|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
67.60.98.130
In Reply to: RE: : RE: *Power Consumption* Core i3-530 32nm + GA-H55M-UD2H posted by jackwong96 on May 19, 2010 at 08:26:24
Jack,
I'm very interested in more details on your motherboard power supply setup. I'm working up a linear supply inspired by Gene & Mikhalov & have it working with the GA-G31M-S2L & ES2L boards, but not with the GA-EG45M-UD2H which seems to have a more sophisticated startup process. I'm happy that there's a technique for doing this with the earlier boards, but have been concerned that we won't be able to port this forward.
Now that I know it can be done, I'm very curious what you're doing to make it work with this board & if you have to do any sophisticated timing or communicating with the motherboard.
TIA!
Greg in Mississippi
Everything matters!
Follow Ups:
GStew,My PS Setup
I used a PICO ATX PS to turn on a 12V relay which in turn power up the regulators to the MB. Of course, 5V should be provided to Pin 8 (PWR_OK) and Pin 9 (5V SB) on the MB before powering up.
The PICO PS is supplied by a 12V regulator, which is always on when AC is applied, whereas the 12 V relay shares the same regulator.
Pin 16 (PWR_ON)(on ATX 2.x 24Pin) of the MB should be connected to the same pin of the PICO PS. When the power-on button is pressed, Pin 16 (PWR_ON) is pulled down by the MB and the PICO starts up and turns on a 4-points relay. The relay then connects the regulated supplies, i.e. 3.3V, 5V and 12V, to the MB. The 5V PS to Pin 8 (PWR_OK) and Pin 9 (5V SB) is provided by the PICO PS, as they do not affect SQ.
I don't use a PCI sound card and there is no need for a clear -12V, therefore, the -12V is also provide by the PICO PS. You may of course provide regulated PS to the 5V SB and -12V, but one more one relay will be needed.
Alternative Method
If you do not plan to use a PICO PS, the following method has been tested and found working. The same theory applies, i.e. the Pin 16 (PWR_ON) (on ATX 2.x 24Pin PS) should be pulled down to 0V in order to start up a MB.
- 5V supply should be provided to Pin 8 (PWR_OK) and Pin 9 (5V SB) on the MB. The MB is ready.
- either use a multi-points power switch or relay (controlled by 12V or 5V from your PS). Connect the MB 12V, 5V &3.3V PINs to NO of the relay also connect the respective regulated PS to the NC. (except Pin 8 (PWR_OK) and Pin 9 (5V SB) of the MB which should be connected to the PS direct and they should be turned on automatically when AC is applied to the power transformer).
- one set of connecting points of the multi-pointf power switch or relay should be used to connect Pin 16 (PWR_ON) on the MB to the COM (any black wire or 0V). They should stay disconnected as long as the power switch is not turned on.
- When the power switch or relay is turned on, Pin 16 (PWR_ON)is pulled down to 0V, simultaneously, all necessry voltages from the regulators are supplied to the MB. This will turn on the MB. The swtich remains on (connected).
Using Batteries
If you are using linear regulators for your PS, I highly recommend adding batteries to power them as the SQ improvement is significant. For lead acid batteries, simply added two small regulators for charging, one for 12 V (set to 13.6V and add a small resistor to control the charging current, say 0.22 - 0.33 ohm), and one for the 6V battery sets. You will need two sets of 6V batteries, a smaller set for 3.3V, and a larger set for 5V. They can be charged by the same regulator (6.8V, 3-4A). A relay should be used to connected the regulated PS to the MB. The battery should be connected to the moving pin of the the relay, the 12V, 5V and 3.3V wires from the MB should connected to the On_pin (which will be off unless is power is supplied to the relay. The Off-pin should be connected to the chargers so that they will charge the batteries when the computer is off. The relay configure varies slightly for the charging circuit -the batteries connect to the contact switch, NO is connected to the regulators which are connected to the MB, and NC to the chargers.
I have been using battery power supplies whenever practicable, in preamp, poweramp and DAC. I have tried different types of PS circuits, be it regulated or choke filters but nothing come close to batteries in performance. It is worth trying and I recommend to start with battery supplied DAC because the current demand is low.
Hope this help.
Edits: 05/27/10 05/27/10 05/27/10
Jack, thanks for all the great details above.
I'd had great results with a serious Pico-PSU-based Hybrid supply which provided SIGNIFICANT sonic gains for my cMP^2 setup, so I've been very excited about installing a fully linear supply... and frustrated that it hadn't worked consistently!
After reviewing what you shared and going back over the original thread where Gene discussed his linear supply, I've gotten mine working consistently now. What I missed was the need to put 5v on the PWR_OK AFTER pressing the power button... I'd been doing it the other way around, which worked some of the time on the ES2L board, but only occasionally on the S2L board.
Now that I'm using this sequence, I'm getting a very consistent startup:
1st - Apply power to the 12v P4 and the ATX 3.3v, 5v, 12v, and -12v rails AND to the 5vSB (no time delays needed)
2nd - Press power button
3rd - Apply 5v to the ATX PWR_OK
This does not work with the GA-G31M-UD2H board, however. It seems to have a startup sequence where it powers-down the PSU and starts it up again along the way... not sure why. I hope the new boards that you, cics, and others are trying don't do this.
But as the UD2H does not sonically equal the S2L boards in my setup, I have not spent much time debugging it.
I actually have a reverse-battery setup... The 12v P4 and the 3.3v, 5v, and 12v ATX voltages each have their own transformer. But inspired by Gene's setup. I'm using two 9v batteries in series with a 7912 chip to get my -12v and a single 9v battery with a 7805 for my 5vSB & PWR_OK. I've found that once the startup process has intiated, I can turn these off with no problem, hence no need to have a more constant source. Now that I have a consistent startup, I will try it tonight with a single 9v battery and no regulator on the -12v line... this worked for Gene and I suspect it will work ok for me now. I also don't need a constant -12v as my sound card setup has separate power rails.
Of course, this is all on a board on my bench. I've got a few days off around the US Memorial Day holiday, so I'll be able to install it into my cMP setup and listen... this will be the start of PSU Follies Strike Back!'.
Later!
Greg in Mississippi
P.S. I will give batteries a try. I have significant battery-charging infrastructure around as I am a model airplane flier too and most of my powered planes use electric power. I also have some serious Lithium-Polymer packs in various sizes that will be interesting alternatives to the Pb-acid setups... I've heard that different battery types produce different sonic results and have not heard anyone report on these.
I have been cautious about them partly because of the additional infrastructure and care required AND due to some of the comments by people I respect who have tried them and not liked them... but enough have said they are the best to at least give them a try.
Also, I am VERY curious about your battery-powered direct-heated tube amps. Got pix?
Everything matters!
Congratulations on your success!
I believe the 5v should be applied to the ATX PWR_OK before or simultaneously with the 12v P4 and the ATX 3.3v..... This will send an OK signal to the MB.
Sonically, lithium packs sound best, next is NiCd, HiMh and than Pb-acid. I have compared them carefully and used different types depending on applications. More clarity and focus with lithium. Pb-acid is noisy and needs bypass caps.
The sonical difference is contributed to internal resistance and noise of different types of battery. Lower the better.
I will post some photos of my DHT DAC analoge stage cum preamp later. The two stages have been condensed to one and it was a big step forward.
Jack,On the S2L board, it hasn't worked consistently for me to apply ATX PWR_OK before pressing the power button. This was also discussed in this thread and sub-thread:
http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=65451
But it may be a difference across different boards.
I can play with this a bit, tho. I have the 5vSB & -12v on one switch and the PWR_OK on a second switch, so I can vary the sequencing.In any case, I am pretty pumped and excited about getting this working consistently. It has been a tease, powering on the test board sometimes, but not every time.
BTW, this is all with no relays or PicoPSU. I when I plug the beast in, it directly energizes the 3.3v, 5v, and 12v rails. The -12v and 5v control voltages are turned on and off by switches.
I'm happy that this is working with such a simple setup.
Also, very happy to hear your comments on the sound of different batteries. I personally wouldn't use NiMH or NiCD unless I had to... too much self-discharge when not being used. I have LiPo packs in various capacities and every cell-count combo from 2-5 cells and all are pretty high-current for their capacity (some packs I have list peak draws of over 100 amps... I use these on larger electric helicopters).
I'll report on sound with this supply after I have a chance to listen a bit. I also have plans to do some listening to various regulators and various power sources for the P24... I am skeptical of the current recommendation to go with a 2nd computer power supply for that, I found a significant improvement in SQ when I went to linear supplies for my 'dirty' connections... 12v screen, 5v USB, and 5v HDD. This surprised me, but even more surprising was that the quality of the caps I used around those regulators also made a difference... so I expect a good linear to be better than an SMPS for the P4 also.
Again, thanks!
Greg in Mississippi
Everything matters!
Edits: 05/27/10
Quick update... just as Gene said, using a 9v battery to provide the voltage for the-12v connection allows a good startup... at least with the S2L board.
That simplifies things quite a bit too, removing 1 9v battery and the associated regulator & caps.
So it's ended up as a very simple construct... 4 regulated supplies (3.3v, 5v, 12v, 12v), 2 9v batteries (one used as-is to provide -9v, the other regulated down to 5v), two switches (one SPST, the other DPST, and the associated wiring, connectors, fuse, and heatsink.
Left it on for awhile to allow it to warm up... I was worried a bit about possibly having a heatsink that was too small. Max reg temps a bit high at 130F, but not too bad. I can run with that.
I'll put it in for listening tomorrow... and start a new sub-thread then too! I hope it sounds as good as I'm expecting!
Greg in Mississippi
Everything matters!
Correction re: My PS Setup
Line 4: "When the power-on button is pressed, PWR_OK is pulled down.....",
PWR_OK should read PWR_ON.
Correction re Aternative Method
First line: i.e. PWR_OK should read i.e. Power_ON.
Jack,If you select and view a single post, you can edit it. That way you can correct your original post with details AND then delete the add-on's... Just a thought.
Greg in Mississippi
Everything matters!
Edits: 05/27/10
I have made amendment and some refinements.
Are you kidding? "Hope this helps"?
This is GREAT. Will digest and possibly come back with questions.
THANKS FOR SHARING!
Greg in Mississippi
Everything matters!
I've abandoned the alternate mobo option. It unfortunately requires both P4 and P24 to be sourced from same PSU. Also BIOS allows for voltage increase only. So that leaves us with 2 Gigabyte options: H55M-UD2H & H55M-S2H. There's a mini-itx version (GA-H55N-USB3) which unfortunately doesn't have a PCI slot.
I'd like more input on this new technology before changing recommended hardware and settings.
I did some search. Asrock and Foxcon H55M MBs appeared not suitable for the intended purpose as the voltage settings are insufficient or not low enough. MSI H55M P31 seems to be a good choice with all necessary voltage settings.
According to a MB review MSI H55M E33 performs better then its Gigabyte, Asus counterparts in overclocking. The result is indicative of the PS design and quality of the MBs.
I have asked MSI for the lowest voltage value of CPU VCore, DRAM and GPU.
Take a look at this Intel S3420GP mobo which supports Core i3-530/540. I like its simplicity and clean design. Its a bit large but there's less clutter like no onboard audio.
Sever boards are designed for high quality long running 7x24 performance. Not sure about BIOS. Intel tested this mobo with XP SP2 Pro.
For stability, Intel MBs do not usually provide low voltage setting in BIOS and may not be a good candidate for CMP^2.
With a view to improving current consumption and SQ, I tried and successfully removed the sound and network chips together with some regulators on board my GA-G31M-S2C and an older Intel MB. I found that despite the devices are being turned off in BIOS, the regulators are still running. The chips were removed by using a hot air gun and solder iron. The important thing is that after removal of the chips, the MBs boot up without any problem.
I believe the BIOS setting is crucial and I like the flexiblity and stability of the GA H55M UD2H. Next step, I will try to remove unnecessary chips from this MB.
...is it possible to take some pics and show us which chips you removed?
I support the request. It is very interesting.
See photos. Haven't got time yet to test the performace and current.
Edits: 05/31/10 05/31/10
You measured 2.34A on 5V which is surprisingly high. Would be keen to see new load after these changes?
I'm going to build a CMP machine for another friend so I bought him a H55M-S2H for my experiment. I removed everything around the sound and network Chip, including regulators and transistors. See photo.
Current Consumption decreased by about 0.1A after such removal. Capacitor leakage current should also be taken into account. The 5V is now drawing 2.25A @96K. Sound quality slightly improved.
I can go on to remove the Jmicron IDE controller but I'd better stop here as this MB belong to my friend.
Edits: 06/01/10
can you please point out where the chips are removed? hard to see.
Photo.
The optical output and an OP amp were removed.The CPU fan and System fan PS was cut off.
Edits: 06/01/10 06/01/10
Two more chips, the DVI to HDMI Convertor chips for the DVI and HMDI output, were removed from the MB, with much hardship.
Not sure about any reduction in current consumption, that's not the point. The importance lies with the good improvement in SQ brought about by the risky but worthwhile surgical operation. As you can image, two noisy video chips were severed.
The last thing left, I belive, are a regulator and the Jmicron368 IDE controller. See pix below.
Edits: 06/02/10 06/02/10 06/02/10
thank you
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: