Hi Folks,
I've discovered that the hardware clock on one of my machines is running about 2/3 slow: loses about 20min in each hour.
For the time being, I'm writing the system time to it every so often, and the system time is updated hourly by NTP.
But the principal irritation of it is that the system clock is set from the HW clock on bootup, so I have to make provision to over-ride that.
I don't know of any important system functions that directly consult the HW clock rather than the system clock. Do you?
Also -- and here I'd really welcome good ideas -- does anyone have any suggestions for why this might be happening, and/or for what to so about it?
With thanks, Ted.
-------------------------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@nessie.mcc.ac.uk Fax-to-email: +44 (0)870 094 0861 Date: 27-May-06 Time: 13:34:41 ------------------------------ XFMail ------------------------------
On 5/27/06, Ted Harding Ted.Harding@nessie.mcc.ac.uk wrote:
I've discovered that the hardware clock on one of my machines is running about 2/3 slow: loses about 20min in each hour. For the time being, I'm writing the system time to it every so often, and the system time is updated hourly by NTP.
I have no cure, except to run ntpd all the time - it will notice your clock drifting and correct it.
But the principal irritation of it is that the system clock is set from the HW clock on bootup, so I have to make provision to over-ride that.
Don't turn the computer off?
Does anyone have any suggestions for why this might be happening, and/or for what to so about it?
Probably worth checking the bios battery.
Tim.
On 27-May-06 Tim Green wrote:
On 5/27/06, Ted Harding Ted.Harding@nessie.mcc.ac.uk wrote:
I've discovered that the hardware clock on one of my machines is running about 2/3 slow: loses about 20min in each hour. For the time being, I'm writing the system time to it every so often, and the system time is updated hourly by NTP.
I have no cure, except to run ntpd all the time - it will notice your clock drifting and correct it.
But the principal irritation of it is that the system clock is set from the HW clock on bootup, so I have to make provision to over-ride that.
Don't turn the computer off?
Normally I don't, except occasionally. but recently I've been doing some "maintenance" and have rebooted several times (which is how I noticed what's goinf on).
Does anyone have any suggestions for why this might be happening, and/or for what to so about it?
Probably worth checking the bios battery.
I thought of that, but couldn't see one anywhere. However, a little while ago, looking over the mother board I saw a block-like item attached to it, and guessing that this might be the clock I gave it a firm press with my thumb. Since then, the HW clock seems to be staying in sync with the system time (to a second or two), so maybe I've cured it!
Thanks for the comments, Tim. Ted.
-------------------------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@nessie.mcc.ac.uk Fax-to-email: +44 (0)870 094 0861 Date: 27-May-06 Time: 18:00:25 ------------------------------ XFMail ------------------------------
On Sat, 2006-05-27 at 18:04 +0100, Ted.Harding@nessie.mcc.ac.uk wrote:
I thought of that, but couldn't see one anywhere. However, a little while ago, looking over the mother board I saw a block-like item attached to it, and guessing that this might be the clock I gave it a firm press with my thumb. Since then, the HW clock seems to be staying in sync with the system time (to a second or two), so maybe I've cured it!
The black box is a small "cap" over a socketed Real Time Clock chip (usually a Dallas) the cap contains a small 3 volt battery which is soldered to two of the chip legs that have been bent upwards. You are supposed to replace the device whole, but some of them are getting a bit hard to find, in which case you have to carefully prise of the top and replace the battery.
Just to slightly complicate matters. These devices contain a battery backed area of CMOS memory for the Bios configuration settings (not the BIOS itself that's on an EEPROM) Some systems need this Ram preloaded with some information before they will save to it. So on certain mainboards if you change the device or change the battery the BIOS will be unable to save it's configuration data and will run with defaults every time you boot the system.
In your case however I am not sure why pressing on it has cleared the fault. I think there may be a write enable pin, perhaps if this isn't connected properly the system cannot update the HW clock ?
On 28-May-06 Wayne Stallwood wrote:
On Sat, 2006-05-27 at 18:04 +0100, Ted.Harding@nessie.mcc.ac.uk wrote:
I thought of that, but couldn't see one anywhere. However, a little while ago, looking over the mother board I saw a block-like item attached to it, and guessing that this might be the clock I gave it a firm press with my thumb. Since then, the HW clock seems to be staying in sync with the system time (to a second or two), so maybe I've cured it!
The black box is a small "cap" over a socketed Real Time Clock chip (usually a Dallas) the cap contains a small 3 volt battery which is soldered to two of the chip legs that have been bent upwards. You are supposed to replace the device whole, but some of them are getting a bit hard to find, in which case you have to carefully prise of the top and replace the battery.
Just to slightly complicate matters. These devices contain a battery backed area of CMOS memory for the Bios configuration settings (not the BIOS itself that's on an EEPROM) Some systems need this Ram preloaded with some information before they will save to it. So on certain mainboards if you change the device or change the battery the BIOS will be unable to save it's configuration data and will run with defaults every time you boot the system.
Many thanks for all this info, Wayne. Unsuspected by me! It's good to learn by being told by an expert that's it's something I'm better off not fiddling with, rather than finding out the hard way. (Still keeping good time, by the way ... ).
In your case however I am not sure why pressing on it has cleared the fault. I think there may be a write enable pin, perhaps if this isn't connected properly the system cannot update the HW clock ?
In the words of an historic invoice from a maintenance engineer who, called in to fix a flaky box, gave it a thump on the side and got it working instantly:
To: Hitting the box . . . . . . . . £ 0.02 To: Knowing where to hit it . . . . £49.98 ------ TOTAL £50.00 ======
(Just my 2p-worth)
Cheers, Ted.
-------------------------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@nessie.mcc.ac.uk Fax-to-email: +44 (0)870 094 0861 Date: 28-May-06 Time: 10:47:32 ------------------------------ XFMail ------------------------------
On 28-May-06 Wayne Stallwood wrote:
[...] The black box is a small "cap" over a socketed Real Time Clock chip (usually a Dallas) the cap contains a small 3 volt battery which is soldered to two of the chip legs that have been bent upwards. ... [...] In your case however I am not sure why pressing on it has cleared the fault. I think there may be a write enable pin, perhaps if this isn't connected properly the system cannot update the HW clock ?
Now that I think about your description of how things are arranged, I suspect that the contacts between battery and chip pins may have degraded over the years (the m/board dates from 1995), and possibly leaning on the cap has improved the contact.
??
Cheers, Ted.
-------------------------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@nessie.mcc.ac.uk Fax-to-email: +44 (0)870 094 0861 Date: 28-May-06 Time: 11:06:03 ------------------------------ XFMail ------------------------------
On Sun, 2006-05-28 at 11:06 +0100, Ted.Harding@nessie.mcc.ac.uk wrote:
Now that I think about your description of how things are arranged, I suspect that the contacts between battery and chip pins may have degraded over the years (the m/board dates from 1995), and possibly leaning on the cap has improved the contact.
??
I wouldn't have thought so, the battery contacts are spot welded to the battery casing and are actually quite hard to remove. I can't imagine them coming undone by themselves.
These contacts are then soldered straight onto a couple of pins on the chip...You don't usually get dry joints on a leg to leg joint.
It's doing pretty well, the battery in those things is rated for about 5 years, unless of course it has already been replaced once.
-
Ted.Harding＠nessie.mcc.ac.uk -
Tim Green -
Wayne Stallwood