K7D voltages

I'm running a pair of 2500 Bartons at 15x140=2100 at stock voltage (1.65 V) in my K7D Master-L. PSU is a Hyena (which I suspect translates to generic) 500W supply, with the following ratings:
3.3V : 28A
5V : 50A
12V : 18A

The thing I'm worried about is fluctuations in the voltages suppied by the motherboard. In the format min, average, max:
VCore (1.65V ideally) : 1.49, 1.61, 1.70
2.5V: 2.35, 2.48, 2.59
3.3V: 3.22, 3.23, 3.25
5V: 4.89, 4.90, 4.92
12V: 12.04, 12.07, 12.10
Given the stability of the voltages comping out of the PSU (though slightly on the low side for 3.3V and 5V), and the large fluctuations in the voltages from the board, I'm thinking that this is the K7D's voltage stability problem that I remember reading about somewhere.

The fluctuations don't seem to entirely be load related, as I can load up two copies of BurnK7 and have almost no change in the voltage. But then when it's just sitting there idle, the vcore drop a several hundredths of a volt and then go back up again.

So, is this voltage instability instability a design problem of the board (using two high-power CPUs in a board that was designed back in the palomino days), or a manufacturing problem? IOW, do I return it for a replacement or not

As far as actual program stability goes, the setup is fine. I've had it running two copies of prime95 for nearly a week and there's been no problems. The main thing stopping me going higher is that I've got unlapped Thermaltake Silent Boosts as HSFs and only one 80mm case fan (excluding the two fans - 80mm and 120mm - in the PSU) which causes load temps to hover around the 72 Celsius mark (the wall thermometer reads 30 deg C at the moment, which isn't helping either). So running at 1.725V is a little hopeful temperature wise. These two problems are going to be fixed in the next couple of days, with two 120mm fans (21db ones only, I'm averse to noise ) and a lapping of the heatsinks.

EDIT: Motherboard temps are 41 deg C under load (11 above room temp to put it in perspective ...)

I have a K7D as well and those voltages look similar to mine. I have no stability issues so far (about 8mos.). Your case temp is really high, think about taking a holesaw to it and improving the airflow.

I have three K7D's (all rock-solid) and all three report that kind of voltage fluctuation. I think the key, here, is where the board reports voltages from. Different boards report the voltage from different places on the board. If you actually stick a voltmeter on a Molex connector, you will probably see VERY little variance. Wherever it is that the board reads voltages from, is a very fluctuating place.

Bottom-line: if you don't have stability problems, then don't worry about it. If you do, I would suspect the generic PSU, before the board.

Mine is the same. It locks up every day or two at stock speeds and full load, more frequently if I overclock. Load temps are usually around 45C. CPu's are L5 modded 2400+'s.

5v and 12v are strong and solid. Vcore drops now and then as much as .08v that I seen.

I'm going to try the latest unofficial bios, 1.91.

Arkaine23 said:

Mine is the same. It locks up every day or two at stock speeds and full load, more frequently if I overclock. Load temps are usually around 45C. CPu's are L5 modded 2400+'s.

5v and 12v are strong and solid. Vcore drops now and then as much as .08v that I seen.

I'm going to try the latest unofficial bios, 1.91.

Click to expand...

try upping your vcore
I don't believe the bios adds much

I'm at 1.775 currently, and it behaves the same at 1.675.

Dips of 0.08V are small compared to mine Had a record dip sometime today ... vcore went down to 1.47V. That's 0.15V (9.26%) below average, and 0.18V (10.91%) below where it should be. The 2.5V reading also dipped to 2.30V, which is 7.26% below average, and 8% below spec.

From the PSU, biggest dip was to 4.84V on the 5V line (1.4% below average, 3.2% below spec), 3.22 on the 3.3V line (0.3% below average and 2.4% below spec), and 11.98V on the 12V line (0.66% below average and 0.16 below spec).

So the dips from the onboard regulators are, at best, nearly 7 times larger percentage wise, and pretty much 2 times larger raw value wise.

I think the only reason why I'm not having problems is that I'm not pushing the chips hard enough

Mine are running at full load 24/7. And I have no way of knowing how low the dips are when it actually freezes up.

Here are my High/Low numbers for my main rig, which has been running for two days, since my last reboot.

Vcore Low 1.22V
Vcore High 1.46V
Vcore Average 1.35
(actual setting 1.40V)

Vmem Low 2.30V
Vmem High 2.54V
Vmem Average 2.44V

3.3V Low 3.26V
3.3V High 3.33
3.3V Average 3.31

5V Low 4.78V
5V High 4.84V
5V Average 4.81V

12V Low 11.92V
12V High 12.04
12V Average 11.98V



This is all running on a Fortron FSP300 PSU that I have been testing for the past week, for stability. So far, no problems, whatsoever. These Fortron PSU's are powerhouses.

I have an Antec 550 and my poor 5V line has been in the 4.55-4.60 range lately. It measures that at the molex connectors. I'm so tempted to try a Fortron too.

For comparrison my Vcore range is 1.47-1.71, avg 1.61, actual 1.625. This is with dual 2100Mhz Barton XP-M's.


-Rav

My Fortran 400w supplies a very steady 4.99v to 5v and 12.20v to 12v. I believe these boards pull vcore out of the 12v line, not the 5v line.