WEBVTT NOTE This file was exported by MacCaption version 7.0.13 to comply with the WebVTT specification dated March 27, 2017. 00:00:00.334 --> 00:00:03.737 align:center line:-1 position:50% size:65% We talked a bit about characterizing transmitters. 00:00:03.737 --> 00:00:08.842 align:center line:-1 position:50% size:56% Joel gave a whole talk on oscilloscopes which are the tool that we use these days. 00:00:08.842 --> 00:00:14.147 align:center line:-1 position:50% size:42% Usually it's eye-diagram based, measured with a scope. 00:00:14.147 --> 00:00:21.989 align:center line:-1 position:50% size:61% In the old days when I first started with BERTs about, a long time ago, like 17 years ago, 00:00:21.989 --> 00:00:29.396 align:center line:-1 position:50% size:72% BERT receivers in some cases were better than scopes at characterizing some transmit phenomenon, 00:00:29.396 --> 00:00:32.332 align:center line:-1 position:50% size:43% but those days are gone, scopes have gotten really good, 00:00:32.332 --> 00:00:40.040 align:center line:-1 position:50% size:65% so virtually every high-speed channel transmitter is characterized with a scope. 00:00:40.040 --> 00:00:41.775 align:center line:-1 position:50% size:27% Just relatively quick. 00:00:41.775 --> 00:00:46.413 align:center line:-1 position:50% size:49% It's intuitive, like Joel was telling you, it's time versus voltage. 00:00:46.413 --> 00:00:53.153 align:center line:-1 position:50% size:70% You can look at the single-valued waveform or you can overlay those and make an eye diagram. 00:00:53.153 --> 00:00:55.889 align:center line:-1 position:50% size:53% Often it involves testing against a mask, 00:00:55.889 --> 00:01:01.762 align:center line:-1 position:50% size:64% so you can understand that if you have nice margins around that mask, 00:01:01.762 --> 00:01:06.733 align:center line:-1 position:50% size:38% you have a good transmitter. 00:01:06.733 --> 00:01:10.804 align:center line:-1 position:50% size:45% Receivers, the test is BER-based. 00:01:10.804 --> 00:01:19.613 align:center line:-1 position:50% size:62% The figure of merit for a receiver is what kind of BER can you get in that channel? 00:01:19.613 --> 00:01:24.251 align:center line:-1 position:50% size:60% The output waveform of the receiver, so the receiver is going to make bit decisions 00:01:24.251 --> 00:01:29.723 align:center line:-1 position:50% size:56% and somehow transmit those bit decisions back to a measuring instrument. 00:01:29.723 --> 00:01:31.558 align:center line:-1 position:50% size:35% We'll talk more about that. 00:01:31.558 --> 00:01:36.964 align:center line:-1 position:50% size:49% The output waveform of the receiver gives you no idea whatsoever. 00:01:36.964 --> 00:01:42.169 align:center line:-1 position:50% size:46% You could have this beautiful eye where virtually every bit is in error, 00:01:42.169 --> 00:01:44.571 align:center line:-1 position:50% size:47% but it'll still look great on the scope. 00:01:44.571 --> 00:01:49.710 align:center line:-1 position:50% size:37% You'd see this pristine eye coming out of your receiver 00:01:49.710 --> 00:01:53.146 align:center line:-1 position:50% size:57% after the slicer has made the bit decisions. 00:01:53.146 --> 00:01:58.018 align:center line:-1 position:50% size:66% You'll have no idea unless you measure the BER. 00:01:58.018 --> 00:02:01.154 align:center line:-1 position:50% size:50% You have to measure that the correct decisions were made 00:02:01.154 --> 00:02:05.492 align:center line:-1 position:50% size:40% and that it will not only tell you if the decisions were made 00:02:05.492 --> 00:02:09.029 align:center line:-1 position:50% size:63% but how well your clock recovery is performing. 00:02:09.029 --> 00:02:13.200 align:center line:-1 position:50% size:56% As I've learned over the years, it's a lot less intuitive than transmitter testing. 00:02:13.200 --> 00:02:17.838 align:center line:-1 position:50% size:49% People have a gut feel, and they've gone through EE school 00:02:17.838 --> 00:02:21.308 align:center line:-1 position:50% size:64% and have been using scopes their whole career. 00:02:21.308 --> 00:02:28.281 align:center line:-1 position:50% size:59% There aren't that many BERT-based classes in EE school. 00:02:28.281 --> 00:02:32.419 align:center line:-1 position:50% size:59% We use BER instead of eye measurements. 00:02:32.419 --> 00:02:36.590 align:center line:-1 position:50% size:62% What we do is we send a known input pattern, 00:02:36.590 --> 00:02:41.128 align:center line:-1 position:50% size:37% so this could be PRBS or it could be a user pattern 00:02:41.128 --> 00:02:44.064 align:center line:-1 position:50% size:60% or a pattern that the industry has determined 00:02:44.064 --> 00:02:48.135 align:center line:-1 position:50% size:39% that's not a PRBS pattern but a memory-based pattern, 00:02:48.135 --> 00:02:50.637 align:center line:-1 position:50% size:44% but you know what the pattern is. 00:02:53.106 --> 00:02:56.510 align:center line:-1 position:50% size:36% You put it into the receiver, it makes the decisions, 00:02:56.510 --> 00:02:59.346 align:center line:-1 position:50% size:54% and it somehow outputs those decisions 00:02:59.346 --> 00:03:05.285 align:center line:-1 position:50% size:62% and you see that the decisions that are output, there are a couple differences here. 00:03:05.285 --> 00:03:12.159 align:center line:-1 position:50% size:50% That guy was supposed to be a 1, and that guy was supposed to be a 1. 00:03:12.159 --> 00:03:16.496 align:center line:-1 position:50% size:67% Anyway, you see that there were some differences between the output pattern and input pattern. 00:03:16.496 --> 00:03:19.399 align:center line:-1 position:50% size:40% Those we record as bit errors. 00:03:19.399 --> 00:03:22.803 align:center line:-1 position:50% size:19% What is BER? 00:03:22.803 --> 00:03:33.713 align:center line:-1 position:50% size:61% You simply ratio the bits in error over the total bits transmitted, and that's BER. 00:03:33.713 --> 00:03:36.983 align:center line:-1 position:50% size:47% Number of bits received in error over the number of bits transferred.