|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
85.147.82.48
In Reply to: RE: cMP - the open source high-end Memory Player posted by cics on December 30, 2007 at 05:42:01
For those who want too place smoothing caps on the power supply line to smoothen the ripple voltage.When placing smoothing caps, one needs to know (or guess) the current and the estimated ripple voltage on that supply line. The ripple voltage can be found in product reviews on the www. The actual working current has to be measured.
In some posts I saw measurements of the current on the P4 pin of about 0.7 amp when working. And 1 amp during boot and startup.
I found these mentioned currents very high. So I measured them myself.Mobo: Gigabyte GA-G31M-S2L Bios F9
Intel: E7200 core2duo CPU ratio: 7 Busspeed: 160 CPU voltage: 1.0000 VoltReadings when playing music:
Foobar2000
SRC 96000 0,35 - 0,36 amp
SoX 96000 0,26 - 0,27 amp
PPHS 96000 0,25 - 0,26 amp
--- 44100 0,24 - 0,25 ampcPlay
SCR 96000 0,29 - 0,32 amp
--- 44100 0,24 ampWhen CPU voltage is lowered to 0,90000 volt than all readings are 0,04 amps less.
Playing dvd’s with GOM player uses 0,27 – 0,29 amp.
How the measurements were done:
A NAIM Hicap acts as 24 Volt power supply.
The 24 volt is fed into a dc-dc adapter to make it 12 volt.
Then this 12 volt is fed in to a DC-DC ATX power Supply
(http://www.mini-box.com/PW-200M-DC-DC-power-supply)
The PW-200m only powers the processor through the P4 pin.
The multimeter was placed between the HICAP and the 24/12 dc-dc adapter.Other power sources while the PC was working:
The mobo pin20/24 pin is powered by an Antec EarthWatts 430.
Sphericals (HDD, USB-ports, DVD, etc.) are powered from an external PC power supply outside the PC-case. (In: ac 230 V, Out: dc ± 50 W, 5 V / 3 A, 12 V / 3 A). I don’t mind DC wires inside my PC. But I definitely do not like 230 Volt AC wires inside the PC near mobo, soundcard, etc. So I didn’t choose the granite mini power supplies.The PC is calibrated according the art of building computer transports except for:
- the network is still working
- the video is still working (1920 x 1080 ) and displaying too Sony Bravia KDL-32W4000
- EIST function is off. Corresponding XP power settings are also off.ESI juli@ or LynxL22 dig i/o -> Lavry Black DA10 -> Mogami Gold -> Klein & Hummel O300
Edits: 04/06/09Follow Ups:
hfavandepas wrote:
In some posts, I saw measurements of the current on the P4 pin of about 0.7 amp when working. And 1 amp during boot and startup. I found these mentioned currents very high. So I measured them myself.
The only reported measurements of CPU current draw on this list in recent months that I know of have been mine so I assume it's these you are referring to. If so, your report is not quite accurate. In January, I wrote:
“I repeated . . . the measurements I did of the current drawn by the E7200 chip [on the G31 motherboard]. On power up, it draws about 1.5 amps. As ‘underclocked’ BIOS settings kick in during POST, this falls to about 0.6 amps. Once the OS is loaded and while playing music (with a NOS DAC), the chip draws between 0.4 and 0.5 amps or between five and six watts."
The meter I used was an old Avometer DVM 2001 with 1000-milliamp (ma) and 10-amp ranges. For obvious reasons, I started on the 10 amp range, saw the 1.5 power-on peak and, well, left it there.
In the light of your post, I dug out my other meter, a rather newer Fluke 8060 with a 2000 ma range, making it safe for this test.
I'm using a 12 volt linear supply to drive the CPU (and a "pico-processor" à la mode GStew for the P24 line). Except that I use a USB NOS DAC (so no soundcard) and run at 140 MHz and a nominal Vcore of 0.75, I think the systems are comparable.
I repeated my tests using the Fluke on its 2000 ma scale. The current peaked soon after power-up for about 30 seconds at around 1500/1600 ma (though it went off the scale on one occasion), falling by degrees to settle at 330/340 ma when playing.music.
This figures seems much the same as yours though I confess I'm less willing than you to place confidence in figures showing tiny differences when doing this or that.
The difference between my January measurements and the ones I've just done I'd put down to
1. Using a different meter on a different scale - a 10-amp scale is always going to be rough. I was happy enough to get any data at the time as I couldn't find anything else anywhere else and all I wanted was a ball-park figure. I got one.
2. Since taking those measurements, I've changed a deal of settings and no longer recall what is different between the two setups. The measurements are broadly in line with each other, much the same as yours and adequate for purpose.
What might be of interest is that if you set Vcore to "Normal" (i.e. 1.5 volts or about double what I used above), the system settles at 1070 ma when playing the same track as above. In short, setting Vcore with care can cut the current draw by almost two-thirds – not to be sneezed at.
Best
Dave
.
cics wrote:
What VID are you using to get Vcore of 0.75V (and host @140)? Noticed that with EIST off, VID can go lower!
Now that you pin me down, I have to say that I'm not sure what VID is in this context and my note may have been muddled as a result.
Whatever, in the BIOS's MIT menu, there is a setting for CPU Voltage Control . This is set to 0.75000. (Just below that is a note that the "normal" CPU Vcore is 1.15 v.)
With this setting (I've just checked), CPU-Z reports that Vcore is 0.704 but I have also noticed that the BIOS menu "PC Health Status" says that it is 0.724.
I know what VID is in general terms but as to exactly what these figures mean and which Vcore reading is right, I've no idea.
I have run the board with CPU Voltage Control set as low as 0.65 but, especially when cold, it tended to restore defaults on powering up so I settled on 0.75 and left it at that. (Some report SQ differences after making minute changes to CPU Voltage Control but I can't hear any.)
I have a hunch you're right about things being more stable with EIST off but I've not tested it systematically.
Dave
VID is an 8 bit code set from BIOS to CPU. Each VID reference relates to Vcc specifications as per Intel. You have VID of 0.75000 set - I'm currently testing 0.73750 (lower values are unstable). The "normal voltage" BIOS message is a CPU setting (each is set by Intel). I have one E7200 at 1.15 and another at 1.21 (which does 0.83750).
Can you measure overall power consumption using your fluke on main 12V supply to P24 & P4 (i.e. use one PSU).
cics wrote:
Can you measure overall power consumption using your fluke on main 12V supply to P24 & P4 (i.e. use one PSU).
Just to be absolutely clear, the figure below refers to the Gigabyte G21 MoBo.
When playing flac-format music data without upsampling with the CPU clock at 140MHz, VID at 0.75000, PSU at precisely 12v, EIST off, 512 MB Kingston "ValueRAM" and a "Full Monty" cMP2 setup, the motherboard and CPU typically draw about 1630 ma.
Allowing for dissipation in the "pico-PSU", call it 1600 ma or about 19 watts. Last time I measured this, it came out slightly higher (see cics's earlier request for this measurement). I don't think I've tweaked anything significant in the meantime but, as the first measurement was taken with a different meter on its 10 amp scale, I'd tend to put more faith in this one.
BTW, after a tedious to-and-fro session, it's clear that my setup at least sounds better with EIST OFF . Who'd have believed it mattered?
That'll do me for current measurement for a while - it seems oddly emotive for such a humdrum topic.
Best
Dave
Thanks.
Got SQ improvement with VID 0.73750 (Vc as per CPU-Z shows 0.704V). As expected, CPU temperature dropped nicely by 4-6°C. Now testing VID 0.73125 (Vc is 0.688V) which is the lowest stable VID in my case.
Setting Host Freq to 150 and upsampling to 192k @145db SNR should yield overall power consumption below 20W (with 256MB RAM)!
Hi dave,
Since we have the same mobo and processor it’s interesting to compare these measurements.
On my system when booting the current on the P4 pin in my setup is between 0,32 and 0,37 amp with one short peak at 0,40 amp. When the boot process has ended and windows is at rest than the current on P4 is 0,20 amps. (With Bios at: CPU ratio: 7 Busspeed: 160 CPU voltage: 1,0000 Volt)
You’re reading are; 1,5 amp at power up, then go down too 0,6 amps and while playing music you’re reading are about 0,4 – 0,5 amps on your system.
May be I’m not doing my measurements properly, but your readings are twice as high as my readings. I think that’s surprising.
So are there any other inmates who did some current measurements?
Best
Mark
Data:
Foobar2000 in optimized XP setup:
SRC 96000 0,35 - 0,36 amp
SoX 96000 0,26 - 0,27 amp
PPHS 96000 0,25 - 0,26 amp
--- 44100 0,24 - 0,25 amp
cPlay in optimized XP setup (note: not in cMP enviroment)
SCR 96000 0,29 - 0,32 amp
--- 44100 0,24 amp
When CPU voltage is lowered to 0,90000 volt than all readings are 0,04 amps less
LynxL22 dig i/o -> Lavry Black DA10 -> Mogami Gold -> Klein & Hummel O300
hfavandepas wrote:
. . . your readings are twice as high as my readings.
Sorry, but I don't see how you reach that conclusion. Yesterday, I wrote:
"The current peaked soon after power-up for about 30 seconds at around 1500/1600 ma (though it went off the scale on one occasion), falling by degrees to settle at 330/340 ma when playing.music."
You report readings ranging from 240 to 360 ma. From where I'm sitting, the steady-state (i.e. while playing music) values are pretty much the same.
I repeated the measurements today (I'd left the meter in place.) The steady-state reading has fallen to 320 ma. I'm not getting excited about that - it's a difference of three per cent. (It could be up to 350 ma tomorrow, making it +/- 3%. I'd be delighted if my meter - it's a good one - is accurate to +/- 2% when reading current.)
The 1500 ma peak on a cold restart was still to be seen today, though not for so long. Presumably that's because I'd left the BIOS at clock=140 MHz; Vcore=0.75v. (I wasn't expecting a "Spanish Inquisition".)
You'll see only a short peak if you do a warm restart. To test it properly, you need to store your settings, put the BIOS back to defaults and do a cold start. Then, if you don't get a peak roughly comparable to mine, something funny IS going on.
Whatever, the values you report all lie, give or take, within +/- 100 ma of mine - some are higher, some lower.
(What I didn't report yesterday was that I get a short peak of ~400 ma at the start of each track, presumably as cPlay decodes flac data.)
You don't say what scale you are using (though your data is in amps) or suggest the calibre of your meter. If you are using a 10-amp range, I'd be a little concerned about the accuracy of readings so close to the low end of a low-resolution scale.
You report 0.24 amps running cPlay at 44.1. Are you sure? That's less than three watts - judging by the way it eats batteries, I have a (OK, OK, it's very old) calculator that seems to need more than that.
Seriously, these are very informal "experiments". We have different meters (neither, I'd hazard, recently calibrated) possibly on different scales; neither of us recorded the temperatures of the CPU or the meter or checked how long the device has been running. We are running different software on a different configuration with different target data, etc etc.
I confess I haven't a clue as to how normal manufacturing tolerances might be reflected in crude tests like these.
I just don't see any "Eureka" moments coming from comparing these data - they seem pretty well in line. In short, we can report with reasonable confidence that an under-clocked E7200 chip draws between quarter and half an amp from the P4 line when playing music.
Anything else risks reading into the data what is not there.
Best
Dave
Hi Dave,
Lets skip this discussion.
Somehow you feel offended or annoyed if somebody else is posting data that is not inline with yours. You also don’t seem to have any interest if there are other inmates who performed current measurements on the P4 pin.
Every body can read our former posts and draw there own conclusions based on that. Or even better: perform there own current measurements.
LynxL22 dig i/o -> Lavry Black DA10 -> Mogami Gold -> Klein & Hummel O300
hfavandepas wrote:
Somehow you feel offended or annoyed if somebody else is posting data that is not inline with yours.
I'm not offended (or even annoyed) that you did the measurements - this list is short on empirical data. My point was that, on examination, the two "data sets" were in fact pretty much the same and that the differences you claim to have noted were based on (twice) misreading what I'd reported. What's your beef with that?
What did disappoint me was that you seemed to have done the measurements sloppily and then implied that the fault was mine. Why not, instead of throwing a hissy fit, answer my queries?
You also don’t seem to have any interest if there are other inmates who performed current measurements on the P4 pin.
You could be right - what consenting adults do in private with their P4 pins doesn't interest me. However, I did, a while back, report some data for ripple voltages in the context of adding smoothing caps to MoBo power supplies. Bill (Old Listener) pointed out that my figures were way different from those on respected review sites and queried if I had done them properly. I did them again them and found that I had been sloppy - Bill was right and I was wrong. Why don't you try and do the same?
Meantime, skip the abuse.
Dave
Hi Dave,
All I did was mentioning that in past I had read a post on: current measurements on the P4 pin.
I wrote:
“ In some posts I saw measurements of the current on the P4 pin of about 0.7 amp when working. And 1 amp during boot and startup. I found these mentioned currents very high. So I measured them myself. “
I did this by heart. But my memory wasn’t far off.
Your post:
Posted by Ryelands on December 27, 2008 at 11:06:42
………..
Incidentally, it also makes it easy (ish) directly to measure the CPU’s current draw, something I’ve never seen reported (though I’m sure it’s done often enough). So far, I’ve measured it only briefly and only on the 65nm Biostar board.
I had assumed it would vary wildly but it was surprisingly stable: an E1200 processor drew 1.4 amps on power-up and during POST but, as the OS loaded, it dropped to and stayed at between 0.7 and 0.8 amps and rose again by about 100 milliamps when running cPlay (without upsampling).
………..
Since I found these mentioned currents very high (for my Gibabyte GA-G31M-S2L mobo + E7200 processor), I decided to measure the currents on my P4 pin myself and also post the results on this forum.
So owners of a Biostar board + E1200 processor can use your readings.
And owners of a GA-G31M-S2L + E7200 processor can use my readings.
No more, no less.
After that you jump on me with pretty strong language and lengthy discussions on a wide variety of topics. The rest one can read in the former episodes of this post.
I now consider this case / post closed.
LynxL22 dig i/o -> Lavry Black DA10 -> Mogami Gold -> Klein & Hummel O300
do you run a cpu fan or are you using a fanless cooler? if so which one
Hi Theo,
I use a fanless cooler. This one:
http://www.nexustek.nl/NXS-LOW-7000_silent-7cm-high-CPU-Cooler.htm .
From which I detached the fan part.
If I remember correctly I red in one of your posts lately, that you still use your processor fan.
If you run your processor at voltages lower than 1.0000 volt and you also under-clock, then I would go for fanless.
It made an improvement in SQ in my setup.
Some background:
At first, I ran the E7300 processor a few weeks with the original cooler that came with the processor.
I couldn’t find the type of fanless cooler that Cics shows in his AOB manual. And when I finally did find it, I found it horribly expensive.
But in those few weeks I had seen that, when under-volted and under clocked, the CPU doesn’t run hot. So I decided that other cooler models with enough heatpipes and a large exchange surface, probably also would do the job.
Very strange discovery:
There was an improvement in sound quality, when I changed from the original cooler too the fanless cooler.
The strange part is: …… I never powered the original coolerfan from the mobo. While using the original fan for a few weeks, I powered the original coolerfan from the second external PSU that also powers the HDD, USB, DVD. So I expected to find no difference in sound quality when removing the orginal fan. Because it wasn’t powered by the mobo, but powered by the second external PSU.
Somewhat strange:
This couldn’t be repeated with the coolerfan in my Zahlman ZM600-HP PSU.
Running the Zahlman coolerfan from the second PSU also made a little improvement in SQ.
But completely removing the fan from the PSU unit (just like removing the fan from the processor) didn’t make any further improvements in SQ.
So if you can, get rid of the processor fan.
:-)
Even though the processor coolerfan was powered by a second external PSU, the original fan and the processor ‘didn’t like each other’ in my setup.
ESI juli@ or LynxL22 dig i/o -> Lavry Black DA10 -> Mogami Gold -> Klein & Hummel O300
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: