Hacker News Reader

Comment by schlowmo | Hacker News Reader

This is no powerplant-level story, but sometimes there are high startup currents were you didn't expect them:

On one of my first IT jobs at a big manufacturing company my team was tasked to find out why there are regular power outages in some printer rooms (there were rooms with shared printers on each floor of the office building). There were always some tripped circuit breakers and the facility management had to dispatch someone to put them back on. Between those incidents were always some weeks were nothing happened, but when it happened it affected a lot of printer rooms.

In the end we found a monthly cronjob on a central printing server which triggered a testpage print on all connected printers. Took us quite some time since no one ever saw those test pages. Never underestimate the needed current for a room full of colour laser printers coming to live all at once.

View on HN

Laser printers use a lot of power for a few seconds when coming out of standby.

Part of the printing process involves passing the paper covered in toner through a hot roller, to fuse (melt) the powder toner (ink) onto the page. That roller has to be up to temperature to print. It is normally heated by a powerful (ie. 1 kilowatt) light bulb inside a hollow roller. Sometimes if you peek through the vents in the printer, you can actually see the light it makes.

The light bulb is pulsed on and off to maintain the right temperature - but when coming out of standby it is solidly on for ~10 seconds. Manufacturers want their printers to warm up from standby quickly, so they put very powerful heaters in them, even though the steady state heat requirement isn't awfully much while printing.

Id THAT what that light is?! It always bothered in the back of my mind. It didn’t look very laser-y to me. Especially since the laser was supposed to be infrared.

This is why you never put a laserprinter on a UPS.

What will happen to the UPS? A lot of discharge cycles?

I think the idea is that you have other stuff on the UPS too, and you'd prefer that stuff stay powered during an outage, but if your printer is on the UPS, it'll chew up all the power.

It's more that consumer-grade UPSes will all trip due to overload as soon as you hit print.

Wait. Why? My laser printer is on a UPS. What is the problem I am not aware of?

Mine will always trip the alarm. Voltage draw too high.

Surely it’s current draw? Wouldn’t the printer always operate at line voltage?

Depending on your supply (for instance in an old house), the printer current consumption can cause a voltage drop on nearby circuits, and this voltage drop can in turn trigger the UPS.

Yes. The voltage alarm in this case would be the voltage dipping too low (voltage sag) due to high current.

It would becomes a power event on the UPS.

Not only you should not plug it in a UPS (the initial surge when a laser printer is on would likely surpass the power limit of your UPS too), a laser printer and an UPS should not be under the same circuit breaker. I know because due to a limitation to where I lived before, I have to put both under the same circuit breaker. Every time I print something, it is a power event (where I can hear that the UPS temporarily switched to battery powered mode.)

My UPS has a plug for the printer. It does only surge protection. This is convenient to have the full IT equipment of the office protected with just one box.

Does your UPS have a plug for laser printers, or just for printers?

It might be assuming an inkjet, which have much lower power draws: an inkjet will draw maybe 50W. A laser printer are generally in the mid-hundreds at least and some models can get into the high hundreds to low kW, you need a pretty heavy UPS to handle that and still power other systems.

For laser printer, it is specified on the plug. It protects only against power surge coming from the utility provider. It does not provide power to the printer, this is just a protection if your house is hit by lightning.

Laser printer is known to cause a surge event too. Normally there's a limited amount of surge protection available and routine surge would gradually used up the pool.

However, surge protection may not be as important. I heard an argument that it is basically snake oil (because an actual surge would normally be much more powerful and beyond a typical capacity they have, and some causes fire.)

Running a printer off a 2000W gas-powered generator, it always gives me a little chuckle to hit print and hear the engine rev.

In our office we were so over the rated current for the building that after the breaker tripped (which it inevitably did) it wasn't possible to just switch it back on. The moment you did all the servers and PCs went back on and it tripped again. You'd have to go around and pull out plugs all over the office then switch the breaker, then switch them all back on one by one. We also had regular electrical fires. Those were the good old days.

Having enough inrush current to trip a breaker doesn't seem so terrible or entirely unexpected, but electrical fires? That's certainly bad news, the breakers are supposed to be sufficient to protect the wiring.

> Having enough inrush current to trip a breaker doesn't seem so terrible or entirely unexpected

It should still not happen, since breakers are supposed to deal with inrush currents. A quick look at a random circuit breaker manufacturer page tells me that this particular breaker model is meant to instantly trip once the current is 3 to 5 times larger than the nominal current; less than that, it should take several seconds to trip, giving enough time for the inrush current to cease. So either the breaker (and the wiring) is underdimensioned, or the device is using too much power.

(IIRC, the trick is that most breakers have two independent trip mechanisms: a thermal one which has a built-in heat-dependent delay, and a magnetic one which is instant.)

> (IIRC, the trick is that most breakers have two independent trip mechanisms: a thermal one which has a built-in heat-dependent delay, and a magnetic one which is instant.)

precisely it. the magnetic one uses the fact that ac makes an electromagnet and they tune it so it snaps away quickly if there's a short.

Right isn't that a time delay fuse is used for?

Right, I don't think inrush should pop a breaker, and I'd replace that breaker to see if it's just got an over-sensitive magnetic trip (IME they get more and more sensitive the more they pop). But it's a safe failure, at least. If the thermal breaker isn't working, that's dangerous.

Most server BIOSes have an option for a random delay on power loss recovery to prevent this exact scenario.

True, but most PC BIOSes don't, and PCs can outnumber servers 100 to 1, unless you're talking about a datacenter.

A single black and white laser printer made the lights in my dorm room dim for a fraction of a second. So yeah, that thing must've pulled quite a few amps.

Also: don't put a laser printer in your bedroom. It's unhealthy. Only learned about that after the fact.

> Also: don't put a laser printer in your bedroom. It's unhealthy.

3D printers are even worse, depending on the filament type (ABS is worst?). Always ventilate!

A few papers printed over the course of years won't kill you. What will are the conditions of working adjacent to the office copier, 8-10 hours a day, for years.

Get an air purifier to capture particulates. (Supposedly, houseplants help too.)

That's why i keep my 3D printer next to my toilet these days. The joys of living in a small apartment.

I’ve heard of people hacking system startup procedures so 15 hard drives didn’t try to spin up at the same time.

This was a feature on SCSI disk controllers. I remember one controller that had dip switches to set the spin up sequence number, and then you would configure the controller to wait for all the drives to be spinning before it tried to bring the array online.

I'm going from memory here but each Ultra 320 SCSI HDD had a startup current of almost 2 Amps so if you had a disk shelf with 24 drives and stack a few shelves in each rack you could do some serious power damage if you didn't plan the startup sequence right.

Staggered spinup is still a feature on virtually all modern hardware/RAID disk controllers.

On a per-machine basis, many server motherboards have out-of-the-box BIOS support for this feature. At least they used to. It's been a long time since I've built a server and mechanical hard drives are less common than they used to be.

I had a home hacked 1TB+ server, using 5.25in 23GB 8lb monster drives salvaged from a long life as a TV video bank (long, long ago, when dinosaurs walked the earth). There was 60+ actual spindles as i recall.

The drive array was powered by 6x, 400W ATX server supplies with my own wiring harness. This was enough to keep them running but they had to be sequenced carefully to keep from overdrawing the power supplies.

This was all on an UltraSPARC 6k so there was plenty of support for that; bringing up the system always sounded like multiple jet takeoffs tho. Took 15min. When the rack of 10k RPM "quick cache" disks spun up it was like a chorus of the whines of the damned.

I had a MicroVAX in my bedroom which booted with a tick-tack-tick-tack going faster and faster culminating in a crescendo where it sounded like the discs synced up or something.

I'd then login to a prompt and type DIR before I turned it off again. I just pulled the power switch, I had no idea how to do a proper shutdown.

I spooled up my SIMH vax just to find out... the proper command is

  @SYS$SYSTEM:SHUTDOWN

There was some home computer - either an original Apple II or a Commodore PET, I don't remember - where if you splurged for the fancy second disk drive, the computer could be destroyed by a rogue program spinning up both drives at once. And since every program ran at the same protection level as the OS (because there were none), it was either two MOVs or two POKEs to the hardware registers to make it happen.

Reply on news.ycombinator.com