Babelfish XJ , or JX .. or whatever (Aleph X servo for Greedy Boyz)
Pretty mach pasting a text and just few pictures from my blog post ( still not fully finished) :
It started not by plan , but by chance .. several (or even more than several) months ago , I tried to help one of our Greedy Boyz at Diyaudio.com/Pass Labs subforum , to improve things with his Aleph X monoblocks.
Problem was simple but not so simple . in first incarnation , amp was sounding divine , but it was unreliable and eventually it gave up in flames . or at least with some gray smoke puffing out .
As usually . ZM at work :
As always with these beasts , culprit was in badly matched output mosfets and generally somewhat lower quality of build execution , at least that's how I understood .
Later , when our friend Robert found another guy willing to make things better (Robert himself being much more versatile in other things than electronics) , amps were repaired/made better but magic was gone - they sounded thinner , lifeless in comparison to starting iteration .
Anyway , first I tried in thread to help Robert with some advices , later we proceeded with e-mail correspondence , made some progress (to my satisfaction , I had opportunity to show how small changes in strategically important positions are much more important than type of wire , Preputium-made of resistors and caps , thus making much more prominent changes sound wise )...... but all our efforts concluded somewhat badly , with amps again producing gray smoke puff ...... ya know - pretty common thing when playing Bad Telephone Chain game ......
Even before that , Robert wrote to me that he is already preparing to make another pair , this time even bigger , of Aleph X ; he's saying that Aleph X is his preferred type of amps , after being in opportunity to hear older and newer Pass Labs amplifiers , including .8 series.
He eventually asked me to make them for him , he providing already bought cases and whatnot .....
My comment was - first - I'm not making knockouts of PL amps (Aleph X was PL commercial model , even if disguised with XA200 title) and second - him living in Poland (EU) , Mighty Moi living in Serbia (non EU) , circulation of parts and later finished amps looks extremely complicated and expensive.
During next few weeks , we cross-checked our watches and came to satisfactory arrangement - I'm going to make Babelfish XJ (JX)- Jazz standard Aleph X iteration with few important twists , so hopefully worth trying and making. We also found least cumbersome solution for transporting goods between us.
Now ..... little history.
Almost 16 years ago , Gray Rolins came to clever idea (alone ? bugging Papa in private mail, as we all do more or less?) to bridge two regular Aleph channels ....... well , read by your self exact story HERE .
As far as I remember (origins), properly visible schematic is this one :
Several iterations of amp were made , in various sizes and complexity , several threads to my knowledge most comprehensive being this one , also extremely worth mentioning is Erics web page , dedicated to his Aleph X incarnation.
There were few major problems why , even if Aleph X was probably most massive project , considering number of pcbs made in Group Buys , there was even tinier number of finished amps lets see:
-Human Nature Problem One -people are having troubles to break one whole in smaller logical modules , so circuit understanding seems unreachable;
-Human Nature Problem Two-ambition project is scalable starting big is much more troublesome than starting small to gain experience and knowledge, then building from that; that way is , in the end , even cheaper . then the other way (ending without any amp)
-Technical Problem One too many mosfets , lousy matched;
-Technical Problem Two not enough Aluminum or even not enough (enameled) Copper
For ones , able to step or jump above (around ) mentioned obstacles, there was ( or they still are) few more down the road
Amp being bridged , means both output posts floating (being and not being referenced to GND) so we are having two different DC Offsets to take care of;
First one (Boyz calling it Relative Offset, which I never remember) is most important, being DC Offset between Pos and Neg output of amp, exactly where we usually tie our speaker cable ends. Lets just say that any Aleph X is pretty nicely behaving there, input LTP stage besides serving other important things, also serving as Servo for exactly that .
Second one (DC Offset) is much more problematic to obtain at right value and to maintain- DC voltage between our floating speaker posts and GND (Boyz calling it Absolute Offset , which I also never remember) . Simply put if both outputs are at approx. zero (GND) potential , they can equally swing from zero to rails (positive and negative) thus our output signal is having max. amplitude equal in both directions ,so possible clipping is symmetrical.
That one , and few other problems (maybe smaller , maybe not), read at Erics page. I dont have intention to cover even small part of things already covered by him . why doing things twice , when you mustnt.
So , I had intention to try/verify few small ideas I had/borrow/stole during years, applicable to Aleph X circuit, so being enough different and possible to call it Babelfish iteration.
-Trivial 1 make it with JFet LTP input stage .
-Trivial 2 cascode that JFet LTP input stage . not so much because of Herr Mueller , but to keep them cooler, later possible to cover broader range of rail voltages.
-Little Less Trivial 1- put emiter follower in between input LTP and gargantuan bank of output mosfets (later about why gargantuan).
-Little Less Trivial 2 cascode that emiter follower .. just for heck of it
-Whoopsie Daisy 1 address issue of Absolute offset in simplest possible way
-Whoopsie Daisy 2 address useful arrangement of easy way to influence THD spectra .
Wheel is turning . Robert sent cases to me , I was LTSpice-ing like mad , trying to implement all details ; I did tried them one by one in vivo , but never combined.
So , after countless hours of PC time (yes , Im slow) , here is LTSpice schematic , intended for first prototype step ; Big Bada$$ mosfet banks omitted for clarity , each quadrant of 10 parallel mosfets shown as single one :
Then Ikebana prototype , measurements and some tweaking of values ; it passed , so it was time for pcb work.
Robert insisted on some fancy resistors , in most important positions ,so I made that – placement for both usual metal films or fancy Shinkoh/Kiwame.
Front End Eagle schematic and Front End Eagle pcb screenshot:
Output Stage Eagle schematic and Output Stage Eagle pcb screenshot:
few more lines from blog post :
Servo is doing its job , both relative and absolute offset practically rock steady from very first moment of powering up, same as Iq .
Changes from stone cold to temperature equilibrium are approx. in 5% range, so I didnt even bother to write exact figures down.
Will post tomorrow more tech data/details , same as explanation of arrangement used for relatively easy THD Spectra shaping . probably few more pics of amps with finally closed top lids .
Did I said that theyre Gorgeously sounding ?
I'm lazy to post more pics here , so just go to my blog in case that you're having too much time on your hands .......
So , few more sentences about really most important details of entire build will follow ....
In a meantime , just observe red marked positions on schm in post #3
Cute DAC :devilr:
By my math, 10 MOSfets per QUADrant = 40 mosfets per channel.
Congrats on a worthy V/I converter :)
Mighty Moi practicing small ..... one day to make proper MC stage
hotlink test :
hotlink test 2 (my blog did have hotlink block , now switched off)
- how to tighten mosfet without twisting on goo :clown:
I think I found a photo from testing the thermal load....
:D :D :D
|All times are GMT. The time now is 10:49 PM.|
Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio