Home made SOLT for connector evaluation
时间:04-05
整理:3721RD
点击:
Hi all,
I'm just about to design a "connector evaluation card" to correlate some model-based s-parameters for a connector pair we're going to use for ~8Gb/s transmission (so I'll want data that's good to maybe 20GHz...the extent of my N5230A). What I want to have at the end of the process are s-parameter data for the mated connector pair only, which I can incorporate into a larger model of the system.
My intention is to make up a set of little cards with each half of the connector pair where the signal nets are brought out to end launch SMAs. I then plug and play and get s-parameter data for each mated pair PLUS THE ADDITIONAL END LAUNCH AND TRACE.
In order to de-embed the s-parameters to the connector pin, I could either mathematically subtract a simulated model of the traces in my RF sim tool, OR use the de-embed function on the PNA, OR construct additional SOLT can standards into my design and cal the VNA to them before making the tests. I quite fancy the last option, but I'm worried about how good my home-brew cal standards will be (specifically the "open").
Any advice, or (more usefully) actual experience in doing this?
Many thanks in advance...
David
I'm just about to design a "connector evaluation card" to correlate some model-based s-parameters for a connector pair we're going to use for ~8Gb/s transmission (so I'll want data that's good to maybe 20GHz...the extent of my N5230A). What I want to have at the end of the process are s-parameter data for the mated connector pair only, which I can incorporate into a larger model of the system.
My intention is to make up a set of little cards with each half of the connector pair where the signal nets are brought out to end launch SMAs. I then plug and play and get s-parameter data for each mated pair PLUS THE ADDITIONAL END LAUNCH AND TRACE.
In order to de-embed the s-parameters to the connector pin, I could either mathematically subtract a simulated model of the traces in my RF sim tool, OR use the de-embed function on the PNA, OR construct additional SOLT can standards into my design and cal the VNA to them before making the tests. I quite fancy the last option, but I'm worried about how good my home-brew cal standards will be (specifically the "open").
Any advice, or (more usefully) actual experience in doing this?
Many thanks in advance...
David