[c-nsp] SFP high power alarm

[Juergen Marenda]

On Tue, Aug 21, 2012 at 06:05:32PM +0200, Gert Doering wrote:
> Hi,
> 
> On Tue, Aug 21, 2012 at 12:26:50PM +0000, John Brown wrote:
> > Put a real optical power meter on the fibers and adjust with pads as
> > needed to get your levels within specs.
> 
> That's *RX* power.  Not TX power.  TX power is something that is measured
> inside the SFP - an the question "how can TX power go high" is a valid
> one.


>From an X2-Datasheet:

VII. DOM Parameters

Values Parameter    min. max. Unit

Transponder Temperature Monitor Accuracy 1) -5 +5 °C
Laser Bias Current Monitor Accuracy 2) -10 +10 %
Transmit Power Monitor Accuracy 3) -3 +3 dB
Receive Power Monitor Accuracy 3) -3 +3 dB

1) 0 to 70°C case temperature.
2) 0 to 12.5 mA.
3) -8.2 dBm to +0.5 dBm

... so that may be just a measurement error.

An other SFP+ shows +-2dB TX or RX Power Monitor Accurancy,
and +-10% TX BIAS Accurancy.

Did not find an explanation of how optical modules
determine optical output power, some kind of
handshake with the remote side , increasing output power
so that the remote side _sees_ sufficient light?

But *cleaning* the fibers and plugs is a very good idea(TM),
with (too much) power the reciever gets blind;
somtimes the dirt beetween fiber-fiber ore fiber-optics
get grilled/emailed onto the glas.

--
Juergen