Replace KT133MB, processor & memory or just replace nForce2MB?

[videobruce]

Should I replace an existing nForce2 MB a nForce2 ultra 400 MB and reuse the processor & memory or replace a older KT133 MB along with the processor & memory (big difference in cost)?
The nForce MB has a issue with the clock (gains 6 seconds a day, sometimes more) and hibernation (doesn't work through the serial port), both features I need for a program to work corectly. No, I don't want to run a 'time' program in the background all the time to 'band aid' the problem.
The KT133 MB would usually be used for web surfing, jpeg viewing/editing, maybe some games, but NOT anything intensive and some CD burning.
The nForce2MB is for a HTPC with a HD tuner card and a DVD burner.

Thanks..........

[ajrettke]

You can get refurbed NF2 boards from newegg for 30-40 bucks. That's what I would do. I'm kind of wondering though, do you have the KT133 board and SDRAM as well already?

Welcome to the forums

[RoadWarrior]

No matter how much you pay for a mobo, you can still have issues with the clock, it's one of those "luck of the draw" kind of things coupled with very low cost components, and changing environments. For instance, should the timing crystal be specified at 20C room temperature, Joe Sixpacks overloaded Prescott case temperature of 40C, or an overclockers well vented machine at 25C, or an extreme overclockers chilled air rig at 5C???? Plus if you turn your computer on and off every day, leading to thermal cycling, the poor clock components can drift quite a ways off thier specified ranges.

6 seconds a day is actually a lot less than a lot of people have to put up with!

Any IT pros with a system that is time critical will put an NTP client on it, set to update as often as every half hour, to make sure things don't get too far out of whack.

I can understand though, that if you want an HTPC to wake up at a specified time, and record a program, that a drifting clock is not very helpful, and an NTP client will only help when it's actually on, or connected to the internet. I'm just saying there's a possibility that you could replace it and then have a new board act worse!

As for the serial port hibernation problem, that could be a general problem with too low a standby current available from your PSU, or it could be that messing around with the wake up options in BIOS and changing it to trigger on IRQ activity for the IRQ of your serial port, may have a positive effect.

I understand that PBS channels have a time signal on them, it may be possible to get a driver for your tuner card, or a TSR, to update the time from that signal, when the machine is on, if you don't plan to network this machine. I know, another bandaid, but given the abysmal state of PC clock hardware in general, you're going to be very lucky not to need one.

I dearly hope that recent news of a new type of atomic clock that is small enough to go in a watch, will mean that such clocks will be available for PCs in the near future to save us all from this pain and misery.

I do have one idea of how you might work around this whole issue and still be able to leave the machine hibernating most of the time between recordings, I don't know how to arrange all the details of this, it's just some way to think of approaching it.

If you run a NTP (Network Time Protocol) client on your general usage computer that is always on, so that THAT clock is accurate, then if your HTPC is networked to it, at a certain time every day, (or twice a day) you can get the main computer to wake up the HTPC with a WOL (Wake on LAN) packet, and update it's clock to the correct time... presuming that being as much as 6 seconds off won't screw anything up, it being more of a worry if you forget about it for a few days and it ends up missing the whole first minute of your favourite series. I know, it's not perfect, but best I can think of.

I think you should take a look at this...
http://www.beaglesoft.com/mainfaqclock.htm
Which asserts that 5-6 seconds a day is typical and as much as 15 seconds is common. It also has a lot of insight into the timekeeping processes of PCs.

Another source, quoting an accuracy of 0.005% for the crystal tolerance asserts this can mean variations of 4.5seconds a day within the tolerance of the crystal, add to this that the software clock can only set itself to within a second, and you get 5.5s damn near 6, as an answer for "how bad can these things be"

I really sympathise with your problem and realise how frustrating it is, but there is not any likelyhood that you'll improve your PCs timekeeping with anything other than luck by buying a new motherboad. :-/

So, take all that out of the equation in your decision and decide for yourself which is more worthwhile for your other purposes,

regards,

Road Warrior

[videobruce]

You can get refurbed NF2 boards from newegg NF7 or NF7-S v2's? Big difference. Didn't see any when I looked, but will look again.Which asserts that 5-6 seconds a day is typical and as much as 15 seconds is common. It also has a lot of insight into the timekeeping processes of PCs.Why can a regular clock that sells for $5 or $10 keep time? The time issue isn't just 6 seconds, its 5,10 and 15 minutes a day when the problem is happening. Some program/process is wrecking havoc in the system, but it doesn't ahppen all the time (no pun intended). It'd been two days off, two days on. The past two days it's been ok, NOW the clock is 750 seconds FAST. When I used it yesterday it was 9 seconds fast. It think it was used last night by my wife I beleive (e-mail mostly).
The clock gains minutes within an hour!
I tried the usual, viruses, spyware, checked for processes, tried different programs, no lock with anything. I run the same programs all the time (usually 3 or 4) and tried different combinations, but nothing consistant. Thanks for that link BTW, though I see they talk about LOOSING time, NOT gaining which I am aware of.
My first MB/system back in '99 kepted great time and as far as I know it still does (I sold it to a friend).

As far as the hibernation, the serial port works as far as the IR receiver goes, it's when the box is off, whatever command that is suppose to be isssued, isn't. The vendor says it's a TCP/IP deal.........

BTW, hello neighbor......... ;)

[RoadWarrior]

Best way to check if it's hardware or software screwing it up is to check how it keeps time when switched off. If it's drifting less than 6 secs a day (losing or gaining) when it's switched off, you're hardware clock is about as good as you can expect. If the major drifts only happen when it's on, then it's surely a software problem......

Some things to look at.
Is the system timer sharing an IRQ with anything? It should be fixed on IRQ 0 on it's own, but ACPI can do some freaky things sometimes.
Are you using any DOS games or programs? Some of those can play havoc with the system timer.
Try a fresh CMOS battery, I've seen systems GAIN time with dying batteries, a fresh one solved it. Oddly the clock would screw up before it lost any CMOS settings or gave any other sign of needing one.
I recall some issues with an older version of Winamp some time ago, forgot what version but it kept screwing up the time, so you might want to try a different version or use a different player to see if that helps.
USB webcams, wireless NICs, drives, etc might be i/o binding the CPU and causing it to miss timer interupts.

The likely problem with the IR receiver is that, normally a serial port wakeup event would be triggered by the +/-12V signalling of the remote serial device on the serial cable, since the voltages used to drive the serial port are local to the IR reciever and the device is in standby, then I don't think that the IR signal is enough to wake it up, because there's no external voltage fluctuation to trigger it. .... or something along those lines.

regards,

Road Warrior

[videobruce]

The likely problem with the IR receiver is that, normally a serial port wakeup event would be triggered by the +/-12V signalling of the remote serial device on the serial cable, since the voltages used to drive the serial port are local to the IR reciever and the device is in standby, then I don't think that the IR signal is enough to wake it up, because there's no external voltage fluctuation to trigger it. .... or something along those lines.I was told it was because the IR remore through the serial port uses TCP/IP protocol making it different from a progarm that will wake the system back up (which DOES work fine here!).
This 'command' that wakes the system back up, it is stored in the bios correct? If so doesn't that mean the command isn't getting there in the first place? IOW's it's either not being generated or it isn't making the connection.

BTW, the battery has been changed twice in a year and it is definitely a software issue, but it is intermittent. I run the same porograms all the time and some days it does it, others it doesn't. :bang head

[RoadWarrior]

Wierd, the TCP/IP protocol is handled on the PC at a higher level in software. So if the device requires TCP/IP to communicate then yeah, it's not going to do a darn thing while the PC is on standby, asleep, hibernating or off. That would be an odd way to implement a remote though, I didn't think it was possible to use TCP/IP for a one way connection such as a remote control, because it requires the ability to request lost packets etc.. This only makes sense if you are using some device such as a PDA as a remote control, where there is a 2 way communication link established. If you have a simple, dedicated remote I am inclined to think the tech was BSing you about TCP/IP.

Road Warrior

[videobruce]

The remote doesn't turn on the computer, it just would turn )shiydown) the computer if you wish (option).
As far as how the regular commands are handled with the remote I don't know, but the command to wake is set up before the unit is shutdown via TCP/IP, or at least that is the way I understand it.