| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
Vipul,
I don't think measuring nominal sensitivity is impossible. In fact, its the other way around. For qualifying any receiver, we take a golden Tx - basically a test equipment and put say a small fiber length into Rx and get the sensitivity - no ISI, no DJ.
For stressed sensitivity, I have asked the guys who make the phy chips and many of them don't even test stressed sensitivity -its a complicated setup with limiting amps, noise sources etc.
-----Original Message-----
From: Vipul Bhatt [mailto:vipul.bhatt@xxxxxxxxxxx]
Sent: Friday, October 12, 2001 11:41 AM
To: '802.3ae Serial'
Subject: RE: [802.3ae_Serial] Consistent treatment of sensitivities and
margins
Peter,
You raise a good point. It seems to me that we have a lose-both-ways
situation:
- Nominal sensitivity is impossible to measure because it requires
the test signal input to be free of ISI and DJ. Yet, outside of a
small group of Serial PMD experts, that's the specification most
people are comfortable with.
- Stressed sensitivity is a pain in the neck to measure, even though
the idea behind it is worthy of keeping. The test setup is complex,
time consuming and unreliable.
I think one solution is to break out of this either-or deadlock. One
possibility may be to identify a means to calculate nominal
sensitivity from known measurements of input signal's
characteristics. (Yes, measurement imperfection will remain a
problem, but that's a separate issue.) We can provide some kind of a
formula or a table that provides the required sensitivity as a
function of test signal's rise time, jitter and inner eye height.
It lets the user make a pass/fail decision on a receiver under test,
but only after the test signal input is characterized. The crafting
of such a table will take some work, but it's a one time effort.
Regards,
Vipul
vipul.bhatt@xxxxxxxxxxx
408-542-4113
==============================
<snip>
> So, if people think that this is not a problem,
> especially for the 1300 and
> 1550 nm PMDs, we should consider eliminating the stressed
> rece9iver
> sensitivity and making the nominal receiver sensitivity
> normative. The
> 850nm case has to be considered more carefully.
>
> Peter
<snip>