GPON Optical Levels

We are deploying our first GPON network, and are trying to get a sense of
what is a good optical level per ONT. The equipment install guide says the

Measure the levels of the 1550 dBm and 1490 dBm receive optical inputs using

an optical power meter. The input range should be between -27.0 dBm to -8


Attached are the measurements I can see and graph from the ONT side. Trying
to get a sense of what range we would graph as good (green) measured within
the clear, what we could consider yellow as working but out of optimal
range, and what we could consider poor (red) range.

What about the Laser Bias Current, Optics Module Voltage, and Optics Module

Here is the link to the attachment as it looks like the list does not allow
uploading attachment:

A good general rule of thumb for optics is +-4. Try to stay 4 db off the bottom, and 4db off the top. The sweet-spot RSL is between 1/2 and 2/3 of the optical budget. For PON, typical OLT/ONT optics run Receive Max of -8, and Min of -27 with an optical de-assert at -31. The sweet-spot is between -17 and -20, with a min/max of -12 to -23.

A typical PON system will have a TX power between +3 and +5, with a 32 way split, add 15dB of insertion loss. You should see -12 on the output of the first splitter (less the optical loss of the fiber between the OLT and the Splitter est .3/km)

There should never be a need to attenuate a PON port unless you're working directly on the OLT (You should work through a splitter even directly in the CO/HE)

Here is a small secret: If you attenuate the PON port so the signal level
at the OLT is as low as possible, your network will be more robust against
people connecting P2P fiber media converters to your network. This is
because these devices typically have TX power that is significantly less
than your ONUs. By attenuate the signal you can bring the rogue signal
below the detection threshold of the OLT.

In my experience one way to do this is to always use 1:128 splits. Even if
you are only going to connect less than 32 ONUs, you will find that 1:128
can be more robust. In fact I discovered this little trick after we started
using 1:128 splits (for flexibility, not because we actually connect that
many clients). Because the fiber plant is shared with other service
providers, we used to have OLT crashes due to people connecting a media
converter they got from another service provider. I found that we had zero
of these issues where we had 1:128 splits.

Btw we use class C+ optics. This might work out differently if you have
class B optics.



Wouldnt a 1:32 splitter + a 9dB attenuation be the same? From
the price/performance the 1:128 splitter should be much more expensive.

We have seen happening the same converting a former p2p footprint to
GPON and somebody got a left over p2p device connected. But this was
a single incident so far.