|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
76.8.236.2
In Reply to: RE: cMP - the open source high-end Memory Player posted by cics on December 30, 2007 at 05:42:01
10/10/09 Updates...I've the following info to my Juli@ Follies post below:
1. Info about one more regulator chip on the analog card which appears to power what appear to be the mixer chips, U24 and U26.
2. Info about the balanced outputs and the unbalanced and balanced inputs.
3. Info about the use of 5v on the digital card
4. Preliminary notes on where the outputs from the DAC go.
This all was prompted by:1. I'm modifying a Juli@'s analog balanced outputs for another inmate, so I needed info on the balanced side.
2. I suspect there are some additional gains available by upgrading that last regulator chip that provides the negative rail to the mixer chips... the positive rail for those chips are covered by U10.
3. Replacing the output muting transistors with relay-based ones is a common modification for circuits such as this, providing some sonic gains.
4. I strongly suspect if one is using the Juli@ just as an output device, there are some serious sonic gains available by bypassing the mixers and output circuitry and taking the DAC output to a simplified and more direct output stage. I have this on my queue of Juli@ mods to try.
I hope this information will be useful for others!
Ok... I haven't abused one of my Juli@'s with my soldering iron... yet. But it's just a matter of time
What I did do is spend a bit of time poking around the Juli@ and wanted to report what I found... and also gather info others have posted into one place.
What's prompting this is the imminent arrival (I hope) of an I2S-connected DAC. It's not a commercial product, but just a tweaked set of boards as used by one of the EQ modifiers. I'm hopeful that this DAC will work well driven by the Juli@ as it is based on the AK4399, a similar DAC to the AK4358 that's already used in the Juli@.
I needed to find the reset pin in the J3 header, one that I haven't found identified in previous posts on various forums in the past. So I pulled out my trusty meter and started poking around... and then wanted to confirm some of the other pins... and then find the +-12v pins... And then kept going... and this is what I found:
J3 connector signal pinout (note much of this is taken from the Head-Fi post http://www.head-fi.org/forums/f46/easy-i2s-juli-pci-sound-card-242285/):J3 Pin 1 -> 1724 PSDOUT[0] ( I2S data out) -> AK4358 SDTI1 (I2S data in)
J3 Pin 5 -> 1724 PSYNC ( I2S Word Clock ) -> AK4358 LRCK (I2S Left/Right Word clock)
J3 Pin 7 -> 1724 PBCLK ( I2S bit clock out ) -> AK4358 BICK ( I2S bit clock in )
J3 Pin 9 - > Xilinx on the digital board -> AK4358 MCLK ( I2S Master Clock in )
J3 Pin 11 - > Xilinx & AK4114 on the digital board - > AK4358 PCN (Reset)
J1 connector power supply pinout (note much of this is taken from the Head-Fi post cited above and some is taken from this Asylum post http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=35497):J1 Pins 1 & 3 -> +12v
J1 Pins 7 & 9 > -12v
J1 Pin 2 -> +5v
J1 Pin 10 -> +3.3v
J1 Pins 4, 5, 6, 8 -> grounds
PCI power supply connections (note some of this is stated in this Asylum post http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=35497 . Also, a good source for the PCI pinout is here http://pinouts.ru/Slots/PCI_pinout.shtml):First of all, the Juli@ only uses the +5v and +-12 lines from the PCI bus. If you look carefully at the pins specified as 3.3v on the Juli@ edge connection, you'll see that the conductive fingers are not connected to the board. This simplifies hooking the Juli@ up to a separate power supply... here's the PS pins:
B1 - > -12v
A2 - > +12v
B5, A5, B6, A8, B61, A61, B62, A62 - > +5vWhile there are a number of +5v pins that will need intercepting to inject the 5v from a separate power supply, it's a LOT easier than if you also had to connect up twelve +3.3v lines too!
Digital card power supply notes (note some of this is stated in this Asylum post http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=35497):U1 regulates 5v to 3.3v, with the first pin connected to ground, the second the output, and the third to the 5v bus. C1 & C3 are input filtering capacitors and C2 is the output filtering capacitor.
Also note these comments from a poster titled 'marcello' (from a post on this forum, sorry, I didn't dig it up to cite directly):
"Upgrading digital section:
Locate a good "audiophile"clock oscillator manufacturer. There are multiple offers available on the internet, varying from LC, Audiocom, etc. In NL, Audioart is making a VERY good clock but I'm not sure if they are selling it separately. If not I can intermediate.
I will come with my own clock this summer. 12V and the lowest noise and jitter you've heard of.
Two clocks are needed, one for 22.5792MHz and 1 for 24.576MHz. Replace the existing crystals (x1 and X2)with the corresponding clock oscillators. Power these external clock oscillators by a very good linear supply, or even a lead acid battery if the required voltage is 12Vdc.
Solder a 220uF/10V SG or SH OsCon capacitor from U6, AKM's 4114 chip AVDD pin 38 to ground. Best is to do this on the underside of the board, across BC25. Negative to the solder pad nearest to the "BC25" text.
Solder a 220uF/10V SG or SH OsCon capacitor across C8; positive to the pad towards text "C9", negative to pad towards text "C8".
These measures take care of the most basic limitations of the sound quality of the Juli@ card."
Analog card digital power supply notes:U10 regulates +12v down to 5v for the digital chips and the mixer chips on the analog card, with the same pinout as U1 above, G O I. C13 is the +12v input filter capacitor and C14 the +5v output filtering capacitor. C16, C46, C17, and C85 are local filters for this rail.
Analog card analog power supply notesU11 regulates +12v down to +9v, with pin 1 the +12v input (the pin towards the center of the card), pin 2 the output, and the tab ground. C18 is the +12v input filter and C19 the +9v output filter.
U13 regulates -12v down to -9v. with the tab the -12v input, pin 1 the ground (the pin towards the center of the card) and pin 2 the output. C22 is the -12v input filter and C22 -9v output filter.
U14 regulates -9v down to -5v(?), with a pinout of G I O. C24 is the -9v input filter and I haven't found (although I haven't looked very hard!) an output filter. It provides -5v to the chips I believe do the mixing, U24 & U26. They get their +5 from U10.
Analog card analog input circuitry notesU15 and U16 are the unbalanced input opamps and I believe U17, U19, U20, and U21 are the balanced, I haven't traced this section much. C29 C34, C37, and C42 are the unbalanced input coupling caps and I don't believe the balanced input has coupling caps. I suspect Q1 and Q3 are the balanced output muting transistors and D2, D3, D4, and D5 provide some protection for the unbalanced inputs, but these are just guesses. The balanced and unbalanced input circuits look to be very different from each other.
Analog card analog output circuitry notesU22 and U23 are the unbalanced output opamps and U25 and U27 are the balanced. C70 and C73 are the unbalanced output coupling caps and C76, C79, C81, and C83 are the balanced output coupling caps. Q5 and Q6 are the unbalanced output muting transistors, while Q7, Q8, Q9, and Q10 mute the balanced outputs.
Analog card DAC output circuitry notesAlthough I haven't documented the details yet, I've traced the outputs from the AK4358 DAC to two vertical rows of components on the backside of the card, the rows starting with R71 and R76. The 4 sets of outputs (R&L, +&-) are found on the side of the components away from the part identifications.
Separate Power Supply Voltage Insertion Points (note that this is inspired by and includes the information posted by Alfred in this post http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=49122 ):DVCC1 - 5v into both digital and analog cards (Note that the 5v IS used by several chips on digital card in addition to regulator U1, such as U2 and U8. I've never tried it, but I suspect it will not work correctly if you don't supply 5v to the card in addition to the 3.3v. So if you feed it 3.3v separate from the PCI buss, either leave the 5v connections to the PCI buss or feed the card 5v separately (Note this is different than what I published before!).
Also be aware that the analog card uses both the 5v and the 3.3v, along with +12v & -12v.
P3V3D1 - 3.3v out of U1 regulator. Can be used as an insertion point for 3.3v to both the digital and analog cards by removing U1
DGND1 - Digital card ground
AGND1 & AGND2 - Analog card grounds. Use 1 if you are using the balanced outs, 2 if you are using the RCAs
P12V1 - +12v into the analog card
NVA1 - -12v into the analog card
P5VA1 - +5v out of regulator U10, which regulates from the +12v rail
PVA1 - +9v out of regulator U11, which regulates from the +12v rail
C19 - The side of this capacitor away from the center of the board is also the +9v out of regulator Ull
C23 - The side of this capacitor away from the center of the board is the -9v out of regulator U13
R29 (open pad on my Juli@) - the side of this set of resistor pads away from the 'R29' is an insertion point for the +12v
R32 (open pads on my Juli@) - the side of this set of resistor pads towards the center of the board is an insertion point for -12v
With all this information, modifying the Juli@ will be a snap.So what do I plan to do with it?
Sorry, that'll have to wait for another night, I need to go back to work for awhile.
Greg in Mississippi
Edits: 03/30/09 03/30/09 03/31/09 03/31/09 04/01/09 04/01/09 07/08/09 10/11/09Follow Ups:
Hi.
As per your suggestion, I purchased the Dexa 3.3v regulator from Parts Connexion. It arrived yesterday. Before hooking it up on the Juli@, I need a small clarification. On the Dexa, one side is labelled OUT & the other end labeled as "N". Does N stand for Negative(ground pin), or has it been mis-printed for "IN".
Also is it appropriate to remove the onboard regulator & substitute it with the Dexa, or to use that Dexa in making a Linear 3.3v supply & directly feed it to the card bypassing the onboard regulator.
Please advice. Thanks
First, attached are the pictures I promised from a Juli@ I'm modifying for another inmate... came off a bit neater than the one I did for myself because it's for someone else!Note the Dexa regulator for the 3.3v laid down over the board to minimize the footprint.
BUT, there's more! I'm getting ready to add a BNC connector for him to bypass the breakout connector. I spent a little time tracing the connections identified in an earlier post to their sources so I can wire it as directly as possible... and the SPDIF output terminates at the high 6-pin component labeled 'HanRun HY600652' just to the left of the breakout cable connector. Looking up HanRun's website (http://www.hanrun.com/hren/sortinfo.asp?ClassID=91), it is likely a interface transformer. Checking both the + and - SPDIF connections, NEITHER connect to the ground. The SPDIF is transformer-coupled!
Two thoughts about this... first, a second transformer is redundant. Second, if we can find the parameters of this unit, upgrades may be possible.
I still think that an I2S connection is going to be the best bet. But for those that are tied to SPDIF-input DACs, this news should be a bit of piece-of-mind (some decoupling is already present!) and offer a possible upgrade path!
Greg in Mississippi
P.S. The SPDIF input is also transformer-coupled, it's component is just to the right of the output transformer in the picture.
P.P. S. Lifting the output pins on the transformer from the board will remove the connections from the transformer to the breakout connector from the circuit... likely another worthwhile tweak!
Edits: 05/11/09
Just finished testing it. It came out very nice... the sound with my Peter Daniel NOS DAC at 44.1, even through a BNC-RCA adapter, was much more tactile and impactive bypassing the breakout cable. This was the first time I was tempted to go that way instead of 192 upsampling through the Juli@! I may have to do this with one of my Juli@ cards!
It also came out very neat... see attached pix.
Note that the nut was scrounged from a fuse holder. You have to remove it when you put it into the computer and then put it back on... carefully because it wants to cross-thread. But do put it back on... it is a delicate connection at the board. And the washer is a rubber faucet washer.
Also, if you do this, be sure the card is completely seated in the PCI connector when you mount it... because it isn't as firmly mounted to the hanger, it can go in and only be partly seated. I don't know if it will survive getting turned on that way... be aware of that.The BNC connector I used is available through Digikey and you can find it by searching for part number A97562-ND. To have room to solder the connections to the legs of the output transformer you have to grind a bit off the corner of the plastic base of the connector that faces the component side of the board ... you can see this in the first picture.
You can't get more direct without hard-wiring!
Greg in Mississippi
Edits: 05/13/09 05/13/09 05/15/09 05/15/09
Hi Greg,
Just before I place my order of partsconnexion...
How did you connect the Dexa Regulator to the pins on the Juli@? From your picture, I only see two connections to the Dexa (looks like GND and input to me), and one (3.3v output?) seems to be missing. Is the output connected somewhere else, or simply hidden in the picture?
Thank you in advance!
You can see it better in this picture... the output pin is conneected to the pad where the heatsink flange of the regulator was (same as output).
The nice thing about doing it this way is that they form a nice tripod that helps support the regulator and prevent pulling up the pads.
Several things to know about doing this mod:
1. Replace/add the caps first and test to make sure it's ok. Then remove the two caps near the regulator before pulling the old one and putting on the new one... then put the caps back on.
2. The pads are delicate. When removing the caps and regulator, use a tiny screwdriver to pry them up, but be sure that the solder is melted before moving them... it should take almost no force to lift. On each component being removed, use your soldering iron to heat the joint on one one end/leg, GENTLY lift it, then go to the next.
3. The connection pins on the Dexa regulator have to be moved from one side of the board to the other to make it fit well. Plan how they need to fit make it work before putting them back in.
4. Check that you are hooking up the capacitors AND the Dexa regulator with the correct polarity at least 3 times before you hook them up and at least 3 times after. YOU WILL NOT have a chance to correct it if you hook them up wrong and power it up.
5. Make sure each cap and the regulator is secured so that they can't be hit and tear off the pads. For the regulator, I used two layers of original Dynamat on top of the controller chip (labeled 'ESI' and then a layer of double-stick to secure it.
Let me know how else I can help
Greg in Mississippi
P.S. I really want to try the Paul Hynes series and shunt regulators here too... about the same package, they should fit too.
How significantly do the capacitor sizes affect sound quality? I plan on using BG NX 1500uf/10v as input capacitors, and a spare BG N 100uf/50v for the output capacitor. Are they too big, or too small?
TIA!
p.s. Are you using a 180uf/20v Os-Con for the output cap? I think the Black Gates are probably STD 1000uf/16v.
I haven't spent a lot of time listening to various capacitor combos... for these caps, I used how Peter Daniel setup his premium NOS DAC as a starting point for selecting which caps went where and roughly sizing them. I did see a couple of references to some comments he made on selecting caps... See here:http://www.diyaudio.com/forums/showthread.php?postid=574431#post574431
and here:
http://www.diyaudio.com/forums/showthread.php?postid=581230#post581230
And have used those comments as some general guidance.
I liked the sound of his DAC when I was using it with my Shigaclone CD drive and he uses parts similar to what I've seen some other modifiers use.
To summarize the basic starting point, Peter's NOS DAC uses 1000uf Black Gate standards before the regulators, 10uf-50uf Black Gate N/NX after the regulator, 4.7uf/50v Black Gate N for coupling (unavailable now), Oscon for purely digital PSs.
I think the parts you plan to use will be fine, the sizes are certainly in-the-ballpark and the N/NX are better than what I used.
Later!
Greg
P.S. Yes, I used the 180uf/20v Oscons from the Parts Connexion and 1000uf/16v BG standards.
Edits: 05/13/09
Thank you, the information is very helpful!
I find it interesting that Peter Daniel thinks more highly of the STD 1000uF than Ns of the same size for power supply applications. The combinations do seem promising, however.
I might try a power supply with the 1000 STD / 50 N combo in the future, and see how it works out.
Thank you taking the time to explain, I think I know where to position the pin now.
I'm glad that I read your instructions before proceeding any further. Though admittedly, there's still a good chance that I'll frying/breaking the Juli@, even with the additional help.
I know that your time is pretty tied up, but will you consider modding the Juli@ for "another inmate" in the future?
LOL.
I can fit it in... I'm getting pretty good at it and doing it fast.
Let's see how the BNC mod works first.
Greg in Mississippi
Thank you so much!
By the way, have you tried using RAMDisk on your computer transport? I think I hear some subtle improvements by loading cPlay from the virtual RAM drive instead of my HDD. Might be worth trying out.
NT
.
I say the epilogue because I'm basically out of significant mods to the Juli@.The last thing I did this week was to remove the redundant Dexa regulators between the raw DC and the on-board Dexa regulators... so now I:
1. come directly from the a raw +-18v supply to the on-board +-9v regulators for the output stages (this used to be fed from a +-12v regulator pair)...
2. come directly from a raw +11v supply to an on-board +5v regulator for the DAC (this used to be fed from the same +12v regulator above)...
3. come directly from another raw +11v supply to an off-board +5v regulator for the digital board which then feeds the on-board +3.3v regulator.
My goal here was to reduce the number of cascaded regulators, which one of my tweak-mentors suggests will not sound as good as these regulators used solo per voltage rail. I need to go back and forth a couple of times to confirm which I think is better, but both are very good.
At this point, the sound I'm getting is breakthrough compared to what I had at the beginning of this odessey. Detail, naturalness, impact, density, dynamics, complexity, composure when things gets loud and/or complex are all several levels above using the stock Juli@... which was no slouch itself in the linear/hybrid-powered cMP environment.
I have a few more tricks to try at this point...
1. Finding & cutting superfluous +-12v & +-9v links... I don't use the balanced outputs, so I could un-power those chips. Also since I'm feeding the input voltages directly into the regulators, all of the upstream on-card bus traces are now just sources for interference and noise pickup.
2. Alternative RCA outputs, such as the WBT Next-gen's.
3. Determining if there are transistor-switched muting circuits & disable them.
4. Adding a fourth raw DC supply to feed the 3.3v regulator separate from the digital 5v.
5. Battery-power, especially direct battery power for the 3.3v digital supply as suggested by Alfred.
6. Alternative regulators, such as the Paul Hynes & Teddy Reg units.
7. Physically shield the card within the PC environment.
While I think these are all worthwhile avenues to pursue, I strongly suspect I have made the changes which will provide the greatest improvements... except possibly the battery power and shielding.
So I'll continue to tweak my Juli@ as time allows, but given where I've gotten, I will start to turn my attention to other areas of possible improvement... First, I have some ideas for a custom-fabricated case that will provide better physical and electrical isolation... and second, start to look at the up-coming alternative motherboards and possibly motherboard tweaking.
And of course, watch for SSD prices to fall.
Any other suggestions?
Greg in MississippiP.S. I'll try to post some pictures of the seperately-powered Juli@ this weekend.
Edits: 05/01/09
We're still looking forward to the pictures you promised!
Here they are.
Hard to see much, it's all very 'busy' now.
I'll be working on another Juli@'s digital board this weekend and will take pics as it's done... it'll be easier to see the add-on regulator there.
I think it's a good idea to move the linear supplies out of the computer. I noticed an improvement in SQ after doing so.
Agreed that moving them out is a good plan. But there's always that tradeoff between distance (less radiated field interference) and longer connecting cable (more cable losses & cable quality becomes more important).Also moving them out can reduce the amount of transformer-caused vibration that is fed back into the electronics.
Trying to deal with all of this effectively, I'm slowly gathering ideas for a custom case. Current thoughts include:
1. Three tiers with a butcher-block platform at the base of each tier.
2. PS's mounted onto base of bottom tier. That tier will sit directly on the floor on either spikes or isolation feet and have no solid connection to the upper tiers.
3. Middle tier and top tier bases will be mounted onto a carbon-fiber-based support/surround structure, basically a tube shaped to fit around the butcher block bases. The motherboard mounts on the base of the middle tier with the DAC on the top tier (may hang below to shorten I2S connections).
4. The support/surround structure tube will fit over, but be spaced away from the bottom base and have the top two butcher block bases mounted to it. It will be multi-layer... current plan is copper shield on the inside, then damping material, with a carbon fiber exterior.
5. Each tier will have own shielding zone, with the copper layer on the inside of the support/surround separate for each tier. Then there'll be a copper layer on the top of the base beneath the tier and another on the bottom of the base above. The top-middle-bottom shield layers will be tied together for each zone.
6. A small mobile-type touchscreen will mount on the outside of the support/surround tube in the middle of the middle tier.
This is an 'over the summer' project for me & I'm still tossing around ideas for it.
Greg in Mississippi
P.S. I should note that the GA-EG45M-UD2H motherboard is a bit deeper than the older GA-G31M-S2L, so I'll be shortening the base of the Juli@'s supply & possibly rotating it 90 degrees to make room for the new board.
Edits: 05/09/09
Hey GS,
What about vibration?
How about something like Jon Risch's sandbag tweak on the timing crystal or some rope caulk on the card like all those $39 toshiba dvd player mods had?
Sorry, I do that type of thing so automatically nowadays that I don't always mention it.
I damped the crystals when I did the first mod to the Juli@'s cards... look closely at the picture below in this thread and you can see the small squares of damping materials. Then when I've replaced caps and regulators, I secure and damp them with a strip of Dynamat or 3M damping material from Michael Percy.
Also I've damped the Zalman case top and sides with pieces of Dynamat Extreme and damped all the heatsinks on my add-ons supplies and regulators and the motherboard.
And the Zalman sits on a damped stand on Herbie's damping feet with a collection of Herbie's damping pucks on the top.
My impression of all of this was that the effect was subtle... the largest improvements came from siting of the Zalman on the damping feet and the case damping. More significant was getting the HDD on damping feet with a weight block on top of it and positioning the SATA and power cables to minimize vibrations fed back into the structure.
Greg in Mississippi
P.S. I've also wrapped the HDD in ERS cloth plus a swatch of the stuff on the back of the Zalman touch-screen. Again, the effects were subtle if any.
Well... life happened and I didn't get the time to hookup the separate power supplies to the Juli@ until late this week.And I wasn't happy with it at first. No, it didn't sound bad... but I expected to be blown away and all I got was a little additional detail with a tradeoff of slightly diminished bass and dynamics. Not what I expected out of serious separate supplies. Yah, it could have been component break-in, but I doubted it. Heck, even replacing the caps on the Juli@ made a larger difference.
But this evening, I had a thought and went with it... and it's all changed. My raw supplies were regulated to 5v/+-12v using Dexa discrete regulator boards. Then they went into the 3.3v/5v/+-9v chip regulators on the Juli@. Tonight I replaced the the 3.3v and 5v regulators on the Juli@ with additional Dexa boards. Now I'm getting the improvements in bass, detail, density, texture, and dynamics I expected at first. I guess that the combination of the high-bandwidth Dexa into the somewhat lower bandwidth chip regulators was not ideal. This cMP's sound is VERY compelling and emotive now... and I'm a very happy camper!
And its only gonna get better as it breaks in. Woohoo!
If I get time tomorrow, I'll replace the +-9v regs on the Juli@ that power the output opamps with two more Dexa. Could be good!
Later this week, I'll post a report on what I did to connect up the supplies (I had a learning or two there) and post some pictures.
Between you and me, my cMP is not very pretty... my Zalman case is now FULL with 7 separate linear supplies. I've been mulling over ideas for a custom case... I'll need to move on that if I want to do much more tweaking with this beast! It's also a poster-child example on the benefits of SMPSs... My setup takes up 3-4 times the volume, weighs 5-6 times as much, and produces 2-3 times more heat than an equivalent SMPS.
But it sounds so good now!
Greg in Mississippi
P.S. The difference in sound between the modified Juli@ and having it powered by properly functioning separate supplies reminds me of an early tube-audio exprience I had a few years back. I'd I built up a phono stage that used 6DJ8 tubes and outfitted it with recent-manufacture tubes. Then I found a pair of NOS Mullard 7308s. Woah! I did not imagine it could sound that good. But the best thing here is that none of this setup will age and need to be replaced (except maybe the harddrive someday).
Edits: 04/25/09 04/26/09
Sunday morning update...
It was one of those 'listen until it's way-too-late and I fall asleeep' nights.
The cMP with linear, linear/hybrid, and seperately-powered Juli@ is sounding very, very good.
The improvement in sound quality powering the Juli@ with separate linear supplies and replacing the 3.3v & 5v regulators with high-bandwidth units is in the same magnitude as powering the motherboard with linear/hybrid supplies, perhaps even larger.
Of course, one could look at this and say that I'm just getting to the same place as anyone who's using a separate DAC with their cMP. But as Alfred showed us, an I2S-connected DAC has some potential for greater performance than SPDIF.
Me, I don't care what they say. I'm in listening heaven!
Greg in Mississippi
Hi Greg,
latest update of Julia PS:
I fed the digital section without any regulator directly with LIpofer Batteries. They deliver 3,3 Volt with an inner resistance of 5mOhm-
Sound is phantastic and shows the bad influence of any regulator.
Alfred,Great update. I'm glad to hear you've made further gains.
Personally, I'm cautious about powering things like sound cards which expect a fairly limited range of input voltages directly from batteries which go through a somewhat larger range between fully charged and near empty (easily over a volt for Li-ion and Li-Poly).
Also, I'm not convinced of the evils of regulators, given they are done well.
But I think the key learning here is that computer audio, just like amps, preamps, CD players, whatever, benefits from improvements to the hardware. As best as I can tell, few people are doing things like providing better power supplies for their audio computers the way we are. Personally, I am amazed at the level of improvement in sound quality that I've gained over the past 2 months with some simple and not-too-expensive upgrades.
And of course, there are many ways to make those improvements... battery power, AC power supplies, tweaks to the SMPSs, etc. Just like tubes and solid state!
Let's keep tweaking & finding better ways!
Greg in Mississippi
Edits: 04/26/09
2nd update... first, the breakin continues, but I have a tale to tell about it.After the upgraded Juli@ had been running about a week, I was not happy with the sound. It was muffled and confused sounding. Not what I'd expect from a sprinkling of Black Gates. So I swapped out the modified Juli@ for my un-modified control... no joy, still sounding muffled and confused.
Oh, I remember, a few days after tweaking the Juli@, I decided it was time to finally replace the Holco series Rs in my stepped shunt attenuater with those nude Vishay's I'd had around since last summer. Could the Vishay's be bad? No, they weren't, but when I looked at it, I saw solder joints that weren't pretty. I guess I was tired when I made those joints. So I cleaned the solder off and resoldered... and the magic was back.
The sound of the modified Juli@ after two weeks of burn-in builds on the early strength of a competently & confidently served wealth of details (adding a lot of texture and 'dense-ness' to the sound) and adds an even quieter background, stronger dynamics, stronger deep bass, very well-defined and natural trebles, and a much-improved ability to handle complexity with aplomb.
IF you were building a cMP setup and didn't have a separate DAC, the stock Juli@'s analog outputs by themselves are not bad after being fully burned in... and this modification takes it to a whole new level.
Next there's powering the Juli@ from power separate from the motherboard. Danged, Alfred beat me to it for the 3.3v on the digital card, but I did get a full-house power supply built for just this purpose today (see attached picture).
It's a tweak-ish supply... no magnetic hardware, MSR860 diodes, Jensen 4-pole caps, Dexa regulators for the +-12v, 5v, & 3.3v, and minimal wiring. With a little time-management tomorrow, I may be able to cut the PCI power fingers and wire it in.
I'll keep you posted.
Greg in Mississippi
Edits: 04/12/09
Just a quick update... life interceded and I had no soldering-iron time this weekend, so no update on a separately-powered Juli@ using both the digital and analog sides.
Hopefully this coming weekend will be a bit more sane.
OTOH, the breakin of the previous weekend's mods continue... I'm not hearing that the parts are fully in-song yet, still going thru that love-it/hate-it rollercoaster stage.
Greg in Mississippi
Quick update...
First, as the parts run-in, the modified card is sounding better and better. Especially impressive last night was the bass, dynamics, improved ability to maintain composure when the sound got complex, and naturalness of the treble. I didn't do a comparison with the stock Juli@ at that time, but...
Second, I did do a very quick and dirty comparison Monday night. This was mainly because I took out the modified Juli@ to change the mod a bit... I replaced the Oscons before the 3.3v regulator with 1000uf Black Gates... and replaced the 1000uf Black Gate after that regulator with an Oscon. Theoretically, this is a better configuration... and bass was improved immediately, but I didn't notice any other differences at that time.
Third, inspired by Alfred's post locating an insert-point for 3.3v on the digital card, I spent some time finding and I'll be updating my Juli@ info post above with other insertion points to make adding separate supplies to the cards a bit easier.
Then finally, I'm moving in the direction of adding those supplies this weekend. I have all the parts in-house now. I've been down with a cold the last couple of days, but if I feel better tomorrow evening, I'll start assembling the raw DC side of the beast.
Based on Promethk's & Alfred's comments, I have very high hopes for this... although I wonder if I'll see a smaller SQ lift than others sinced I'm starting with linear & linear/hybrid supplies on the computer.
Later!
Greg in Mississippi
Significant Juli@ abuse occurred last night.I implemented upgrades to all of the power supply rails included local reservoirs & upscaled parts before and after each regulator, using Black Gates on the analog card and Sanyo OsCons on the digital board.
Note that I removed the balanced connectors to make it easier to add the resevoir caps. This is not a must, but did improve access to the connection points.
I also replaced the output coupling caps at the Juli@'s analog outputs.
With only a couple of hours of run-in, it sounds very good and very, very promising. The presentation is 'denser' with more texture and detail, but even with the added details, you can more clearly hear it all as a whole AND pick out individual instruments and singer. The spectral balance seems more even top-to-bottom even though I didn't sense an imbalance before. Treble details are more delicate and audible while strangly seeming less pronounced. Dynamics also increased, downward into microdynamics as much as large-scale dynamics.
I first did the analog card, then the digital card. One interesting observation was that I think I heard larger sonic difference with the digital card upgraded than just the analog card.
I do have an un-modified Juli@ and will do some back and forth, mix and match comparisons after the modified cards have had a chance to run-in a bit.
Next is to setup separate linear supplies for the 5v, & +-12v supplies (the Juli@ does not connect to the computer's 3.3v line) and power the Juli@ separate from the computer supplies.
Greg in Mississippi
P.S. One thing that I think would make a worthwhile difference on the Juli@ analog outputs is to leave the side of the card that you aren't using not-powered. As best as I can tell without digging, both the balanced and un-balanced sides are powered all the time. I'm hoping to find some resistors that can be removed or traces that can be cut to do this.
Edits: 03/29/09 03/29/09 03/29/09
Would you consider doing these mods in exchange for an original juli@ card plus some $? I just cannot imagine doing that type of miniature soldering. You don't have to set a price on AA. But just curious if you would entertain doing it. If so we could settle on a price in private.
Let's take this private.
I'll email you later.
Greg in Mississippi
Thank you for sharing sir!
How large are the Sanyo OsCons used for your Juli@ mod? Are there any particular reasons (other than price and size) for using them over BG Ns on the digital board?
Did you replace all the original caps on the digital board? The pictures wouldn't load in my browser for some weird reason.
If you send me your email address, I'll send you the pix.
On the Oscons, I used them...
1. Because I'd heard they were good for digital power supply bypassing and I hadn't used them before.
2. Because this post referenced another post that recommended Oscons:
http://www.audioasylum.com/forums/pcaudio/messages/35892.html
3. To minimize the impact on my dwindling supply of Black Gates.
I used 180uf/20v Oscons that are currently on-sale at the Parts Connexion. BG N's should work as well or better.
On the digital side, I replaced C1, C2, and C3, which are the caps at the input and output of the 3.3v regulator. Then I added the two caps as suggested in the post I referenced above... note that soldering the one across BC25 is a VERY delicate operation. You want to have your cap secured by double-stick tape and the leads bent to exactly where they need to be before soldering. Then using thin solder and a fine-tipped iron, make a very quick solder joint at one end of BC25, let it cool, then do the one at the other end. Otherwise you can lift and/or damage BC25.
I hope this all helps!
Greg in Mississippi
I've posted this in another thread.
Thanks sir.
I'd used some of those posts as starting points for the information I posted. My hope is to take the Juli@ even farther than described in these posts... ultimately including powering it from a separate set of linear power supplies. Even though I have a separate DAC card coming, I believe that the Juli@ is capable of much better performance than what one gets with a stock card powered from the computer's busses.
At a high level, Here's my roadmap...
Phase I - Improve PS filtering by replacing the input and output caps on the two digital and two analog regulators.
Phase II - Improve analog signal transfer by replacing the output coupling caps.
Phase III - Clean up the power supply by cutting the links to the motherboard's power buses and feed the card with separate linear analog & digital supplies.
Phase IV - Further improve the power by replacing the on-board chip regulators with upgraded discrete regulators.
I have parts either in-house or on order to do the first two phases and hope to be able to report on them after this weekend.
Greg in Mississippi
Are you running analogue out, no exrernal dac?
Theob,Yes, I'm using the analog out of the Juli@. I'm actually very impressed with the SQ I'm getting, but I know it can be improved. I suspect that with better power supply bypassing and output coupling, it will rival many outboard DACs... and with separate linear supplies and upgraded regulators, it will beat most of them.
We'll just have to see. That's one of the main reasons I got a spare Juli@ card.
Later!
Greg in Mississippi
P.S. I should also say that the DAC I'm getting is just a circuit board with a seperate board for regulators. Since it will be connected via the I2S connection, which others have said is not very strong in the Juli@, I'll be locating it inside the computer case just beside the motherboard. Since it will start out with an upgraded chip in a minimalistic implementation, higher-end regulators and a separate set of power supplies, I do expect it will beat the Juli@'s internal DAC, even after modifying the Juli@'s analog circuitry.
Edits: 03/25/09
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: