Is the monitor physically close to the boards as they are tested ? Degauss operation will cause a pretty strong magnetic field spike at its beginning and it could induce currents in nearby metal things... but I don't have many ideas here, since multilpe video cards still exhibit the same thing which should rule out any cable related problem. But you can try to degauss the monitor without the cable connected and see if the boards still reboot.

Degaussing is usually connected directly to the AC line, so it's quite far away from the VGA signal in the monitor. Can you make sure that your problem is related to the monitor being connected to the computer? Keep your monitor and your victim computer at the same position as they are now, and connected to the same AC sockets as they are now, but do not connect the VGA cable. Then turn on your monitor while the 486 computer is running. Most modern monitors should degauss anyway, even if not connected to a computer, to display the "no video input" message on screen. If your PC still reboots, the problem is not caused by voltage/current spikes through the VGA cable, but by radiated emission from the degaussing coil or by conducted emission through the AC cables.

Sudden reboots caused by electromagnetic interference or brownouts (a degaussing monitor might load down the AC for some tenths of a second to cause temporary undervoltage on the AC socket) might in fact be caused by worn capacitors, so after narrowing down the cause for your issue, you might have some clue what capacitors to check.

The grayscale effect might have a different cause: Originally, some pins in the 15-pin connector were used to identify the monitor type (the monitor should ground some, and leave the remaining ones unconnected). In particular, mono VGA monitors are supposed to ground 12, whereas color VGA monitors are supposed to ground pin 11. Pin 12 got repurposed for DDC later on. Newer (DDC capable) graphics cards provide +5V on pin 9 to provide power to the DDC interface even if the monitor is powered off. Your ET4000 doesn't, as it pre-dates the DCC standard. I'd guess the ET4000 misinterprets the DDC signal at pin 12 for the "mono monitor identification". As your ET4000 card doesn't power the DDC circuit, the circuit that generates the DDC signal can only be powered while the monitor is on, so this might explain the difference between booting with monitor on and booting with monitor off.

In the past, I've seen little sparks appearing when attaching the VGA cables of a monitor to the DE-15 connector of some graphics cards..

Maybe there is a grounding issue?
PCs have switching PSUs, which I believe unlike real PSUs (with transformers) can accidently have a hot side as ground.

If the monitor has an opposite ground polarity and degauses, does this cause a reset perhaps?

Because, attaching these VGA cables in my example (flying sparks) often caused resets, too.

Also, if that's the cause, how to fix this? 😕Inserting the AC plug of one device the other way round? Say that of the CRT?
If it has a hot chassis, like a vintage TV, this should work?

Jo22 wrote on 2021-08-06, 07:49:

In the past, I've seen little sparks appearing when attaching the VGA cables of a monitor to the DE-15 connector of some graphics cards..

Maybe there is a grounding issue?
PCs have switching PSUs, which I believe unlike real PSUs (with transformers) can accidently have a hot side as ground.

Grounding issue is one of the theories I have in mind, too.

PC PSUs are "galvanically isolated", just like transformer PSUs. The output doesn't need to have a fixed ground. But as we know, the PC case, which is metallic at least at the slot covers, is connected to ground. Having ground connected to the hot side would make the PC case hot, which must not happen. This is why standard PC supplies come with three-conductor AC cables and need to be connected to a grounded outlet. The ground used on the cold side of the supply is connected to the protective earth connector of the AC outlet. If that connection is hot, you've got serious problems with your electrical installation. The same applies for most CRT monitors I know: They also are connected to a grounded outlet (it might be the pass-through outlet on an PC/XT/AT power supply), so the signal ground in monitors and computers is supposed to be at the same level.

With LCD monitors, that changed. Especially LCD monitors with external power supplies often come with 2-pin power cords that provides a floating output. "Floating" means it is not yet fixed to ground being at a certain level like earth or hot or something else, but you can pull it externally anywhere you like. This floating ouptut might float to half of the line voltage by default if it isn't pulled somewhere else, because it is often softly coupled to the AC line through interference suppression capacitors. Connecting the cable of an ungrounded LCD monitor to a PC starts providing earth as ground to that monitor through the cable shield, and some sparks while connecting are likely.

In same places of the world (Germany for example), it was common to have neutral (the non-hot power wire) and protective earth (the ground wire that should be connected to the case) on the same conductor in the electrical installation, and have protective earth to ground bridged in the AC socket. In this kind of installation, current spikes on AC (like degaussing, which can cause momentary currents up to 30A in 250V countries, I expect it to be even higher in 120V countries) can pull the combined protective earth/neutral a considerable amount away from actual earth potential.

So while in theory, ground should be at an identical level in the monitor and the PC, because ground level is provided through the protective earth conductor that shouldn't carry significant current, and especially DC ground level shouldn't be affected by AC consumption, in practice degaussing might cause ground of the monitor to drift away from earth under certain circumstances, and providing that shifted ground to the computer through the VGA cable can cause conflicting ground levels in the PC (one from the AC socket, another one from the VGA socket), which can upset voltage monitoring logic, so momentary loss of "POWER GOOD" isn't impossible. And loss of POWER GOOD might trigger a power-on reset circuit.

This is why I proposed to test whether the VGA connection itself is causing the issue. If disconnecting the VGA cable makes the issue go away, grounding problems, as you suspect, are likely. If the reboot issue stays when you disconnect the VGA cable, we need to look into momentary AC voltage drop or radiated magnetic interference instead.

mkarcher wrote on 2021-08-06, 09:55:

In same places of the world (Germany for example), it was common to have neutral (the non-hot power wire) and protective earth (the ground wire that should be connected to the case) on the same conductor in the electrical installation, and have protective earth to ground bridged in the AC socket. In this kind of installation, current spikes on AC (like degaussing, which can cause momentary currents up to 30A in 250V countries, I expect it to be even higher in 120V countries) can pull the combined protective earth/neutral a considerable amount away from actual earth potential.

Thank you very much for the explanation! 🙂

Yes, I think that might be true.
As far as I remember, neutral and protective earth come together in the fuse box at some point also.

The outlets have three different pins (two holes, 2 outer contacts wired together).
One hole has phase (brown), the other one has neutral (blue) and the outer contacts are protective earth (yellow+green).

So if a fully wired plug is used (the bulky schuko type) instead of that flat, 2pin euro connector, the grounding can be done via protective earth.

Back in the 1960s or so, TVs had hot-chassis still due to missing transformers and the use of the old 2pin plugs that lacked protective earth.

As a workaround, people used external isolating transformers with their TV sets sometimes.
These transformers also protected the TVs from voltage peaks and other dangers.
- TVs back then used to be very sensible (Edit: false friend; meant to say sensitive).
The tubes etc. were directly AC powered sometimes.

To my understanding, switching PSUs nowadays have a pseudo galvanic insulation.
They have a 100khz miniature transformer (ferrite ring) that does the insulation.
One side is connected to the control circuit, the other to the output?

Another issue I can think of:
Maybe an high capacity electrolytic capacity has failed in one of the devices' PSU.
If it was a blocking capacitor that was wired against ground, it could explain why such strange things like resets happen.
Without the blocking capacitor, the ground connection of one the devices could have accidently become hot. Anyway, that's likely not "hot enough" to harm people, but it might affect sensitive electronics.

I hope what I wrote was understandable and makes sense anyhow.
I'm struggling a bit with writing proper English at the moment, I'm afraid. 😅
Even Google translator might be doing a better job, I'm afraid. 😂

Edit: The other things you said are also excellent. Above, I simply answered to the block I was referring to, hope you don't mind.
The "power good" circuit in a PC is also an explanation for resets or shut offs, yes.

Tiido wrote on 2021-08-06, 05:47:

Is the monitor physically close to the boards as they are tested ? Degauss operation will cause a pretty strong magnetic field spike at its beginning and it could induce currents in nearby metal things... but I don't have many ideas here, since multilpe video cards still exhibit the same thing which should rule out any cable related problem. But you can try to degauss the monitor without the cable connected and see if the boards still reboot.

I mean it's nearby, but further away than it would have been sitting on top of a desktop case. I should add the board is not in a case, although my other boards all tolerate operating like that without rebooting when the monitor degausses.

mkarcher wrote on 2021-08-06, 06:38:

Degaussing is usually connected directly to the AC line, so it's quite far away from the VGA signal in the monitor. Can you make sure that your problem is related to the monitor being connected to the computer? Keep your monitor and your victim computer at the same position as they are now, and connected to the same AC sockets as they are now, but do not connect the VGA cable. Then turn on your monitor while the 486 computer is running. Most modern monitors should degauss anyway, even if not connected to a computer, to display the "no video input" message on screen. If your PC still reboots, the problem is not caused by voltage/current spikes through the VGA cable, but by radiated emission from the degaussing coil or by conducted emission through the AC cables.

Alright, now confirmed the VGA cable need not be connected to video card to cause the motherboard to reset. The monitor doesn't have a "no video input" message so I did have to connect it to a laptop, but doing that and powering it on did generate a fresh POST beep from the 486 board nonetheless. So it's either physical proximity or something happening through AC lines as you mention. Might it be a quirk of the UM498 chipset?

mkarcher wrote on 2021-08-06, 06:38:

The grayscale effect might have a different cause: Originally, some pins in the 15-pin connector were used to identify the monitor type (the monitor should ground some, and leave the remaining ones unconnected). In particular, mono VGA monitors are supposed to ground 12, whereas color VGA monitors are supposed to ground pin 11. Pin 12 got repurposed for DDC later on. Newer (DDC capable) graphics cards provide +5V on pin 9 to provide power to the DDC interface even if the monitor is powered off. Your ET4000 doesn't, as it pre-dates the DCC standard. I'd guess the ET4000 misinterprets the DDC signal at pin 12 for the "mono monitor identification". As your ET4000 card doesn't power the DDC circuit, the circuit that generates the DDC signal can only be powered while the monitor is on, so this might explain the difference between booting with monitor on and booting with monitor off.

I did have to entirely unplug the monitor to force-boot in color, not just power it off, but still this all makes sense. I should note I just got the board and this is only the second time I had everything set up. The first time, I never ran into this issue. This time, it was happening on most, but not all boots. So it does make sense that the card is trying to interpret some kind of noise on pin 12 as connected to ground or not and it's a crapshoot as to whether you get color or mono. I suppose the easiest solution here is to get one of those vga extension cables and pluck pin 12 out of it and keep it together with that card.

edit: but don't you think they thought of this backward compat issue when creating the ddc standard?

jakethompson1 wrote on 2021-08-06, 16:53:

mkarcher wrote on 2021-08-06, 06:38:

Degaussing is usually connected directly to the AC line, so it's quite far away from the VGA signal in the monitor. Can you make sure that your problem is related to the monitor being connected to the computer? Keep your monitor and your victim computer at the same position as they are now, and connected to the same AC sockets as they are now, but do not connect the VGA cable. Then turn on your monitor while the 486 computer is running. Most modern monitors should degauss anyway, even if not connected to a computer, to display the "no video input" message on screen. If your PC still reboots, the problem is not caused by voltage/current spikes through the VGA cable, but by radiated emission from the degaussing coil or by conducted emission through the AC cables.

Alright, now confirmed the VGA cable need not be connected to video card to cause the motherboard to reset. The monitor doesn't have a "no video input" message so I did have to connect it to a laptop, but doing that and powering it on did generate a fresh POST beep from the 486 board nonetheless. So it's either physical proximity or something happening through AC lines as you mention. Might it be a quirk of the UM498 chipset?

I don't think it is related to your mainboard chipset. You get a strong low-frequency (50/60Hz) magnetic field from the monitor while degaussing, which might induce spurious voltages in the board, causing the reset, but I think this is unlikely. I'd rather think the primary electrolytics of your AT power supply are worn out, and the power supply stumbles over the reduced AC voltage. Remeber how I said 30-50A peak current during degauss? Assuming one ohm supply impedance, your supply voltage will drop by up to 50V for 10 to 20 milliseconds.

To test whether the problem is caused by induction: Try to increase the distance between the monitor and the board, without changing anything about the AC setup (don't use other outlets, don't use extension cords). If increasing the distance helps, induction is likely the cause. As a counter check (maybe do that one first), leave the monitor and board where it is right now, but use an extension cord to power the monitor from an outlet in a different room. If powering the monitor from a different outlet helps, the problem is most likely caused by the voltage dip in the AC during degaussing.

If you find out that induction is the cause of the spontaneous reboot, maybe some caps on the mainboard (e.g. on the reset line) are worn out or missing. Oh, and if you have a reset button connected: Try disconnecting the reset button. The wires connecting the reset button to the board might work as some kind of antenna and pick up the degaussing field. If you find out that the problem comes through the AC line, try a different PC supply (if you got one).

jakethompson1 wrote on 2021-08-06, 16:53:

mkarcher wrote on 2021-08-06, 06:38:

The grayscale effect might have a different cause: Originally, some pins in the 15-pin connector were used to identify the monitor type (the monitor should ground some, and leave the remaining ones unconnected). In particular, mono VGA monitors are supposed to ground 12, whereas color VGA monitors are supposed to ground pin 11. Pin 12 got repurposed for DDC later on.

edit: but don't you think they thought of this backward compat issue when creating the ddc standard?

They kind-of thought about it. Color monitors are supposed to ground pin 11, mono monitors don't ground pin 11. They didn't reclaim pin 11 for DDC, so even DDC-compatible color monitors can ground pin 11. So if the card uses pin 11 to detect the monitor type, it works with all kinds of monitors. The issue I suggest only is relevant for cards that use pin 12 for monitor type detection, which seem to be a small minority.

mkarcher wrote on 2021-08-06, 17:45:

I don't think it is related to your mainboard chipset. You get a strong low-frequency (50/60Hz) magnetic field from the monitor while degaussing, which might induce spurious voltages in the board, causing the reset, but I think this is unlikely. I'd rather think the primary electrolytics of your AT power supply are worn out, and the power supply stumbles over the reduced AC voltage. Remeber how I said 30-50A peak current during degauss? Assuming one ohm supply impedance, your supply voltage will drop by up to 50V for 10 to 20 milliseconds.

I'm using a relatively new ATX supply with a converter. And the reason I mention chipset is that it doesn't happen on any PCI boards, but happens with two different UM498 VLB boards and not on an SiS 471 VLB board.

mkarcher wrote on 2021-08-06, 17:45:

If you find out that induction is the cause of the spontaneous reboot, maybe some caps on the mainboard (e.g. on the reset line) are worn out or missing. Oh, and if you have a reset button connected: Try disconnecting the reset button. The wires connecting the reset button to the board might work as some kind of antenna and pick up the degaussing field. If you find out that the problem comes through the AC line, try a different PC supply (if you got one).

Don't have a reset button connected. I'll have to experiment with some of the other steps later.

Ok, so this is related to power connection. Degauss causes a momentary and intense power use surge on the AC voltage coming from the wall and depending on your house/room wiring and components used, it can cause a brownout and this is exactly what seems to be happening.

A different PSU can fix the problem, and so can plugging the monitor to a different outlet.

Jo22 wrote on 2021-08-06, 16:10:

Thank you very much for the explanation! 🙂 […]
Show full quote

mkarcher wrote on 2021-08-06, 09:55:

In same places of the world (Germany for example), it was common to have neutral (the non-hot power wire) and protective earth (the ground wire that should be connected to the case) on the same conductor in the electrical installation, and have protective earth to ground bridged in the AC socket. In this kind of installation, current spikes on AC (like degaussing, which can cause momentary currents up to 30A in 250V countries, I expect it to be even higher in 120V countries) can pull the combined protective earth/neutral a considerable amount away from actual earth potential.

Thank you very much for the explanation! 🙂

Yes, I think that might be true.
As far as I remember, neutral and protective earth come together in the fuse box at some point also.

The outlets have three different pins (two holes, 2 outer contacts wired together).
One hole has phase (brown), the other one has neutral (blue) and the outer contacts are protective earth (yellow+green).

Yeah, thats for newer installations. The idea is to have a low-impedance ground path connected to the fuse box, so the current in the neutral doesn't shift ground significantly at that point. And starting at the fuse box, protective earth is a dedicated wire that only ever carries current in failure condition, so it stays very near to true ground, even if the return current causes some voltage over the resistance of the neutral wire in your house installation.

Older installations didn't have any green/yellow wire, but connected the brown wire to the clamps for the outer contacts in the socket.

Jo22 wrote on 2021-08-06, 16:10:

So if a fully wired plug is used (the bulky schuko type) instead of that flat, 2pin euro connector, the grounding can be done via protective earth.

Actually, according to current code, the grounding must be done via protective earth. If your device uses a 2-pin connector, you need a kind of failure-resistant isolation between mains and any metal parts you could touch - i.e. those parts are floating. In no way it is legal to ground devices using the neutral (brown) wire, even if it is a device that is permanently connected to the line, without a plug you could plug in reversed.

Jo22 wrote on 2021-08-06, 16:10:

Back in the 1960s or so, TVs had hot-chassis still due to missing transformers and the use of the old 2pin plugs that lacked protective earth.

Oh, the days of the P-series tubes/valves with constant current heating... I didn't live back then, was the antenna socket live in those sets, or did they use a HF transformer to couple the antenna into the set?

Jo22 wrote on 2021-08-06, 16:10:

To my understanding, switching PSUs nowadays have a pseudo galvanic insulation.
They have a 100khz miniature transformer (ferrite ring) that does the insulation.
One side is connected to the control circuit, the other to the output?

Yeah, the thing about the 100kHz mini transformer is correct. But it still provides full galvanic isolation, just as the big 50/60Hz transformers.