Semaphore
Premium
join:2003-11-18
Arnprior On. | reply to pmurdock
Re: hub or switch at solar powered POP?
I agree with NWN & Daggar... I did say it does work..... electronically, and from a network perspective, it's terrible. It won't use any power, but it will likely induce high collisions. |
|
daggardale
Premium
join:2003-11-07
Gurnee, IL
 ·Vonage
| reply to pmurdock
I used a simple splitter device like this a few years back. The collision rate was terrible, but it worked. Here's a link to a cheap one, if you're interested.
»www.cablesnmor.com/index.asp?Pag···odID=698 |
|
nwn
Premium
join:2004-03-05
Centerville, IN
| reply to pmurdock
Looking for trouble here. The ports are not designed for this. They are designed to connect to 1 other device. This may work, if you can get your new device to put the ethernet port into a high-impedance state, so it looks invisible, when not operating. I doubt that it will work, though, unless the other device does the same.
Ethernet works by toggling the voltage from 0 volts to x volts between the transmit pair wires. If one device tries to drive the voltage to 0 and the other is not transmitting, the second device will do its best to maintain the x voltage. The current sourced from device 2 will go way up, E = IR, and to get to 0 volts R will be 0.
This is the reason for hubs, so every device can manage its own transmit pair.
--
Scott |
|
pmurdock
Paul Murdock
join:2001-09-13
Riverton, UT
3 edits | reply to Semaphore
Ok.. so what we have is the new device - the battery voltage device will only ever be active for 5 seconds each hour to transmit it's voltage back to the data center.. So in essence I guess the two devices never need to talk to each other. Here is a drawing that I have made to represent the design..
Basically what will happen is a server in the NOC will send out a UDP request to the battery monitor asking for it to send back the voltage every hour. It will only need to respond with a very small packet and then go back to sleep (drawing microamps current - now that is what I call small!  )
Bear in mind also this is a far-away repeater in the middle of nowhere.. with a very limited potential client base.. I'd be happy with a 1Mbps connection to this location and not worry about saturation..
cheers,
Paul |
|
Semaphore
Premium
join:2003-11-18
Arnprior On.
| reply to pmurdock
If the required layout is that two devices need to talk to one device then a passive Ethernet splitter would work. Yes they make them. I haven't installed one in over a year but they allow two PC's to share a single Ethernet drop. You can't get Full Duplex but you can get 100Mbps. Yes collisions are problem. No the to devices can't talk to each other directly. Ugly but it does "work". |
|
pmurdock
Paul Murdock
join:2001-09-13
Riverton, UT
1 edit | reply to John Galt
It will be run from a 12V sealed gel cell 285 Amp/hour battery that is charged by some photovoltaic cells.. So the 7.5 v supply is better than the 10V minimum requirement.. however 52 mA versus 300 mA doesn't make sense.. well the equation makes sense..
for 52 mA switch
P = IV
2 watts = (I)*10 volts -> I = 2/10 -> I = 0.2 amps
for ~300 mA switch
P = IV
1.5 watts = (I)*7.5 volts -> I = 1.5/7.5 -> I = 0.2
so it appears that the parvus one is better.. though 200 mA is a horrible draw to have for both of the switches..  is that a continuous draw..?
I would think that the draw would be higher at a higher voltage since voltage is like "pressure" if you applying a higher pressure more current would want to flow through the system. So why only draw 52 mA @ 24VDC and draw MORE current (~200 mA) @ 10VDC? This is what I don't see.
cheers,
Paul |
|
John Galt
Forward, March
Premium
join:2004-09-30
Happy Camp | reply to pmurdock
What is your system voltage?
--
A is A |
|
pmurdock
Paul Murdock
join:2001-09-13
Riverton, UT
| reply to John Galt
Here's another one I found.. made by Parvus
»www.parvus.com/products/IODataco···itch104/
Says it is a 1.5 Watt, as opposed to the 2 Watt usage by the other one. Though at 5 volts, the P=IV equation says it will draw 300 mA.. how can that be lower than the 52 mA device that draws 2 Watts.?!?
cheers,
Paul |
|
John Galt
Forward, March
Premium
join:2004-09-30
Happy Camp
 ·CenturyLink
| reply to pmurdock
said by pmurdock  :Does anyone have a feel for what a "low power" switch should be.. is 52 mA really the best we can do? That's pretty low...!
--
A is A |
|
pmurdock
Paul Murdock
join:2001-09-13
Riverton, UT
| reply to jdmarti1
looks like it is 7.5 Watts.. maybe I'm better off going with that ethernet switch up at the top that consumes 2 Watts or 52 mA @ 24VDC.
I see what you mean about tying which wires together.. that does present a problem. As it is now, they are a cross over direct connect.
Does anyone have a feel for what a "low power" switch should be.. is 52 mA really the best we can do?
cheers,
Paul |
|
jdmarti1
Jack
join:2004-06-15
Oilton, OK
| reply to pmurdock
My biggest concern would be to create a chance for collisions on my backhaul. That is a very important link. This little unit only takes one watt. I am unsure what voltages you have on your power supply, so you either use the wall wart with this or use the 7.5v.
»www.trendnet.com/products/TE100-S5P.htm
--
»magicwisp.com |
|
public
join:2002-01-19
Santa Clara, CA
·DSL EXTREME
| reply to pmurdock
said by pmurdock :I'm not so sure about that.. A hub is useful where you need to repeat the signal over larger distances. However, the ethernet protocol is a CSMA/ CD protocol, which implies "C"ollision "D"etection.. this means that a number of transmitters theorectically could be on the EXACT same wires and still be able to communicate as I understand it. When a collision occurs a random backoff period happens and then the two transmit again at different times.. simply by splicing the wires together indeed I will create the opportunities for collisions to occur, but it still should work? cheers, Paul That applies to original thick wire ethernet.
Read up on 10baseT requirements. |
|
pmurdock
Paul Murdock
join:2001-09-13
Riverton, UT
| reply to public
I'm not so sure about that.. A hub is useful where you need to repeat the signal over larger distances.
However, the ethernet protocol is a CSMA/CD protocol, which implies "C"ollision "D"etection.. this means that a number of transmitters theorectically could be on the EXACT same wires and still be able to communicate as I understand it. When a collision occurs a random backoff period happens and then the two transmit again at different times.. simply by splicing the wires together indeed I will create the opportunities for collisions to occur, but it still should work?
cheers,
Paul |
|
public
join:2002-01-19
Santa Clara, CA
·DSL EXTREME
| reply to pmurdock
said by pmurdock :isn't that all I really need though is a way to splice the cables together? cheers, Paul No. Ethernet twisted pair needs a hub. |
|
pmurdock
Paul Murdock
join:2001-09-13
Riverton, UT | reply to BigCreek
isn't that all I really need though is a way to splice the cables together?
cheers,
Paul |
|
BigCreek
God Is Good.
Premium
join:2002-06-25
Heber Springs, AR
| reply to pmurdock
I believe that's just a patch panel. Note the first description line:
quote:
Use in Conjunction with an Active 10/100 Base-T Switching Hub
Here's the manu's page:
»www.hometech.com/techwire/vanco.html
--
SBC Pro Static DSL; Linux. Terrific wife & kids; live on a farm by Big Creek. Software & network consultant. |
|
pmurdock
Paul Murdock
join:2001-09-13
Riverton, UT
| reply to BigCreek
I was thinking of something like this..
»cgi.ebay.com/ws/eBayISAPI.dll?Vi···ame=WDVW
and then just wire up the ports to all be the same wires.. should work.. ?
cheers,
Paul |
|
