Yeah approx $400 for a pair. Mainly chip shortage messiness, I think I can get the cost down to less than $200 a pair in April 2023.
Regarding VSDL, never used it myself; I guess it would come down to space/power considerations. I haven't seen VDSL used much in the industrial automation, single pair ethernet seems to be the preferred choice.
You can get full products that do higher speeds over longer distances at lower cost, search term "vdsl repeater".
It's possible that this product handles bad wiring better as it's ethernet rather than VDSL2, but I think those devices are a much better fit.
One thing that might be better with an ethernet based standard is the ability to add multiple stations to the same link, relying on ethernet's error correction.
What's the difference in price between 1km of fiber vs 1km or single pair wire? It might play a role in this solution being cheaper than immediately apparent.
The fiber is much cheaper because it's so much more common.
For example:
I/O fiber is about 0.10 cents a foot or less for 2 fiber cable. (12 fiber is only 20 cents a foot).
OSP gel-filled direct burial is 0.55 cents a foot or less for 2 fiber cable. (12 fiber is only 65 cents). This is micro-armored OSP cable, resistant to chewing/damage as well.
Basic 18 awg single pair shielded direct burial wire is about 0.80 cents a foot. Can't even find a non-expensive armored version
10km single fiber transceivers at 10gbps are 40 bucks.
10km dual fiber transceivers at 10gbps at 29 bucks.
It would be much cheaper, more effective, and a better result to just use fiber for 1km here.
You could easily bury 12 fiber cable and use it for the price you will pay to use single-twisted-pair.
I've got a 3 pair run[1] to my gate with a keypad gate controller. Two pair are used for voice communications, and one pair is idle. The gate controller has 10 base T ethernet (for management), but you need two pair, and it's maybe a touch outside the distance spec for 10baseT. I'm not quite ready to spend $350 to get it on ethernet though.
[1] Actually, it's 4 pair to an in-ground junction box, where it splices to 3 pair to the gate controller. For $350, I have maybe some other other options around the junction box (I've already got some passive 10baseT <-> single pair devices, but they don't work at the full distance. Probably would work at the junction box, if I had a powered ethernet switch there, but I need weatherproofing.
When everyone prices fiber, they fail to include the termination costs.
How much are the termination costs on that fiber, per run?
Also, you'd never install single pair direct burial, and almost never shielded (unless its audio and not phone type) - it'd be 4/6/12 pair direct bury gel-filled cable.
Termination is basically a non-cost anymore. The price a pack of 10 sc/apc field terminations (IE that a random DIY'ers could do) is <20 bucks, and to your question, the same as a price of a pack of 10 ez-r45 connectors.
The loss and ease of mechanical connectors has gotten good enough (0.15 db or better) that most of the folks i know will only fusion splice when they really have to.
I agree it wouldn't make sense to do single pair or single fiber, but that's the actually the comparison at which 10baset-1l is at all competitive.
I have buried a lot of fiber, and a lot of network cable over the years :)
I was just trying to be fair and present the best case for 10baseT1L
In practice you can do 12/24 strand fiber (or 6 strand fiber + power + whatever) for less than the price of whatever particular set of 18 awg twisted pair (which is what the spec requires at 1km) + other things you wanted.
The cable would be smaller, and if you use A3 or B3 fiber, it would be more flexible/support a lower bend radius than the 18 awg twisted pair. By far. IT's not even close.
Besides what DannyBee says about termination being a lot cheaper and easier when you can tolerate small losses (which is most people):
Big infrastructure often orders the cables pre-terminated, eg an ISP will order their arial cable with built-in termination at certain intervals. No muss, no fuss.
You're right about single pair though. If you're putting in two strands might as well do 6. If you need 6 might as well do 24. Never hurts to have spares and extra capacity. Especially if it involves stringing poles or digging holes.
Yeah, from a cost/redundancy/etc perspective it rarely would make sense to run 1.
I will get pre-terminated for MTP runs, but not for anything else. Fusion splicing connectors just isn't slow (and if i need to go quick, i'll just splice a pigtail onto it until it fails)
I agree for redundancy, etc, better to just add more until loss-of-ROI occurs, but will say doing something for extra capacity for home users/even business users feels like massive overkill these days.
Even 2km 100G duplex transceivers for SMF are <$200 a piece now.
Cost is dropping crazily fast.
(bidi simplex ones now also exist on fs.com as of a month or two ago, but are still way too expensive)
It's dollars per foot, yes.
Honestly though, since they are all in the same relative scale, and the point is what was cheaper and by what percent, this feels like pedantry.
Put another way - outside of being technically correct, why does it matter to the point at hand?
I pay about $350 per kilometer for 24 count flat all dielectric G.657A1 fibre with 2 x 2.0mm fibreglass strength members. Add about $100 for shipping from overseas.
It comes down to whether the cable is there already. If you were laying it, you'd probably go for fibre. But if you've got 1km of some poor quality cable there already, then spending $200 to get usable bandwidth might be a lot cheaper than laying something new.
The cost of laying wire is virtually the same as laying fiber, so the difference really comes down to the material costs. The only good reason to use wire would be existing installations, as you sidestep that cost entirely.
But perhaps the wire is already there. I don't think anybody advocates for building entirely new datalinks using this. Laying another 1km of cable is much more expensive than 2x200 USD to reuse existing cable.
Whatever EUROCARE said about existing installations applies.
Though what I had in mind was that when I was handling optical fibre as a layman (we have fibre to the living room here in Singapore), that fibre is a lot more delicate, so would require a bit more care to handle than copper. But I guess that's probably not a major cost factor, if eg you have to dig a trench for your cable.
However, with StarTech's product, you have a shot at faster speeds if you're going a shorter distance. It can do 75 Mb/s at 300 meters and up to 100 Mb/s at shorter distances.
I would really like to find a similar box that can bond multiple pairs for faster speeds but I haven't found such a device yet.
And they come with an enclosure and power supply. And a better sort of "step up / step down" bandwidth based on distance...see the chart at the bottom.
I certainly believe SPE is the future but not as an immediate replacement to ethernet for the following reasons:
- BASE-T1 PHYs and connectors are going to be more expensive than cheapo BASE-T PHYs and RJ-45s for a while.
- The entire industry is based around BASE-T, try finding off-the-shelf mobile imaging hardware (or other embedded sensors) that supports BASE-T1.
- BASE-T technology is less fragile, ultimately a single pair is more susceptible to signal loss / reflections and needs more complex encoding/transmission schemes to overcome this. (For example, 100BASE-T uses 25MHz, two states, four pairs. 100BASE-T1 uses a 33.33MHz, three states on a single pair. This reduces the tolerable noise floor and requires the use of better cabling. 100Mbit ethernet is damn-near indestructible I've found. (There is another alternative here which I will get to in a moment).
Despite these issues, SPE's benefits will win out for mobile applications that have Size and Weight requirements. GigaBit speeds (or even up to 10 gigabit) on a single pair is just too damn tempting and can translate directly into better products. Want an example? Imaging you're making a payload for a drone; you probably use slip rings and your electrical contacts are limited. Using SPE means you can cram more data down the same slip rings, resulting in higher resolution imaging compared to a competitor using standard ethernet.
My understanding is that 10BASE-T1L is designed for long reach applications like factory automation, to connect between clusters of devices using 10BASE-T1S (all sitting on the same pair of wires). 100BASE-T1 and 1000BASE-T1 are for ideally for mobile applications.
Anyway this is all theoretical, it so happens that I run a hardware startup called BotBlox where I design compact ethernet hardware. I've designed a compact 1000BASE-T to 1000BASE-T1 converter, partly as a working prototype to get used to this technology. I believe there needs to be affordable, compact bridging hardware BASE-T and BASE-T1, and I intend to develop it.
My current tests with 1000BASE-T1 is that the range is not near 40m, I managed to get around 10m before it dies. I need to do more testing on this to see whether it's the cabling, and the firmware I made.
I have a 10BASE-T to 10BASE-T1L board in manufacture because I'm ridiculously excited about the prospect of 1.6km copper transmission. We'll see how this goes.
Anyway, there's a link below. The chip shortage has destroyed my ability to scale up on this board, hence the high price. I expect once chip shortages abate, achieving a $100 price point on this kind of board is achievable.
One point I completely forgot was Gigabit Home Network, (G.hn) as a alternative to 1000BASE-T1. It essentially does the same thing and can achieve long ranges. Ultimately this technology is designed to pump 1000Mbps down nasty home wiring, so using it on a single pair of wires is probably going to result in a very good range.
The encoding scheme here is OFDM, aka, break the signal down into multiple signals at different frequencies, and send it down the wire.
It's more robust than SPE, but it requires more costly and larger ICs than SPE. When you can define a connector and cable, SPE is still better in terms of cost and size I believe.
G.hn could have its uses in retrofitting older wiring systems.
BTW, since I have the hardware on hand, let me know if y'all want me to conduct specific tests. I have a beauty of an oscilloscope that can get very good sampling resolution on this.
Regarding VSDL, never used it myself; I guess it would come down to space/power considerations. I haven't seen VDSL used much in the industrial automation, single pair ethernet seems to be the preferred choice.