This post is a bit out of the ordinary as the subject is more on network tech issues (albeit old tech) rather than my usual woodworking or renovation topics. I’m going to describe the work I did to extend the life of my main network switch by talking about my Nortel Baystack 5520 fan rescue.
As some of you may remember, when I did the renovation of our main bathroom, I put in a wireless access point. If not, and you’re interested, you can see the post here. I had the wall open so it was a good time to run some cat-5 cable up and get some additional Wi-Fi coverage for the upstairs.
Some background on the setup
I use Ubiquiti Unifi AC Pro access points, which are PoE, so I need a switch that supports that. In typical overkill fashion, I picked up a pair of old Nortel Baystack 5520-48T-PWR switches on ebay for a song and use them in a hot/cold configuration as my main switches. As Nortel came apart, Avaya picked up the switches and manufactured them under the ERS 5500 designation.
Fans are the Weak Spot
The 5520s are pretty bomb proof in general but do have one specific weakness. In order to deal with the heat generated by supporting PoE over 48 ports, they can produce a lot of heat. As a result, they have a row of 6 40mm x 40mm x 20mm fans running along one side to draw air through.
These fans move a lot of air, which is not a problem but is noisy. The issue is that given the switches are over a decade old, the fans are well past their average failure date. As a result, it is not uncommon for them to just fail and stop spinning. Sometimes you can coax them back into life temporarily but, that is temporary at best and you’ll find they simply grind to a halt later. That’s where mine was at. In the period of about a week, I dropped from having 6 functioning fans to only 2. The switch never hiccuped but, with heat and electronics, it was only a matter of time. Something needed to be done.
My first stop was to look for some replacements and there are good options. Unfortunately, a set of 6 fans new would run me around $150. That was more than I was willing to put into my $25 ebay finds. I though maybe I could do something to refurb the existing fans. Maybe get them running for a while longer until I decide what to do.
Lubricating the Fans
The fans are a mechanical device so it made sense to me that maybe I could just lubricate them to get them running smoothly again. A bit of a long shot but total cost for the experiment: about an hour of my time.
First job is to pull the case open. Takes a total of 20 screws to get the cover off. A bit of a pain but relatively easy to do – no fancy fasteners, just good old Phillips head screws. Once it is off, you have full access to the fans.
Two more screws release each fan.
The fans are pretty self contained. However, if you peel the sticker off the center, you can get access to bearing for the fan.
I gave a little spritz with pot control cleaner and lubricant followed by a spin of the fan. I then fired it up to see what happened. To my surprise and pleasure, my previously unresponsive fan immediately spun up to speed and continued on happily. I gave it a little more lubricant to be safe and then moved on to do the rest of the fans. In the end, I had 6 of 6 spinning along happily again.
Modifying the Case
I’ve left things running for a couple of hours to make sure that they don’t just grind to a halt again. I’m not so naive to think this has fixed my problems. These are my main switches so it is rather inconvenient to have to disassemble just to refresh the lubrication. My solution to this is to make a modification to the case to allow me to re-lubricate the fans without disassembling things.
First use a punch to set the center of each fan cover.
Now drill a hole large enough to expose the bearing section of the underlying fan.
I know that the holes are pretty rough but, after grinding the edges smooth, they serve the purpose.
Finally re-install the cover on the now revived switch.
Final Thoughts
Before anyone gets upset that I’ve exposed the bearings, or am using the wrong lubricant, or such, keep a few things in mind:
- The switch is over a decade old so I don’t feel like I’m reducing some future resale with my modifications.
- It only cost me $25 in the first place so refurbing for over $100 is not very palatable.
- This is not a production environment, just a lab. If things fail out, it isn’t the end of the world.
In the end, it provides me with a simple way to get a bit more life out of my switches. I’ll have to replace them but, in the meantime, I can keep that money in my pocket. There is also a certain sense of satisfaction in having got things back in working order. Whether that last is another question but I’ll post an update later to document the longevity (or lack of) for the project.
Thanks for reading!
