|Parts Where to get, and how to make the best bits. PCB's, caps, transformers, etc.|
Please consider donating to help us continue to serve you.
Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
||Thread Tools||Search this Thread|
|14th June 2008, 09:41 AM||#1|
X-Fi bugs? I have solution
In spring 2006 I finally get my hands on X-Fi Fatal1ty card. The reason for upgrade from my Audigy2zs was mostly the pause bug, BF2 support (I started play BF2 a lot) and the X-ram thing, that provide 64MB soundbuffer that should dramaticaly reduce PCI load - hence give me some speed-up Or so I thought. The first nasty surprise when I actually get the card in my hand was this:
Not only the unnecessary long legs, but mainly the capacitors type - Jamicons. Jamicon cap's are bad, they are known to fail - eg. over time lost most of their specs.
At first, everything work great, except some weird noises when BF2 is loading sounds. (maybe the card did not like OGM playback when there is going writing samples to the X-ram? Or my A9R480 Sapphire Grouper mobo is up to blame?) However after like 2 months are the pause bug back.
When I say pause bug, I mean this. Using any player, play a AC3 5.1 movie while using 5.1 speakers on analog connection. Decoding filter has to be AC3 filter and it has to use 24bit output to generate enough data to trigger the failure. Now randomly pause and unpause the playback. Sooner ot later the channels start play from wrong speakers and/or there is terrible noise from all speakers comming after the unpause. This is a pause bug. I had it on Audigy (1), Audigy2zs and after two months on X-Fi as well.
It was obvious that the X-Fi is need to be recapped and the Jamicons changed to good, preferably audio caps - but NOT for all positions. The catch is, that Creative did use one capacitor type for all positions - eg. for the voltage filtering as well, as for the audio output. That is nonsense at best. Audio and computers capacitors don't mix!
So, they simply put Jamicons everywhere. Not something that I would expect from 260$ pricetag (spring 2006, mind you) product :hitwall:
Now why I started this thread.
Well, to begin with, some weeks ago (to 14. 8. 2006) I started to play Mafia - a original Czech game of the 1930 NewYork organized crime. You simply work your way up in the organization at prohibition times, drive old cars, steal, murder and finally seek salvation from the hands of cops... During play it sometimes in these briefing talking scenes is seems to starting to lost some samples. I thought that the Czech programmers engine simply could have some quirks with 64MB of sample buffer on X-Fi Fatal1ty. After all, as the game release there was not such a thing as X-Fi, right? However the more I play and the closer to the end I was, the more often it happening. And it kinda suxx to go on mission and not know the briefing, right? However as I finished the game, the problems are gone, so...
So then I started play Prey And when shortly in game in one spot appears more sounds at once (your girl scream, mechanic move, buttons, alien...), the whole computer freeze. I remember the Jamicons and quickly pull the X-Fi out and replaced the big Jamicon (originaly a 220uF 16V) with Panny FM 470uF 16V as well, as two 100uF Jamicons with 120uF Panny FM ones.
I finished whole Prey in few days of hard gaming w/o crash or any problem. I went back to Mafia and - whoa, samples aren't lost anymore again. My apology, Czech programmers. I did not trusted in yours skils... :redface: In adition, the pause bug is - GONE!
Pause bug - long story:
So I added the frontpannel later (moth or so), but in like a week, problems and pausebug are there again. Frontpannel is full of Jamicons too, and it obviously accelerated the dying of Jamicons on card or so. I pulled it out, yet the pause bug is still there. No crashing, luckily.
I would like to hear from X-Fi and perhaps also Audigy2(zs) users about their experience and possibly about their sound bugs. There is so many people on Creative forum that claim to have similar issues as me, so... I wonder in how many cases this could be avoided just by using good caps and not Jamicon crap...
To completely recap X-Fi you need:
Remove and short 16x 22uF 16V Jamicon decoupling caps - C23, C50, C76, C77, C28, C55, C85, C83, C61, C32, C62, C26, C67, C36, C34 and C68. http://img120.imageshack.us/img120/1...gcapsgoan9.jpg
1x Samxon GC 1000uF 6.3V (d8) Big Pope
2x Samxon GD 470uF 6.3V (d6.3) Big Pope
17x Panny FM 150uF 6.3V (d5) P12917-ND Digi-key
1x Panny FM 100uF 10V (d5) P12919-ND Digi-key
2x Panny FM 68uF 16V (d5) P12921-ND Digi-key
4x Panny FM 47uF 25V (d5) P12923-ND Digi-key
6x Black Gate N 4.7uF 50V (d5) - d6.3 can fit Sonic craft
2x Black Gate C 4.7uF 50V (d5) - d6.3 can fit Sonic craft
(you can ship the Black Gates and only remove the original caps off the card, if you are not plan to use the 10pin Creative connector and the AUX-IN connector near it)
3x 100uF 16V d6.3 (Jamicon) - P12922-ND
5x 47uf 16V d5 (Teapo) - P11196-ND
35x 22uF 16V d4 (Jamicon) - P11213-ND
5x 10uF 16V d4 (Jamicon) - P11212-ND
11x 4.7uF 50V d4 (Jamicon) - P10315-ND
3x 1uF 50V d4 (Wincap) - P10312-ND
(digi-key.com product numbers used)
|14th June 2008, 09:43 AM||#2|
This is a picture of X-Fi Fatal1ty by user Raggingbone that continue to have crashes and stuff with the card:
No wonder when the Jamicon for main X-Fi chip is alredy leaking!
(X-Fi Fatal1ty Sanyo Os-con SEPC 820uF 2,5V for main chip voltage filtering used)
This attempt not helped me to get rid of the pause bug - eg. pausing and unpausing movie with any player, using AC3 filter 24bit output cause switched channels or terrible noise after unpause.
Bridging the polymer with 10uF SMD ceramic helped a bit more, but it is still like 50/50 chance to get a pause bug there. Panny FM 470uF with 4,7uF ceramic helped me much more before, damn.
Next try - 1000uF Samxon GC bridged with 4.7uF CMD ceramic (Tayo Yuden).
Nah, next try - try to study how to voltage regulation for the main X-Fi chip is done in the first place
So, the volage regulation for the main X-Fi chip is made by TI PS54352 chip.
The recommended use is this:
From this is quite obvious, that Creative design is very bellow specs one and for example - the imput filtering elyte capacity should be at least 100uF and as close to the chip a 10uF (at least) ceramics low-ESR cap. The Creative imput filtering elyte are 22uF Jamicon (!) and I did not yet measure the ceramic one, but I fear it is not "at least" 10uF as it should be.
Measured 15uF. I should desolder it, I think I measure something else with it also...
Same for the output. I see bulk capacity there, a 330uF elyt, 100uF another low esr cap and 0,1uF ceramics. Creative used a 220uF Jamicon witch is not low ESR cap in any way, shape or form...
So now it is clear, why I get the best results with mediocre 470uF Panny FM 16V cap (mediocre because using 16V cap on 1.25V a "bit" soften it's specs) with bridged 4,7uF ceramics are best and why the ultra-low-ESR polymer was not a great there.
My fault. Now I also going to replace the imput filtering "bulk" capacity - by Creative 22uF 16V Jamicon - by me a 150uF 6.3V Nichicon HE :dev: Yet the usage of 470uF Panny FM bridged with 0,1uF ceramic cap is not so great also. I still having the "pause bug"...
Considering what might cause the "pause bug" I do wonder, how well Creative followed up the DAC recommended connections:
PS. I just replaced the there output opamps from the old useless (they was used even in Audigy (1) !!!) MC4558C ( http://www.alldatasheet.com/datashee.../MC4558CD.html ) to the praised LM4562 ( http://www.alldatasheet.com/datashee...SC/LM4562.html ) opamps. It was kinda easy, however the changes are minimal. You cannot hear any difference in mp3 bellow 320kBi, that is for sure. When we talking about DTS 768kBi, well, ten there finally is some difference But not so major. It is obvious that complete recap of the Jamicon crap caps has to be done as well, as the audiojacks has to be gold-plated for the subwoofer connection, the whole subwoofer has to be recapped and possibly modified also, so there is a notable difference.
Ask the Nartional Semiconductor for there sample pieces of the LM4562MA in the SOIC NARROW version for free there:
And you can replace the there 4558 "things" used for the output channels :dev: There is also another opamps for mic in, line in and some more down there... So maybe a 6 opamps is not a bad idea.
PS2. parts list for SB0460 - X-FI Fatal1ty:
parts list for SB0550 - X-FI Elite Pro:
...a good start
A few interesting (but not mine, Bichi's work) X-Fi mod pictures:
|14th June 2008, 09:44 AM||#3|
Middle stage of my experiments - Panny FC 22uF 25V used, before simple wires are used:
Panny FC 22uF 25V instad of Jamicons 22uF 16V - not bad :xd: Finally the Opamps exchange produce hearing difference. W/O caps exchange = no difference! But only wires are better - give better audio details :dev:
(and noticably stronger bass line, witch is somewhat extreme sometimes...)
So, my mods on X-Fi so far are:
C177 - 220uF 16V Jamicon, main chip voltage (1.2V) filtering:
470uF 16V Panny FM d8 - good, but still pause big
3300uF 10V Samxon RS d10 - card did not work at all (!!!)
1000uF 6.3V Samxon GC - good, pause bug reduced
C172 - 22uF 16V Jamicon, main chip regultor voltage imput (5V) filtering -> 150uF 6.3V Nichicon HE
DAC - Cirrus Logic CS4382 - http://www.cirrus.com/en/pubs/proDat.../CS4382_F1.pdf
C16 - 100uF 16V Jamicon, DAC voltage (3.3V) filtering -> 470uF 6.3V Samxon GD d6.3
C107 - 47uF 16V Jamicon, DAC voltage (3.3V) filtering -> 47uF 25V Panny FM d5
C91 - 10uF 16V Jamicon, DAC voltage (5?V) filtering -> 150uF 6.3V Nichicon HE d5
C119 - 10uF 16V Jamicon, DAC voltage (5?V) filtering -> 150uF 6.3V Nichicon HE d5
C46 - 100uF 16V Jamicon, opamps voltage (5V) filtering -> 470uF 6.3V Samxon GD d6.3
Unknown - at back of the card
Cxx - 47uF 16V Jamicon -> 47uF 25V Panny FM d5
C186 - 47uF 16V Jamicon -> 47uF 25V Panny FM d5
C205 - 22uF 16V Jamicon -> 22uF 25V Panny FC d5
New opamps LM4562 (old are NJM4556)
Voltage regulator 1.2V (CA20K1 - main chip) - PS54352
input (5V) cap filter - C172, output cap filter - C177
Voltage regulator 3.3V (???) - AMS1117
Voltage regulator 5V (opamps + DAC VA) - UA78M05C
input (12V) cap filter -
all the 22uF 16V Jamicons shorted as recommended:
Too radical? As intermediate step you can try replace these Jamicons with Panny FC 22uF 25V caps - and hear the difference. In fact, there is almost no difference after opamps exchange, but huge after caps exchange WITH already exchanged opamps...!
So far I did:
Exchanged 4 opamps to the LM4562 ones
Removed 16 pcs of coupling caps C23, C50, C76, C77, C28, C55, C85, C83, C61, C32, C62, C26, C67, C36, C34 and C68 and shorted them all with wire to get rid of the cursed Jamicons and improve the quality of sound output
Replaced C177 to Samxon GC 1000uF 6.3V
Replaced C172 to Nichicon HE 150uF 6.3V
Replaced C16 to Samxon GD 470uF 6.3V
Replaced C46 to Samxon GD 470uF 6.3V
Replaced C91 to Nichicon HE 150uF 6.3V
Replaced C119 to Nichicon HE 150uF 6.3V
Replaced C107 to Panasonic FM 47uF 25V
Replaced C186 to Panasonic FM 47uF 25V
Replaced Cxxx to Panasonic FM 47uF 25V
Interesting X-Fi Elite Pro mods:
X-Fi Fatal1ty 12V opamps mod:
Most important picture ever - X-Fi and voltages
Hard work, but there we go:
In picture it looks this way - green are caps that has audio on them with stereo output, WinAMP playing...
C177 (1.23V) - 220uF 16V Jamicon -> 1000uF 6.3V Samxon GC C16 (0.7V) - 100uF 16V Jamicon -> 470uF 6.3V Samxon GD C46 (5V) - 100uF 16V Jamicon -> 470uF 6.3V Samxon GD C107 (4.7V) - 47uF 16V Jamicon -> 150uF 6.3V Panny FM C186 (5V) - 47uF 16V Jamicon -> 47uF 25V Panny FM C205 (0V) - 47uF 16V Jamicon -> 47uF 25V Panny FM C20 (2.47V) - 22uF 16V Jamicon -> 150uF 6.3V Panny FM C33 (3.3V) - 22uF 16V Jamicon -> 150uF 6.3V Panny FM C56 (-5V) - 22uF 16V Jamicon -> 150uF 6.3V Panny FM C72 (-5V) - 22uF 16V Jamicon -> 150uF 6.3V Panny FM C74 (12V) - 22uF 25V Jamicon -> 68uF 16V Panny FM C75 (-12V) - 22uF 25V Jamicon -> 68uF 16V Panny FM C101 (5V) - 22uF 16V Jamicon -> 150uF 6.3V Panny FM C114 (5V) - 22uF 16V Jamicon -> 150uF 6.3V Panny FM C124 (3.3V) - 22uF 16V Jamicon -> 150uF 6.3V Panny FM C172 (5V) - 22uF 16V Jamicon -> 150uF 6.3V Panny FM C206 (0V) - 22uF 16V Jamicon -> 47uF 25V Panny FM C209 (0V) - 22uF 25V Jamicon -> 47uF 25V Panny FM C27 (5V) - 10uF 16V Jamicon -> 150uF 6.3V Panny FM C43 (8.8V) - 10uF 16V Jamicon -> 100uF 10V Panny FM C91 (5V) - 10uF 16V Jamicon -> 150uF 6.3V Panny FM C108 (2.16V) - 10uF 16V Jamicon -> 150uF 6.3V Panny FM C115 (2.4V) - 10uF 16V Jamicon -> 150uF 6.3V Panny FM C119 (3.3V) - 10uF 16V Jamicon -> 150uF 6.3V Panny FM C123 (2.3V) - 10uF 16V Jamicon -> 150uF 6.3V Panny FM C136 (5V) - 10uF 16V Jamicon -> 150uF 6.3V Panny FM C151 (3.3V) - 10uF 16V Jamicon -> 150uF 6.3V Panny FM C18 (2.4V aud) - 4.7uF 16V Jamicon BIPOLAR -> Black Gate N 4.7uF 50V C19 (0V aud) - 4.7uF 16V Jamicon BIPOLAR -> Black Gate N 4.7uF 50V C21 (2.39 aud) - 4.7uF 16V Jamicon BIPOLAR -> Black Gate N 4.7uF 50V C22 (2.47) - 4.7uF 16V Jamicon BIPOLAR -> Black Gate N 4.7uF 50V C48 (2.26V) - 4.7uF 16V Jamicon BIPOLAR -> Black Gate N 4.7uF 50V C49 (2.26V) - 4.7uF 16V Jamicon BIPOLAR -> Black Gate N 4.7uF 50V C102 (2.4V aud) - 4.7uF 50V Jamicon -> Black Gate C 4.7uF 50V C104 (2.4V aud) - 4.7uF 50V Jamicon -> Black Gate C 4.7uF 50V Remove and short 16x 22uF 16V Jamicon decoupling caps - C23, C50, C76, C77, C28, C55, C85, C83, C61, C32, C62, C26, C67, C36, C34 and C68. It is possible to remove all the marked with green audio caps C18, C19, C21, C22, C48, C49, C102 and C104 if you did not use the AUX IN analog 2 channel CD in and the 10 pins Creative connector, for witch the 6 pcs of the bipolar caps are used to separate the AC3 6 channels signal (5.1). 35 caps total ------------- 1x Samxon GC 1000uF 6.3V (d8) 2x Samxon GD 470uF 6.3V (d6.3) 1x Panny FM 100uF 10V (d5) P12919-ND 4x Panny FM 47uF 25V (d5) P12923-ND 2x Panny FM 68uF 16V (d5) P12921-ND 17x Panny FM 150uF 6.3V (d5) P12917-ND 6x Black Gate N 4.7uF 50V (d5) - d6.3 can fit 2x Black Gate C 4.7uF 50V (d5) - d6.3 can fit
|14th June 2008, 09:46 AM||#4|
Why not use tantalum caps
Read conclusion about noise in tantalum cap
The most important sources of fluctuation consist in regenerative microbreaks, fluctuation of polarisation and mechanical strain. The frequency dependence of noise spectral density in mHz region gives information on slow irreversible processes of tantalum pentoxide crystal-isation and oxide reduction. The self-healing process can improve sample quality due to leak-age current and noise reduction. :redface:
Low Noise Balanced Microphone Preamp
I also recommend against the use of tantalum capacitors, and regular readers will notice that I have not suggested them for any project (although there was one suggestion that you could use them if you wanted to). The only capacitor fault I have ever had to track down with an intermittent short circuit was a tantalum bead type - it was neither fun, nor easy to find :-(
So, using tantalum caps in X-Fi are bad idea, alrough they might look as good choice at first and I must admit, I was considered them at first as well, but then I looked more deeply and contacted experts and they all recommended against it.
I had enought of waiting and I decided to test, if the d6.3 caps (remember, the original ones are just d4) will fit there. So I took my poor X-Fi (I wish I get my hand on some dead X-Fi for this testings, but...) and started soldering on it again, even when I did not have to caps in my hands yet, sadly.
So, I desoldered them all and cleaned the holes nicely, because I knew I will solder them right back, but first I need to test the d6.3 caps there...
So far, so good. Then I get the caps - I picked up a 8 pcs of - luckily to still have a few of them here and there - 120uF 16V Panny FM caps that are d6.3 - tought they are not bipolar and not audio caps either, but I was not going to solder them there - I just wanted to made some progress and to make sure we can fit a d6.3 caps there... So I just stick them thru:
And they fit! :wahaha: Hoooray!
And then I make the most important discovery yet. I suspected that before, but I was not sure. These 6 bipolar caps are for the Creative 10-pin connector 6 in/out channels separation.
The two ones are not bipolar, but normal caps and they are for decoupling of the CD-in (AUX IN) input.
This might not sound so important to you yet, but... I asked myself "What if I did not even use these connectors?" "Do I really need the caps there at all, or for the BEST possible quality is better to get rid of them completely?"
I mean - there is one thing better that good audio cap in the passing of audio signal. And that is wire - or no cap in this case.
So out of a pure luck and based just on a intuition I tried the X-Fi w/o these caps and quess what - it played and playing well still! :dev: :wahaha:
This is an important solution and very cheap audio improvement only for people who not use either of these connectors!
Amazingly and not unsurprisingly, once the Jamicons are out of the audio loop now completely, the detail level in music increased hearably once again.
However there is a side issue to that improvement. Now I can hear some compression distortions even at 320kBi, witch is considered as very high bitrate! Also, any more importantly, many many and many my mp3 music files contain some slight errors in music, that are now also possible to hear well...
In short, you might want to consider doing these mods (I'm being very frank here) because you will hear more that you might wish to hear :redface:
PS. old but interesting article: Creative's Sound Blaster X-Fi audio processor http://techreport.com/articles.x/8884
|14th June 2008, 09:48 AM||#5|
Final look - the thread should be renamed to "X-Fi holocaust", as I removed everything I did not use/like
Bad caps Jamicons are OFF my machine, HOOORAY!
The dynamic seems to be improved slightly, but that could be just in my head...
However replacing the suxxking Jami-crap working as DAC voltage stabilization and be connected on the FILT+ pin COULD change/improve the sound...
(bear in mind that my audio output AFTER the X-Fi is reasonably suxxking low-end one, and I did not have so good ears... so maybe someone else spot much more noticable difference - or dismis my claim of slight improvment there)
Voltage stabilizing for opamps is very important.
Voltage filering for the main X-Fi chip
Useless stuff is gone now.
Better voltage stabilization for the Wolfenson audio codec can't hurt too :P
removed everything I did not use
PS. moders should rather consider rather AD8599 opamps, it has much better audio: http://www.analog.com/en/prod/0,,759_786_AD8599,00.html
PS2. another things to consider - get rid of the output muting transistors
|14th June 2008, 10:24 AM||#6|
Join Date: Sep 2006
The level of detail and explanation of changes is impressive. Thanks for the write-up. You make me want to dig through my soundcard (not x-fi) to find room for improvements.
|21st June 2008, 09:54 AM||#7|
Definitively you can. This mod is not over yet anyway.
There is the muting transistors and other crap just after these opamps that is asking for removal, because of the distortion and add DC offset they introduce...
Also the opamps will be replaced with AD8599 ones anyway.
And I also would like to replace the MIC input opamp too...
A elektric schematic of the card will be worth gold... or at least a damaged spare X-Fi to play with and measure and measure...
|13th July 2008, 07:54 PM||#8|
Removing X-Fi muting transistors.
Having transistors in the audio loop is definitively a big NO-NO, just go ask any audiophile of Hi-Fists and he did not even need to be an extreme one to tell you this. So, I concentrated my efforts this way. First I produced this picture of what is short direct contact on my X-Fi Fatal1ty: http://i232.photobucket.com/albums/e...ps_outputs.jpg
And todays I go deeper. It is same for each output. After opamp, there is a 33ohm resistor to protect the opamp from shortcut on the end. It has minimal impact on the signal, so, keep it. Later there are two muting transistors, for the positive and negative flow:
And after them a two small caps to the ground to kill the possible high frequency interference. For L and R channels the card utilize 4 separate transistors. For the rest a dual transistor is used to save space - two (or coudl that be four?) ones in one.
Regardless, the removal is rather easy. First get rid of them:
Then solder a shorts there:
...and you are free to test them. As you probably noticed, I for the work desolder the C46 and C27. And also as you sure noticed, I replaced the LM4562 opamps with AD8599 ones.
After hearing the sound, I was like "Holly ****...!"
There is no words to describe how much better it sounds. The oversharped thick sound of the LM4562 is gone (LM4562 also like to pick a lot RFI) and the sound is rich and full - and yet more detailed - very likely thanks to these muting transistors removed!
Fantastic change, damn I'm glad I did it!
As everything in life, there is a price for this. Not only this is NOT easy mod (do NOT try that, unless you are soldering MASTER, and I'm not kidding) but it also has consequences. Not only you want to delete all, even the 320kBi mp3 files now, but upon the driver loading in windows boot, there is notable click in the speakers. Same when changing audio mode. Not louder that the amp is set for, so a minor price for such wonderfull, rich sound.
There is a bigger price and I was quick to discover it.
I remember it well from measuring the voltages on the caps on X-Fi. When I touch the opamps, well, then the X-Fi started to oscilate like MAD and the resulting sound noise, even on small testing speakers, are unbearable.
So, to get to it - I plug my rear speakers into the SUB/CENTER jack, so upon discovering that, I was like fixing this. So I unplug the jack... and the moment it started. The X-Fi, no longer protected with these muting transistors, start oscilating like mad. From all 6 speakers it output SO horrible noise and SO strong, that it was like when F18 is about to land on your head.
I can't remember anything comparable in my life.
My stepbro run to my room asking WTF... so you get the picture.
The oscilating noise, when you change your speakers, does not stop till reboot.
For me it is fair price to pay. I just can't now hotplug the speakers of fiddle with them "on the fly", like I used to. This is kinda sad and limiting, but what one can do. I'm ceratainly not going to put these cursed transistors back, no way. I love the sound now way too much.
|13th July 2008, 09:08 PM||#9|
Join Date: Sep 2006
That's really cool. I have a few issues with bypassing the output filter on the Sondigo soundcard as well. Leaving an amp on unattended is a bad idea I found my speakers pulsing full excursion both ways during a thunderstorm even though the computer did not restart or have any issues. Restarting the computer was required to stop it.
|18th August 2008, 09:57 PM||#10|
Well, we having few thunderstorms here and there, but no problemo anymore. I just cannot pull the jacks on the run from the card anymore. That it is. I think is is a pretty good price to pay for the quality
My modification of X-Fi Fatal1ty, UK edition w/o heatsink on main chip and with caps wires bent to sides and then soldered from the bottom of the PCB for user Tez, Head-Fi forum:
...and his card is working just great!
|Thread Tools||Search this Thread|
|Thread||Thread Starter||Forum||Replies||Last Post|
|Bugs||Jason||Forum Problems||265||10th January 2013 10:10 AM|
|Best Solution ??||cybtech||Multi-Way||5||6th July 2004 11:26 AM|
|New To Site?||Need Help?|