Nortel 5520-48T-PWR noise reduction -18dB

Let op: Tweakers stopt per 2023 met Tweakblogs. In dit artikel leggen we uit waarom we hiervoor hebben gekozen.

By ArnieNFW on woensdag 28 november 2018 20:14 - Comments (14)
Category: Divers, Views: 3.598

Background

Years ago I bought many Nortel 5520 switches for work. When I saw these switches being sold on a Dutch Auction site, I couldn't resist and bought one.
Not a very bright idea since I have a small hobby room and the 5520 is a enterprise datacenter switch meaning: it has a high power usage and creates an awful lot of noise. And of course I don't have real use for it.

The high power usage I can't solve; I should have bought the 24T version, which was available. But the 2 rows of 24 RJ45 connectors in a 1U form are just too beautiful. Here it is:

The noise problem can be solved. I went from 53dB to 35dB (as my app says: a quiet library). Unbelievable! Below the steps you have to take in order to get almost 20dB reduction.

A picture of the internals; quite impressive.

The start

We start with 53dB.

Old and new fan details

The 5520 switch uses Delta Electronics fans:

DELTA ELECTRONICS model EFB0412VHD
40x40x20mm
9000rpm
32.5dBA
9.25cfm (airflow)
12V, 2.16W

Since I will be using the switch only for experimenting and not for 24/7 use, I don't want to spent a lot of money on it. After some research I ended up buying SUNON fans, based on price and noise level. Specs:

SUNON model KDE1204PKV3
40x40x20mm
3000rpm
15.5dBA
8.9cfm (airflow)
12V, 0.4W

Price was excellent at Aliexpress, €6.12 for 2 fans.
https://nl.aliexpress.com...r-2-stks/32841514403.html
(no hidden sponsoring, I paid for them myself)

A picture of the old and new fans:

The first fan mounted. I rotated it 90 degrees, to keep the wires on the same place.

Connector differences old versus new fan

The connectors are the same model, but sadly the pin wiring is not the same:

(old up, new down)

Luckily it is easy to change. Gently push the metal of the wire inside the connector (see picture) and gently pull the wire at the same time. It's easier than it sounds.

A detail of the wire connection:

Push the wire back in the right place, but exactly the same orientation it came out. You feel a little "click".

The actual mounting of the fans is easy. 2 screws, that's it. Here is a picture of all new fans mounted:

And a detail of the wires:

The result

And this is the result. Perfect for me!

06-'19 Spatbord clip - plastic (KOGA / Gazelle) - 3D Print
01-'16 Windows 10 Store eindelijk werkend

Comments

By woekele, donderdag 29 november 2018 06:31

By vanaalten, donderdag 29 november 2018 09:58

When I started reading, I thought "more quiet fan, so probably far less airflow, so less heat removal"...

But, 9.25 vs 8.9cfm airflow is a very minor difference - if it manages to move this amount of air, cooling is roughly the same so indeed a good improvement.

By ArnieNFW, donderdag 29 november 2018 10:16

vanaalten wrote on Thursday 29 November 2018 @ 09:58:
When I started reading, I thought "more quiet fan, so probably far less airflow, so less heat removal"...

But, 9.25 vs 8.9cfm airflow is a very minor difference - if it manages to move this amount of air, cooling is roughly the same so indeed a good improvement.

Indeed, I had the same feeling, but am still confused. First of all the 9.25 versus 8.9 is a minor difference, But... how is it possible that a fan which looks 100% the same (they all do, even the blade shape looks the same to me) and which runs 6000rpm slower (3000 vesus 9000) has around the same airflow as the original?

So maybe the airflow spec is a little overrated. I can't measure it though

By the way, the switch is made to deliver POE on each port. However you can switch off POE per port which I did and I also disabled 75% of the ports. So heat should not be a problem. It consumes 86watt with no connections.

I should run it for a few hours and have a look at the internal temperature sensor. I'll update the text when I have this information.

[Comment edited on donderdag 29 november 2018 10:17]

By heuveltje, donderdag 29 november 2018 11:26

Stil curious what are you going to do with em ?
At 86W thats atleast 160E per year if u run em 24/7 . I couldnt justify that for myself

By ArnieNFW, vrijdag 30 november 2018 07:57

heuveltje wrote on Thursday 29 November 2018 @ 11:26:
Stil curious what are you going to do with em ?
At 86W thats atleast 160E per year if u run em 24/7 . I couldnt justify that for myself

The answers are hidden in the text:

Not a very bright idea since I have a small hobby room and the 5520 is a enterprise datacenter switch meaning: it has a high power usage and creates an awful lot of noise. And of course I don't have real use for it.
Since I will be using the switch only for experimenting and not for 24/7 use

I experiment a lot with IoT devices, microcontrollers, arduinos and single board computers (pine64, onion, raspberry). It's nice to have 48 Ethernet ports available when I need them, Indeed I could better have bought a 8-ports TPLINK unmanaged switch. Cheaper to buy and much lower power usage.
[br]
But.... I learned to update the firmware (quite different than other devices I had). Had a lot of fun replacing the fan's. And can now also experiment with bundled Ethernet ports to my NAS, Server and client computer. If only I had time available.

By DJ, vrijdag 30 november 2018 10:33

Those SUNON fans are great and quiet. I once did the same as you, but then to a Cisco 3620 router (very old, and long gone by now!). It was connected to my Chello cable internet (now ZIGGO). I could obtain a speed of 2Mbit, which was the max of UPC at the time (somewhere in the early 2000's) . . .

By ArnieNFW, vrijdag 30 november 2018 12:11

DJ wrote on Friday 30 November 2018 @ 10:33:
Those SUNON fans are great and quiet. I once did the same as you, but then to a Cisco 3620 router (very old, and long gone by now!). It was connected to my Chello cable internet (now ZIGGO). I could obtain a speed of 2Mbit, which was the max of UPC at the time (somewhere in the early 2000's) . . .

I think there is not a lot of innovation on FANs. Although this one reads MAG-LEV which sounds really cool. (and creates less noise than ball bearing, I presume).

[Comment edited on vrijdag 30 november 2018 12:28]

By ArnieNFW, vrijdag 30 november 2018 12:29

[double post]

[Comment edited on vrijdag 30 november 2018 12:31]

By ArnieNFW, vrijdag 30 november 2018 12:31

[double post - having problem with my browser cache]

[Comment edited on vrijdag 30 november 2018 12:32]

By Hkuit, vrijdag 30 november 2018 13:52

waarom reageert iedereen in het Engels

?
Leuk verhaal, goede foto's maar natuurlijk niet heel spectaculair; je hebt wat fans vervangen en stuitte op een probleempje met de stekkertjes wat je hebt opgelost.

Ik vind het vooral grappig om te zien dat je een apparaat aanpast waarvan je zelf al aangeeft dat het volstrekt overbemeten en onzuinig is. Ik heb ook nog steeds het plan om hier een oude Jaguar XJ voor de deur te zetten

By Martijn.C.V, vrijdag 30 november 2018 14:24

ArnieNFW wrote on Thursday 29 November 2018 @ 10:16:
[...]

Indeed, I had the same feeling, but am still confused. First of all the 9.25 versus 8.9 is a minor difference, But... how is it possible that a fan which looks 100% the same (they all do, even the blade shape looks the same to me) and which runs 6000rpm slower (3000 vesus 9000) has around the same airflow as the original?

So maybe the airflow spec is a little overrated. I can't measure it though

By the way, the switch is made to deliver POE on each port. However you can switch off POE per port which I did and I also disabled 75% of the ports. So heat should not be a problem. It consumes 86watt with no connections.

I should run it for a few hours and have a look at the internal temperature sensor. I'll update the text when I have this information.

Het principe is eigenlijk heel simpel: Bij de eerste paar duizend RPM is er voor de ventilator voldoende lucht beschikbaar om aan te trekken, want een klein laagdrukgebied maakt waar nieuwe lucht rustig heen kan stromen.
Simpel gezegd: Hoe hoger de RPM, hoe meer de lucht in de ruimte zn best moet doen om voor de ventilator te 'stromen', dus het gebied voor de ventilator is een laagdrukgebied, er is simpelweg minder lucht om efficiënt aan te zuigen.

By ArnieNFW, vrijdag 30 november 2018 16:09

Hkuit wrote on Friday 30 November 2018 @ 13:52:
waarom reageert iedereen in het Engels ?
Leuk verhaal, goede foto's maar natuurlijk niet heel spectaculair; je hebt wat fans vervangen en stuitte op een probleempje met de stekkertjes wat je hebt opgelost.

Ik vind het vooral grappig om te zien dat je een apparaat aanpast waarvan je zelf al aangeeft dat het volstrekt overbemeten en onzuinig is. Ik heb ook nog steeds het plan om hier een oude Jaguar XJ voor de deur te zetten

Ik claim ook niet dat het spectaculair is; waarom je dit dus kwijt wilt is me even onduidelijk. Internet staat vol met onzin; dat melden is onbegonnen werk. En verder is er op internet wel wat te vinden over het vervangen van de fans van dit type switch, maar niet heel veel. Zo er er iemand die 12cm fans bovenop monteert middels een gat in de behuizing. In het engels zijn meer mensen in staat te lezen(zien) hoe ik het opgelost heb.

Over zinnigheid hoeven we het denk ik verder niet te hebben. Niet omdat het moet, maar omdat het kan.

Martijn.C.V wrote on Friday 30 November 2018 @ 14:24:
[...]
Simpel gezegd: Hoe hoger de RPM, hoe meer de lucht in de ruimte zn best moet doen om voor de ventilator te 'stromen', dus het gebied voor de ventilator is een laagdrukgebied, er is simpelweg minder lucht om efficiënt aan te zuigen.

Bedankt voor je uitleg; begrijpelijk.

By PcDealer, zondag 2 december 2018 17:27

Hoe meer leds hoe meer beter

By Rowdy.nl, vrijdag 11 januari 2019 10:24

Interesting. Dan kan ik mijn Dell PowerConnect ook weer eens dichtschroeven. Ik heb gewoon de behuizing opengeschroeft en de ene fan er uit getrokken. (genoeg omgevingsflow, maar wel een rommel in de kast)

Comments are closed

Web(b)log van Arnoud