103G stops boot process

[colbec]

Well I can repeat this behaviour in two 1.8 Intel boxes - without the 103G inserted the boxes boot fine. With the card in place, which was working fine in one of the boxes, the boxes do not boot at all. Fan comes on, drive lights active as they are probed for disks, but then no beeps at all and no progress on boot, fan keeps blowing at high speed when normally it would throttle back until temp issues require more ventilation, and the box can only be shut down with holding the on/off switch in for several seconds or unplugging the power cord.

I have tried different slots on both machines, and inserting a different type of pci card is not a problem, even one which requires its own power supply, so the power supply itself seems to be putting out lots of juice.

Normally no beeps means power supply or system board problem. Perhaps inserting the card creates a short. Anyone seen this issue before?

 

[WyrTwister]

Well I can repeat this behaviour in two 1.8 Intel boxes - without the 103G inserted the boxes boot fine. With the card in place, which was working fine in one of the boxes, the boxes do not boot at all. Fan comes on, drive lights active as they are probed for disks, but then no beeps at all and no progress on boot, fan keeps blowing at high speed when normally it would throttle back until temp issues require more ventilation, and the box can only be shut down with holding the on/off switch in for several seconds or unplugging the power cord.

I have tried different slots on both machines, and inserting a different type of pci card is not a problem, even one which requires its own power supply, so the power supply itself seems to be putting out lots of juice.

Normally no beeps means power supply or system board problem. Perhaps inserting the card creates a short. Anyone seen this issue before?

Is this with or w/o the sat coax connected ?

Check to see if the card is seated all the way in the slot .

Check to see if anything on the card is shorting against anything else .

Wyr

 

[colbec]

Coax connected or not, same effect. The pc box has one of those blue plastic clips that hold all pci cards and dust covers in place. My experience is that unless the cards are seated all the way you can't get that clip back in place properly - I have reseated it several times, seems to slide in good and solid up to the neck, and tried several different slots. Card looks clear all round, plenty of room not to interfere with other cards or caps sticking up.

 

[Anole]

Dr. Mengele ?

Well, there seems to be a bit of history with this DVB card.
If I recall, you got it somewhat working in a 400mhz machine.
Maybe you could try it there, again.?
Because the alternative, based on what you've said, is that you've now had a second DVB card die.

 

[colbec]

Ah, Anole, to compare a poor grieving parent who despite the best care and attention has lost two children to the same disease to such an icon... See, here is the blood from your rapier wit...

The 400 MHz in question is currently testing out a BSC621-2 so is not currently available. In the meantime let us run a thought experiment and say that it does work in that box. What does that then say about the boxes it does not work in? I guess that brings us to motherboard compatibility, which I believe would be a new issue for this card.

Seems to me that some return policies are destined to fail. If you are obliged to return to the place of purchase then any exchange is likely from the same production batch and just maybe suffers from the same issues.

 

[Anole]

returns -
Yea, couple of years ago, I returned (rush-shipped $40) a pair of USB DVB receivers just under their 30-day time limit.
And got back two more that were just as heat-sensitive.
To which my buddy and I cut 'em open , providing an entirely different solution.

I was just hoping we could come up with some alternatives for you and your burden.
The first step, would be to determine if it's the card, or some interaction with your computer (including software, if you had gotten further).
If it worked back in your 400mhz, that would be one data point.
Without that success, . . . it'll be a real cloak 'n dagger task.

You didn't go into why the 103 card stopped working in one of your IBM? 1.8 machines.
Apparently it did work at one time. What happened.?

If the two 1.8's are very similar, then the fact that the card won't work in either is less significant.
I've found commercial PCs to sometimes be more proprietary than you'd think.
Usually, my machine collection all have different motherboards from different vendors, with different chipsets, and often quite different CPU chips.
So, moving a card from one to another is a real test.

If you're going to restrict the experiments to what you can do with the two 1.8's, then here are a few things ya might try:
- remove all I/O cards
- if you have optional on-board video, remove video card!
- try powering on the machine with no cards, and no video; then try with your DVB card and no video card; see if the beep pattern is same or different.
- try a master reset to the BIOS
- pull the motherboard battery, and short the jumpers to reset the CMOS
- repeat all tests above on the other 1.8ghz box.


That should keep you busy for the afternoon.
Hope some answers from the above tests will expose a solvable problem, and not confirm your poor parenting skills.

 

[colbec]

Anole, thanks for suggestions, I plugged the card into the 400 MHz machine and the same thing happened, nothing. CDRoms spin, fan goes on high and stays there, no visible signs of booting on screen or from audible signs/symptoms. Remove card and everything proceeds as normal. All 3 boxes are Intel but there is quite a generation gap between the 1.8s and the 400.

No options to remove on the 1.8 computers, no onboard video. As mentioned before, other PCI cards in the same slots work fine.

Now, as to what has changed. Well here I am stumped. It was working in the 400 and at least one of the 1.8s. The other 1.8 was pulled out of a buddy's working setup as a known good machine. So the card now stops all 3 machines from booting.

One observation: when it was working, the card seemed to get hot at one end, the end furthest from the coax inputs. Not on the conexant chip where you might expect the heat. Now, if the card stays in the computer as you wait for it to boot, the card seems to get hot on a part of the card that does not seem to be populated with chips or resistors. I have no conclusions to draw from this since I have no knowledge of how these cards are built.

 

[Anole]

don't make me say it . . .

"He's dead, Jim"

If you have a picture of the board (even from their web site) and want to mark on it where it gets hot, we can discuss it more.
But, I think you've pretty well settled the matter.

Are there any jumpers or configuration flash , or anything similar on the card?
You could probe the motherboard power connector, with the card in (then out), and power turned on.

Maybe it's time to search other forums... or a general Google search.
Perhaps others have had similar troubles.

 

[TheKrell]

You could probe the motherboard power connector, with the card in (then out), and power turned on.

Hear hear. I predict that the card is shorting your 5v supply. That's why your fans will run but nothing else does.

How doithe outer edges of the card look? Is it possible to push it into the PCI slot slightly to one or another end?

 

[colbec]

I will check the motherboard power connector later when there is more time and I can get myself organized sufficiently to understand what I am doing. There are no jumpers visible to me.

The outer edges of the card look good. I guess that imperfect edges would indicate internal heating followed by expansion and cracking, or are you thinking of something else? I have never allowed the card to stay in the box for very long in its present state. If it does not boot I switch off toot sweet.

Regarding the card in the slot, it is a pretty tight fit and does not allow any lateral movement.

The new heating area seems to be directly over the middle of the longer span of the gold connector tabs, just to the side of the main chip in the middle of the card.

Thanks for your interest.

 

[colbec]

Voltages

I tried to check voltages - the main motherboard power connector is buried right beneath the connections into the drives, so I did not start disassembling that part.

What I did do is check voltages at an unused molex connector and found 12 and 5 volts both with the card in and the card out. This may be meaningless in this context if the short only affects the power at the mb and not the other power delivery lines.

Presumably I have to leave the main power connection into the motherboard intact otherwise there will be nothing to measure. Is it possible for me to check voltages say via a serial port pin?

 

[TheKrell]

Well, I'll be darned. Seems the card is not shorting your 5V supply after all. I believe you could check it on a serial port pin, but that will likely tell the same story. I don't think the supply has multiple independent sources such that some 5V leads would remain up while others go down.

What I was thinking when I wrote my earlier question was whether it was possible to insert the card into the pci slot offset one way or another. A card couldn't stand much offset before the prongs either bridge two traces or line up with the adjacent traces on the card. I don't think that's possible if the card edges and PCI slot are both in good condition.

 

[colbec]

However I think your theory is sound. Inserting the card is the problem, so there is a short created which does not allow the boot to continue. This could be anything, not necessarily involving voltage supply. It does not appear to be at the prongs, but could be inside the sandwich of the card itself. Intuitively it makes sense to me that cards that are prone to heating will eventually uncover defects in manufacturing.

 

[Anole]

<taps>

I don't think the supply has multiple independent sources such that some 5V leads would remain up while others go down.

What I was thinking when I wrote my earlier question was whether it was possible to insert the card into the pci slot offset one way or another.
A card couldn't stand much offset before the prongs either bridge two traces or line up with the adjacent traces on the card.

Pretty much what I had on mind.
Some fancy power supplies many now have some somewhat independent outputs, but I'd consider it unlikely.
And if I couldn't probe the motherboard connectors while running, I'd sure hit the extra drive connectors...

I know some video cards fit their slots pretty badly.
But now we are talking about a stock PCI card.
So, it doesn't sound like a fruitful path, here.

The alternative, is that something on the card has died, dragging a data/address or other pin high or low.
The heating may be the result or the cause of the problem.

But as I suggested earlier, I think he's dead...

 

[colbec]

In the interests of reporting the truth here, I have more to add. In a sense one step forward and another one back, but sideways.

After ignoring the card for a week and hoping it would get better by itself, I plugged it in to an empty box (one in which it had failed before) and it booted. Break out the champagne. Shut the machine down again and rebooted, failed to start with the same issues as before, fans on high and stay there with no boot. Put the champagne back in the fridge.

So I thought, hmmm fridge, hmmm what about card in fridge? Put card on bottom shelf of fridge for 15 mins, plugged it into the box again and booted fine. Left for several minutes, shutdown and rebooted, failed. Put card back in fridge for 15 minutes, replaced in box and booted fine again.

Kept it running this time, attached the satellite input and tuned E*7, AMC5 and Hispasat 1C1D no problem at all including the weak channel Solo Tango. Left it running for two hours, no issues at all with heat or failure. Only problem was changing the video cable, ProgDVB does not like to have the input changed while it is running.

After running for 3 hours to get the card nice and warm I shut the box down again for the final test of the theory that a nice cold card will start in the box but a warm one won't. Box started up fine again with the warm card. Sometimes it can be very hard to establish a consistent pattern of behaviour!

No doubt this is not the end of the story.

 

[colbec]

Shutdown the box and unplugged it from main power supply and left it overnight for 12 hours. On restart, it failed. Put the card in the fridge for 15 minutes, replaced the card and the box booted.

I have noted that this card appears to be reporting itself to the O/S (XP Pro) as a "Genia Networking" card under the network adapters . Is this normal? I am using the mainboard network chip and have no other network cards installed.

I will now repeat the experiment with the box shut off for an extended period but this time leave the box plugged into the main power. This maybe keeps a chip warm that might help.

 

[CharredPC]

Sounds like a finicky bit of hardware! Just to throw another random possibility into the mix, I've had a similar problem with a PCI card. It turned out that it had at least one bad solder joint. Putting it in the fridge apparently made the solder condense (shrink) momentarily, which tightened the connection point and made it function enough to start working. Another theory is that the warm card in the cool fridge created slight condensation at the joint and allowed for current to move...

Anyhow, I ended up carefully redoing a few dozen solder spots on the board (heated a second to re-liquefy each and ensured the wire or part poked through visibly) and never had a problem again. YMMV, but if you or a friend is handy with a solder gun it couldn't hurt. A lot of seemingly random electronic dysfunction is due to this issue in my experience. I've also owned a 103g, and (at least on mine) all solder points were accessible- no "sandwich boards," thank goodness!

 

[colbec]

Thanks, CharredPC, this is the reason I put the card in the fridge. I have heard about bad connections before and how minute gaps can render things unreliable. The problem could be either a bad connection as you describe or a connection that is too good, ie one where the solder when warm leans over and touches another line that it should not. Your comment on the board not being a sandwich type is helpful information. As you say it makes the issue slightly easier to diagnose.

I just tried booting that box again and it worked fine after 6 hours. The difference being that this time I left it plugged in, so that various parts of the mainboard remain active even though the PC itself is not running.

I'm inclining in this case to a combination of the bad connection and card identification. If the PC goes cold, ie all memory in chips that can is allowed to die and it forgets that it has seen the 103G before, then on boot it says "Hey there is a new card" and asks it what it is. The card when warm is brain dead in this department. It would not know how to respond. If the card is cold and that connection is, at least for a short time, functioning, then it can identify itself. If the PC is warm, ie the cards are 'remembered' then there is no need for the PC to poll the card and the bad connection is not discovered.

I should read some more about the bios and pci resource allocation. It may be that I have a setting that is not compatible with that type of card.

Edit: nope, I left it plugged in overnight but in the morning it failed to boot. Whatever is decaying/changing does so after 6-8 hours and without regard to presence of minimum power on the mainboard.

 

[colbec]

Follow-up

As a follow up on this item, the card is now working and I cannot put it down to anything specific I have done.

1. I contacted the retailer and Digiwave for support regarding warranty and did not hear from either of them. As far as I am concerned 'warranty' on this product through this retailer/wholesaler means nothing.

2. Realizing I was stuck with the card, I found an accommodation with it by ensuring that when the computer was shut down it went into hibernation and was not fully shut off. Then the card would not stop the computer on rebooting. Total shut down would mean the card would need the fridge treatment.

3. From this point the major glitch was that on computer restart from hibernation ProgDVB could not tune channels until the computer had been rebooted. On reboot it could tune channels again. This was consistent.

4. After many days of this repeated treatment I have discovered that I can once again shut the computer down fully, unplug it, leave it overnight to get cold and it will boot fine in the morning. I am at a loss to find a reason for this return to its behaviour straight out of the box.

Now I am getting HD channels and AC-3 no problem at all, and diseqc switches are handled fine. I could almost say it has become a very good card. Just one more problem which I think deserves its own thread - screen resolution.

 

[colbec]

After 6 months or so of intermittent use, watching HD from time to time no problem, yesterday the system returned to the above odd behaviour. Freezing the cards for a while did not work. Pulling the CMOS battery (Anole recommendation above) at first did not work, but pulling the battery and leaving the box open overnight did work. Now it is booting fine again.

I thought physics was supposed to be reasonably predictable?