricap or Hovland, and even then its not earth shattering)
Tantalum – sweet and juicy, my favorite signal path resistor. No detectable roll off and sounds fairly open. Usually used for all signal path resistors in tube input stage (note: in the pic they’re stock, I was experimenting and didn’t have time to de/resolder before we took pics).
Caddock MK132 – very nice extension, didn’t hear any veiling. To my ears sounds dry in the midrange and boring at the input, but performs well after the tube output and leading to the base of the output FETs (where they reside in the pic and now).
Holco Old Stock (now discontinued) – A lot like the Tants but with upper frequency roll off. Thick midrange that mates well with the Hovland to my ears. Used as stated above as well as the JFET base and emitter between the tube grids. Call me shallow but i think they look cool (they're the black resistors). Nice and cheap too.
Resista/Roederstein – nice and linear, very nice (cheap) resistor but doesn’t sound as open as the Tants or Holcos to me. Used for closed loop in the bias circuit.
Vishay/Dale – sounds hard to me at the upper registers and a tad veiled, but otherwise gives up little to the Roederstein. Didn’t feel it was the best choice for the signal path but am using it into and out of the bias circuit. I’m looking to try others at those spots however.

Other:
Gold plated, ceramic body tube sockets – they’re uneven because I underestimated the size of the replacement caps. I’m not sure what I’ll do about that until I decide on a chassis.
Tubes used are Great Britain Gold Pinned Mullard 7308 and 6922, 1964 NOS Phillips “Amperex factory” e88cc SQ (pictured) and Amperex 7308 PQ. All tubes with the exception of the Amperex 7308 were NOS until they met The Maestrobator.


How does it sound?

Good

Backgrounds are midnight black, bass response up the ying yang, and this amp is fast. While I feel like I know the amp (and what its capable of) as well as anyone I also don’t remember what the stock board sounds like, there’s just been too many changes (my resoldering the stock resistors and caps in the input stage was an attempt to compare them to the Tants). However the current incarnation of the amp is giving me almost all I’m asking for – there’s a hardness on horns and in the upper registers that bugs me and I’m trying to focus on, but I need to try different headphones before I attribute it to the amp.

Realistically someone else needs to judge its sound, I'm way too close to the project. I think the current combination is the absolute best I can get though.

Can anyone willing to take the time do the mods?

Yes! Seriously, following traces is fun. I need to check if anyone has the intellectual rights to the schematic, if not I’ll post them in this thread. In the meantime please feel free ask questions and I'll chip my two cents in. One warning: my board had absolutely terrible soldering - should you take on the project invest in a good desolderer and practice on an old board (desoldering is harder than soldering to me), it's worth the saved aggrivation.

Is the project done?

No. Edwin has been kind enough to agree to get together when we both have time so I can try his r10s and he can try the amp.
Zanth may be sending his HP-1000s so I can do more experimentation in the search for neutrality. I still need to add a preamp section, but there's no rush since I'm only interested in this project as a headphone amp and my next DIY project will probably be a transformer based volume control. I will add input selection (with the lowest impact on sound possible) eventually, as well as balanced ins and outs (which the Gold Reference board has provisions for, although its not fully balanced). And of course I'm still debating case.

Why didn't you double blind test?

a)No one I know is willing to desolder resistors one at a time, try different resistors, but in new resistors, and then try the first set of resistors. (I'm the only person I know who's psychotic enough to do it).

b)I trust my ears.

bonus pics:

The Maestrobator (thunderclap) gets in your face:

The Maestrobator chillin and thinking about space and time:

The Maestrobator planing world domination:

And a picture of me with hot chicks (the true yield of DIY audio). They keep leaving and going to Gilmore's house though. Bitches.

yadda yadda
carlo.

(sorry for the super long post.)

note: this is not an ad. if i decide to sell anything i'll do the mall-fi thing. please don't ask ad type questions and expect ad like answers

EDITS: The Lynks are obviously to the bottom right of the closeup pic. "Double Blind" should be "blind".

 

[erix]

Carlo,

You beat me to it and I'm glad! I was thinking I needed to get to my Melos board soon - perhaps when the snow flys!

Great Job!

ok,
erix

 

[aeberbach]

Nice job Carlo. I've got a board too but a lot less complete than yours. Still, no damage to to PCB itself and definitely savable.

If you happen to have made a listing of parts that corresponds to the numbers on the PCB it would be of immeasurable help...

 

[geom_tol]

Excellent work Carlo!!

And great pics.

 

[carlo]

erix,

yeah, but looking at your past work you're going to be tough competition for nicest melos case

aeberbach,

unfortunately i don't have a scanner or a digital camera (i borrowed the one i took pics with), but there is a stuffing diagram and a parts list in the paper documentation. i have no clue if posting that stuff pose problems in terms of intellectual rights and property, but i don't want to do it unless i know its not going to be a headache down the line. i'm hoping there's others here who can pass it along to you (*cough*) behind the scenes where the mods don't have reason to freak out in protection of head-fi (*cough, cough*).

the tricky thing with the documentation is that the schematic values don't match the parts list, but its not a big deal since the parts list and the board stuffing diagram match and both are dated after the schematic (by a three year difference no less) and following that my stock board worked just dandy.

now to mod the amp is a different story - its a case of following the (often times insane) traces of the board to see where items go... a process that took me weeks; the only thing that helps here is the schematic, its not surprising to see a signal trace go from the left side of the board to the right and then head towards the upper left of the board - the board stuffing diagram is absolutely useless and there's no way to accomplish it without actually following the traces.

geom,

thanks

 

[kelly]

Carlo
Very cool writeup. I may have to read it a few times to digest it.

I've been anticipating this for a long time and it's good to finally see it.

Tweak the amp to the HP-1000, not to the R10. And if you want the amp to live up to its name, get Diane to wear the HP-1000 and pet the Melos gently.

 

[carlo]

some other info i remembered:
should the schematics come into your possesion, there's a few quirks that may be a bit tricky. the resistors into the tube grid and out of the tube are 2k on the board, not 1k. the resistors off the heater supply are 38 or 39 ohms on the board (i forget which). when working with the input stage notice that the signal goes to the far pair of the 6922/7308's dual tiodes. if you look for the stock input resistors on The Maestrobator closeup (they're all over the place, but they're the light blue resistors. keep in mind the four per channel are in series to each other when contemplating the insanity of these traces) it goes through two of the 470k resistors under the heater and through one 2k resistor, you don't really have anything going on with the opposite triode (assuming you're running single ended input and output for the preamp). note: i'm talking about the 6922 as a dual triode (individual) tube.

there's been some speculation that the headphone amplification and the preamp amplification stages are the same on the amp. they're not. the four yellow topped caps in the rear are the output caps off the preamp, basically everything north of the top ground trace is preamp stuff. the headphone stage is a fet connected to another fet emitter to gate (the name of which escapes me at the moment).

this is off the top of my head without the schematic btw, so i may be adding and revising information as it comes to me. once again, this'll make more sense once you work the traces and if you have the intent to mod (otherwise its just putting together a puzzle, you don't need the schematic just the board stuffing diagram)

kelly,

thanks man. the hp-1000 and its history with the gold reference/maestro's predecessor (SHA-1) obviously makes it the headphone to try when modifying the amp. your advice will be taken.

warning: excessive maestrobation has been linked to vision impairment.

 

[erix]

carlo,

I agree on the mystery routing on the board. I've spent many (enjoyable) hours just following traces on the board and comparing them to the schematics we have.

There are a few things that puzzle me on the board. The first is the four sections with MOSFETS in the high voltage supply. I *think* it's some kind of stabilzation or timing thing for the tubes. I see you put big 'ol Vishays there - why?

A funny thing about those MOSFETS - they are the same ones used in the big heatsinks - they just sanded the numbers off...

I continue to get lost around the resistors at the tubes - the values spec out but I just don't see the schematic when I follow them - I guess it'll come to me!

Your powersupply - using a toroid for the HV side and another for the low voltage/filiments? or are you running everything off the one tranny (through the isolation tranny)?

Looking forward to reading more about these boardds coming to life! Who here has one?

ok,
erix

 

[carlo]

erix,

zzz and i have been staring at the same part of the high voltage supply and we still don't get it. since it leads to the tube plate load i'm guessing it has something to do with stabilization, but its over my head. i just checked the part numbers: i'll be dammed

i tried the vishay/dales there because i was experimenting with what would happen if i played with resistors in that part of the circuit. i liked that the vishay/dale don't have magnetic end caps , were cheap enough to try, and when i tested them they went four values deep in matching (very impressed), but if there's a difference from stock i didn't hear it.

yeah, the resistor trace at the tube input and output is confusing. the 2k (1k on the schematic) resistor that looks like its from the blocking caps to the output gate is really before the caps (from the tubes). these are the caddocks (they look like gray boxes) south of the tubes in the close up pic, may help a bit. after it clicks you'll think "why?"

power supply runs like this:
iec->slow blow fuse->1:1 tranny->switch (double pole double throw, jump one of the poles top to bottom and use it for the high voltage "warm up" switch, make one end normal "on")->

(high voltage supply)->emi filter which also serves as an inductive load->board
(low voltage supply)->toroidal->board

note that the high voltage supply takes a 117v input. all this stuff is planned to go into a seperate chassis and shielded for rfi by the way (which will sit by the board chassis). overkill sure, but two problems i had were 1) i couldn't find a toroidal that had both of the values needed and 2) i don't want to have to use a line conditioner/whatever for the amp, just plug it straight in and forget about it. easier way to go about it is to just get two transformers or get a custom transformer (good luck).

interested as well about other builders' experiences,
carlo.

edited caddock placement due to mental drop out.

 

[geom_tol]

The output transistor pairs are called Darlingtons.

I've also wondered about the mosfets in the power supply. Maybe it's a common high voltage ps technique we don't know about?

The layout of the board is so sad, and I have often wondered what they were thinking when they did it. Probably didn't spend much time at all laying it out.

The older amp boards are slightly better, they don't have the preamp outputs so there's less components to muddle it up.

 

[Budgie]

(Sorry, finger slipped.)

 

[Budgie]

Very cool, Carlo. I hope you will keep us posted and get pics when you case the board.
(I had a similer exesperience with the Solens. I don't think I will ever use them for anything but power supplies again.)

 

[eric343]

VERY VERY cool, Carlo!

I much appreciate your reviews of the resistors and capacitors, also. That will be seriously helpful later on...

And it sounds like the board was designed by me in a previous life

 

[geom_tol]

Carlo,
I think the transformer used is a common one. Get a transformer with two 120 V primaries and two 12 V secondaries. Hook up one of the primaries to the ac power. The second primary can now be used for the high power. Wire the two secondaries in series to give 24v.
I think melos used a Avel Lindberg d4011, I have one of them and the wires are all the correct colors comapred to the wiring schematics.

 

[carlo]

budgie,
will do man. for what its worth i think the icw absolutely kills the solens for not much more money (icw is a non "audiophile" cap company who's biigest claim to fame at the moment is b&w's use of them in their top of the line crossovers). thick wire and huge though, so hard to manage unless you're designing around them.

eric343,
thanks for the kind words man, i'd love to hear cap comparisons with the meta

geom,
man, i spent a lot of time confused about that. ultimately i decided to follow hammond's connecton faq (http://www.hammondmfg.com/5CHook.htm), i went with both primaries paralleled (since its 120v and not 240v) with the phase relationship in mind.

best,
carlo.