Jump to content
IGNORED

12-volt-only motherboard and linear power supply?


Recommended Posts

Would it be possible to build a server using a 12-volt-only motherboard and linear power supply?

 

Reading on Wikipedia about the ATX power standards, I encountered this interesting section:

 

https://en.wikipedia.org/wiki/Power_supply_unit_(computer)#12.C2.A0V.E2.80.93only_supplies

 

"12 V–only supplies

 

Since 2011, Fujitsu and other tier-1 manufacturers have been manufacturing systems containing motherboard variants that require only a 12 V supply from a custom-made PSU, which is typically rated at 250–300 W. DC-DC conversion, providing 5 V and 3.3 V, is done on the motherboard; the proposal is that 5 V and 12 V supply for other devices, such as HDDs, will be picked up at the motherboard rather than from the PSU itself, although this does not appear to be fully implemented as of January 2012.

 

The reasons given for this approach to power supply are that it eliminates cross-load problems, simplifies and reduces internal wiring that can affect airflow and cooling, reduces costs, increases power supply efficiency, and reduces noise by bringing the power supply fan speed under the control of the motherboard.

 

At least two of Dell's business PCs introduced in 2013, the Optiplex 9020 and Precision T1700, ship with 12 V–only power supplies and implement 5 V and 3.3 V conversion exclusively on the motherboard."

 

Now, unlike ATX linear power supplies that seem to be made by only a couple of companies (HDPlex, Teradak... is that all?), 12V-only linear power supplies seem to be a commodity. For instance:

 

http://www.acopian.com/single-l-goldbox-m.html

 

http://www.mouser.com/Power/Power-Supplies/Linear-Power-Supplies/_/N-brwqi?P=1yzrc3dZ1yx5k7vZ1yxt7ffZ1yxt791&Ns=Output+Power|1

 

...and others. Many are made for medical devices (a big market segment) and need pretty tight regulation.

 

As for the motherboards, a quick search last night found that the Dell Precision T1700 mobo is readily available from a variety of sources. It's Ivy Bridge and so will take some of the i7 and Xeon processors, and supports some of the PCIe one needs for fast SSDs.

 

Any thoughts on this route? I get the sense that people already want to power a fancy USB card with a separate external linear power supply. So why not just skirt the ATX standard, I wonder? I am no electronics wizard, but is this feasible in principle for DIY?

Link to comment

It looks like the mobo of two current Dell Precision Workstations (the 3420 and 3620) are also these 12V-only boards. I say "looks like" because I can't find documentation that says so... but the power connectors on these two boards are also the 4-pin and 8-pin connectors that you find on the 1700 mobo (which is 12V-only) and there is no 24-pin ATX connector to be seen.  The 3420 and 3620 mobos have c236 chipsets, and so support most of the latest bells and whistles. I see the 3620 mobo for sale various places, around a c-note in price.

Link to comment

I use Dn2800mt motherboards for Roon server and Roon Bridge(s).

They have a 12 volt power plug.

They have been the best sounding motherboards for me.

The problem is they can play DSD but they don't have the power to upsample to DSD.  They can upsample PCM. So I use Aul Converter to create DSD256 files.

I use one with Windows Server 2012 and AO.  I am also playing with Audio Linux.

I run Ubuntu 16.04 server with Roon bridges. Low-latency kernels. Basic just ssh.

 

I use this for my headphone system. Holo Spring Dac Level 3. Crack amp with tubes/mods and HD650 cans.

 

You can get them for around $60 on eBay if you make offers. Need to add ram, drive and linear supply. $90.  Best to try to find one complete.

 

Jeff

Qobuz - HQPlayer(d) 5 - M1 Mini - RPI 5 8gb - Nuc i7 - Signature Silver DC - Keces P3 PS - Lush USB - Holo Spring Dac - SRC-DX - Chord Mojo2 - Bottlehead Crack Upgraded w/ GEC or Tung-Sol 6as7/vt231 - Triode Wire Labs AC - HD-650 - GR Research V2 - Dennis Murphy Pioneer BS22 - B&W 602 S2 - Apple Music Spatial 7.1.4

Link to comment

Looking at the manuals for those Supermicro X11SSV boards, I'm beginning to suspect that we want to be careful about the "12-volt-only" capabilities. Specifically these boards also have a 4-pin power connector (designated jp1) whose pinout has both a 12v and 5v pin. The manuals say this connection is needed for onboard HDD connections.

 

This MAY mean that with a 12-volt-only connection to the 8-pin JPW2 connector, and nothing else, you wouldn't be able to successfully connect (say) a SATA3 drive. Maybe you are stuck with an SSD via the M.2 slot, which might be fine but most of us would want to be able to have some TBs of SATA3 too.

 

Thoughts? Am I confused?

Link to comment

@SwissBear and @jkelly I am thinking that if one went with an LGA1151 board such as the Gigabyte board SwissBear pointed at, one would want to choose a low wattage processor to keep things cool (ha ha) in a fanless setup. On the other hand, as jkelly points out, too weak a processor and one wouldn't be able to do some of the things folks here like to do.

 

So here is a list of the lowest-wattage CPUs supported by the Gigabyte GA-H110TN. These are 35W processors. That is in the neighborhood of (I think at least) seven times the Watts of the pico processors one finds in a solution like jkelly's Dn2800mt. So first--is 35W too hot to get away with a fanless setup (big badass heatsink for convection cooling)? If so, question two is among the 35W processors listed below, which seem just powerful enough for the typical chores this group asks of a processor?

 

CPU model Freq l3 cache core name
Core i3-7300T 3.50GHz 4MB KabyLake
Core i3-7100T 3.40GHz 3MB KabyLake
Pentium G4600T 3.00GHz 3MB KabyLake
Core i7-7700T 2.90GHz 8MB KabyLake
Pentium G4560T 2.90GHz 3MB KabyLake
Core i5-7600T 2.80GHz 6MB KabyLake
Core i5-7500T 2.70GHz 6MB KabyLake
Celeron G3930T 2.70GHz 2MB KabyLake
Core i5-7400T 2.40GHz 6MB KabyLake
Core i3-6300T 3.30GHz 4MB Skylake
Core i3-6100T 3.20GHz 3MB Skylake
Pentium G4500T 3.00GHz 3MB Skylake
Pentium G4400T 2.90GHz 3MB Skylake
Core i7-6700T 2.80GHz 8MB Skylake
Core i5-6600T 2.70GHz 6MB Skylake
Celeron G3900T 2.60GHz 2MB Skylake
Core i5-6500T 2.50GHz 6MB Skylake
Core i5-6400T 2.20GHz 6MB Skylake
Link to comment
  • 2 weeks later...

Most all of the current Intel NUC motherboards will run from just a 12V supply.

 

You are much better off putting the supply external to the case.  But then I am biased as I build and sell a popular choke-filtered, dual-output, 5-7 amp LPS.  

 

Also, having the DC-DC switching converters on the motherboard does not offer much advantage over choosing the ATX mobo you want and plugging in a picoPSU to drop the 12V to 5V and 3.3.  Comes complete with the harness and everything.

 

 

Link to comment

I'm confused......

 

My Windows 10 server (laptop, powered by its battery) runs minimserver, music is stored locally on a ssd, streaming files to a sms-200. Does it even makes sence to optimise the server (lpsu, AO, passive cooling, etc) in this setup? 

 

Some people on this forum say it does, other say the server can be as noisy as can be. If it does, there are lots of tweaks to be done. If it doesn't, it would save a lot of money.

Link to comment
57 minutes ago, Lebouwsky said:

I'm confused......

 

My Windows 10 server (laptop, powered by its battery) runs minimserver, music is stored locally on a ssd, streaming files to a sms-200. Does it even makes sence to optimise the server (lpsu, AO, passive cooling, etc) in this setup? 

 

Some people on this forum say it does, other say the server can be as noisy as can be. If it does, there are lots of tweaks to be done. If it doesn't, it would save a lot of money.

There is no easy answer to that question...

I suppose it is a bit case specific, as most things these days.

However, since I have a similar server scenario to yours, my experience so far has been that hardware-wise, I cannot for the life of me detect any strong server sound signature in a typical server/endpoint scenario.

However, I can detect small differences which most usually have to do with two things: cleaner power and software: using a good psu and trimming down W10 or using Linux perhaps as a means of optimizing the PC's performance can result in  a small, but noticeable improvement.

I haven't obviously tried everything at the hardware side, and YMMV of course, but that has been my experience so far.

For me, endpoint optimization has resulted in a more clear and noticeable improvement.

I think that if I were to mention everything by order of magnitude, it would look like this:

Hardware

1. linear psu for sms-200

2. linear psu for network switch and modem/router

3.  ethernet filter/isolator

Running linear psu's made a dramatic difference in my system. The isolator less so, but it did improve the high frequencies quite noticeably.

I had decent Audioquest and Supra ethernet cables to begin with, so I didn't test those.

Software

1. Using the xtr family filters in HQ Player has made the strongest impact on SQ I have ever experienced from all software tweaks I've ever done combined and multiplied by one hundred!!!

2. Using NAA on its own

3. Trimming down W10 (like AO) and using Fidelizer Pro

I will continue to experiment however, keep an open mind and check out everything (..well, almost everything) the good folks in this forum try.

I know from experience that  opinions, as well as "undisputed truths" for that matter, change more frequently than not.

The old Socratic paradox "I know that I know nothing" might prove quite useful when approaching computer audio...

 

 

 

Link to comment
6 hours ago, Lebouwsky said:

I'm confused......

Frankly, me too. What I will say is that for me this is a hobby with multiple dimensions. One of the dimensions is technical exploration. I'm willing to wander down a fruitless path or two provided it doesn't cost me too much, and I get substantial yuks from goofing around with computers. All in good fun.

Link to comment
On 22-4-2017 at 7:18 PM, tedwoods said:

There is no easy answer to that question...

I suppose it is a bit case specific, as most things these days.

However, since I have a similar server scenario to yours, my experience so far has been that hardware-wise, I cannot for the life of me detect any strong server sound signature in a typical server/endpoint scenario.

However, I can detect small differences which most usually have to do with two things: cleaner power and software: using a good psu and trimming down W10 or using Linux perhaps as a means of optimizing the PC's performance can result in  a small, but noticeable improvement.

I haven't obviously tried everything at the hardware side, and YMMV of course, but that has been my experience so far.

For me, endpoint optimization has resulted in a more clear and noticeable improvement.

I think that if I were to mention everything by order of magnitude, it would look like this:

Hardware

1. linear psu for sms-200

2. linear psu for network switch and modem/router

3.  ethernet filter/isolator

Running linear psu's made a dramatic difference in my system. The isolator less so, but it did improve the high frequencies quite noticeably.

I had decent Audioquest and Supra ethernet cables to begin with, so I didn't test those.

Software

1. Using the xtr family filters in HQ Player has made the strongest impact on SQ I have ever experienced from all software tweaks I've ever done combined and multiplied by one hundred!!!

2. Using NAA on its own

3. Trimming down W10 (like AO) and using Fidelizer Pro

I will continue to experiment however, keep an open mind and check out everything (..well, almost everything) the good folks in this forum try.

I know from experience that  opinions, as well as "undisputed truths" for that matter, change more frequently than not.

The old Socratic paradox "I know that I know nothing" might prove quite useful when approaching computer audio...

 

 

 

Thank you for sharing, really helpfull. May I ask, do you do DSD? And have you tryed the bridged ethernet connection?

Link to comment
12 hours ago, Lebouwsky said:

Thank you for sharing, really helpfull. May I ask, do you do DSD? And have you tryed the bridged ethernet connection?

I do DSD exlusively, as I have built my system around my Lampizator Euforia DAC, which is DSD only.

As a musician, I am finding myself drawn towards DSD more.

It sounds more "real" to me as far timbre and attack are concerned, but also somehow manages to capture the feeling of the original analog master tapes better.

Not all DSD is created equal of course: I have heard DSD which is too soft or too airy, as of course PCM that's too dry or mechanical, but among the best variants of each, at least the ones I got to hear under ideal conditions so far, DSD has my vote.

I also find RBCD material, which comprises most of my library, quite un-listenable without some form of upsampling, which also means that I have to run DSP and when it comes to that, the logic and functionality of DSP done at the server side via software also gets my vote and the necessity of using of a server/endpoint scenario is undisputed.

I have indeed tried the direct connection (using the DHCP method).

It sounds a bit more "raw" to me, as opposed to to the more "tactile" sound of going through a switch.

In all honesty, I cannot yet pinpoint what factors create what and how they interact with each other, so the jury is still out on that.

Link to comment

DSD dacs can't read PCM, they need to convert PCM into DSD. There's no such thing as a hybrid PCM/DSD dac. In an article I've red that most DSD dacs are not very good in doing this PCM to dsd conversion.

 

That might explain why PCM files, upsampled (by for example hqplayer) to a DSD file sound better on a DSD dac compared to that same PCM file but converter to DSD by the DSD dac. That might explain why people prefer DSD, right? This does not apply to "real DSD files"

 

If that is true, one can only judge if PCM sounds better when upsampled to DSD and converted by a true DSD DAC, if it is compared to that same PCM file but converted by a true PCM DAC of the same league.

 

Somehow I've got the feeling if one is in the possession of a good dac like the Yggdrasil or an AMR dp777-se you're just fine. It may be different when you've got a big "true dsd" library.

 

I absolutely have no clue what it's got to do with a 12 volt motherboard. What I do know is that I don't like to be pushed towerds a powerfull (and expensive) i7 server, combined with having to invest in another dsd dac. I'm biased, I know.

 

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...