Interesting AGP warning LED problem
in [Asus]
|
Prev: Asus P4R8L Drivers required......Urgently!!
Next: nvata.sys blue screen, Microsoft IDE driver speed (lack of!)
From: kimiraikkonen on 11 Jan 2008 06:41 On Jan 11, 1:27 pm, Paul <nos...(a)needed.com> wrote: > kimiraikkonen wrote: > > > Hi Paul, > > Thanks for taking care and good explanations. Here is the current > > situation: > > > Existing AGP card is inserted, mobo boots up and comes to Windows with > > no problem, but the AGP warning led is not stable, after some computer > > operation it begins to lit for some reason, if i remove and re-insert > > card, it doesn't lit. But during normal operation it doesn't lit for a > > while, then suddenly it lits but this light has no effect on graphic > > or other computer operation. I can play 3D games and surf Windows but > > sometimes the screen may go off and come off for a short time.(don't > > know the reason also). > > > Additionaly, another interesting point, if i press or touch the AGP > > card slightly with my fingers while it's inserted and operating on > > Windows, "sometimes" but not always AGP warning led comes and goes for > > an instant. I mean i don't it's a fluky/chance that when i touch the > > seated card to make sure it's completely seated, AGP warning LED lits > > for a time. But this is rare. > > > However, the burn smoke on VGA card's GPU is not present, but after > > some testing i can report if additional problems occurs about > > displaying. > > > Never had such problems with that good mobo so far, I trusted Asus > > components over the years, today i'm so surprised because of an > > unknown hardware reason. > > > Thanks! > > I would not recommend poking the video card while power is applied. > It would not take too much movement of the AGP card, before there > would be trouble. AGP uses two rows of contacts, and a slight slip > would cause the wrong pins to get connected together. > > I do not have a complete circuit diagram for "AGP warn". I only > traced the circuit, from the TYPEDET# pin to a transistor, on my > P4B board. I did not trace the whole circuit, as that would have > taken forever. It is very time consuming to trace the copper, and > without identification for all the components, I still might not > understand the whole thing. I stopped tracing the circuit, once > I could see they were using TYPEDET#. (When I first got my P4B > motherboard, I didn't understand how the AGP warn would work. > So I traced the circuit just a little bit.) > > The idea of the circuit, is supposed to be that the ATX power > supply gets switched off, if the red LED is on. Your red LED > operating state, doesn't seem to affect the power supply like > it should. As far as I know, there are two transistors in the > circuit. But how they work together, I don't know. > > Paul I agree, make sense. But still confused why the computer boots up although AGP_warn lits with the card inserted, and also lits when there's no AGP card on the system. You mentioned a kind of driver corruption, is it frequent? Or does it seem hardware AGP_warn circuit problem? Does AGP_warn LED indicate a serious problem at the moment or just an annoyance? Thanks.
From: Paul on 11 Jan 2008 07:06 kimiraikkonen wrote: > > I agree, make sense. But still confused why the computer boots up > although AGP_warn lits with the card inserted, and also lits when > there's no AGP card on the system. > > You mentioned a kind of driver corruption, is it frequent? Or does it > seem hardware AGP_warn circuit problem? Does AGP_warn LED indicate a > serious problem at the moment or just an annoyance? > > Thanks. Based on your description, the "AGP Warn" circuit is not working at all. It seems to have more than one fault in it. If the LED lights, the computer power should go off (and it did not). The LED should not light when the AGP video card is removed (it did). I would say the circuit has no predictive ability at all now - it is worthless to you. I am confused as to how the circuit could go quite that wrong. I've heard of the light getting stuck in the ON state, but never as many different symptoms as yours has. Paul
From: kimiraikkonen on 11 Jan 2008 07:11 On Jan 11, 2:06 pm, Paul <nos...(a)needed.com> wrote: > kimiraikkonen wrote: > > > I agree, make sense. But still confused why the computer boots up > > although AGP_warn lits with the card inserted, and also lits when > > there's no AGP card on the system. > > > You mentioned a kind of driver corruption, is it frequent? Or does it > > seem hardware AGP_warn circuit problem? Does AGP_warn LED indicate a > > serious problem at the moment or just an annoyance? > > > Thanks. > > Based on your description, the "AGP Warn" circuit > is not working at all. It seems to have more > than one fault in it. If the LED lights, the computer > power should go off (and it did not). The LED should > not light when the AGP video card is removed (it did). > I would say the circuit has no predictive ability > at all now - it is worthless to you. > > I am confused as to how the circuit could go > quite that wrong. I've heard of the light getting > stuck in the ON state, but never as many different > symptoms as yours has. > > Paul Confusing than you guess beleive :-( Until now, i had two determinations about the AGP led situation. Currently it lits if: 1-No AGP card is installed. And it does its job, you cannot power up system and behaves as if there's a card on the slot, actually there's no card. Weird. 2-Correct AGP card installed(sometimes) and, sometimes after some computer operation or if you touch the card slightly during powered- on.
From: Paul on 11 Jan 2008 07:30 kimiraikkonen wrote: > > Confusing than you guess beleive :-( > > Until now, i had two determinations about the AGP led situation. > > Currently it lits if: > > 1-No AGP card is installed. And it does its job, you cannot power up > system and behaves as if there's a card on the slot, actually there's > no card. Weird. > 2-Correct AGP card installed(sometimes) and, sometimes after some > computer operation or if you touch the card slightly during powered- > on. The circuit needs two inputs. One is TYPEDET#, to sense whether the card is an illegal 3.3V one or not. But the other signal, I never traced. Somehow, the circuit senses whether an AGP card is installed. It almost sounds like that input could be malfunctioning, and controlling the circuit more than it should. Remember, that his circuit is based on simple transistors and resistors, An incorrect voltage somewhere could upset the bias and so on. It is not like a lot of the other stuff on the motherboard, running on pure logic levels. It is an analog circuit. The circuit must be running off +5VSB, because otherwise it would not be able to gate off the PS_ON# signal to the ATX power supply. This is a partial block diagram. On my motherboard, the transistors were near the AGP slot. The transistors are tiny (SOT-23 ?). http://www.national.com/images/pkg/mf03a.jpg +5VSB | +------+-----+ +--------+ TYPEDET# ---->| Transistor | | PS_ON# | | Circuit |------->| logic |-----> PS_ON# CARD_PRESENT ? ---->| | | | on main +------------+ +--------+ ATX power connector Paul
From: kimiraikkonen on 11 Jan 2008 07:36
On Jan 11, 2:30 pm, Paul <nos...(a)needed.com> wrote: > kimiraikkonen wrote: > > > Confusing than you guess beleive :-( > > > Until now, i had two determinations about the AGP led situation. > > > Currently it lits if: > > > 1-No AGP card is installed. And it does its job, you cannot power up > > system and behaves as if there's a card on the slot, actually there's > > no card. Weird. > > 2-Correct AGP card installed(sometimes) and, sometimes after some > > computer operation or if you touch the card slightly during powered- > > on. > > The circuit needs two inputs. One is TYPEDET#, to sense whether the > card is an illegal 3.3V one or not. But the other signal, I never > traced. Somehow, the circuit senses whether an AGP card is installed. > It almost sounds like that input could be malfunctioning, and > controlling the circuit more than it should. > > Remember, that his circuit is based on simple transistors and > resistors, An incorrect voltage somewhere could upset the bias > and so on. It is not like a lot of the other stuff on the motherboard, > running on pure logic levels. It is an analog circuit. > > The circuit must be running off +5VSB, because otherwise it would > not be able to gate off the PS_ON# signal to the ATX power supply. > This is a partial block diagram. On my motherboard, the transistors > were near the AGP slot. The transistors are tiny (SOT-23 ?). > > http://www.national.com/images/pkg/mf03a.jpg > > +5VSB > | > +------+-----+ +--------+ > TYPEDET# ---->| Transistor | | PS_ON# | > | Circuit |------->| logic |-----> PS_ON# > CARD_PRESENT ? ---->| | | | on main > +------------+ +--------+ ATX power > connector > > Paul Hi Paul, Thanks for your polite and friendly care so far, however i wanted to mail from private but it seems it's fake due to spam, doesn't matter. Currently, the aim of opening the topic is that i sometimes have display problem like display goes off and comes back for a while. System doesn't complain as beep or reset like there's no AGP card when this problem occurs, only monitor doesn't display anything, even i can see my shared files on the network through a different computer. After waiting for a while display comes back. That problem is the main purpose of the topic but the problem brought me to different places like AGP warn Led or else which are not may be the reason. Don't know but my current monitor is almost 10 years CRT one, maybe it may be the reason for this problem. Could be? I'm using 1024x768 @ 60hz for many years with no such problem. But it may be dying. Do you think that "display goes off and comes back" problem because of monitor? Thanks! |