Question Will changing components with power supply on cause damage?

dsc106

Senior member
May 31, 2012
320
10
81
Computer off, but power supply still plugged in and on with motherboard LED light on. How bad is changing a PCIe card, GPU, or RAM module in this state?

I know this is not best practice, that’s not the question. Question is, if done accidentally, how likely is it to actually damage something? And would the damage be instant all or nothing damage, or a more invisible works now, but damage to component lifespan or performance over time?

High context answers welcome - the question here is both academic in nature, but also practical, unfortunately.
 

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
Do the components still work? If so, count yourself lucky and don't do it again as you can damage/kill something by doing that.

It takes two seconds to disconnect the power cord from the power supply.
 

mindless1

Diamond Member
Aug 11, 2001
8,031
1,440
126
In theory it is possible to damage something from having a moment where there is positive power differential and no ground so current backflows. On many connectors (better designs) they take this into account so the ground makes connection before the rest but understandably this is not as much the case with components never intended to be hot plugged.

Immediate damage is more likely but certain things like transistors, whether individually packaged or integrated into an IC chip, can be stressed and fail later.

The odds of that are low enough that I wouldn't worry too much about it, just test whatever you risked.
 
  • Like
Reactions: dsc106

MalVeauX

Senior member
Dec 19, 2008
653
176
116
Heya,

If the system is powered off, but the supply is still powered (ie, it's switch is not off, and is still plugged in), it's not probable that you will damage any components. It's possible, but not probable. It's not ideal practice of course to do this unless the component you're using has the ability to toggle power with specific switches (think hot swappable hard drive systems for example; but this is not just straight connections, there are more components involved between the power source and the actual component). So, why is it possible to damage a component? Well, what's to guarantee that all the switches, relays, etc, are actually not powered when you pull a component? Also, what's to say that when you do that, only partial connection is made and so it doesn't have +/-/ground/etc and you're just feeding it a positive signal? This is why things end up damaged, from thermal load as the electricity (current) has to be able to pass and leave and when it cannot do that, it doesn't just sit there. You won't know something isn't working properly until you pull a component or add one and it gets cooked. And the PSU itself can fry the whole system, everything connected to it.

So best practice is to disable the power switch to the PSU itself on the back before doing anything inside that machine with any components. If you want to ensure that cannot fail, unplug it completely (with the switch off). Do your work in there. Then plug it back in, and switch back on the PSU. Then power on the system after that.

Very best,
 
  • Like
Reactions: dsc106

dsc106

Senior member
May 31, 2012
320
10
81
Thank you for these very detailed answers. I’ve been a bit stressed because I was troubleshooting my new premium build (specs in sig) - it wouldn’t boot, and I was swapping components. Turns out I suspect the issue is that I need a Ryzen 3000 chip to update my boards bios.

In all my years I’ve never swapped ram and GPU with standby power to motherboard but it was late and there was a lot of back and forth. Total lapse in the moment.

When we say possible but not probable, do we mean like... a 1% chance? 10% chance? 0.1%?

As it is, I cannot test the components until I get the 3000 chip here... so I have to sweat this for a week! One question I have in the meantime is, assuming this all works and boots later (currently all fans and lights still spin up - I just can’t get into bios, or run any mem tests etc) - these are all brand new, expensive premium parts. I’m concerned I potentially shortened their life by stressing them?

If it were you, would you swap them out with new replacement parts as available (I am well within return policies), or just use them as is in the rig if they pass all tests? Should I worry much about some invisible damage via stress on parts? There were no sparks, no weird smells, or anything else of that nature. Also it’s a high quality PSU and motherboard, so I assume some level of protections that may help buffer this and idiot proof things a bit?

Finally, I am also worried that other parts may have had stress applied - IE if swapping the RAM could somehow send voltage spike back to CPU or nvme SSD or something like that.

Thank you very much for taking the time and energy here to help me both understand the science of this (much appreciated!) and advise on things. I’ve been sweating bullets here as I moved very slow and methodically through the whole build, and had things perfect after being very careful... only to get surprised with code 0d and no boot (likely due to new 5000 series)... and pulled this dummy move!
 

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
When we say possible but not probable, do we mean like... a 1% chance? 10% chance? 0.1%?
Nobody can tell you that. It will either work, or it won't.

It's pretty much the same advice you've already been told on Reddit.

If it were you, would you swap them out with new replacement parts as available (I am well within return policies), or just use them as is in the rig if they pass all tests?
Manufacturer's warranties don't cover misuse by users. Returning components you damaged through a PEBKAC moment would basically be committing fraud, as I doubt you would disclose removing components while it was still connected to power.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,229
9,990
126
As it is, I cannot test the components until I get the 3000 chip here... so I have to sweat this for a week! One question I have in the meantime is, assuming this all works and boots later (currently all fans and lights still spin up - I just can’t get into bios, or run any mem tests etc) - these are all brand new, expensive premium parts. I’m concerned I potentially shortened their life by stressing them?

On one of my boards in the past, when I didn't have a BIOS flashed that would accept that CPU, the board was pretty-much dead-dead. No diag lights, even. Totally dead.

I'm actually a little more concerned for you, that if you're getting fans and lights, that you may have damaged a component, and it's not just a BIOS issue. Possibly.
 

dsc106

Senior member
May 31, 2012
320
10
81
I'm actually a little more concerned for you, that if you're getting fans and lights, that you may have damaged a component, and it's not just a BIOS issue. Possibly.

Well, I was getting this in the first place after meticulously assembling the build, which is what sparked the troubleshooting. So if that's the case, the components came defective. After a number of part swaps, I got mixed up on one go with the PSU standby power still kicking to the motherboard, but there were no noises, sparks, smells, etc. I just realized after changing a stick that the mobo LED was still on. But all error codes and other operations remained consistent before and after.
 

mindless1

Diamond Member
Aug 11, 2001
8,031
1,440
126
I wouldn't worry too much about shortening their life, I mean odds are low you did damage, especially with no smell of smoke or sparks or burn marks, and odds are lower still that the damage takes a while to surface, but there is still a chance you had a problem from the get-go, not just needing a certain CPU.

I'm suggesting that if you get the CPU you think you need, and it still doesn't work, odds are some part came to you defective, but if it works but then dies within hours to a few days, I'd take blame for it and eat the loss. If it took weeks to months to fail, odds are higher that it's not your fault, reasonable to assume the part was defective and seek warranty replacement.

It's more or less just a waiting game. No need to return any parts that still work, and if something you didn't install with live power is dead, I'd assume it came that way.
 
  • Like
Reactions: dsc106

dsc106

Senior member
May 31, 2012
320
10
81
Gotcha, thank you mindless. That all makes sense and is very helpful. Also good to hear that I would not experience a shortened lifespan problem 2+ years down the road due to this (my main concern if all boots up and tests out) and that any problem, if any, should be more immediately diagnosable in the low chance that something went awry here.

The code I received was code 0d which is "reserved for future codes". The best I can make of my MSI motherboard serial number seems to indicate a manufacture date early enough that it may well not have the BIOS required to support 5000 series, so I will troubleshoot with the chip swap first and then go from there.

I called the local dept. store and told them about the issue, and the employee there seemed unphased and that accidently leaving the PSU on during a hardware change was something that was pretty common. He said of course you want to avoid it, but that he's only ever seen a problem occur one time and that it was with visible sparks and a pop. So, that was reassuring.

I can't imagine that with millions of PC builders and many users far less cautious than I, that this sort of thing isn't happening all the time?
 

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
I can't imagine that with millions of PC builders and many users far less cautious than I, that this sort of thing isn't happening all the time?
Maybe if a person's first build and they haven't taken the time to watch a build video on YouTube. Electricity 101: Before working on anything when it comes to electricity, you disconnect from the power first.

It even warns people of doing this on the first page of just about every motherboard manual out there. From my Asus manual:

4.jpg

I guess if anything, you should remember this rule before installing/removing any more components going forward.
 

dsc106

Senior member
May 31, 2012
320
10
81
Geez, UsandThem, did you read my posts? From the very beginning I stated I was well aware this is a problem and thus concerned. I’ve built over 50 rigs, accidents happen. What’s with the antagonist attitude?
 

VirtualLarry

No Lifer
Aug 25, 2001
56,229
9,990
126
Geez, UsandThem, did you read my posts? From the very beginning I stated I was well aware this is a problem and thus concerned. I’ve built over 50 rigs, accidents happen. What’s with the antagonist attitude?
I think that he's just trying to make a strong point, for those users that might come across this thread from a Google search, and decide to think it's "not so bad" to plug/un-plug RAM or PCI-E devices with the power plugged in and switched on in back.
 

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
Geez, UsandThem, did you read my posts? From the very beginning I stated I was well aware this is a problem and thus concerned. I’ve built over 50 rigs, accidents happen. What’s with the antagonist attitude?
No attitude. Just a difference in opinion.

Based on the comment I quoted, you seem to believe to be a common occurrence with builders, and I disagreed with that.
 

dsc106

Senior member
May 31, 2012
320
10
81
Thank you for the clarification. First time in 20 years I’ve ever done such, hence the anxiety around it. To any future googlers, clearly you do not want to deal with the stress of wondering even if it works out fine.

As far as my belief, no, I do not know if it is a common occurrence. That is why I asked it with a question. It is however what the person at the dept store, who does this day in and day out, said from his personal experience. That is a singular data point, and I would be curious for academic reasons to know what others have observed.

It is my plan to hopefully proceed another 20 years without such a brain fart moment.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,229
9,990
126
Look, if you didn't see sparks, or magic smoke escape, you're PROBABLY OK.

I've done it too, and I was kicking myself afterwards. It happens. Just don't do it "regularly", because it can cause fried components, and RAM these days IS fairly sensitive. PCI-E devices less-so, but it could still damage them.
 

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
It is my plan to hopefully proceed another 20 years without such a brain fart moment.
We all have our own personal brain fart moments, but I have built/fixed so many PCs over 25 years, I have a system/method in place in order to prevent mistakes.

Most of brain fart moments come via forgetting to do something on my wife's "honey do" when I told her I would have it finished. :p
 

dsc106

Senior member
May 31, 2012
320
10
81
Thanks, all. Another academic question: does the SCP (Short Circuit Protection) feature on the Corsair AX1600i come into play at all in the scenario described here?
 

mindless1

Diamond Member
Aug 11, 2001
8,031
1,440
126
Not necessarily, connecting electricity to the wrong pins or not having ground first so voltage potential is higher where it shouldn't be, can cause a problem before overcurrent protection would kick in. Corsair calling it SCP is just trying to market a feature that any decent PSU has.
 

dsc106

Senior member
May 31, 2012
320
10
81
An update for anyone interested. After using a Ryzen 3000 series chip to update the BIOS and swapping back to the 5950x, everything went on flawlessly. No detectable damage seems to have occurred. I ran memory test for 14 hours, zero errors, prime 95 and other benchmark software is stable. System in normal use thus far looks all stable with no errors reported.