• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 32 subscribers
  • Views 3241 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

probe setup for eye diagram: differential or single-ended?

Mike Valvo
Intellectual 470 points

hello, 

My colleague is claiming that it is better to use 2 single-ended probes rather than using a differential probe when generating LVDS eye diagrams on an oscilloscope.  When using 2 single-ended probes, use the math function to subtract one signal from the other and it will give the "true" difference between the +/- pair at the receiver.  My colleague claims that differential probes are meant for probing signals with respect to a floating ground and is not meant for obtaining the difference between 2 signals. 

Can someone comment on this?  is there any truth to what he is saying?

  • 0
    TI__Mastermind 29635 points
    Hi Mike,

    I have tried both ways in the lab in the past, and I have actually preferred the differential probes over the single-ended probes where you use the "math" function to obtain the difference. When you measure LVDS signals, the nature of the signal's positive and negative pair imply that meaningful information is conveyed by measuring the difference between two signals. You should be able to accomplish this either by subtracting the signals on two single-ended probes (one probe for '+' to GND, another probe for '-' to GND) with the "math" function or by attaching a single differential probe to the channel and probing directly across '+' and '-' lines. So, from my experience, whether you use the single-ended approach or the differential probe, the result should be the same.

    I favor the differential probe for a few reasons:
    - I've found it convenient that with the differential probe, you do not need to provide a separate reference clip to GND like you would need for both single-ended probes. This makes probing easier, physically, around the board, and it also prevents from the off-chance that your GND reference clips are in fact not clipped to identical reference levels on the board.
    - I have also found the differential probe to be convenient for functions that can only be performed with a single channel rather than with the math function. For example, with one of the Tektronix DPOxxx high-speed real-time scopes, I am able to use the "Fast Acquisition" button to obtain a snapshot with good persistence for the channel I have attached the differential probe to. This cannot be done for signals obtained via the math function.

    One caveat I'd like to warn about before taking this measurements is that LVDS signaling only works when there is a 100-ohm resistor across the '+' and '-' LVDS signals. Therefore, if you are probing and there is not a 100-ohm resistor provided, either internally on an IC or externally on the board, be sure to add one before taking your LVDS measurement.

    Thanks,

    Michael
  • 0 in reply to Michael Lu (Santa Clara)
    Intellectual 470 points
    thanks Michael,
    Did you notice any difference in the eye diagrams when comparing differential vs single-ended? did they have a different shape, voltage level, skew, etc?

    regarding the 100-ohm resistor: yes, generally I will have that resistor. But, in one case, I will be trying capacitive coupling as mentioned in the LVDS user manual. 2 resistors set up a 1.2V bias. would the differential probe not work when probing the receiver inputs in that case?
  • 0 in reply to Mike Valvo
    TI__Mastermind 29635 points
    Hi Mike,

    Thanks for your response. I was out on vacation the past two weeks and am just getting caught back up on forum posts.

    In my opinion, I would say the only noticeable advantage of taking single-ended measurements and then using the math function, is that you can verify the 1.2 V bias (or other voltage bias) on each single-ended probe, whereas if you use a differential probe, you can only observe nonidealities in the voltage bias when the resulting waveform is not centered around 0 V. In terms of different shape, voltage level, skew, it is dependent on the capacitance of the probe. For high speed, you want a fairly low capacitance probe (only a 1-2 pF, if not less).

    In that one case you mentioned in your second question, if there is not a 100-ohm resistor between the + and - receiver inputs, you will measure some kind of AC signal when applying the differential probe, but the amplitude you measure will not be correct. Since you effectively have an open-circuit (high impedance) between the + and - input, I would expect the voltage swing measured by the differential probe to be approximately double what it should normally be if you had used a 100-ohm resistor to match the impedance of the 50-ohm media characteristic impedance.

    Thanks,

    Michael