cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
jamesgabauer_og
Observer
Observer
1,074 Views
Registered: ‎01-29-2019

Scope/Waveform Eye Diagram

Hi,

I am using a GTP transceiver on an AC701 and I am trying to run eye scans through Matlab using a modified version of xapp1198.

I am able to collect the correct data and display the results for a statistical eye. However, I was wondering what the configuration for the GTP is that supports the scope/waveform eye diagram. It is mentioned throughout the documentation but I cannot seem to get a result so far.

Thank you

0 Kudos
6 Replies
eschidl
Xilinx Employee
Xilinx Employee
1,045 Views
Registered: ‎10-19-2011

Hi @jamesgabauer_og ,

we do not provide code for a waveform or scope display.
But several customers implemented something like it.
You would probably need to make use of the ARMed branch of the eye scan state machine and apply the trigger at the right point with using periodic data.

------------------------------------------------------------------------------
Don't forget to reply, give kudo and accept as solution
------------------------------------------------------------------------------
0 Kudos
jamesgabauer_og
Observer
Observer
1,031 Views
Registered: ‎01-29-2019

Hi,

Thank you for the quick response. Is there anywhere I can look to for direction on this, I have found the documentation to be a little confusing. I think one of the main things I have had trouble with is what rdata and sdata actually represent, are they segments of the bit stream? measures from an ADC? I think understanding exactly what those 40 bit values represent would help a lot.

From what you are saying it sounds like I would need to be implementing a test pattern to be sent over the interface for a scope view, is that right?

Thank you again for your help

James

0 Kudos
eschidl
Xilinx Employee
Xilinx Employee
1,016 Views
Registered: ‎10-19-2011

Hi @jamesgabauer_og ,

rdata and sdata are described at the bottom of page 151 and top of page152 in ug482.
Also look at the description for ES_ERRDET_EN on page 155.

So rdata always contains 80bit of consecutive data sampled with the data sampler at the time of trigger.
And sdata contains either the error information for these 80bits or the sampled data of the scan sampler at the same time.

You can only use one offset point at the same time. To scan a waveform you would need to repeat that measurement at different offsets over the same waveform segment. So you need to know at what time to trigger to catch the same sequence.

------------------------------------------------------------------------------
Don't forget to reply, give kudo and accept as solution
------------------------------------------------------------------------------
jamesgabauer_og
Observer
Observer
992 Views
Registered: ‎01-29-2019

Ok, that does make a lot of sense for rdata and sdata. I feel like I should have realised that when I read it.

I still don't fully understand how this is meant to work or why I would need t be using the ARMed state but I will keep looking into it.

Thank you again for your help

0 Kudos
jamesgabauer_og
Observer
Observer
955 Views
Registered: ‎01-29-2019

One last thing

UG482 suggests that I need to unmask a single bit of sdata and accumulate errors on the previous cycles data. I have not done this but I have gotten what resembles an eye diagram by getting sdata and rdata from a qualifier. I don't see how the error accumulation would help in this case because if errdet_en is disabled this won't actually be accumulating errors 

0 Kudos
clive
Visitor
Visitor
555 Views
Registered: ‎05-12-2020

At the risk of re-opening an old post, but I have exactly the same question for the GTY transceiver..

In UG578 for the GTY transceiver there are various comments about "scope mode" eyes in addition to the statistical/BER eye. These comments appear in a number of other places offering a tantalising carrot but I'm wondering if these are actually progressive mis-quotes from earlier documents (eg UG482 for GTP and earlier). The same pictures appear in many places but the earlier you look, the less "scope" claims are made. There appear to be no published examples. I'm starting to think this scope mode (at least as described in the GTY and inferred in the GTP) is just a combination of progressive mis-understandings, mis-quotes and wishful thinking.

To at least partly clarify the picture, would it be possible to confirm whether when ES_ERRDET_EN is FALSE the RX_EYESCAN_VS_NEG_DIR, RX_EYESCAN_VS_CODE (ES_VERT_OFFSET on the GTP) and RX_EYESCAN_VS_RANGE values have any impact on the digitised value in SDATA and hence whether scanning the vertical offset can change the counts in the "error" counter? Likewise for the horizontal offset ES_HORZ_OFFSET.

And then whether that functionality is consistent across the different transceivers?

If the answer to that is yes, then there is a real posibility of deriving something interesting and scope like, but if not, then the only path would seem to be reading the rdata and/or sdata with fixed data patterns. This might give a waveform picture, but I struggle to see how that would give anything like a "scope view".

0 Kudos