Yes. That's why the reflectance is so high for the twist-on connectors. They're basically a connector with a mechanical splice built-in.
In one of the ITU specs I've been skimming lately it says a good fusion splice must have -95 reflectance or better, which is essentially none.
Mechanical splices can be all over the place. I have a spec sheet for the AFL twist-on saying reflectance is -50 max and -55 typical. I think -55 is optimistic but do-able. A max of -50 seems silly. I don't think there's really a maximum. There's always
a gap in a mechanical splice, and the bigger the gap is the more reflectance you'll have.
Well, I couldn't let it go. It dawned on me this morning why -32dB optical return loss figure actually matters. The OLT will get reflections from every output leg of every splitter. Start adding all of that up and you can start to approach the specified
limits of total optical return loss.
The logarithmic scales aren't always intuitive. One bad value can hurt really bad. In this example, my one "Bad connector" with -40 reflectance has the impact of adding roughly 300 SC-APC connectors.
From: Adam Moffett <
[email protected]>
Sent: Thursday, August 21, 2025 5:34 PM
To: AnimalFarm Microwave Users Group <
[email protected]>
Subject: Re: [AFMUG] Help me Grok this - tolerance to reflected power
I think I figured out the limits.....or a limit.
ITU G.9807 is saying that the minimum optical return loss for 10Gig operation is 32dB.
Return loss is the opposite of reflectance. -65dB reflectance = 65dB return loss. You sum the power of all the reflections to get the total optical return loss on the path. A simple way is every time you double the number of connectors you get another 3dB
of reflectance (better formulas exist of course).
If I have 8 of those -65dB APC connectors then my total ORL (all else being perfect) is 56dB. It would take a whole lot of good connectors to get you to 32dB ORL. Or one really bad connector. I suppose the twist-ons aren't a big problem as long as they're
done decently.
Technically Raleigh scattering also adds to the total ORL, but it's negligible compared to the connectors.
-Adam
It’s not much of an issue on PON as far as I can tell. If you are running RFOG (analog cable added to the fiber) I think it’s a much bigger issue. I don’t know of very many companies still running RFOG.
Mark
I think the "tolerance to reflected power" was the wrong direction to look. That seems like it's more about not damaging the equipment. I'll back out of the rabbit hole and just ask what I'm trying to figure out:
How much does reflectance actually hurt you on PON? Google results are vague.
Example: You generally need to have two connectors at the OLT (device to patch panel), and there's one connector at the ONT. If you fusion spliced everything else so there were no additional reflective events, would there be a measurable BER or stability difference
compared to having a coupler in the NID, connectorized splitters in the field, etc.
Another example: An AFL Fastconnect twist-on SC/APC field termination has -50dB reflectance according to the spec sheet. A factory terminated connector or splice-on connector is -65dB. How much does the twist-on hurt you?
-Adam
From: Adam Moffett <
[email protected]>
Sent: Thursday, August 21, 2025 11:54 AM
To: AnimalFarm Microwave Users Group <
[email protected]>
Subject: Help me Grok this - tolerance to reflected power
Definition from ITU G.987:
3.3.15 tolerance to reflected power (transmitter): A transmitter parameter that characterizes the
maximum admissible ratio of the average reflected optical transmit power incident at the transmitter
to the average optical transmit power.
If my average transmit power is +6dBm and the tolerance to reflected power is -15dB, then 6 - 15 = -9, so reflected power as strong as -9dBm won't harm the transmitter?
Is it that easy or am I misunderstanding the definition?
-Adam
