Hello everybody,
We are now going in purchase of one of 9000 or 90000 Oscilloscopes, and I wolud like to ask communicaty about High speed serial decoding feature in this oscilloscopes N5384A or E2688.
Our problem is that we would like to use this feature to investigate our burst mode transfer over fiber optic.
Our link has the speed is 640mbit, and the problem is that frames are coming in bursts. E.g. there are long idle times on line, without any edge transition. To recover the clock, the burst has 140 0-1 transitions and then specific align character. Afterwards all data is 8b/10b coded. the frame is up to 20bytes in size
So my question is - is it possible to trigger on such stream and decode it with High speed software? I read in datasheet, that " When you choose PLL clock recovery, the clock recovery algorithm requires some time at the start of each record to lock to the data. This interval cannot be viewed or analyzed". I actually would like to see my 140 transitions in front of the stream on the screen. Is that possible?
Another question - if I know my frequency and choose constant-frequency CDR, will Oscilloscope be able to synchronize from first several edges and search for alignment character?
I know that my description is a little bit confusing, so I'm ready for specific questions.
Thank you.
We are now going in purchase of one of 9000 or 90000 Oscilloscopes, and I wolud like to ask communicaty about High speed serial decoding feature in this oscilloscopes N5384A or E2688.
Our problem is that we would like to use this feature to investigate our burst mode transfer over fiber optic.
Our link has the speed is 640mbit, and the problem is that frames are coming in bursts. E.g. there are long idle times on line, without any edge transition. To recover the clock, the burst has 140 0-1 transitions and then specific align character. Afterwards all data is 8b/10b coded. the frame is up to 20bytes in size
So my question is - is it possible to trigger on such stream and decode it with High speed software? I read in datasheet, that " When you choose PLL clock recovery, the clock recovery algorithm requires some time at the start of each record to lock to the data. This interval cannot be viewed or analyzed". I actually would like to see my 140 transitions in front of the stream on the screen. Is that possible?
Another question - if I know my frequency and choose constant-frequency CDR, will Oscilloscope be able to synchronize from first several edges and search for alignment character?
I know that my description is a little bit confusing, so I'm ready for specific questions.
Thank you.
> Our link has the speed is 640mbit, and the problem is that frames are coming in bursts. E.g. there are long idle times on line, without any edge transition. To recover the clock, the burst has 140 0-1 transitions and then specific align character. Afterwards all data is 8b/10b coded. the frame is up to 20bytes in size
The SW uses the standard 8B/10B Sync characters to get synchronized, I think it's the K28.5. If you are not using a standard sync character, the SW won't be able to sync, and the 8B/10B decode won't work.
> So my question is - is it possible to trigger on such stream and decode it with High speed software?
You will probably be able to use the standard edge trigger, combined with a holdoff. It depends on what the signal does between bursts. You don't say if this is differential or single-ended, so it's not possible to tell. You won't be able to use the serial trigger capability, because that depends on getting the clock recovery right.
> I read in datasheet, that " When you choose PLL clock recovery, the clock recovery algorithm requires some time at the start of each record to lock to the data. This interval cannot be viewed or analyzed". I actually would like to see my 140 transitions in front of the stream on the screen. Is that possible?
Whether you can see it on screen is dependent on getting the trigger right. It might take some "messing around". It's a short enough packet that you might be able to use segmented mode and capture a bunch of these packets in a single run. From what you've said, there appear to be about 350 bits in the complete packet (140 statup, 10 sync, up to 200 data). Sampling at 20G, that's just over 10K samples per packet, so with 20M depth, you could capture ~1000 packets at once, the page through them.
> Another question - if I know my frequency and choose constant-frequency CDR, will Oscilloscope be able to synchronize from first several edges and search for alignment character?
Possibly. Constant Frequency clock recovery depends on finding a starting and ending edge. It looks at the time between them, and tries to figure out how many integer bit periods that represents. Then it builds a perfect clock between those 2 points. Since you are running at "only" 640Mb/sec, you should be OK.
> I know that my description is a little bit confusing, so I'm ready for specific questions.
No more or less confusing than other questions I've seen on this forum :-)
Let me know if this helps, or if you need more clarification.
Al