Date
1 - 7 of 7
Is it possible to get an eye diagram on a TDS784?
|
dnmeeks
InstaVu seems like the right place to start, but I don't know a way to trigger properly (on both edges). Has anyone done this on this vintage scope?
Thanks Dan
|
|
|
|
Mark Goldberg
You really should have active data if you want a worst case eye diagram.
toggle quoted messageShow quoted text
Intersymbol interference, jitter and line reflections will vary depending on the data. Assuming your data has 2 or more states, transitions at fixed intervals,is random and has some sort of preamble that is fixed, you can trigger on one edge in the preamble and then look at the eye later down the data where it is random and see the worst case with a long persistence setting. I don't know if this will work for your data. Hold off may help triggering where you want. Regards, Mark
On Fri, Aug 17, 2018 at 9:46 AM, dnmeeks <dan-meeks@...> wrote:
InstaVu seems like the right place to start, but I don't know a way to
|
|
|
|
EricJ
Yes it's definitely possible on TDS7xx. It's a bit better on the later versions (i.e. 7xxD vs 7xxA) because InstaVu was improved to include grading. Intensity is visible through color mapping. You want to trigger on only one waveform, then change horizontal position to 3-5 events further down the line and use your persistence adjustment to fine tune your eye.
toggle quoted messageShow quoted text
--Eric Sent from my Samsung Galaxy smartphone.
-------- Original message --------From: Mark Goldberg <marklgoldberg@...> Date: 8/17/18 12:28 PM (GMT-06:00) To: TekScopes@groups.io Subject: Re: [TekScopes] Is it possible to get an eye diagram on a TDS784?
You really should have active data if you want a worst case eye diagram. Intersymbol interference, jitter and line reflections will vary depending on the data. Assuming your data has 2 or more states, transitions at fixed intervals,is random and has some sort of preamble that is fixed, you can trigger on one edge in the preamble and then look at the eye later down the data where it is random and see the worst case with a long persistence setting. I don't know if this will work for your data. Hold off may help triggering where you want. Regards, Mark On Fri, Aug 17, 2018 at 9:46 AM, dnmeeks <dan-meeks@...> wrote: InstaVu seems like the right place to start, but I don't know a way to
|
|
|
|
Jim Ford
Usually you want to trigger on the clock. That way all the data transitions overlap and form the eye. Eye closure top to bottom indicates noise, ringing, and other amplitude anomalies. Eye closure side to side indicates jitter and other time-based anomalies. But you probably knew that.
toggle quoted messageShow quoted text
Sent from my Verizon, Samsung Galaxy smartphone
-------- Original message --------From: Mark Goldberg <marklgoldberg@...> Date: 8/17/18 10:28 AM (GMT-08:00) To: TekScopes@groups.io Subject: Re: [TekScopes] Is it possible to get an eye diagram on a TDS784?
You really should have active data if you want a worst case eye diagram. Intersymbol interference, jitter and line reflections will vary depending on the data. Assuming your data has 2 or more states, transitions at fixed intervals,is random and has some sort of preamble that is fixed, you can trigger on one edge in the preamble and then look at the eye later down the data where it is random and see the worst case with a long persistence setting. I don't know if this will work for your data. Hold off may help triggering where you want. Regards, Mark On Fri, Aug 17, 2018 at 9:46 AM, dnmeeks <dan-meeks@...> wrote: InstaVu seems like the right place to start, but I don't know a way to
|
|
|
|
Mark Goldberg
Add one assumption to my statement:
toggle quoted messageShow quoted text
Assuming your data is self clocking with no clock available, has 2 or more states, transitions at fixed intervals,is random and has some sort of preamble that is fixed, you can trigger on one edge in the preamble... I assumed a clock was not available, which was the case for most of the work I did, so the data was self clocking and synced to a preamble. Of course, if a clock is available, that is better to trigger on it and all transitions can overlap, but I did not have that luxury and wanted to show how it could be done even in the no clock case. If you have a sufficient persistence, you can still see the worst case. If there is a reflection in the line that is several bits away, it's effects won't be seen until later in the data. Regards, Mark
On Fri, Aug 17, 2018 at 11:53 AM, Jim Ford <james.ford@...> wrote:
Usually you want to trigger on the clock. That way all the data
|
|
|
|
Jim Ford
True, Mark.
toggle quoted messageShow quoted text
It has been several decades since I have created or observed an eye diagram, after all. I was thinking of how data patterns with long strings of ones or zeroes might not show the entire picture. And I discount RZ, PAM, and other non-NRZ data formats, since I haven't worked with them. Mainly frequency domain work these days, although for time-domain work I do prefer Tektronix products. Thanks for adding to the discussion. You have obviously had your "hands dirty" more recently than I have! Sent from my Verizon, Samsung Galaxy smartphone
-------- Original message --------From: Mark Goldberg <marklgoldberg@...> Date: 8/17/18 12:16 PM (GMT-08:00) To: TekScopes@groups.io Subject: Re: [TekScopes] Is it possible to get an eye diagram on a TDS784?
Add one assumption to my statement: Assuming your data is self clocking with no clock available, has 2 or more states, transitions at fixed intervals,is random and has some sort of preamble that is fixed, you can trigger on one edge in the preamble... I assumed a clock was not available, which was the case for most of the work I did, so the data was self clocking and synced to a preamble. Of course, if a clock is available, that is better to trigger on it and all transitions can overlap, but I did not have that luxury and wanted to show how it could be done even in the no clock case. If you have a sufficient persistence, you can still see the worst case. If there is a reflection in the line that is several bits away, it's effects won't be seen until later in the data. Regards, Mark On Fri, Aug 17, 2018 at 11:53 AM, Jim Ford <james.ford@...> wrote: Usually you want to trigger on the clock. That way all the data
|
|
|
|
Kevin Oconnor
Well an eye diagram is really just the envelope of all hi/lo & early/late levels & transitions in a data stream. Assuming that you can trigger off something relative to the stream, even old scopes with analog or digital persistence can do a decent job showing an eye.
My HP54503A monochrome 500Mhz has a trig/display mode that will do that. Of course some of the more fancy Tek/HP scope with color planes can colorized the eye with more interesting data. Which I had one of those sometimes! Kjo
|
|
|