You have a poltergeist living in the 5152 !

I have a PC that often wakes up sleep mode when I turn another device nearby. It is a modern computer and I use cheap USB-to ps/2 converter so I can use an old AT keyboard with it. Likely the USB-to ps/2 converter is lacking proper shielding so whenever a current is induced to it from the outside it signals that as key being pressed on the keyboard to the computer.

Are the printers shared with the other PC via Windows?

Electro magnetic interference in the USB, probably. Maybe. Dunno.

Any description of a bizarre thing is a bit hard work to read. The USB printer jumps into life when the old machine is switched on?

What happens if you just disconnect and reconnect the USB on the printer? Does it jump into life in the same way?

I have two computers here that suffer from USB keyboards vanishing, almost certainly due to EMI, Linux says so sometimes.

I use a thing called inputexec on a couple of Linux machines to make a keyboard into a one-press for an action, a monitoring computer and the kitchen music player.

The monitoring computer is near some poor shielded transformers (in plastic, not metal), and if they switch on/off then a USB keyboard can have a disconnect/reconnect event. That makes inputexec exit. Investigating that problem led me to USB/EMI problems. I bodged a solution by putting inputexec in an infinite loop with a sleep (some readers will have had their cringe gland implode seeing that, sorry).

The kernel messages also sometimes suggest EMI is the problem.

The kitchen music player can have a keyboard disconnect/reconnect when I run a new washing machine, probably because of the motor or a 2kW heater turning on and off.

Moving the transformers further from the monitoring machine helped, as maybe did buying a shielded USB cable. The kitchen computer I just ctrl-alt-del reboot it if the keyboard stops working with inputexec.

Could be some kind of jolt down the power, could be LAN, but I think induced power in the USB from an old computer. Ahhhh, got an old CRT on that? Big, inefficient electromagnets!

Some form of sheath current, maybe? 🤷‍♂️
https://en.wikipedia.org/wiki/Sheath_current

Curious little "problem"

If they are both booted without any network connected does the same thing still happen?

Maybe some kind of wake on lan behavior?

Are the printers shared on the network through any of the systems?

Run Wireshark and capture some frames when it occurs to rule the network out. Look for broadcast packets.

keenmaster486 wrote on 2023-12-05, 16:46:

- all three are wired via into my home network.

Maybe the problem is the connection via .

I suspect an AC related issue. Maybe some sort of ground loop, a power peak/current peak.

In many cases, cheap switching-psus are a culprit.
The invention of the switching-psu caused more trouble than it's worth (imho)..

Edit: If it's a ground loop issue, turning the power plug may help.

Network cards (Ethernet, anyway) are isolated on each end using transformers. Often referred to as "magnetics" in network card parlance. This is done to prevent ground loops, etc.

Jasin Natael wrote on 2023-12-05, 17:22:

Are the printers shared with the other PC via Windows?

Yah I'd suspect windows restarting the print service when it detects printer or something doing it.

kingcake wrote on 2023-12-05, 18:57:

Network cards (Ethernet, anyway) are isolated on each end using transformers. Often referred to as "magnetics" in network card parlance. This is done to prevent ground loops, etc.

Yes. Power supplies, too, if they are the good linear type with a transformer (and are not tying both grounds together).

Unfortunately, switching-psus are often badly made. Just look what's inside USB wall psus. They're often terrible designed and righteous dangerous.
Sometimes, there's a 60Hz/50Hz signal reaching the DC side, for example.

Edit: I've checked the inter webs, the MX-80 has a traditional PSU. So far so good.
*If* it's the culprit for the malfunction, a new set of capacitors may help to provide cleaner DC power.

Edit: It's not just the electrolyte caps, but the filter caps on the AC side.
Historical, these filtering caps often fail after a longer period of time.

https://mansfield-devine.com/speculatrix/2018 … -friend-part-1/

But maybe the problem is the other printer, also.
It might be possible that RF noise emitted by its PSU sneaks into the parallel cable somehow.
Depending on its lenght and physical location, it might act as an aerial for all sorts of RFI.
Adding ferrites to the cable may help.

Sounds like it's soon to be a Maximum Overdrive situation.

You know what to do.
https://www.youtube.com/watch?v=N9wsjroVlu8

BitWrangler wrote on 2023-12-05, 18:58:

Jasin Natael wrote on 2023-12-05, 17:22:

Are the printers shared with the other PC via Windows?

Yah I'd suspect windows restarting the print service when it detects printer or something doing it.

Yep, pretty much what I was getting at.

^@DosFreak Thanks, YMMD! 😁

Jasin Natael wrote on 2023-12-05, 21:33:

BitWrangler wrote on 2023-12-05, 18:58:

Jasin Natael wrote on 2023-12-05, 17:22:

Are the printers shared with the other PC via Windows?

Yah I'd suspect windows restarting the print service when it detects printer or something doing it.

Yep, pretty much what I was getting at.

What does the LED on the ethernet card inside the 5150 say?
If there's an imminent activity happening in same moment, it should flash or change blinking pattern?

PS: An oscillograph (scope) would really be helpful here. It could be used to visualize the data stream on the ethernet cable, as well be helping to check the AC line*.

Edit: A so called "logic analyzer" would be helpful for the checking data stream, too. Some modern scopes have it built-in.

*Probing the AC line is dangerous and not recommended for the novice.
It can help to check mains quality, though. If there's any RF noise on the power line.

The Deskjet is connected via USB to my primary modern computer and via parallel to a 386 machine

1 host | 1 printer ..... I could be wrong, but I think that's non-standard hookup (unless of course the deskjet manual says you can hookup 2 hosts to the 1 printer)?

I forgot to mention that this happens whether the computers are on or not - although I haven't tried it with the 386 on. I suspect if I turned it on, it would actually stop happening.

It's definitively going through the AC power lines. I've verified that.

Yes, probably bad filter caps somewhere or something (perhaps in the 386's PSU), it's just bizarre that it causes the Deskjet to think it is receiving a new print job.

Is it possible you're having very quick brownouts? If the voltage sags below a certain point the switching power supply will restart, or the watchdog will reset the device.