Hi,
can this mux be used crosspatched to let it work on single strand fiber (obviously halving the capacity)? We'll use this technique with other products and it'll work fine, but I see that mikrotik's one explicitly distincts mux from demux. Is it just an indication, or is the demux side physically different from the mux side ?
BR,
Simone.
Re: CWDM-MUX8A
Yes, you can, but
- You'll need an additional splitter to combine both RX and TX, since you're planning using single strand.
- You'll have additional losses due to the combined RX/TX and connectors
- You're only using half of the wavelengths, since you need the others at the other side to be demuxed.
It can be done, but I recommend a seperate mux/demux when you're planning on using a single strand. Especially when you're having issues its much easier to debug a 'build-for-it' setup instead of 'misusing' a 2-strand mux/demux.
Also, I recommend to use 2 adjecent channels for a link, eg 1270/1290 for a link, and not 1270/1610. Since CWDM uses a wide range of optical spectrum this can cause some issues, especially on longer links.
Finally, make a good design. Most fibers are specified on either 1310 or 1550nm (or both) and when using 1270 or 1610nm they have a much higher attenuation. this, along with your mux losses, has to be taken into account when calculating your power budget as well.
Nowadays, I would only consider CWDM on
- Links where you need to mix 1 and 10Gbit (1Gbit DWDM is pretty expensive compared to CWDM)
- Short links (<30-40km)
- Links which you're 100% sure they never need to be extended in the future.
Also make some calculations for DWDM. This is more expensive, but gives you much more future options:
Upgrade to higher link speeds, more channels (40 instead of 16 when using 2-strand), ability to amplify when you need to cover a longer fiber span, and much more equal attenuation on all wavelenghts.
- You'll need an additional splitter to combine both RX and TX, since you're planning using single strand.
- You'll have additional losses due to the combined RX/TX and connectors
- You're only using half of the wavelengths, since you need the others at the other side to be demuxed.
It can be done, but I recommend a seperate mux/demux when you're planning on using a single strand. Especially when you're having issues its much easier to debug a 'build-for-it' setup instead of 'misusing' a 2-strand mux/demux.
Also, I recommend to use 2 adjecent channels for a link, eg 1270/1290 for a link, and not 1270/1610. Since CWDM uses a wide range of optical spectrum this can cause some issues, especially on longer links.
Finally, make a good design. Most fibers are specified on either 1310 or 1550nm (or both) and when using 1270 or 1610nm they have a much higher attenuation. this, along with your mux losses, has to be taken into account when calculating your power budget as well.
Nowadays, I would only consider CWDM on
- Links where you need to mix 1 and 10Gbit (1Gbit DWDM is pretty expensive compared to CWDM)
- Short links (<30-40km)
- Links which you're 100% sure they never need to be extended in the future.
Also make some calculations for DWDM. This is more expensive, but gives you much more future options:
Upgrade to higher link speeds, more channels (40 instead of 16 when using 2-strand), ability to amplify when you need to cover a longer fiber span, and much more equal attenuation on all wavelenghts.
Re: CWDM-MUX8A
So the CWDM-MUX8A need 2 fibers to transmision between locations right?
Re: CWDM-MUX8A
That would be usual application since most commonly CWDM SFPs use same wavelength for both TX and RX ... at least they are declared as such. However I've seen CWDM systems which used single fibre strand between the sites ... MUXes and SFPs came in complementary pairs. Not MT one though.
OTOH I've seen many SFPs with no affinity for RX wavelength (e.g. standard 1310 nm wide-band and a 1590 nm CWDM SFPs cross-connected to each other worked just fine) and if your SFPs are this kind, then you could build setup with single strand between sites and using standard (dual-strand) MUXes and SFPs ... you only have to be careful to use SFPs with different TX wavelengths on both sites.
OTOH I've seen many SFPs with no affinity for RX wavelength (e.g. standard 1310 nm wide-band and a 1590 nm CWDM SFPs cross-connected to each other worked just fine) and if your SFPs are this kind, then you could build setup with single strand between sites and using standard (dual-strand) MUXes and SFPs ... you only have to be careful to use SFPs with different TX wavelengths on both sites.