Switch Seletion for Hub

[gatorback]

This community is innovative and clever: I look forward to good-interesting responses.

Goal: identify criteria for choosing an unmanaged switch. Seeking innovative ideas & products for cleaner (low-profile) / neater cabling + switching / cooling

Context:
* located in garage: can get warm in summer (90 Farenheit)
* active cooling (fan or similar) desired
* nearest power supply is in ceiling: used for garage door opener
* switch will need 6th port to bridge-in WAN
* 100 Megabit ports in photo: wired with 4 conductors of CAT5E
* other 4 conductors wired for telephone
* telephone conductors are not used and can be re-purposed for network
* 5 ports shown in photo (black port difficult to see)


Decisions:
1) Locate of switch: power outlet or ethernet panel in photo?
2) If using a USB 5V fan-cooler, then a 5V switch https://www.amazon.com/gp/product/B008PC1MSO/?tag=snbforums-20 could share the same USB 5V power supply?

 

[L&LD]

Is this garage unheated? Will condensation play a part in fall/winter/spring?

 

[System Error Message]

physical location/size has no meaning to the net or the virtual world. All that matters is your requirements such as do you need managed features? What sort of protection do you want or is it just to switch data around regardless if its good or bad?

Cooling, active or passive doesnt matter. Perhaps you could get better cooling and save power by having all passives and using one big fan to cool multiple devices at the same time. All that matters is airflow, ambient temperature and how good the cooling design is on the design (such as gigantic heatsink that is exposed to ease heat transfer). You could also replace thermal paste as well but if it uses pads it will be more difficult as you will have to find better pad/thermal glue which isnt easy.

Another thing is ports. How many ports do you need and what bandwidth? think about the future, how many more ports will you have within a few years before you will be replacing your switch?

There are many more features a switch has and it all depends on your requirements and wants. Maybe you want fancy LEDs, or POE with active and passive support and support for all those active protocols.

Humidity is bad for electronics and heat so this is mportant to consider. As long as ambient temperature is good and low humidity this is good for many electronics except electrolytic capacitors that do well within a middle temperature but have bad capacitance/ESR at low temps and fail at high temps. Its a reason why PSUs tend to fail commonly with the caps being the cause as electrolytic are considered to be the least reliable but the cheapest when you need capacity. So if you've sorted out your environment for electronics hardware than look at the featuresets of the device.

Ambient temperature should be at least 20C below their rated operating temperature. So if rated ambient temperature is 50C than your max room temperature should be below 30C.

 

[L&LD]

Ambient temperature should be at least 20C below their rated operating temperature. So if rated ambient temperature is 50C than your max room temperature should be below 30C.

That doesn't seem right? Why don't they just rate it for 20C cooler then?

 

[System Error Message]

That doesn't seem right? Why don't they just rate it for 20C cooler then?

Temperature in device chassis will be hotter than outside.

 

[L&LD]

Temperature in device chassis will be hotter than outside.

You're confused then. The temperature inside the device's chassis is what the manufacturer has control of.

The ambient temperature is what the user 'may' have control of. There is no sliding score that needs to be applied.

The manufacturer has already taken that into consideration (hopefully). And if they haven't? Then they are giving us the wrong spec's to decide and plan on our purchases.

 

[System Error Message]

You're confused then. The temperature inside the device's chassis is what the manufacturer has control of.

The ambient temperature is what the user 'may' have control of. There is no sliding score that needs to be applied.

The manufacturer has already taken that into consideration (hopefully). And if they haven't? Then they are giving us the wrong spec's to decide and plan on our purchases.

The rating is based on the lowest spec electronic component with the temperature around it. The spec specifies ambient temperature in chassis. Temperature in chassis will always be higher than room temperature in a consistent way. This i noticed because in my router the temperature reported was always 50C which is why the PSU failed fast. After using external PSU temperature dropped to 45C.

 

[sfx2000]

This community is innovative and clever: I look forward to good-interesting responses.

Goal: identify criteria for choosing an unmanaged switch. Seeking innovative ideas & products for cleaner (low-profile) / neater cabling + switching / cooling

Context:
* located in garage: can get warm in summer (90 Farenheit)
* active cooling (fan or similar) desired
* nearest power supply is in ceiling: used for garage door opener
* switch will need 6th port to bridge-in WAN
* 100 Megabit ports in photo: wired with 4 conductors of CAT5E
* other 4 conductors wired for telephone
* telephone conductors are not used and can be re-purposed for network
* 5 ports shown in photo (black port difficult to see)

Might consider a local expert - check Craigslist or something similar - asking a web forum might give you tidbits of info, but you've got a special case - if I were local, I'd be happy to help out at cost...

 

[L&LD]

The rating is based on the lowest spec electronic component with the temperature around it. The spec specifies ambient temperature in chassis. Temperature in chassis will always be higher than room temperature in a consistent way. This i noticed because in my router the temperature reported was always 50C which is why the PSU failed fast. After using external PSU temperature dropped to 45C.

I find it very hard (to impossible) to believe we're to account for the inefficiencies of the manufacturer's thermal design decisions.

And to be sure, it is not consistent in any way.

 

[System Error Message]

I find it very hard (to impossible) to believe we're to account for the inefficiencies of the manufacturer's thermal design decisions.

And to be sure, it is not consistent in any way.

What matters to a component is its operating temperature so take a passive component for example and it will always be higher than the temperature of the air around it. Usually operating temperature would mean the temperature of air around the component so that would mean the temperature inside the chassis. I've taken up electronics and the specs and tolerances of specific components do align with the quoted values of electronic related products meaning the temperature of air inside the chassis matters which will always be hotter than the temperature of air outside the chassis. If you still want to disagree about this try texas instrument's designer and take a look at the specs of various components which will have their limits written from every resister, capacitor, even the transistors and silicons as well and you will often see the values to be the same as quoted on a finished product, usually taking the worst spec of the components.

Open up a switch, even for a passive 5 port switch those internal components (whether or not they have heatsink) will always keep the air in the chassis or around them hotter than outside so what matters is the temperature inside the chassis and not the room temperature. Just because something is rated to operate up to 50C doesnt mean it would operate in the desert at 50C, only means the air inside chassis must not exceed 50C.

You cant control the difference between the temperature inside and outside the chassis, it could be hidden in the wall for example entirely enclosed. If your motherboard temperature read around or above 50C wouldnt you be worried?

 

[CaptainSTX]

Use an extension cord from the ceiling outlet to where the switch is needed. Mounting the switch lower in the garage will keep it slightly cooler.

Given that an 8 port unmanaged switch can be purchased for $25 the premature failure won't break the bank. Purchase two and keep a spare for a quick change out. Small switches don't use a lot of power so they don't generate a lot of heat. A fan might help but not much as blowing hot air over the switch isn't going to have much of a cooling effect.

 

[gatorback]

Is there a good source for operating temperature data, so that one can use this information in purchase decisions? This is an excellent thread: I appreciate the genuine interest (and excellence).

Yikes! my HP Elitebook routinely idles at 57C and can spike up to 70 and that's after a recent cleaning: http://askubuntu.com/questions/801905/hp-laptop-temperature-expectations

I agree that the cost is insignificant and having a backup is a sensible solution, however, the challenge is that this is in the remote house in another state.

 

[L&LD]

What matters to a component is its operating temperature so take a passive component for example and it will always be higher than the temperature of the air around it. Usually operating temperature would mean the temperature of air around the component so that would mean the temperature inside the chassis. I've taken up electronics and the specs and tolerances of specific components do align with the quoted values of electronic related products meaning the temperature of air inside the chassis matters which will always be hotter than the temperature of air outside the chassis. If you still want to disagree about this try texas instrument's designer and take a look at the specs of various components which will have their limits written from every resister, capacitor, even the transistors and silicons as well and you will often see the values to be the same as quoted on a finished product, usually taking the worst spec of the components.

Open up a switch, even for a passive 5 port switch those internal components (whether or not they have heatsink) will always keep the air in the chassis or around them hotter than outside so what matters is the temperature inside the chassis and not the room temperature. Just because something is rated to operate up to 50C doesnt mean it would operate in the desert at 50C, only means the air inside chassis must not exceed 50C.

You cant control the difference between the temperature inside and outside the chassis, it could be hidden in the wall for example entirely enclosed. If your motherboard temperature read around or above 50C wouldnt you be worried?

Now you are backpedaling.

Of course the components are rated for a certain temperature. And that maximum temperature will be the actual temperature the component is subject to, in the enclosure it shipped with and at the maximum rated load of the device.

I am not arguing those points. We're in agreement here.

Your initial post stated that we somehow have to predict (by an offset of about 20C) between what manufacturers quote as 'ambient temperatures' and what every other normal human being defines 'ambient temperatures' to be. Namely, room temperatures.

If a manufacturer states that a device can operate safely up to 50C in ambient temperatures, then I expect it to have the cooling capacity and designed to operate at up to those room temperatures.

The 20C difference that you feel is 'consistent' between actual room temperature and internal ambient temperatures of devices is not a constant at all. This is simply false information.

The ambient temperature is the room temperature. Anything else is your misunderstanding.

 

[thiggins]

A Gigabit industrial switch isn't that expensive. The linked one is rated -40 DegreeC to 75 DegreeC

 

[System Error Message]

Now you are backpedaling.

Of course the components are rated for a certain temperature. And that maximum temperature will be the actual temperature the component is subject to, in the enclosure it shipped with and at the maximum rated load of the device.

I am not arguing those points. We're in agreement here.

Your initial post stated that we somehow have to predict (by an offset of about 20C) between what manufacturers quote as 'ambient temperatures' and what every other normal human being defines 'ambient temperatures' to be. Namely, room temperatures.

If a manufacturer states that a device can operate safely up to 50C in ambient temperatures, then I expect it to have the cooling capacity and designed to operate at up to those room temperatures.

The 20C difference that you feel is 'consistent' between actual room temperature and internal ambient temperatures of devices is not a constant at all. This is simply false information.

The ambient temperature is the room temperature. Anything else is your misunderstanding.

20C is an estimate. The problem in a garage during summer is how hot it can get.

 

[L&LD]

20C is an estimate. The problem in a garage during summer is how hot it can get.

From 'consistent' to 'estimate'. Okay.

 

[System Error Message]

From 'consistent' to 'estimate'. Okay.

Its an estimate that is consistent. My router reports 50C when my room temperature is 30C on the board, not CPU and it has poor chassis cooling. board temperature is always consistently 20C higher than room temperature (which i use a thermometer). So 20C difference is a good estimate.

 

[sfx2000]

Use an extension cord from the ceiling outlet to where the switch is needed. Mounting the switch lower in the garage will keep it slightly cooler.

Given that an 8 port unmanaged switch can be purchased for $25 the premature failure won't break the bank. Purchase two and keep a spare for a quick change out. Small switches don't use a lot of power so they don't generate a lot of heat. A fan might help but not much as blowing hot air over the switch isn't going to have much of a cooling effect.

I agree with the Captain here - 8 port unmanaged switches are fairly affordable these days - one thought would be the Netgear GS108 unmanaged switch - metal housing, wall mountable, and they're pretty tough... and right now, they're pretty cheap, and the ProSafe line has a good reputation...

26 bucks on the Amazon right now...

https://www.amazon.com/dp/B00MPVR50A/?tag=snbforums-20

 

[Trip]

Might be splitting hairs, but I've found the Cisco SG110 and HP 1420 to be a bit less flaky than the Netgears and TP-Links. Moderate sample size of maybe a couple hundred switches and there's a bit of pedigree bias in my product selection, so take it on face value.

 

[sfx2000]

Cisco SG110 and HP 1420

That's a different class of gear, a lot more spendy, and they're excellent space heaters...

They are robust, but probably overkill for OP's intent