Seeking an explanation or theory

I would suspect wrong bios settings. remove the battery for 15 minutes,or find the bios reset strap, and let the bios initializee itself to factory settings. Dont forget to put the reset strap back in its old position.

I appreciate your suggestion, but the computer power never really was applied in the failure I described. Hence the bios settings could not have come into play since there was no power to read them. Also, without having changed any of them, they worked for a boot when the power cord had not been removed with any combination of RAM.

Now I could be missing something. If so, I am open to what I am overlooking. My theory has to do with the standby voltage not coming up soon enough to reset the logic properly, but I do not understand why the size of the RAM has anything to do with it.

There are a number of things the bios has to "read" from a memory stick in order to properly use it. Therefore the re-initializing.

I think you are, the bios is mis-configured perhaps, and resettng it to factory would perhaps correct the problem.

Maybe it doesn't?

up.

perhaps not properly seated causing short.

Either you blew something by plugging the ram in while the standby power was on or the new ram requires more standby power than the power supply can produce. Standby power is low and you 300W psu may not have a higher current rating for standby than the 250W one, it may be that only the 5v or 12v rails can give more power. The labels on the psu usually have the ratings for each rail.

Have you tried both 1GB sticks with a 512MB one and they both give the same symptoms?

To the extent that I understand your problem, I think it's a non-problem.

In general, it's not enough to simply shut down the machine. Before doing any work, you should throw the "standby" switch on the back to "off", and wait for the green LED to go out. Then, and only then, should you swap cards, update memory, etc.

If the machine still boots correctly when its original memory is installed, then there is nothing wrong with the machine. If the machine does not boot correctly when the new memory is installed (the standby power off), then either the memory is defective, or it is not the appropriate memory for that machine.

In short (no joke intended), I don't think you have a problem. At least, not with the computer itself. I would not, at this time, worry about "why".

This makes the most sense. I am not sure the standby current is listed for either PS on the label. Each rail is, but not the standby current.

As for blowing something, both sticks work just fine. In fact, I took one stick I had replaced with a 512MB one and placed it in my current computer when I saw a possibility existed for a problem when a total power shut down occurred. Both sticks are working just fine right now.

Standby power is low and you 300W psu may not have

Yes, it did not matter which 1GB stick was used. They behaved the same way.

I agree, I do not have a problem. I am just curious as to why the computer behaved as it did. Call me overly aggressive, but I like to come away from a problem with an explanation to a solution. In most cases when you find the solution you are able to understand why it behaved as it did.

I do not dispute your point. That IS the best way. But in this case a complete removal of power (such as a power failure) forces the power supply to continually cycle with no way to interrupt it. Even the reset button would not interrupt it.

I am not trying to be argumentative, I appreciate your input and that of everyone else. But if the memory were defective, why would it operate correctly with either 1 GB stick, but not both??

As I said, I am curious as to why it behaved as it did.

Good question. Does the computer support two 1GB sticks of that memory type? Also, does it matter which stick goes in which slot?

You needn't reply; I'm just asking questions.

And I didn't fully understand what you were getting at.

I don't mind replying if you don't mind reading it. Both 1 GB sticks DO work in the computer and pass a rigorous test regardless of their slot. The problem only exists when the computer is applied power from a totally off condition. That is with no standby voltage built up and applied to the motherboard.

If I were to install the 2 GB of RAM with standby voltage applied to the MB, it would run just fine.

Something I have noticed repeatedly, with modern PCs having ACPI-aware BIOSen and "soft" power supply switching, is the following:

- PC was shut down properly, unplugged, and serviced

- PC is plugged back into the wall

- There is an obvious surge of current - lights dim in the room for a moment, fans start to spin, power LED comes on, etc.

- PC apparently shuts down, stops drawing power... fans stop, lights go out, etc.

- Pressing the power-on button causes the motherboard and peripherals to power up and boot. In systems of this sort, the power-supply switch is "soft"... it's actually just a logic signal to some circuitry on the motherboard, which is powered by the PSU's low-current "standby" supply, and which kicks the rest of the system awake. A similar kick can be provided by the "wake on LAN" signal from the Ethernet adapter.

These motherboards also tend to have a BIOS setting which tells them what to do when power is applied from a cold start: remain off, or power up, or restore the previous power state.

I believe that the system design is such that if the motherboard is powered down completely (i.e. no standby power available from the PSU), it will automatically tug on the "power on, please!" line to the PSU. Thus, when the PSU is plugged it, it'll immediately switch on and provide full power to the motherboard, which goes through a full reset, initializes the CPU, and starts up the BIOS.

The BIOS then presumably goes through some amount of its normal POST processing, reads the system configuration out of NVRAM, and decides whether it ought to power up. If so, it just continues with the boot. If not, it configures the motherboard to generate another power-supply wakeup event if the POWER button is pressed, and then switches off the main power supply (leaving the standby supply running).

I can well believe that if there's a failure of some critical early-stage step in the POST (including a failure to initialize the DRAM controller or the DRAM itself), then the system might either just stop dead in its tracks, or attempt another power-off/power-on cycle to do a really *hard* reset of the failing hardware.

I suspect that's what you were seeing, if you plugged in some DRAM that the motherboard "didn't like" for one reason or another.

Ah! Well...

What happens when you turn on the standby power ("the green LED") and then wait briefly before applying the main power (that is, pressing the main power switch)? This is the usual way PCs are operated. I think.

I can't think of any other way of doing things. You have to turn on the standby power before you start the machine, and it takes at least a second or two to get your hand from the standby power switch to the main power button.

Again, no reply is needed. I don't want to waste your time considering possibly incorrect suggestions.

It is on every psu I've looked at, +5Vsb. It's not very high about 1amp but I did have a motherboard that required 1.5amp minimum.

The power supply comes fully on initially, as another poster explained, to run the cpu and read the BIOS to see what to do on power application. Your psu is trying to do this but the initial surge is too much for it and it shuts down and then has another go. I've seen this when a drive failed once and once when lightning took out the memory on a system.

You could try disconnecting the drives to see what happens but check the 3.3v rail limits on the psu labels between your 250 and 350W psus. It may just be the 5 and 12 volts that have extra current and the 3.3v rails have the same limits. The memory uses the 3.3v rails [1] with on motherboard regulators dropping it to what the memory needs, and it's possible that the new memory is simply taking too much when when first switched on. What I'm trying to say is that just because a power supply is a higher wattage it doesn't necessarily follow that all rails have the limits increased.

I'd want to run a memory test, such as Memtest86(+) for 24 hours though to ensure that the memory wasn't damaged by inserting it with the standby power on. The system gets the settings for the ram from a chip on the ram module and if that's been damaged the defaults may be a little high and overheat things after a while causing failures. If you are running Windows then CPU-Z should be able to read and display the ram configuration for you.

I think this is exactly what is happening. There definitely seems to be a correlation between the initial surge and it failing as it did. If you recall I could overcome this failure by allowing the LED on the PS to turn off and immediately plug in the power cord again. I suspect the standby voltage had started to discharge but not yet fully discharged and that built up voltage overcame the surge.

I do wish I had thought of unplugging the hard drives and trying it, but I have already sent off the computer.

I have AIDA32 on both computers utilizing the PC3200 RAM, and the SPD chip on each stick reads correctly. I did not run the memory test for more than twenty minutes, but both computers are working perfectly with the PC3200 RAM.

I admit that I do not have a problem, but a curiosity. One thing that puzzles me however is the failure of the manual reset I wired in to reset the logic during the failure I described. One would have thought if anything were being read from the bios, that the +5 volts would need to be up. If the +5 were up, then the reset would have also worked. This is why I doubt that the wake by LAN or any bios setting could account for the power starting and stopping every 1/2 second. I don't think the +5 ever reaches that level. In fact I tend to believe the standby voltage is also not up to an adequate level.

At any rate, I got some good suggestions and comments. The next time I encounter such a problem I have a couple more ideas I might explore. Thanks to all who commented.

I shouldn't think it was getting up to read the BIOS or see the reset switch. It would be the psu that's doing the shut down due to overload and nothing to do with the bios.

Hi!

Is this an ASUS board? (read on, you'll see why this might be interesting...)

I have an HP Pavilion from about the same timeframe that behaved precisely the same way. I picked it up off the curb, complete with monitor keyboard and mouse. It is based on an ASUS board.

The powering on from a dead stop after sevreal hours unplugged suggests a weak CMOS battery, which is well within the realm of possibility. This isn't always immediately noticeable, as most ATX motherboards maintain the memory from the standby power coming out of the power supply while they're plugged in. The battery is only used when the system is completely without power.

When the battery goes bad, some boards even lose track of their power state and come back on as though it is the default state.

My HP system did this, and replacing the CMOS battery (about $3, it's a CR2032 cell) solved that problem.

But then the cycling started up. I upgraded the memory in much the same way you did, using PC3200 parts. The system was happy, passed a memory test and then the next time I went to use it, it would only cycle repeatedly with a tiny "flub" noise from the power supply. The cycling indicates a power supply which is going into overload or short circuit protection, coming out of it when the supply shuts down and starting the cycle over again.

I cleaned and *firmly* reseated the memory. It solved the problem.

Every now and again, when the weather changes or the computer is moved (it is not, for obvious reasons, a primary system of mine) I have to reseat the memory or this problem crops up. My guess is that either the memory DIMMs were not quite made properly or there is something slightly strange about the sockets they go into.

William

I really do not know who made the MB for Compaq. Since I no longer have it, I cannot answer that question. I do know it was seen as a "Trigem Glendale," and I have seen some comments about Asus. Perhaps Asus makes it??

The CMOS battery is not the problem, as I replaced the old one with a new one. The old one measured 2.92 volts and since I had it open I thought it was easier to put a new one in.

This isn't always immediately noticeable, as most ATX

You seem to have experienced the same problem as I did.

I cannot argue with the results you achieved. All I can say is re-seating did not solve my problem. In fact even after I had resigned myself to the 1 GB + 512 MB stick combination, I went back to see if the problem would occur again consistently. (I am one of those people who has a hard time accepting a mystery, as I am convinced there IS an explanation, just not known.) It behaved the same way again! It did not matter which socket was used or which 1 GB stick was used. If the combination was two of the 1 GB sticks, it would fail the same way. If I placed one 512 MB stick in combination with a 1 GB stick, it worked.

It might just be that I will have to accept this mystery? It does however make me wonder if the two different speeds had an impact? The thing that upsets all my theories is the ability for it to run with 2 GB if the standby voltage is still applied or not totally discharged.

I do appreciate your comments and your relating how you encountered the same problem. It convinced me that I am not losing my mind!!

Trigem made it. Glendale is the model. Trigem made a lot of older HP & Emachines motherboards. The only difference in some was the bios chip. Their 'Florida' model was in a lot of Win ME models.