Open jvoigts opened 8 years ago
There seem to be decent TVS diode packages fro LVDS lines - thes ones have ~2pF max. http://www.onsemi.com/pub_link/Collateral/ESDR7534-D.PDF
This one looks similar, at 3pF: http://www.semtech.com/images/datasheet/srv05-4.pdf
These ones look pretty decent too - littlefuse has a pretty wide range of these, the 3011 for instance covers 3 pairs per chip, the 3010 does 2. These claim <0.5pF: http://www.littelfuse.com/~/media/electronics/datasheets/tvs_diode_arrays/littelfuse_tvs_diode_array_sp3011_datasheet.pdf.pdf
What level of signal degradation can we accept? Also, we should think about the board space a bit - we have 4*21 LVDS lines to deal with.
Here are a bunch of offerings for LVDS:
http://www.ti.com/lsds/ti/interface/esd-emi-protection-products.page#p158=LVDS
ah yea, that looks great - the TPD8S009 looks good for instance
Yeah, ESD diodes for high speed differential lines are pretty much common nowadays, we just need a couple of multi-channel chips and we're set.
sounds good - should we go ahead woth this one for the schematic for now? Looks like the lowest capacitance one of the bunch, and comes in 6 channel packages. http://www.ti.com/lit/ds/symlink/tpd6e05u06.pdf
Any of these are fine. Go with the one that is good for lvds and most amenable to routing at reasonable pcb tolerances: 5mil trace space. 12 miles holes. At anything less than 1 pf, parasitic capacitance will be totally dominated by our layout, trust me.
Cool, i'll start putting something together later today if i get some downtime. So I take it you have seen my board layouts before..
The pcie breakout/DIO board needs to expose a lot of almost bare FPGA pins. Even tough these pins are somewhat protected by the VHDCI connector design, we should add the maximum level of input protection that ensures that we dont degrade the performance of the lines. This means that galvanic isolation is out, but protection diodes might work.