• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

Hot Booting CC

raoulduke

Well-known member
I found the suggestion to turn the machine on with logic board removed, then hot plug the board to turn it on - not bc of a missing keyboard, just if it failed to soft boot normally.  I can't find the original suggestion (this was months ago).

My board has a SCSI problem.  I think (/am pretty confident) it's trace rot from caps.  But I am curious if there are thoughts on the safety of this hot booting method.  I didn't think about it at the time, but in theory so long as it's stable, you aren't unplugging, you're just applying power.  I don't think it'd spike, though it might decrease slightly under the load - then maybe spike?

I'm not an expert (clearly).  Any thoughts?

[Also, I think the need to hot boot is basically due to dirty contacts anyway.]

 
Last edited by a moderator:

raoulduke

Well-known member
Byrd: why?

Uniserver:  it will do something; it will accomplish the goal of turning it on.  the question is whether that hot plugging can damage the board?  (so far i have not seen evidence that it can; the ADB issue is interesting though, and was actually the genesis of my question.)

 

CelGen

Well-known member
Byrd: why?
Putting words in his mouth, hot plugging entire logic boards is typically an EXTREMELY bad thing to do. The often total lack of interfece buffering, grounding, ESD or surge protection means you'll blow up anything that interfaces along that connector.

 

Paralel

Well-known member
CelGen is correct. Never, ever hotplug a power supply into a LB unless you want to destroy things.

 

raoulduke

Well-known member
Never give a mouse a cookie... I still don't understand why, other than for the reason I originally gave.  CelGen offered an explanation in PM that I'd suggested he put here because I don't understand these explanations.  But that explanation was essentially an analogy to pulling RAM from a system that's on.  I understand why pulling the board is not a good idea.  I can't find a way to extend that logic to putting power to a board.

 

techknight

Well-known member
Unless you have a lab full of equipment to observe the failure analysis aspect of it, youll have to take everyones word for it... 

Any hot-swappable system has pins that are varying in lengths. Why? because there is a certain order sequence in which power and signals have to be applied and broken. 

take any hotswappable sata, or USB, ground pins always extend out the longest, then next in line is power pins, in the correct order, then the data pins last. 

These lengths are there for a reason. To allow hotswapping by safely completing circuits in the correct order, during insertion. 

these logic boards, have NONE of that... 

Lets say the ground doesnt connect, but the 12V and 5V do connect. the 12V may loop itself through various circuits to make its way to ground, and that ground is 5V. 

What you say? 5V isnt ground, thats ludicrous! Ahh.. In the case of a missing 0V reference, 5V is ground to 12V. it is at a lower potential, so youll have the potential difference in voltage of those two, and currents flowing through areas in which it wasnt designed to. 

Hence why ground has the longest spade in hot swap systems. 

Now granted the connections are instantaneous, so the spike would so short it wouldnt affect anything. However, the circuit connections on insertion is completely random. All it takes is the wrong order, or the wrong spikes in the wrong order, and poof!

 
Last edited by a moderator:

raoulduke

Well-known member
[Thanks!  I think our definitions of taking one's word for it may differ, though, because that's precisely what I was looking for.  I still don't think that's what did the SCSI in on mine, though.]

 
Last edited by a moderator:

raoulduke

Well-known member
Well, and I don't want to burden you for your kindness in responding, but in theory if either hot plugging or unplugging the board could blow out the SCSI, what would it blow out?

 

raoulduke

Well-known member
[You already told me to change the caps on this CC... it already hasn't worked, sadly.  / Do you know off-hand which cap(s) might do it?]

 
Last edited by a moderator:

uniserver

Well-known member
ah well if you already re-capped the board, and then washed it good.... then maybe you fried the scsi chip by hot plugging/booting it...

i have a lc575 on the bench with dead scsi chip because the doofus kept hot plugging it with dead caps to make the Mainboard come on

instead of the EGRET safely turning on the machine with the main board properly mounted.

so i changed the caps and it now powers up properly. and everything seems to work, except for the scsi does not work internal or external.

next up is to change the scsi chip... the only problem is i don't think i have replacement on any of my parts boards.

 

Paralel

Well-known member
If the caps were changed already, unless you don't have any confidence in the person who did it, it seems rather unlikely to be the problem.

Also it may not even be caps. check the traces, etc... it could be one of any number of things including a dead SCSI chip.

 
Last edited by a moderator:

techknight

Well-known member
speaking of which I got a braindead LC575 board I never did figure out. the 040 clock IC was corroded to hell, but no traces broken. replaced it anyway. 

its possible it was a victim of a hotswap, no idea. 

 
Top