Hi all, I am new to SI and SigXplorer as well and I have to do simulations on a DDRII chip attached to a freescale MPC8313 placed on my board. I have been playing with differential pairs in SigXplorer and have seen that there are models for differential output buffers and differential input buffers, the problem comes out with the IBIS models supplied by the vendors.For example let's focus on the differential CLK line,the DDRII IBIS model comes with a differential input model but the MPC8313 IBIS model doesn´t have a differential output buffer for the clock, instead you have two single ended models each one for clk+ and clk-. From hardware point of view a differential signal consists of two pins having the same models, but opposite signal phase. And my question is, is there a way in SigXplorer to take two single ended signals and invert or change the phase of one signal respect from the other, is this the right way from a simulation point of view? In "Stimulus Edit" I cannot change the default stimulus signal and I don´t have a way to create a custom stimulus. I need to see the eye opening and reflections but cannot with two single ended models.
Thank's in advance
Not having the model I cannot say for sure it is incorrect. It depends on how you have added the part to SigXP. If you extracted the net from the .brd, then most likely the IBISDevice is not set up properly to understand that these pins are differential. You can tell if they are set up properly if you see Inverse pin information when editing the IBISDevice in the IBIS Device Model Editor.When adding buffers to the SigXP canvas manually, then you need to be sure to select IBISDevice when adding the buffers, This will allow you to select the exact pin you want, and if it is defined differentially, it will bring it in as a differential part. The stimulus is then handled with inverting outputs as expected.Once you have the model correctly, you can set up the simulation such that it runs a lot of pulses (since you don't have the Custom Stimulus capability) . Do this by setting the Measurement Cycle number (in the Analysis Preferences form) to whatever value you want. Then you will have the ability to view it in Eye Mode.
By now I have not extracted the net from the .brd as I am just exploring the models.
I have added the buffers manually to he SigXP canvas, as you mention I have selected the IBISDevice and then I have selected the exact pin that drives the clk+ but it doesn´t bring it as a differential part (as you can see in the attached pic), just bringed as a singled ended pin.
In the other hand the net in the .brd is set as differential with its ECsets configured, so if I extract it from the .brd, will it be correctly displayed in the SigXp canvas as a differential output buffer although the model have not a differential model?
As John said, go into the IBIS Device Model Editor. You get there through the Library editor, not Add Part. Then scroll down to the correct pin, click on it, and associate the differential pin. It must not have been defined in the 8313 IBIS.Donald
Thank's Donald, it seems to work now as a differential pair. It also gives a warning because there are no Logic Threshold Values on Diff Pair Data but that's another thing...
Or you can extract the spice model for BRD file, then you can run the hspice by combination of the bid's spice model with IBIS. You can expect the same results.