|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
219.79.250.28
Reading immate's observations of the effects of bus speed and ram timings on sound quality, which quite naturally suggest noisy power as the culprit. I took a look at the components of Juli@ soundcard digital section and found the following.
The three main components are all powered by 3.3V, namely:
VIA Envy24HT controller
Xilinx XC9536XL CPLD
AK4114 digital audo interface transceiver 32kHz - 192 kHz
There is a low dropout voltage regulator MT1117-3.3A0639F, providing 3.3 V to all the components above from an input voltage of 5V, suggesting to focus on both 3.3V and 5V.
To suppress noise, the best tweak would be to solder high quality caps directly on the power pin of the three chips above. Though the soldering of such tiny SMD components' legs are quite beyond me. The second best would be to solder quality caps on the voltage regulator which could be workable. I take the easy route without soldering and use connection pins of the digital board to the analogue board for mounting the coupling capacitors.
After a bit of research and toying with the multimeter, I figure that the group of 20pins seems to be digital data. The group of 10pins should be power distribution to the analogue board containing 12V, -12V, 5V and 3.3V and several grounds.
I use screw-in type power connectors to connect a blackgate FK 470uF parallel with a multicap RTX 0.1uF to pin 2 (5V) and pin 5 (ground). The blackgate was originally "pluged" into the 12V supply to filter the noise there.
Results:
The traces of glare or tightness very noticible in the voices associated with running the system bus at 200Mhz are reduced/eliminated. There are also other big improvements but I would not go further here as they may as well be due to gradual break-in of the recently built balanced transformer for the DAC.
Disclaimer: I don't have any professional experience in electronics, feel free to correct me!
Follow Ups:
I did the 5 volt capacitor mod...very nice sq pop. Thank you again great idea.
Impressions after several days: much better with time on the new 470 uf cap. The sound is as if the whole soundstage has been illuminated. W/o the mod the sound seems dark and murky (I'm talking small but significant effects) with the mod there is clarity, much much more depth/width better imaging within the soundstage.One more thing dynamics are way better (both at the low and high end).
Thanks again Promethk
Edits: 01/26/11 01/26/11
what size are the screw type connectors?
where can they be sourced?
thanks,
drmike
No screws for me...I soldered.
what are the values of the two white caps?
thanks, drmike
The two caps I use are 470uf (50v) and .01 uf 400v. Remember I am not the op you may want to look at what he posted above. Also I put mine in the 5 volt hole through on the sound card board in upper right corner with chips down and pins on the near side and in the digital ground hole in the card on the left side. You can see it on the other side of the sound card board.
What values of caps did u use in the above picture? And connecting which pins? I am interested in trying out this mod. It would be very helpful if u could furnish details. Thanks
The idea of using coupling caps on the 5 volt pins is a good one. But regarding the 3.3 volt supply a much better solution is to add LiFePo4 batteries as the 3.3 volt supply along with bypass caps. This, next to minlogon, was the most productive tweak I have made to CMP. I believe it was Serge who originally had the idea.
Edit correction it was Alfred (sonics) who originated this idea.
Edits: 01/18/11
Dear promethk,Your idea of using the pins that connect to the analog section of the JULI@ card for supply bypass is ingenuous.
For the benefit of the lazy; though a picture is worth a thousand words in this case would you mind verifying what I think I am seeing?
You say you are using Pin 5 for ground but that does not look like what I am seeing in your picture.
Pin 1 at the rear of the card - looks like a voltage rail
Pin 2 looks like ground from the photo
Pin 3 looks like ground from the photo
Pin 4 looks unused
Pin 5 closest to the front of the card - looks like a voltage railPlease advise if this is correct or, if not, what is correct.
Thanks very much for your efforts and any additional help you are willing to supply.
Rick McInnis
The pin numbers are as follows:
2 4 6 8 10 <-back row
1 3 5 7 9 <-front row with chips facing us
pin 2 is the 5V
pin 5 is one of the ground
pin 10 is the 3.3V
I only bypass the 5V and the 3.3V as I could only identify chips operating on these two voltages in the digital section. The analogue section also use 12V.
I guess I need to look at my card to see what I am missing.
Sorry for being obtuse!
Can't wait to implement your suggestion.
Thanks, again,
Rick McInnis
For the Juli@ tweak, I have inserted some teflon insulation tubes into untweaked pins to insulate them, this may have misled viewer of the pic into believing that the cap's wire goes into them while in fact they are not connected to the caps but just covered by teflon tubes. The telfon tubes are needed as I have remove the plastic cover of two of the wire connectors due to the tight space.
Other possible tweaks are using metalized polypropene caps, Mundorf supreme, Solen MKP, PIO, SIO etc. However, all of them are so big such that direct mounting on the cards seems impossible. Mounting them on the side via connection wires, though I am not sure if the benefit outweights the wire inductance and EMF etc.
One last thing, as there are several ground pins, I may not be using the correct ground pin for this tweak as some circuit could have more than one ground. Maybe someone could enlighten me on how to distinguish the different grounds in the Juli@.
Both GB and Biostar mobos supply 5V to PCI card (rear notch identifies 5V supply).
Looks like you have the newer version of the GB mobo which supports low voltages (default is very high causing the high temps). See here for details. I have mine on 0.95000V with system running without any fans. Power consumption is lowered - would be great if you could measure the differences.
Combined changes to Juli@ and above may give better results.
My current motherboard does not support voltage change nor the fsb frequency change.
Well, the Juli@ card does depend on the 5V from the motherboard. The Juli@ card has its own voltage regulator (the rightmost chip) which convert the 5v into 3.3V for the other chips I stated earlier.
The photo shows the capacitors I've added for coupling both the 5V and the 3.3V.
You are also right by pointing out the importance of isolating the 2.5 harddisk, as such harddisk mainly draw its power from the 5V supply which the Juli@ also depends on.
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: