or Connect
Head-Fi.org › Forums › Equipment Forums › Computer Audio › Hotrodding the X-Fi: A Layman's Guide (No 56k)
New Posts  All Forums:Forum Nav:

Hotrodding the X-Fi: A Layman's Guide (No 56k) - Page 80

post #1186 of 2194
Well, I installed the OPA2822U.

It has a much higher slew rate, and a greatly quicker response time.

I've gotten back the edge and brightness back that I lost with the OPA2132. This amp is very detailed, and isn't forgiving to poor recordings.

I find the sound very clear and clean sounding, but the soundstage is definitely lacking.

Overall though it works well. And the numbers got measurably better as well.

post #1187 of 2194
Did some more interesting power-cap changes: (SB0460)

WOLFSON WM8775 Power Section:
- changed C101 and C124 to 22uf @ 16vdc TANT from Jamicon 22uf @ 16vdc.
- changed C115 to 6.8uf @ 16vdc TANT from Wincap 10uf @ 16vdc.
- changed C123 to 10uf @ 16vdc TANT from Wincap 10uf @ 16vdc.

WM8775 Datasheet Refs:
- note 3, page 6 (VMID)
- C1, (DVDD) C3, (AVDD), C5, (ADCREFP), page 37, figure 29
- note 3, fig 29

CREATIVE CA20K Power Section:
- C172, changed Jamicon 22uf @ 16vdc to Panasonic S, SMT, 33uf @ 25vdc

Seems to have added more, ummm, "kick," to low end, without "muffles, muddyness..."
High still retain "clean, clearity with definition." (small cymbal "ting, ting" as opposed to "tick, tick")

Off to reload food stocks, back in a hour or so.....

AD8599 Line-Out installed with changes above: (do your excellent magic, that is, if you feel up to it...GermaniumAtomic32!)
- RMAA Summary: http://www.esnips.com/doc/208e13ad-6.../X-FI-MOD-038b
- RMAA Detail: http://www.esnips.com/doc/77b34e7f-1.../X-FI-MOD-039b

REF:
WM8775 Datasheet:
http://www.wolfson.co.uk/uploads/doc.../en/WM8775.pdf
post #1188 of 2194
Another nice, cheap power section change:

OP AMP -5vdc SECTION:
C72 - 10uf @ 16vdc TANT from Jamicon
C74, C75 - 33uf @ 35vdc SMT, from Jamicon 22uf @ 25vdc

- Stereo Crosstalk reduced by about 1.9db, -98.2 vs. -96.3
- THD and IMD figures improved as well

Sonically reproducing with excellence, accuracy and clearity, across, vocals, instruments, etc...

Said it before, but I NOW think I'm ready to experiment with coupling capacitors....

SB0460 with AD8599 installed:
Socket Hack close-up: http://www.esnips.com/doc/87ab1b96-f.../X-FI-MOD-033c
RMAA Summary: http://www.esnips.com/doc/6e57a231-9.../X-FI-MOD-040b
RMAA Detail: http://www.esnips.com/doc/d1839503-f.../X-FI-MOD-041b
PIC: http://www.esnips.com/doc/690156f4-f.../X-FI-MOD-042b
post #1189 of 2194
Quote:
Originally Posted by germanium View Post
1.Yes going to try the bigger metalyzed film caps on the 2-220uf caps. can only get 12uf caps at the local store so going to pick them up this weekend. Yes I have changed my mind & am going to leave these bypassed. See below for reason.

2. BG might be better if you arn't going to bypass the power caps but a waste of time & money if you are as the film bypasses are probably substantially superior.

3. 4 of the 47uf caps serve as a ref voltage supply & may not be critical. The 100uf caps are to the intermal regulated supply on the DAC & appear to be critical. Do not change these but bypassing seems to help the clearity. The red marked caps are definately coupling caps & can be shorted.

4.yes definately on the 100uf cap not sure whether the 47uf cap bypass matters but leaving mine bypassed with 1 3.3uf cap.

5. Yes but going to try bigger 12uf bypasses there as well as there may be an improvement there. There was definately a drop in IMD swept distortion at 15KHZ & above though even with just the 3.3uf bypass caps there. Sound is slightly brighter but not overwhelmingly bright. There is definately a fair amount more detail there.

If you look at my tests & compare them to Creatives own tests mine are subtantially better. Bypassing input coupling caps to the ADC helps the noise floor of the ADC & slightly flattens the low frequence response of the ADC. Creative poor noise result is likely because of these caps.

Germanium, thank you very much for your explanation. Another 4 questions!~

1. Have you tested all the 3 power supply cap bypassed by 12uf polymer film cap? What the max cap do you suggest for bypass caps? 18uf? 22uf or even 47uf? Solen is the good choice, right?

2. What capacitance and voltage do you suggest for the 2 opamp power supply caps that I want to replace?

3. You suggested not to touch the 47uf and 100uf caps for DAC to opamp path. Just bypass them with 3.3uf polymer (1 for 47uf and 2 for 100uf). Do you think these 47 and 100 caps are handicaps?

4. Beyond all above, what's your other suggestions for hotrodding?


Thanks!
post #1190 of 2194
Quote:
Originally Posted by germanium View Post
... You see my tests Ter1?

Yes, I see your 24bit 96kHz pictures, very perfect!!!

I posted your pictures on other forums, the response is very positive, but often challenged that "16bit 44.1kHz" is the main parameters we usually use, "24bit 96kHz" is rarely used!!

So can you post the results for "16bit 44.1kHz" and "24bit 48kHz" results pictures?

Lots of people are waiting them.......

Thanks!
post #1191 of 2194
Quote:
Originally Posted by ter1 View Post
Germanium, thank you very much for your explanation. Another 4 questions!~

1. Have you tested all the 3 power supply cap bypassed by 12uf polymer film cap? What the max cap do you suggest for bypass caps? 18uf? 22uf or even 47uf? Solen is the good choice, right?

2. What capacitance and voltage do you suggest for the 2 opamp power supply caps that I want to replace?

3. You suggested not to touch the 47uf and 100uf caps for DAC to opamp path. Just bypass them with 3.3uf polymer (1 for 47uf and 2 for 100uf). Do you think these 47 and 100 caps are handicaps?

4. Beyond all above, what's your other suggestions for hotrodding?


Thanks!
1. Have not tested any 12uf caps yes & only have 2 as they are quite expensive even for cheapies. 11 dollars each. I'm on a very tight budjet for a few weeks so can't afford any more.

2.16-25 volts & no more than 1000uf cap if you are changing electrolytic. If just bypassing I don't see much sense in going over 25uf per cap. Metalyzed films are very effective at 10% of the original caps size & even 5% is probably enough for the purposes as a bypass. Solen is a very good brand. They are also very large but I do believe they have come out with some smaller lower voltage caps that may be suitable. You will have to check thier catalog out.

3. These caps are for an internal voltage regulator for the DAC only. I shorted the coupling caps.

4. Nothing extra at the moment.
post #1192 of 2194
Quote:
Originally Posted by ter1 View Post
Yes, I see your 24bit 96kHz pictures, very perfect!!!

I posted your pictures on other forums, the response is very positive, but often challenged that "16bit 44.1kHz" is the main parameters we usually use, "24bit 96kHz" is rarely used!!

So can you post the results for "16bit 44.1kHz" and "24bit 48kHz" results pictures?

Lots of people are waiting them.......

Thanks!
The pure 16 bit graphs are not really all that good interestingly. Look no better than the lower X-Fi photos, maybe even worse!!! 24 bit 44.1KHZ looks really good though. If you run the record section at 24 bit though & feed the DACs 16 bit it doesn't look to bad. You can actually tell your DACs are at the limit of the 16 bit system when read with 24 bit ADC. The ADC really needs to be run in 24 bit mode though to get the best readings.

24 bit 48 pics really the same as the 96KHz pics but less bandwidth is all. exact same distortion characteristics.

Will post pics though a little later going to listen to music a little now so check back later for the pics.
post #1193 of 2194
16 bit DAC tests with 16 bit ADC reads

http://www.esnips.com/web/16bitRMAAtests

16 bit DAC tests with 24 bit ADC reads

http://www.esnips.com/web/16bitRMAAtests24bitread

You will see in the 24 bit read tests that I am at the 16 bit performance limits for this DAC & possably the real world 16 bi capabilities. In some ways the 16 bit DAC performance is better than the 24 tests on the lower X-Fi card.

The pure 16 bit DAC & 16 bit ADC read is a little worse but within reach of the lower cards 24 bit performance.

All in all not as bad I was thinking as I was thinking the other tests were for 16 bit but then remembered that they were for 24 bit performance on the lower card.

Here is the results list. The first is 16 bit DAC 24 bit ADC read. The second is 16 bit DAC 16 bit ADC read both at 44.1 KHZ. Third is 24 bit DAC & ADC read at 48KHz sample rate.

http://www.esnips.com/doc/bb043116-8...erent-settings

Hope this helps ter1
post #1194 of 2194
SB0460 - Coupling Capacitor Socket Hack:
- socketed Line-IN (C48, C49) and Line-OUT caps (C23, C50, C76, C77)

Basic RMAA and listen tests for correct function and baseline, after adding coupling cap sockets.
RMAA Test Configuration:
- AD8599 op amp
- tantalum bypass changes, to date.
- LINE-IN: (C48, C49) Panasonic A, 4.7uf @ 35vdc SMT from G-Luxon 4.7uf @ 25vdc NP
- LINE-OUT: (C23, C50, C76, C77) Panasonic S, 22uf @ 16vdc, SMT from Jamicon 22uf @ 16vdc
- AUX-IN (C102, C104) Panasonic A, 4.7uf @ 35vdc, SMT from Wincap 4.7uf @ 25vdc (not installed)
- (C18, C19 ??) G-Luxon 4.7uf @ 25vdc (not installed) (need more study, not sure of function)

Odd that Creative used NP for LINE-IN and polar for AUX-IN coupling caps. Both couple to Wolfson WM8775
AIN2L/R, #27/26, AIN4L/R, #22/23 mux inputs, respectively.

Ready for coupling-cap shorting, metal-film, ceramic, value change, peanutbutter-cap, experiments...

Still sounding excellent, but have lost "ear," for now. Too much test listening this past week.

RMAA v6.0.5 Test Results:
Summary: http://www.esnips.com/doc/7e1a1e8a-6.../X-FI-MOD-044b
Detail: http://www.esnips.com/doc/c3346183-b.../X-FI-MOD-045b

Picture:
http://www.esnips.com/doc/eccf8cc2-7.../X-FI-MOD-043b
post #1195 of 2194
Quote:
Originally Posted by germanium View Post
16 bit DAC tests with 16 bit ADC reads

http://www.esnips.com/web/16bitRMAAtests

16 bit DAC tests with 24 bit ADC reads

http://www.esnips.com/web/16bitRMAAtests24bitread

You will see in the 24 bit read tests that I am at the 16 bit performance limits for this DAC & possably the real world 16 bi capabilities. In some ways the 16 bit DAC performance is better than the 24 tests on the lower X-Fi card.

The pure 16 bit DAC & 16 bit ADC read is a little worse but within reach of the lower cards 24 bit performance.

All in all not as bad I was thinking as I was thinking the other tests were for 16 bit but then remembered that they were for 24 bit performance on the lower card.

Here is the results list. The first is 16 bit DAC 24 bit ADC read. The second is 16 bit DAC 16 bit ADC read both at 44.1 KHZ. Third is 24 bit DAC & ADC read at 48KHz sample rate.

http://www.esnips.com/doc/bb043116-8...erent-settings

Hope this helps ter1
Yes, definitely, the recording depth should be 24bit with the same sample rate as in RMAA testing to get the best result. Also, the shared playback depth should be 24bit to get the best too!

For 16bit 44.1kHz RMAA, I got the following result in following condition:
1. bit-matched playback
2. playback shared mode format: 24bit 44.1
3. recording shared mode format: 24bit 44.1
4. master sample rate: 44.1

Frequency response (from 40 Hz to 15 kHz), dB +0.01, -0.06 Excellent (+0.01, -0.07)
Noise level, dB (A) -102.8 Excellent (-103.5)
Dynamic range, dB (A) 95.1 Excellent (94.9)
THD, % 0.0013 Excellent (0.0017)
THD + Noise, dB (A) -89.0 Good
IMD + Noise, % 0.0047 Excellent (0.0050)
Stereo crosstalk, dB -103.3 Excellent (-103.7)
IMD at 10 kHz, % 0.0052 Excellent (0.0055)
General performance Excellent

Seems your results (after mod) are a little bit worse (except noise & crosstalk) than my stock results.

Do you think the real listening are far better than the stock ones?
post #1196 of 2194
Quote:
Originally Posted by bichi View Post
SB0460 - Coupling Capacitor Socket Hack:
- socketed Line-IN (C48, C49) and Line-OUT caps (C23, C50, C76, C77)

Basic RMAA and listen tests for correct function and baseline, after adding coupling cap sockets.
RMAA Test Configuration:
- AD8599 op amp
- tantalum bypass changes, to date.
- LINE-IN: (C48, C49) Panasonic A, 4.7uf @ 35vdc SMT from G-Luxon 4.7uf @ 25vdc NP
- LINE-OUT: (C23, C50, C76, C77) Panasonic S, 22uf @ 16vdc, SMT from Jamicon 22uf @ 16vdc
- AUX-IN (C102, C104) Panasonic A, 4.7uf @ 35vdc, SMT from Wincap 4.7uf @ 25vdc (not installed)
- (C18, C19 ??) G-Luxon 4.7uf @ 25vdc (not installed) (need more study, not sure of function)

Odd that Creative used NP for LINE-IN and polar for AUX-IN coupling caps. Both couple to Wolfson WM8775
AIN2L/R, #27/26, AIN4L/R, #22/23 mux inputs, respectively.

Ready for coupling-cap shorting, metal-film, ceramic, value change, peanutbutter-cap, experiments...

Still sounding excellent, but have lost "ear," for now. Too much test listening this past week.

RMAA v6.0.5 Test Results:
Summary: http://www.esnips.com/doc/7e1a1e8a-6.../X-FI-MOD-044b
Detail: http://www.esnips.com/doc/c3346183-b.../X-FI-MOD-045b

Picture:
http://www.esnips.com/doc/eccf8cc2-7.../X-FI-MOD-043b
Hi, during your various mods to getting better results, I think the it's the limit by your card itself (DAC & ADC, etc). I think it will be huge improvement if you mod elite pro since you're willing to try different mods and looking into reference pdf files!~
post #1197 of 2194
oop looking at you post wrong so deleted this one. Yes the real listening is where it counts & that is substantially better. All tests are within a margine of error for a good stock card so nothing to worry about. Besides the pure 24 bit numbers speak for themselves. It is there that you can tell that I'm actually near or at the limit of the DACs & the DATA sent them when dealing with 16 bit DATA as those tests are much better. Remember also I'm testing in Windows Vista which may account for some very small differences & you have to admit the differences are extreemely small in this case.

Will retest in Windows XP
post #1198 of 2194
The higher distortion definately an artifact of running Windows Vista as is the lower noise figures. Noise is not actually increased at all in XP, just the way it's read is all.

Noise in XP. -95.0 db
THD in XP .0009
IMD + noise in XP .0050
IMD swept in XP .0052
crosstalk in XP -96.0
Dynamic range in XP 95.0 db

You must have been testing in Windows Vista also though given my XP results

http://www.esnips.com/web/WindowsXPR...ts441KHz16bit/

Here you can see that the distortion tests in XP are actually considerably better in 16 bit THD & Dynamic range tests
post #1199 of 2194
Enabled bit matched recording in XP, New results

noise -96.6db
Dynamics 96.7db
THD .0009%
IMD .0042
IMD swept .0043
crosstalk -96.9db

Will retest in Windows Vista with bit matched recording enabled
post #1200 of 2194
No differences in Windows Vista with bit matched recording enabled. Appearently Windows Vista is optimized for 24 bit resolution playback.

I've done some tests in Windows Vista with 24 bit DATA sent to the DACs but read at 16 bit by the ADC. no appearent distortion was added, only noise by reading the 24 bit tests in 16 bit read mode. These AKM ADCs are excellent performers & are perfectly capable of revealing any limitations found elsewhere!!!

Noise effects the distortion tests readout but in looking at the spectrum analysis the distortion is not actually increased at all compared to the full 24 bit read & write tests
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Computer Audio
Head-Fi.org › Forums › Equipment Forums › Computer Audio › Hotrodding the X-Fi: A Layman's Guide (No 56k)