0
\$\begingroup\$

I have a SerDes communication that I am trying to perform the eye diagram to. I have an FPGA with generic IO's set up as LVDS SerDes and I will run them down to 300 MHz instead of my target frequency of 1 GHz. The loop back in the FPGA IO's is Tx+ to Rx+ and Tx- to Rx-, the 100 ohm termination is in the FPGA on die.

enter image description here

I will be using single ended probe and plan on shorting the grounds to use them as differential probes. My question is how do I connect the probes in this setup to get the most reliable eye diagram, would be as the image attached? Any tips to perform an eye diagram test?

enter image description here

\$\endgroup\$

1 Answer 1

1
\$\begingroup\$

Those probes are totally unsuitable for 300 MHz, especially if you actually want to measure things like eye diagram.

If you don't have access to proper low capacitance high frequency probes, then you can build 'a poor man's RF probe' very easily.

schematic

simulate this circuit – Schematic created using CircuitLab

If you want a roving probe, then it's physically convenient to mount the resistor and optional capacitor on the back of a BNC or SMA jack, so it can be used with any cable. Make them as physically short as possible, 0402 components would be my choice. Ground the probe to a close-by ground on the board. This probe will look essentially resistive to the circuit, giving you a small proportional loading, rather than the capacitive loading of the probes you've shown.

The resistor together with its load of 50 ohms (the 'scope seen through the cable) forms a roughly 20:1 attenuator. If needed to isolate the DC loading, the capacitor can be anywhere on the line. However, it's least trouble at the high impedance end where it's in a 1k ohm environment, rather than the 50 ohms anywhere else.

If you can, it's best to solder these components onto the board, not dangle probes around.

Pro Tip. What I used to do with my layouts was to put a few empty 0402 footprints and a grounded pad close to each line I suspected I might want to probe. Then when I needed to, I could just solder the components in (resistor first as shown to isolate the capacitance of its other pad and the capacitor pads) for a very robust, stable, and low capacitance pick-off.

\$\endgroup\$
2
  • \$\begingroup\$ I have P500-010 probes for 500MHz with 10x attenuation, 10Mohms and 12pF loading, would these be suitable? \$\endgroup\$
    – Shannon
    Jul 25, 2022 at 6:56
  • 1
    \$\begingroup\$ 12 pF in a 50 ohm environment (half of your termination resistor) = 600 ps time constant, or a -3dB bandwidth of 250 MHz. Think about the risetimes you will need / get even at 300 MHz, and then ask yourself whether that's adequate for eye diagrams. With care you'll get 1 pF from what I've posted. \$\endgroup\$
    – Neil_UK
    Jul 25, 2022 at 7:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.