Hi Greg, I know it's really rubbish, but I still haven't done it. I also plan on building a linear PSU with multiple transformers and independent regulation for each device, but recently, I have been focusing mainly on developing my own natural capacitors (VNP Caps). This research is mostly done now though, so I should have some time for experimenting with my cMP2 soon.
What voltage regulators did you use in your PSU to handle the high-current rails?
@wlowes - you should definitely try Peter's DAC with these modifications: http://www.diyaudio.com/forums/audio-sector/187748-pushing-limits-tda1543-nos-dac-3.html#post2662219. I assure you, after the mods, it is a different animal altogether, and I have not heard a TDA1541-based DAC that would better it.
Last edited:
Wlowes, I like your DAC too!
Lotta work in there. I also like the multiplicity of power supplies, which IMHO, seem to always improve the sound if they aren't done too badly. That trick really made a huge difference in that CDB650 where I originally used that TDA1541... 5 seperate supplies in that and a ton of various regulators.
I also like how you wear your ground thingies & crystals on your sleeve! I generally remove mine before taking pix so people don't scoff as much!
My TDA1541 is not setup for operating now... I need to figure how how to package in into my cMP transport. It will be a challenge to keep the I2S lines as short as I'd like. It will probably wait until I re-package my cMP... likely later this year. And then I'll need to learn about I2S attenuation & DEM reclocking.
What circuit do you use for the 6N2P output stage? Do you do the I/V conversion with this circuit or with passive R's or something else?
My fav pix of an over-the-top TDA1541-based DAC so far is Thorsten's from this thread: http://www.diyaudio.com/forums/digital-line-level/166807-dac-gallery.html .
Here's a few pix of my cMP in a near-current state. It shows the three linear supplies for the soundcard, DAC, & +- for the output stage (plus one for the muting circuit), the two linear supplies with one powering the HDD & SDD (not added at the time of the picture) and the other powering the USB and the touchscreen, and the four linear supplies providing 3.3v, 5v, 12v, & 12v for the ATX-20 & P4 (plus one powering the 5v turn-on control lines... the 9v battery used for -12v is no longer in that supply as my current MoBo does not need it).
And if you look carefully over the back edge of the picture that includes the case, you'll see the choke-filters on the AC lines!
Uncle_Leon,
Those caps look trick. I need to decide what value I can get away with for output couplers for my cMP's DAC... and then hafta order some! Thanks for doing that research-work!
On the cMP's ATX-20 & P4 regulators, I started out with LT-1083 regs for each spot... they were slightly better using John Bau's optimized circuit instead of the stock one.
Now I have Belleson hi-current regs on both the ATX-20 12v & the P4 12v lines... VERY nice improvement with those. I also tried one of those Belleson's in the ATX-20 5v line... not as successful there, I suspect it either did not have sufficient voltage headroom (3.4v headroom under load) or it may just have not been happy with the current level (see below). I have another transformer for the 5v supply that will give me about 2v more headroom to try... just haven't had time to do it yet.
I also went to LT3080 regs on the HDD/SDD & USB/screen supplies which didn't require the current capability of the LT1083 & these also were upgrades sonically.
In doing a linear supply for the Mobo, it is key to understand your current requirements and stay within the limitations of your supplies & regs. All of my regs are pretty high current, but the transformers for the ATX 12v & 3.3v supplies would be pushed if asked to supply much more than 2a (IMHO). I'm using the GA-G31M-S2L board and with the fully-configured cMP setup, my currents are:
P4 12V .45A-.47A (6w, Hammond 166N24, 2 diodes, 18VDC raw)
P20 12V .08A-.12A (1.2w, Hammond 229D24, 2 diodes, 19VDC raw)
P20 5V 2.2A-2.4A (11w, Hammond 166R12, 2 diodes, 8.4VDC raw)
P20 3.3V .27A-.31A (1w, Hammond 229D10, 2 diodes to give 8.7V)
These are the original transformers I used, I went through 3 Hammond 166R12 to find one that didn't hum after awhile and changed the Hammond 229D10 to another brand (have to check my notes if you are interested) after having two that hummed after awhile.
One nice thing is that the linear supply let me lower the CPU voltage another notch or two down to .72500v... I found that one step lower was possible too, but did not sound as good!
The DAC I'm currently using is based on the AK4399 with high-current video driver chips used as the output stages (this is stuff that I got from a friend who modifies equipment and is proprietary to him, so I can't share that). I have a variant of this card using the AK4397 which I haven't used a lot yet... and another using either the AK4395/AK4396 which I haven't tried yet.
The AK4399 really shines with 192 upsampling via cPlay... I try going back to 44.1 occasionally and never like it as much. Of course, that might be entirely different with other DACs.
The Juli@ is powered by two Dexa regs, the DAC by another two, and the output stages by four Dexa.
I also have the ES9022-based DAC (EUVL's) shown in a post above & a Buffalo-II that I haven't finished mounting on a Juli@ yet. And of course, I am hankering to also try a Daniel/Bobken TDA1543 mounted on a Juli@ & I2S-connected... maybe later this fall for the TDA1543 & Buffalo.
All for the night! Later!
Greg in Mississippi
P.S. Also notice all the raw DC caps are Jensen 4-pole. I didn't try other caps for the DAC or Mobo supplies, but they were significant upgrades over either Panasonic FC or Elna Silmec II for the USB/screen & HDD/SSD supplies!
Lotta work in there. I also like the multiplicity of power supplies, which IMHO, seem to always improve the sound if they aren't done too badly. That trick really made a huge difference in that CDB650 where I originally used that TDA1541... 5 seperate supplies in that and a ton of various regulators.
I also like how you wear your ground thingies & crystals on your sleeve! I generally remove mine before taking pix so people don't scoff as much!
My TDA1541 is not setup for operating now... I need to figure how how to package in into my cMP transport. It will be a challenge to keep the I2S lines as short as I'd like. It will probably wait until I re-package my cMP... likely later this year. And then I'll need to learn about I2S attenuation & DEM reclocking.
What circuit do you use for the 6N2P output stage? Do you do the I/V conversion with this circuit or with passive R's or something else?
My fav pix of an over-the-top TDA1541-based DAC so far is Thorsten's from this thread: http://www.diyaudio.com/forums/digital-line-level/166807-dac-gallery.html .
Here's a few pix of my cMP in a near-current state. It shows the three linear supplies for the soundcard, DAC, & +- for the output stage (plus one for the muting circuit), the two linear supplies with one powering the HDD & SDD (not added at the time of the picture) and the other powering the USB and the touchscreen, and the four linear supplies providing 3.3v, 5v, 12v, & 12v for the ATX-20 & P4 (plus one powering the 5v turn-on control lines... the 9v battery used for -12v is no longer in that supply as my current MoBo does not need it).
And if you look carefully over the back edge of the picture that includes the case, you'll see the choke-filters on the AC lines!
Uncle_Leon,
Those caps look trick. I need to decide what value I can get away with for output couplers for my cMP's DAC... and then hafta order some! Thanks for doing that research-work!
On the cMP's ATX-20 & P4 regulators, I started out with LT-1083 regs for each spot... they were slightly better using John Bau's optimized circuit instead of the stock one.
Now I have Belleson hi-current regs on both the ATX-20 12v & the P4 12v lines... VERY nice improvement with those. I also tried one of those Belleson's in the ATX-20 5v line... not as successful there, I suspect it either did not have sufficient voltage headroom (3.4v headroom under load) or it may just have not been happy with the current level (see below). I have another transformer for the 5v supply that will give me about 2v more headroom to try... just haven't had time to do it yet.
I also went to LT3080 regs on the HDD/SDD & USB/screen supplies which didn't require the current capability of the LT1083 & these also were upgrades sonically.
In doing a linear supply for the Mobo, it is key to understand your current requirements and stay within the limitations of your supplies & regs. All of my regs are pretty high current, but the transformers for the ATX 12v & 3.3v supplies would be pushed if asked to supply much more than 2a (IMHO). I'm using the GA-G31M-S2L board and with the fully-configured cMP setup, my currents are:
P4 12V .45A-.47A (6w, Hammond 166N24, 2 diodes, 18VDC raw)
P20 12V .08A-.12A (1.2w, Hammond 229D24, 2 diodes, 19VDC raw)
P20 5V 2.2A-2.4A (11w, Hammond 166R12, 2 diodes, 8.4VDC raw)
P20 3.3V .27A-.31A (1w, Hammond 229D10, 2 diodes to give 8.7V)
These are the original transformers I used, I went through 3 Hammond 166R12 to find one that didn't hum after awhile and changed the Hammond 229D10 to another brand (have to check my notes if you are interested) after having two that hummed after awhile.
One nice thing is that the linear supply let me lower the CPU voltage another notch or two down to .72500v... I found that one step lower was possible too, but did not sound as good!
The DAC I'm currently using is based on the AK4399 with high-current video driver chips used as the output stages (this is stuff that I got from a friend who modifies equipment and is proprietary to him, so I can't share that). I have a variant of this card using the AK4397 which I haven't used a lot yet... and another using either the AK4395/AK4396 which I haven't tried yet.
The AK4399 really shines with 192 upsampling via cPlay... I try going back to 44.1 occasionally and never like it as much. Of course, that might be entirely different with other DACs.
The Juli@ is powered by two Dexa regs, the DAC by another two, and the output stages by four Dexa.
I also have the ES9022-based DAC (EUVL's) shown in a post above & a Buffalo-II that I haven't finished mounting on a Juli@ yet. And of course, I am hankering to also try a Daniel/Bobken TDA1543 mounted on a Juli@ & I2S-connected... maybe later this fall for the TDA1543 & Buffalo.
All for the night! Later!
Greg in Mississippi
P.S. Also notice all the raw DC caps are Jensen 4-pole. I didn't try other caps for the DAC or Mobo supplies, but they were significant upgrades over either Panasonic FC or Elna Silmec II for the USB/screen & HDD/SSD supplies!
Attachments
Last edited:
Your rig looks really impressive Greg - and that really means something coming from me, because I am not easily impressed. Great inspiration.
I like how you mount your transformers using wood - I also found that advantageous. You might want to try encasing your capacitors in wood too (get a small piece of wood, drill it out a bit and put your capacitor in the hole, wrapped in cotton or silk for a snug fit).
Also, it's great to see that some people still have some Black Gates N left
I'm a bit baffled by the copper pieces - what is their purpose?...
I have not tried Jensens, but I've heard they were very good... I found Panasonic FC to be not so great in digital PSUs, and Silmic was rubbish outright. Have you tried Elna Starget or Sanyo Os-Con? They were both pretty good in my tests, especially the Os-Cons (not as good as Black Gate N or FK though).
I was planning to build my proper "Audio PC" around GA-D425TUD - Gigabyte made an inadvertent step towards improving audio quality of their boards by using thicker copper PCBs I also have a feeling that having graphics integrated with the CPU would help to contain some of their inherent noise. I am a bit torn between the single- and dual-core version though: single core might again be less noisy, but dual core might prove beneficial if the cores are explicitly managed (ie. where we assign one core to the audio application exclusively, and the other to handle everything else). At the minute I am leaning towards the single-core... What CPU are you using?
Great info on the regulators, very helpful.
I'm glad you like my caps... I really think they are something special. I am currently looking into scaling the process of making them, to accomodate values above 20nF - it seems feasible, but unfortunately the costs increase with capacity...
In the meantime, I have pretty much finished working on a new, much improved capacitor type - VNM Cap. Here are some pictures of an early sample I have recently finished - I am sending it to Peter Daniel for evaluation:
I like how you mount your transformers using wood - I also found that advantageous. You might want to try encasing your capacitors in wood too (get a small piece of wood, drill it out a bit and put your capacitor in the hole, wrapped in cotton or silk for a snug fit).
Also, it's great to see that some people still have some Black Gates N left
I'm a bit baffled by the copper pieces - what is their purpose?...
I have not tried Jensens, but I've heard they were very good... I found Panasonic FC to be not so great in digital PSUs, and Silmic was rubbish outright. Have you tried Elna Starget or Sanyo Os-Con? They were both pretty good in my tests, especially the Os-Cons (not as good as Black Gate N or FK though).
I was planning to build my proper "Audio PC" around GA-D425TUD - Gigabyte made an inadvertent step towards improving audio quality of their boards by using thicker copper PCBs
I also have a feeling that having graphics integrated with the CPU would help to contain some of their inherent noise. I am a bit torn between the single- and dual-core version though: single core might again be less noisy, but dual core might prove beneficial if the cores are explicitly managed (ie. where we assign one core to the audio application exclusively, and the other to handle everything else). At the minute I am leaning towards the single-core... What CPU are you using?Great info on the regulators, very helpful.
I'm glad you like my caps... I really think they are something special. I am currently looking into scaling the process of making them, to accomodate values above 20nF - it seems feasible, but unfortunately the costs increase with capacity...
In the meantime, I have pretty much finished working on a new, much improved capacitor type - VNM Cap. Here are some pictures of an early sample I have recently finished - I am sending it to Peter Daniel for evaluation:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I take that as a true compliment...
So thank you. And of course, there's lots more to do... That's what makes it fun!
On what you mentioned...
- Actually bamboo, not wood. I do need to spend some time (SOMEDAY!) listening to the signatures of various subtrates & bases. In the meantime, I depend a lot on the comparisons & opinions of those I who actually compare & listen & I find I can trust (even if I don't agree with everything they say, such as Thorsten's avid dislike of BGs). Hence my appreciation for your work on your line of capacitors... And while I didn't mention this before, my appreciation for the work you did on comparing capacitors on the Shigaclone.
- And on that comparison, I was very hopeful that you would find a cap that would rival the BGs. I do use Oscons in digital circuits (they are hard to see, but there are two of the three I used visible on my modified Juli@ card with the ES9022 DAC card a few posts above), but I have been cautious about using them in analog-ish circuits after Jean-Paul's comments here: http://www.diyaudio.com/forums/chip...g-audio-integrated-opamps-38.html#post2162584 ,& Jean-Paul's & Thorston's comments here: http://www.diyaudio.com/forums/digital-source/15020-use-os-con-vcc.html . And of course, I need to find a good source for the Stargets so I can try them.
- And I do have a small stock of BG's, but I am well aware they won't last forever (or even very long!), so I am keenly interested in quests to find good alternatives.
- And thanks on the tip about imbedding the capacitors. I also saw how Peter D did that with plastic (Delron or Nylon?) & Michael Elliot's (0f Counterpoint & Alta Vista Audio fame) mounting of the caps in his hybrid amps shown here: http://ariaaudio.com/WTXLCapsLarge.jpg . Recognizing that stabilizing the caps against a substrate with a desireable resonance signature was why I do have them tightly mounted against the bamboo base. And I need to start moving towards silk/cotton as another useful filler/insulator/damper... I've heard too many strong recommendations to ignore this.
- Also very interesting MoBo choice. If you aren't upsampling & even better yet, already have your music files in WAV format (mine are all in FLAC... I'm waiting for storage costs to come down more and then I'll pre-convert them to WAV), the Atom solutions look very interesting. And especially that board... I do like the heavy copper (the UD2H board I tried did some very nice things... I still preferred the S2L, mainly due to the improved treble purity after turning off Spread Spectrum, but many would have preferred the 'robustness' of the UD2H)... I also like that it has a separate CPU ATX12 plug to allow you to separate that as a 'dirty' PS source... And I REALLY like it's simplicity in that it has a limited number of SATA, PCI & other ports & no PCI-E or HDMI. I may have to try one! (but bummer, the BIOS does not allow a lot of flexibility... according to the manual.) I do know that I don't know enough about how computers work internally and how that affects the sound to have a useful opinion on single vs dual core and the benefits of integrated graphics.
- I'm currently using an E7200 CPU on an old GA-G31M-S2L board. I intend to do some motherboard mods such as shown in these threads:
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
I laid in a small stock of these boards as they were fairly inexpensive at their end-of-life & I won't feel so bad butchering them. Then after I've gotten some good results (and let some smoke out!), I can duplicate the process on the then-current cMP boards
- The copper bars on the CPU heatsink are the remnants of my first try at a Shigaclone... Note the badly-drilled holes. Luckily I was saved by Peter's simplification of the mechanics. I attached a picture of my version.
They are merely there as an attempt to do some damping of that heatsink.
The copper chunk where the DAC is mounted is really a chunk of bamboo with a copper foil covering as an attempt to shield the DAC & output stages somewhat.
Please keep up the great development work on the caps!
Greg in Mississippi
So thank you. And of course, there's lots more to do... That's what makes it fun!
On what you mentioned...
- Actually bamboo, not wood. I do need to spend some time (SOMEDAY!) listening to the signatures of various subtrates & bases. In the meantime, I depend a lot on the comparisons & opinions of those I who actually compare & listen & I find I can trust (even if I don't agree with everything they say, such as Thorsten's avid dislike of BGs). Hence my appreciation for your work on your line of capacitors... And while I didn't mention this before, my appreciation for the work you did on comparing capacitors on the Shigaclone.
- And on that comparison, I was very hopeful that you would find a cap that would rival the BGs. I do use Oscons in digital circuits (they are hard to see, but there are two of the three I used visible on my modified Juli@ card with the ES9022 DAC card a few posts above), but I have been cautious about using them in analog-ish circuits after Jean-Paul's comments here: http://www.diyaudio.com/forums/chip...g-audio-integrated-opamps-38.html#post2162584 ,& Jean-Paul's & Thorston's comments here: http://www.diyaudio.com/forums/digital-source/15020-use-os-con-vcc.html . And of course, I need to find a good source for the Stargets so I can try them.
- And I do have a small stock of BG's, but I am well aware they won't last forever (or even very long!), so I am keenly interested in quests to find good alternatives.
- And thanks on the tip about imbedding the capacitors. I also saw how Peter D did that with plastic (Delron or Nylon?) & Michael Elliot's (0f Counterpoint & Alta Vista Audio fame) mounting of the caps in his hybrid amps shown here: http://ariaaudio.com/WTXLCapsLarge.jpg . Recognizing that stabilizing the caps against a substrate with a desireable resonance signature was why I do have them tightly mounted against the bamboo base. And I need to start moving towards silk/cotton as another useful filler/insulator/damper... I've heard too many strong recommendations to ignore this.
- Also very interesting MoBo choice. If you aren't upsampling & even better yet, already have your music files in WAV format (mine are all in FLAC... I'm waiting for storage costs to come down more and then I'll pre-convert them to WAV), the Atom solutions look very interesting. And especially that board... I do like the heavy copper (the UD2H board I tried did some very nice things... I still preferred the S2L, mainly due to the improved treble purity after turning off Spread Spectrum, but many would have preferred the 'robustness' of the UD2H)... I also like that it has a separate CPU ATX12 plug to allow you to separate that as a 'dirty' PS source... And I REALLY like it's simplicity in that it has a limited number of SATA, PCI & other ports & no PCI-E or HDMI. I may have to try one! (but bummer, the BIOS does not allow a lot of flexibility... according to the manual.) I do know that I don't know enough about how computers work internally and how that affects the sound to have a useful opinion on single vs dual core and the benefits of integrated graphics.
- I'm currently using an E7200 CPU on an old GA-G31M-S2L board. I intend to do some motherboard mods such as shown in these threads:
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
I laid in a small stock of these boards as they were fairly inexpensive at their end-of-life & I won't feel so bad butchering them. Then after I've gotten some good results (and let some smoke out!), I can duplicate the process on the then-current cMP boards
- The copper bars on the CPU heatsink are the remnants of my first try at a Shigaclone... Note the badly-drilled holes. Luckily I was saved by Peter's simplification of the mechanics. I attached a picture of my version.
They are merely there as an attempt to do some damping of that heatsink.
The copper chunk where the DAC is mounted is really a chunk of bamboo with a copper foil covering as an attempt to shield the DAC & output stages somewhat.
Please keep up the great development work on the caps!
Greg in Mississippi
Attachments
Last edited:
Greg, thanks.. I am learning that power is the key. I am lucky to have the Plitron transformer factory close by. I often find them in surplus stores here in Toronto. This DAC has 3 that cost less than $10 ea.
I use the Lampzator SRPP circuit with 6N2P tubes. Signal direct from the 1541a pin to the grid. R2 is a 90R resistor from the grid to ground. Simple but effective. teachings of Prometheus
Great post btw. I will go through it carefully on the weekend.. You have a wealth of information here.
Walter
I have been interested in building a power supply for my music server for a while now. The supply described in the CMP2 project is over my head. I'd probably need detailed instructions, to just have a shot at getting it to work.
Thus, I've been considering some more simple approaches, like the thread a few pages back, about powering the Juli@ with SLA battery. I found the product below, which looks like it could power the whole machine, but I'm wondering if it's too good to be true and might have some drawbacks.
mini-itx.com - store - power supplies
I believe that 40Ah of a 12 SLA could power my stripped down Core2 server for about 6-7 hrs. I've hesitated going to the i3 series despite the lower power, because a few people who tried them in their server told me they didn't like the sound as much as the Core2. However, the Ivy Bridge is coming in March, with even lower power, so maybe that will be an opportunity to really extend battery life.
I'd appreciate it if anyone can give me some feedback on the product listed above, links to potential PC power supply / battery plans or critiques of my thinking here.
Thanks
Thus, I've been considering some more simple approaches, like the thread a few pages back, about powering the Juli@ with SLA battery. I found the product below, which looks like it could power the whole machine, but I'm wondering if it's too good to be true and might have some drawbacks.
mini-itx.com - store - power supplies
I believe that 40Ah of a 12 SLA could power my stripped down Core2 server for about 6-7 hrs. I've hesitated going to the i3 series despite the lower power, because a few people who tried them in their server told me they didn't like the sound as much as the Core2. However, the Ivy Bridge is coming in March, with even lower power, so maybe that will be an opportunity to really extend battery life.
I'd appreciate it if anyone can give me some feedback on the product listed above, links to potential PC power supply / battery plans or critiques of my thinking here.
Thanks
Brilliant post Greg, thanks.
I have recently bought some bamboo for tests in my capacitors - I too am curious about sonic signatures of different timbers. I also managed to obtain a few ebony pieces for the same purpose - unfortunately ebony is so exceedingly rare - not to mention expensive - that, even if it sounds as good as I'm hoping, it will never be widely available... Bone, shell and ivory are other promising (if a bit extravagant) directions for materials research.
The main reason I did my capacitor tests was the increasing rarity of Black Gates... I was lucky to buy just enough for my personal use, while they could still be bought. But I think it is really unfair that people can't have better sound, now that BG are no more... I was hoping to discover a cap that would be a worthy replacement - unfortunately, while some were pretty good, none could better the FK or N types. If my businesses takes off, I'm hoping to buy the license from Jelmax and restart their production (and improve them, while I'm at it). I don't quite understand why no one has done it yet.
As far as mounting caps goes, I want to go further, and try getting rid of the aluminium can and that nasty black rubber stopper, and embed the actual cap in a drilled-out piece of wood and fill it with a... filler Haven't quite decided on that one - it will take some auditioning to choose; but I will try wax, bio-resin (for casting) and curing oils.
For damping your heatsink, try soldering a few pieces of thick wire along the heatsink, perpendicular to the aluminium "leafs" to join and stiffen them. If your soldering iron is not up to it, some hot melt glue should do the job quite as well, although it will impede heat transfer a little. Oh yeah, and solid copper IS a bas**rd to drill
@ fork - quite a few people have employed PicoPSU successfully in their CMP2s - although it needs to be a fairly low-powered machine. Also check Peter Daniel's modification of it here.
I have recently bought some bamboo for tests in my capacitors - I too am curious about sonic signatures of different timbers. I also managed to obtain a few ebony pieces for the same purpose - unfortunately ebony is so exceedingly rare - not to mention expensive - that, even if it sounds as good as I'm hoping, it will never be widely available... Bone, shell and ivory are other promising (if a bit extravagant) directions for materials research.
The main reason I did my capacitor tests was the increasing rarity of Black Gates... I was lucky to buy just enough for my personal use, while they could still be bought. But I think it is really unfair that people can't have better sound, now that BG are no more... I was hoping to discover a cap that would be a worthy replacement - unfortunately, while some were pretty good, none could better the FK or N types. If my businesses takes off, I'm hoping to buy the license from Jelmax and restart their production (and improve them, while I'm at it). I don't quite understand why no one has done it yet.
As far as mounting caps goes, I want to go further, and try getting rid of the aluminium can and that nasty black rubber stopper, and embed the actual cap in a drilled-out piece of wood and fill it with a... filler
Haven't quite decided on that one - it will take some auditioning to choose; but I will try wax, bio-resin (for casting) and curing oils.For damping your heatsink, try soldering a few pieces of thick wire along the heatsink, perpendicular to the aluminium "leafs" to join and stiffen them. If your soldering iron is not up to it, some hot melt glue should do the job quite as well, although it will impede heat transfer a little. Oh yeah, and solid copper IS a bas**rd to drill
@ fork - quite a few people have employed PicoPSU successfully in their CMP2s - although it needs to be a fairly low-powered machine. Also check Peter Daniel's modification of it here.
Thanks for the tip U Leon. I had heard of Pico, but didn't realize it was exactly what I was thinking of. Greg mentions above that the CPU PS can be dirty, so I might try plugging the case PS into the 4-pin connector and use the Pico on the 24-pin, to extend the battery life. Although the Peter Daniel PS looks like something I could pull off and more convenient than battery, so I'll have give it some thought.
My source on the unfavorable sound of the "i" series chip was Gary Koh, from Genesis Loudspeakers, a friend from my audio club. Gary has posted a document on his website regarding his server efforts. A while back he told me he tried single core processors, the Core2 and the i series and found that the Core2 sounded better than the other two. It's hard to get feedback on this, because not many people have done it, but Gary has very good ears.
It looks like the Ivy Bridge may not be available until Q3-4 of next year. I've seen conflicting reports, but I'm betting on the later date. Too bad, since the 22nm, 35-45 watt models sound really cool, not to mention all the changes in motherboard technology that will be coming with it.
My source on the unfavorable sound of the "i" series chip was Gary Koh, from Genesis Loudspeakers, a friend from my audio club. Gary has posted a document on his website regarding his server efforts. A while back he told me he tried single core processors, the Core2 and the i series and found that the Core2 sounded better than the other two. It's hard to get feedback on this, because not many people have done it, but Gary has very good ears.
It looks like the Ivy Bridge may not be available until Q3-4 of next year. I've seen conflicting reports, but I'm betting on the later date. Too bad, since the 22nm, 35-45 watt models sound really cool, not to mention all the changes in motherboard technology that will be coming with it.
Last edited:
Three responses in one...
Fork...
I was going to point you to the same Peter D post that Uncle Leon listed. I was inspired by those posts & Peter D's use of the PicoPSU to do my first DIY computer supply... What I like to call a 'linear/hybrid' supply because the PicoPSU does have switching supplies that provide the rails other than the 12v. But powered from a good 12v supply such as Peter's and after modifying the PicoPSU with better caps (I used the same BG N's that Peter recommended), it sounded pretty good, easily 60-70 percent of the SQ of a fully linear supply.
My posts on doing that supply are here: Audio Asylum Thread Printer . My supply was very much the same as Peter's, but for the best sound, I recommend the following changes:
1. Use a variable LT-1083 instead of the fixed one. This allows you to bypass the voltage setting resistor with a good capacitor, lowering the output noise. For extra credit, find John Bau's (designer of the Spica time-coherent linear-phase speakers!) DIYAUDIO thread here: http://www.diyaudio.com/forums/power-supplies/143539-another-look-lm317-lm337-regulators.html on optimizing the 317/337 regulators and use his circuit instead of the stock datasheet one (it works for the 1083 too!).
2. Use very good caps for the raw DC filter caps... I used Jenson 4-pole caps, BG FK would also be good here.
3. Build a 2nd regulated 12v supply to power the CPU via the P4 connector. As it is a 'dirty' power consumer, use a separate AC cord & connect it to a separate AC circuit from the rest of your system.
People have reported that you want to stick with the standard PicoPSUs, not the 'wide range' ones designed for battery power... They reported that the wide-range ones did not provide much improvement over a standard computer supply.
I suggest you do a search on both the cMP & cPlay threads and the larger PC Audio Asylum for posts on PicoPSU's for more details on how to best implement them. For example, there's a recent suggestion in the cMP thread on which PicoPSU model sounds best.
Then there is a recent move to provide additional filtering on standard computer supplies using a heroic collection of clamp-on chokes & capacitors... See posts by hfavandepas here: Audio Asylum Thread Printer and here: Audio Asylum Thread Printer . This may provide a large measure of the SQ improvement of a PicoPSU powered by a good supply and may even surpass it, but the same techniques applied to the PicoPSU setup will also provide benefits and I suspect, lift it higher. I also suspect that this technique will provide benefits on a fully-linear supply like I currently use, but before I go there, I want to first repackage the regulators & heatsinks such that they are mounted on the ATX-20/24 & P4 connectors and the connecting wiring between the regulator outputs and plugs is very minimal.
And do note that the current 'official' cMP power recommendation is to use a 2nd computer supply for the P4 & other dirty components powered via a separate power cord to a 2nd AC circuit.
But my experience & impression is that you'll get the best results replacing the SMPS computer supply with good linear ones.
On the SQ of the Core i3 Processors, if you are doing a cMP, that's the current recommended hardware and a number of people have reported on the SQ improvements going to that setup.
BUT, I believe it works as well as it does due to:
1. The various techniques used to minimize power usage such as an slimmed-down Windows XP operating system, specific BIOS settings (including using only one of the cores... Take note, Uncle Leon!), and minimal memory.
2. The improvements in construction of the recommended motherboard (GA-H55M-UD2H) including the heavy copper layers & lots of Oscons or equivalents.
3. The various techniques used to maximize SQ out of this system such as adjusting latency & minimizing the software & operating system produced interruptions... including the use of cMP & cPlay which are designed to leverage & maximize all the above.
If you aren't doing these things, your results will differ. That doesn't mean that there are not other ways to make good-sounding computer music players... But doing a cMP/cPlay setup with the Core I3 HW will work well and sound very good. And you can use 'cMP-techniques' in any computer music player setup... But IMHO, it's not really a cMP^2 setup unless you use a Win XP-based system on one of the recommended hardware setups with the recommended customizations & optimizations. (Off my soapbox now... Sorry!)
Finally, thanks for mentioning Gary Koh's Music Server Project. Interesting & useful reading, both the white paper & the forum. Funny... before I went the computer music route, I was preparing to re-burn my CD's on black blanks. Glad I never went there!
Wlowes...
You made me miss good surplus stores. Here in the rural areas of Mississippi, they don't exist. In Fort Wayne Indiana where I grew up & in Minneapolis where I lived for about 20 years before moving down here, I had access to very good surplus outlets & hamfests. Great deals on Shallco/Daven multiple-position switches for selectors & volume controls, good resistors, capacitors, semiconductors, & transformers, & even some great & reasonably priced NOS tubes were all available. I'm glad I stocked as much as I did back then and wish I'd gotten more! I'm envious on your Plitron deals!
Also, thanks for the info on your I/V. I was planning to try Thorsten's circuit for the TDA1541 on that and the same somewhat modified as an I/V stage on the Buffalo DAC's ES9018.
Uncle Leon...
I'm really impressed with how you are working out of the box. Electrolytic caps outside of the ally case/rubber plug paradigm could be very, very trick. I really look forward to your work on these and on reviving (and hopefully improving on) the Black Gates!
Also thanks for the tips on damping the heatsink. I do have some EAR SD40AL on the sides of the heatsinks which also helps. Long-term, I want to try some smaller and easier-to-damp heatsinks such as the ones Peter D showed here: http://www.diyaudio.com/forums/audi...hip-amp-kits-dacs-chassis-52.html#post1787334 & here: http://www.diyaudio.com/forums/audi...hip-amp-kits-dacs-chassis-50.html#post1706933 . I also attached some other Peter D computer setup pictures that he does not have available online anymore.
Related, my cMP responds very well to vibration-management tweaks. It is on a damped glass platform support stand, sitting on Herbies Iso-cups. I use a combination of the above-mentioned EAR SD40AL and Dynamat Xtreme to damp the case, heatsinks, & electronic components such as capacitors. I have the damping pads from the original Herbie tube dampers on the motherboard mounting posts. And my power supplies and drives are either coupled via brass acorn nuts on the transformer brass mounting bolts or decoupled using various Herbies devices. Power supplies & the drives are plus all the wires are carefully dressed so that they are not touching other components or each other (when possible).
All of this is key the the sound I get... Especially the subcomponent & wire dressing. Just having a wire touching one of the power supplies or a drive will take it from sounding good to just ok.
And I am quite sure I really don't have this arena optimized yet.
Enough for tonight!
Greg in Mississippi
Fork...
I was going to point you to the same Peter D post that Uncle Leon listed. I was inspired by those posts & Peter D's use of the PicoPSU to do my first DIY computer supply... What I like to call a 'linear/hybrid' supply because the PicoPSU does have switching supplies that provide the rails other than the 12v. But powered from a good 12v supply such as Peter's and after modifying the PicoPSU with better caps (I used the same BG N's that Peter recommended), it sounded pretty good, easily 60-70 percent of the SQ of a fully linear supply.
My posts on doing that supply are here: Audio Asylum Thread Printer . My supply was very much the same as Peter's, but for the best sound, I recommend the following changes:
1. Use a variable LT-1083 instead of the fixed one. This allows you to bypass the voltage setting resistor with a good capacitor, lowering the output noise. For extra credit, find John Bau's (designer of the Spica time-coherent linear-phase speakers!) DIYAUDIO thread here: http://www.diyaudio.com/forums/power-supplies/143539-another-look-lm317-lm337-regulators.html on optimizing the 317/337 regulators and use his circuit instead of the stock datasheet one (it works for the 1083 too!).
2. Use very good caps for the raw DC filter caps... I used Jenson 4-pole caps, BG FK would also be good here.
3. Build a 2nd regulated 12v supply to power the CPU via the P4 connector. As it is a 'dirty' power consumer, use a separate AC cord & connect it to a separate AC circuit from the rest of your system.
People have reported that you want to stick with the standard PicoPSUs, not the 'wide range' ones designed for battery power... They reported that the wide-range ones did not provide much improvement over a standard computer supply.
I suggest you do a search on both the cMP & cPlay threads and the larger PC Audio Asylum for posts on PicoPSU's for more details on how to best implement them. For example, there's a recent suggestion in the cMP thread on which PicoPSU model sounds best.
Then there is a recent move to provide additional filtering on standard computer supplies using a heroic collection of clamp-on chokes & capacitors... See posts by hfavandepas here: Audio Asylum Thread Printer and here: Audio Asylum Thread Printer . This may provide a large measure of the SQ improvement of a PicoPSU powered by a good supply and may even surpass it, but the same techniques applied to the PicoPSU setup will also provide benefits and I suspect, lift it higher. I also suspect that this technique will provide benefits on a fully-linear supply like I currently use, but before I go there, I want to first repackage the regulators & heatsinks such that they are mounted on the ATX-20/24 & P4 connectors and the connecting wiring between the regulator outputs and plugs is very minimal.
And do note that the current 'official' cMP power recommendation is to use a 2nd computer supply for the P4 & other dirty components powered via a separate power cord to a 2nd AC circuit.
But my experience & impression is that you'll get the best results replacing the SMPS computer supply with good linear ones.
On the SQ of the Core i3 Processors, if you are doing a cMP, that's the current recommended hardware and a number of people have reported on the SQ improvements going to that setup.
BUT, I believe it works as well as it does due to:
1. The various techniques used to minimize power usage such as an slimmed-down Windows XP operating system, specific BIOS settings (including using only one of the cores... Take note, Uncle Leon!), and minimal memory.
2. The improvements in construction of the recommended motherboard (GA-H55M-UD2H) including the heavy copper layers & lots of Oscons or equivalents.
3. The various techniques used to maximize SQ out of this system such as adjusting latency & minimizing the software & operating system produced interruptions... including the use of cMP & cPlay which are designed to leverage & maximize all the above.
If you aren't doing these things, your results will differ. That doesn't mean that there are not other ways to make good-sounding computer music players... But doing a cMP/cPlay setup with the Core I3 HW will work well and sound very good. And you can use 'cMP-techniques' in any computer music player setup... But IMHO, it's not really a cMP^2 setup unless you use a Win XP-based system on one of the recommended hardware setups with the recommended customizations & optimizations. (Off my soapbox now... Sorry!)
Finally, thanks for mentioning Gary Koh's Music Server Project. Interesting & useful reading, both the white paper & the forum. Funny... before I went the computer music route, I was preparing to re-burn my CD's on black blanks. Glad I never went there!
Wlowes...
You made me miss good surplus stores. Here in the rural areas of Mississippi, they don't exist. In Fort Wayne Indiana where I grew up & in Minneapolis where I lived for about 20 years before moving down here, I had access to very good surplus outlets & hamfests. Great deals on Shallco/Daven multiple-position switches for selectors & volume controls, good resistors, capacitors, semiconductors, & transformers, & even some great & reasonably priced NOS tubes were all available. I'm glad I stocked as much as I did back then and wish I'd gotten more! I'm envious on your Plitron deals!
Also, thanks for the info on your I/V. I was planning to try Thorsten's circuit for the TDA1541 on that and the same somewhat modified as an I/V stage on the Buffalo DAC's ES9018.
Uncle Leon...
I'm really impressed with how you are working out of the box. Electrolytic caps outside of the ally case/rubber plug paradigm could be very, very trick. I really look forward to your work on these and on reviving (and hopefully improving on) the Black Gates!
Also thanks for the tips on damping the heatsink. I do have some EAR SD40AL on the sides of the heatsinks which also helps. Long-term, I want to try some smaller and easier-to-damp heatsinks such as the ones Peter D showed here: http://www.diyaudio.com/forums/audi...hip-amp-kits-dacs-chassis-52.html#post1787334 & here: http://www.diyaudio.com/forums/audi...hip-amp-kits-dacs-chassis-50.html#post1706933 . I also attached some other Peter D computer setup pictures that he does not have available online anymore.
Related, my cMP responds very well to vibration-management tweaks. It is on a damped glass platform support stand, sitting on Herbies Iso-cups. I use a combination of the above-mentioned EAR SD40AL and Dynamat Xtreme to damp the case, heatsinks, & electronic components such as capacitors. I have the damping pads from the original Herbie tube dampers on the motherboard mounting posts. And my power supplies and drives are either coupled via brass acorn nuts on the transformer brass mounting bolts or decoupled using various Herbies devices. Power supplies & the drives are plus all the wires are carefully dressed so that they are not touching other components or each other (when possible).
All of this is key the the sound I get... Especially the subcomponent & wire dressing. Just having a wire touching one of the power supplies or a drive will take it from sounding good to just ok.
And I am quite sure I really don't have this arena optimized yet.
Enough for tonight!
Greg in Mississippi
Attachments
Last edited:
Uncle Leon...
A couple more thoughts on regulators for a fully linear supply...
After trying different regulators in my various supplies over time, this was the progression from good to best:
LT-1083 fixed -> LT-1083 variable w/bypass -> Lt-1083 Bau circuit -> LT-3080 -> Belleson high-current
Some I haven't tried yet, but I suspect will have merit are:
- paralleled LT-3080 for higher current
- AMB Sigma
- Twisted Pear Placid HDs
- Salas low-voltage high-current (see here: http://www.diyaudio.com/forums/power-supplies/168631-5v-high-current-low-noise-regulator.html)
- Paul Hynes'
My only short-term plans for upgrades to my linear supply is to try a transformer that will give me a little more raw DC voltage & re-try the Belleson 5v, hopefully later this fall.
At some point, I'll try a Belleson 3.3v there, but right now it's slated for a shoot-out between Dexa / Belleson / Hynes / LiFePO4 battery sources on the Juli@ digital section for my Buffalo-II.
Greg in Mississippi
P.S. And I forgot to mention that there's significant merit in slightly under-volting the ATX-20/24 & P4 supply voltages... see here:
http://www.audioasylum.com/forums/pcaudio/messages/9/90087.html
There are other posts in the cMP & cPlay threads & a mention in the advanced section of the cMP docs about under-volting the 12v lines from standard SMPS computer supplies.
A couple more thoughts on regulators for a fully linear supply...
After trying different regulators in my various supplies over time, this was the progression from good to best:
LT-1083 fixed -> LT-1083 variable w/bypass -> Lt-1083 Bau circuit -> LT-3080 -> Belleson high-current
Some I haven't tried yet, but I suspect will have merit are:
- paralleled LT-3080 for higher current
- AMB Sigma
- Twisted Pear Placid HDs
- Salas low-voltage high-current (see here: http://www.diyaudio.com/forums/power-supplies/168631-5v-high-current-low-noise-regulator.html)
- Paul Hynes'
My only short-term plans for upgrades to my linear supply is to try a transformer that will give me a little more raw DC voltage & re-try the Belleson 5v, hopefully later this fall.
At some point, I'll try a Belleson 3.3v there, but right now it's slated for a shoot-out between Dexa / Belleson / Hynes / LiFePO4 battery sources on the Juli@ digital section for my Buffalo-II.
Greg in Mississippi
P.S. And I forgot to mention that there's significant merit in slightly under-volting the ATX-20/24 & P4 supply voltages... see here:
http://www.audioasylum.com/forums/pcaudio/messages/9/90087.html
There are other posts in the cMP & cPlay threads & a mention in the advanced section of the cMP docs about under-volting the 12v lines from standard SMPS computer supplies.
Last edited:
P.S. Fork, another thanks to you on the MG Super Shield. This will also likely work on steel cases like the copper plating that Michael Elliot did for his later versions of the Counterpoint SA-100 & SA-220 amps. Did you see the comments on the Asylum by Elizabeth on using anti-static foam for absorbing RF:
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Also see the various comments on the Arcam Stealth Mat & Mask of Silence in the Arcam EQ reviews here: Home Page
From his comments & others, these appear to be ferrite sheets (Mask of Silence) & ERS-style mat (Stealth Mat). I've been hunting for info on how they are implemented by Arcam... got any sources?
I guess the other posts really weren't enough for the night. Sorry for the thought dumps!
Greg in Mississippi
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Also see the various comments on the Arcam Stealth Mat & Mask of Silence in the Arcam EQ reviews here: Home Page
From his comments & others, these appear to be ferrite sheets (Mask of Silence) & ERS-style mat (Stealth Mat). I've been hunting for info on how they are implemented by Arcam... got any sources?
I guess the other posts really weren't enough for the night. Sorry for the thought dumps!
Greg in Mississippi
Hi Greg, thanks for all the work on your cMP2 project. It's been a tremendous source of learning and inspiration for me.
My machine is a dual boot audio server and Blu-ray/HT system. It's more of a standard build, Win7, all passively cooled with a Nightjar fanless PS, using an Entreq cable via async USB to a Wyred DAC 2 (I've also been using the new Burson DA-160 lately). The display is HDMI to a 58" plasma.
I've done a lot of work on Win7 to strip it down and I haven't been able to go any further with it for a while now. I don't know how Win7 stands up to XP, but I've got it turned down to the kernel and a handful of critical services and processes. I cut about 2/3 of the installation away with RT7 Lite, tested all the Autoruns and drivers and found the link below from a guy who tested and documented about 500 of the registry keys for stability (I don't know how he ever got through it). One really cool feature of my audio boot section is that when the display goes to sleep, the plasma screen goes to sleep completely, as if it's off and with no back light or heat, but a tap of the mouse will wake it up; the problem is, I haven't been able to figure out how or why it does this. When I go from my HT boot section to the audio section the difference is not subtle and almost all of those concepts came from the old cMP site. I plan to document what I've learned in a blog that I'm putting together for this fall. I expect my next build will come closer to your setup, with a Juli@ card, out to AES/EBU.
Disable Unnecessary Drivers And Services Windows 7 - Graphix an Stuff!
A tip I got from Gary's build that I haven't heard many talk about is using an externally powered SATA HD enclosure, for the music library. Apparently the SATA is galvanically isolated and he says he was not able to discern between music played from an SSD and this configuration. Gary also said he was able to disable Plug and Play with XP, to great effect, but I wasn't able to get it to work with my DAC and apparently Win7 depends on P&P to make any audio.
As a member of our board for the Pacific Northwest Audio Society, I volunteered to coordinate our new digital playback system. The all-linear PS sounds so cool that this is the route I want to go for the club, if I can get a few members to lend their expertise. Hopefully you'll be online from time to time this fall.
Cheers
My machine is a dual boot audio server and Blu-ray/HT system. It's more of a standard build, Win7, all passively cooled with a Nightjar fanless PS, using an Entreq cable via async USB to a Wyred DAC 2 (I've also been using the new Burson DA-160 lately). The display is HDMI to a 58" plasma.
I've done a lot of work on Win7 to strip it down and I haven't been able to go any further with it for a while now. I don't know how Win7 stands up to XP, but I've got it turned down to the kernel and a handful of critical services and processes. I cut about 2/3 of the installation away with RT7 Lite, tested all the Autoruns and drivers and found the link below from a guy who tested and documented about 500 of the registry keys for stability (I don't know how he ever got through it). One really cool feature of my audio boot section is that when the display goes to sleep, the plasma screen goes to sleep completely, as if it's off and with no back light or heat, but a tap of the mouse will wake it up; the problem is, I haven't been able to figure out how or why it does this. When I go from my HT boot section to the audio section the difference is not subtle and almost all of those concepts came from the old cMP site. I plan to document what I've learned in a blog that I'm putting together for this fall. I expect my next build will come closer to your setup, with a Juli@ card, out to AES/EBU.
Disable Unnecessary Drivers And Services Windows 7 - Graphix an Stuff!
A tip I got from Gary's build that I haven't heard many talk about is using an externally powered SATA HD enclosure, for the music library. Apparently the SATA is galvanically isolated and he says he was not able to discern between music played from an SSD and this configuration. Gary also said he was able to disable Plug and Play with XP, to great effect, but I wasn't able to get it to work with my DAC and apparently Win7 depends on P&P to make any audio.
As a member of our board for the Pacific Northwest Audio Society, I volunteered to coordinate our new digital playback system. The all-linear PS sounds so cool that this is the route I want to go for the club, if I can get a few members to lend their expertise. Hopefully you'll be online from time to time this fall.
Cheers
I thought readers on this thread may benefit from what I found. Instead of modifying the power supply to the computer, I use PCI-e extension card and modified the sound card power supply
Computer is noisy
Computer is noisy
Greg,
I was wondering if I could pick your brains?
I have a Intel D510M mini-itx based pc I'm using for music duties, and currently use a picopsu for supply duties. After various experiences with other bits of audio, I've become convinced that going to a linear supply instead of the picopsu will give me an improvement in SQ. Separate supplies for USB, HDD, SDD, Sound card are also a given.
Building a set of decent linear +12v, -12V, +5V and +3.3V supplies for the motherboard will be straightforward enough (until I win the lottery and have a chat with Paul Hynes!), and I'm lucky enough that Intel publish current consumption for the supply lines for this motherboard, but I was wondering what you did with the control signals?
i.e. On the ATX connector
Pin 8 (Grey) Power Good
Pin 9 (Purple) +5V standby
Pin 13 +3.3V sense
Pin 16 (Green) Power On.
I was thinking that the Power good signal should be on a short delay, to give supply lines a chance to rise? Or I might be over-thinking things (again!)
Any help would be appreciated.
I was wondering if I could pick your brains?
I have a Intel D510M mini-itx based pc I'm using for music duties, and currently use a picopsu for supply duties. After various experiences with other bits of audio, I've become convinced that going to a linear supply instead of the picopsu will give me an improvement in SQ. Separate supplies for USB, HDD, SDD, Sound card are also a given.
Building a set of decent linear +12v, -12V, +5V and +3.3V supplies for the motherboard will be straightforward enough (until I win the lottery and have a chat with Paul Hynes!), and I'm lucky enough that Intel publish current consumption for the supply lines for this motherboard, but I was wondering what you did with the control signals?
i.e. On the ATX connector
Pin 8 (Grey) Power Good
Pin 9 (Purple) +5V standby
Pin 13 +3.3V sense
Pin 16 (Green) Power On.
I was thinking that the Power good signal should be on a short delay, to give supply lines a chance to rise? Or I might be over-thinking things (again!)
Any help would be appreciated.
Last edited:
Sorry!
First, Fork...
Sounds like you've done a lot of the right stuff on slimming down Win 7. Still, since you are already dual-boot, you might want to make a third boot directory & try a stripped-down WinXP setup too, just for reference. For example, the WinXP cMP/cPlay setup with the older hardware runs with only 2 services... The more recent Core I3 board requires 3 services. Still, very slim. But the tips at that link look very impressive.
On the external SATA, is the galvanic isolation part of the SATA connection, the external SATA connection, or a feature of a specific external SATA drive? Very curious... I must admit I don't know a lot about the SATA & eSATA standards.
I think a linear supply group buy/build would be very trick. If you decide to go that way, I'll help in any way I can. I do have some more thoughts on considerations before undertaking that, but that'll have to wait for another day.
Second, ackcheng...
Very, very nice setup you built there. I especially like the cases for the amps & computer transport. I bet it sounds like a million $! Again, more about it and a few questions when I have more time.
Finally, Woodenhead...
See these threads for some good fully-linear supply reading:
Powering a cMP... - GStew - Computer Audio Asylum
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
On the startup, I only have experience with three boards... All Gigabytes, the GA-G31M-S2L, the GA-G31M-ES2L, and the GA-G31M-UD2H board. The ES2L required -12v to start... The others did not. And the UD2H had a somewhat different startup behavior.
To start the S2L and ES2L, here's what I do:
1st - Apply power to the 12v P4 and the ATX 3.3v, 5v, 12v, and -12v (if needed) rails AND to the 5vSB (no time delays needed)
2nd - Press power button (momentary contact switch connected to the motherboard power switch header... In a pinch, I've bridged it with a screwdriver!)
3rd - Flip a switch to apply 5v to the ATX PWR_OK
That's it! If you wanted to make it a bit fancier, you could trigger a 1-second delay relay off of pushing the power button... But that's not really needed.
With the GA-G31M-UD2H board, I add a fourth step:
4th - Wait 8-12 seconds, then switch off the PWR_OK for 1-2 seconds, then switch it back on
As I mentioned above, I have more thoughts on considerations for a computer audio linear supply, but that'll wait for another time.
Later!
Greg in Mississippi
First, Fork...
Sounds like you've done a lot of the right stuff on slimming down Win 7. Still, since you are already dual-boot, you might want to make a third boot directory & try a stripped-down WinXP setup too, just for reference. For example, the WinXP cMP/cPlay setup with the older hardware runs with only 2 services... The more recent Core I3 board requires 3 services. Still, very slim. But the tips at that link look very impressive.
On the external SATA, is the galvanic isolation part of the SATA connection, the external SATA connection, or a feature of a specific external SATA drive? Very curious... I must admit I don't know a lot about the SATA & eSATA standards.
I think a linear supply group buy/build would be very trick. If you decide to go that way, I'll help in any way I can. I do have some more thoughts on considerations before undertaking that, but that'll have to wait for another day.
Second, ackcheng...
Very, very nice setup you built there. I especially like the cases for the amps & computer transport. I bet it sounds like a million $! Again, more about it and a few questions when I have more time.
Finally, Woodenhead...
See these threads for some good fully-linear supply reading:
Powering a cMP... - GStew - Computer Audio Asylum
Audio Asylum Thread Printer
Audio Asylum Thread Printer
Audio Asylum Thread Printer
On the startup, I only have experience with three boards... All Gigabytes, the GA-G31M-S2L, the GA-G31M-ES2L, and the GA-G31M-UD2H board. The ES2L required -12v to start... The others did not. And the UD2H had a somewhat different startup behavior.
To start the S2L and ES2L, here's what I do:
1st - Apply power to the 12v P4 and the ATX 3.3v, 5v, 12v, and -12v (if needed) rails AND to the 5vSB (no time delays needed)
2nd - Press power button (momentary contact switch connected to the motherboard power switch header... In a pinch, I've bridged it with a screwdriver!)
3rd - Flip a switch to apply 5v to the ATX PWR_OK
That's it! If you wanted to make it a bit fancier, you could trigger a 1-second delay relay off of pushing the power button... But that's not really needed.
With the GA-G31M-UD2H board, I add a fourth step:
4th - Wait 8-12 seconds, then switch off the PWR_OK for 1-2 seconds, then switch it back on
As I mentioned above, I have more thoughts on considerations for a computer audio linear supply, but that'll wait for another time.
Later!
Greg in Mississippi
Greg, I’m not sure if Gary reviewed the SATA spec; I’m guessing he did, but I’ll ask him about it again, if he’s in town next month. There’s nothing special about the enclosure connection; using the external enclosure is no different than powering an internal drive with a walwart. In fact you could probably connect the cap of the enclosure to the drive while it’s mounted in the case and connect the walwart there if the wire is long enough to reach. My understanding was that SATA carries only data and no power. Gary specifically mentioned that it had to be SATA connection, not USB. Having a USB input as well would be nice, so you can easily connect another computer to the drive, for ripping.
I have Win 7 running on 5 services and the Resource Monitor shows a total of 12 processes running, with the largest being the Win 7 kernel, at 57 threads. The services are DCOM, P&P, RPC, RPC Endpoint and Power. User Profile has to be enabled, but I use a batch file to disable it after the system boots. I haven’t tried disabling the remaining services in the registry, except P&P (which didn’t work), but I’ve found that the system will crash if these processes are terminated. It could be that I can't turn off P&P because I'm using a USB connection.
One drawback I deal with in my Win 7 setup is that by disabling my Intel graphics driver and using VGASave, there is no sleep feature, so I just press the power button to turn on and off. Not a big deal, but I wonder if XP retains the sleep feature available, in 8-bit mode?
I’d really like to figure out the how the display backlight on/off works, when the display goes to sleep. I have the timer set for 1 minute; there’s something really cool about having the room go dark with the TV still on. Before, when the display went to sleep, my Samsung plasma would always maintain a backlight, even though the screen was blank. I believe it’s the OS and not the TV, because it doesn’t behave the same way in my other boot section.
Regarding the motherboard mods (I'm jumping around here), I guess the UD2H must not have jumpers to disable the circuits that Jack removed. My SuperMicro board has jumpers for the networking and multi-channel audio circuits; I installed a switch to disable both circuits, but didn't hear any noticeable improvement, with my current setup.
As for the Linear PS, I’m not up for organizing a group buy, but if I can figure it out how to do it, I’ll post review of what I did, with the part sources, steps to build and pictures.
One last thing. Is there a thread that goes through the newest CMP build, like the old web posting, in step by step fashion, which seems to have been taken down? I’ve found several threads on the project, but don’t seem to find a master thread.
I have Win 7 running on 5 services and the Resource Monitor shows a total of 12 processes running, with the largest being the Win 7 kernel, at 57 threads. The services are DCOM, P&P, RPC, RPC Endpoint and Power. User Profile has to be enabled, but I use a batch file to disable it after the system boots. I haven’t tried disabling the remaining services in the registry, except P&P (which didn’t work), but I’ve found that the system will crash if these processes are terminated. It could be that I can't turn off P&P because I'm using a USB connection.
One drawback I deal with in my Win 7 setup is that by disabling my Intel graphics driver and using VGASave, there is no sleep feature, so I just press the power button to turn on and off. Not a big deal, but I wonder if XP retains the sleep feature available, in 8-bit mode?
I’d really like to figure out the how the display backlight on/off works, when the display goes to sleep. I have the timer set for 1 minute; there’s something really cool about having the room go dark with the TV still on. Before, when the display went to sleep, my Samsung plasma would always maintain a backlight, even though the screen was blank. I believe it’s the OS and not the TV, because it doesn’t behave the same way in my other boot section.
Regarding the motherboard mods (I'm jumping around here), I guess the UD2H must not have jumpers to disable the circuits that Jack removed. My SuperMicro board has jumpers for the networking and multi-channel audio circuits; I installed a switch to disable both circuits, but didn't hear any noticeable improvement, with my current setup.
As for the Linear PS, I’m not up for organizing a group buy, but if I can figure it out how to do it, I’ll post review of what I did, with the part sources, steps to build and pictures.
One last thing. Is there a thread that goes through the newest CMP build, like the old web posting, in step by step fashion, which seems to have been taken down? I’ve found several threads on the project, but don’t seem to find a master thread.
Ackcheng, thanks for the post. I emailed the company, which is not the Canadian company, but the company named on the device and they pointed me to this link. It seems to me that one could do without this part, using the Juli@, by connecting the PS directly to the card, as Greg did?
PCIEXP.HTM
I've heard of the Belleson devices, but hadn't been to their web page. Someone please correct my logic here if necessary, but this could be the most straightforward approach to a computer PS that might offer SQ comparable to builds like Greg's full linear PS. You used 2 devices for the 3.3V, so I'm assuming it requires require 2 devices each for 3.3v, 5v and 12v, so more expensive ($330) than a complete scratch build, but much simpler. Am I missing something?
Many thanks Greg, great and thorough info.
I am getting closer to building my linear PS now. I wonder what is the current capacity of the Bobken regulator as I want to try this too.
As to the wooden capacitor modification, I have actually done this now, and to fantastic outcome. I am currently upgrading all my caps this way.
I published a detailed guide how to perform the capacitor wooden mod, and my listening impressions on my website, check it out when you have the time.
I am getting closer to building my linear PS now. I wonder what is the current capacity of the Bobken regulator as I want to try this too.
As to the wooden capacitor modification, I have actually done this now, and to fantastic outcome. I am currently upgrading all my caps this way.
I published a detailed guide how to perform the capacitor wooden mod, and my listening impressions on my website, check it out when you have the time.
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.