I have a bunch of LVDS signals on a PCB I want to look at with a scope (without wanting to tear my hair out). They range in speed from hundreds of Hz to ~200 MHz. How can I do this in a way that will be simple to interface with a scope and can be appropriately terminated such that the signal integrity will not be compromised?
\$\begingroup\$
\$\endgroup\$
6
-
2\$\begingroup\$ With the scope probes? \$\endgroup\$– Eugene Sh.Commented Jul 26, 2017 at 14:42
-
1\$\begingroup\$ Are you at the design stage or are you looking at an already finished PCB? \$\endgroup\$– Jack BCommented Jul 26, 2017 at 14:44
-
\$\begingroup\$ connecting to what on the board? Just across the differential signal itself? I need some sort of test point or connector to interface with, no? \$\endgroup\$– scubaCommented Jul 26, 2017 at 14:44
-
\$\begingroup\$ Designing stage \$\endgroup\$– scubaCommented Jul 26, 2017 at 14:45
-
1\$\begingroup\$ Add test points with impedance matched Divider for differential probe or two probes with tips removed and 4 posts for tip/ring probing of two calibrated probes to f limit with top and gnd removed or add coaxial jacks from 50 ohm impedance matched and terminated coax to diff scope +B inverted. Make sure layout uses impedance control with gnd in multilayer matched paths with gnd guarding for minimal EMI, crosstalk and CM noise. Electrical testing on fabrication by supplier is advised to control dielectric capacitance poor tolerance so fab will compensate Z controlled tracks. It's called DFT \$\endgroup\$– D.A.S.Commented Jul 26, 2017 at 15:15
|
Show 1 more comment
1 Answer
\$\begingroup\$
\$\endgroup\$
2
The standard way used in my lab to probe high speed signals is a 1k resistor to the target, directly feeding a 50 ohm coax, grounded on the board close to the signal, into a 50 ohm scope input, optionally with a DC block capacitor in series if the DC loading cannot be tolerated
The rationale behind this
- a 1k resistor has much lower loading capacitance than any inexpensive scope probe (you can buy really nice high frequency low capacity probes at a price)
- any high speed line will likely be 50ohms or a similar low impedance, which will not be troubled by 1k loading
- if you are working on high speed signals, then your scope has 50 ohm inputs
- a 1k resistor per line is a low price to pay (in area as well as cost) for the ability to probe a finished and working board
The coax is terminated by the scope in its characteristic impedance, so presents a 50ohm load to the 1k resistor, resulting in a nominal 26dB signal loss flat from DC to several GHz, depending on the size of the resistor and the tightness of the grounding.
With LVDS, you can probe just one of the lines and assume the other, if you're debugging software, or probe both on two scope channels to make sure the sending hardware is behaving as it should.
-
1\$\begingroup\$ Seems like a nice compromise to avoid $$$diff-probe. Would you add these 1k resistors to a production board? Or only add tombstoned 1k resistors for testing a prototype? \$\endgroup\$ Commented Jul 26, 2017 at 16:13
-
3\$\begingroup\$ @glen_geek The last 6GHz design I worked on had footprints for 0402 1k resistors and U.FL SMD coax sockets on the layout, but they were only populated on the prototypes and trial batch. When they'd been added to a production board to probe, we only bothered to remove the resistor, left the socket in place. There were still places we'd thought we wouldn't need to probe, so tombstoning was used as well. \$\endgroup\$– Neil_UKCommented Jul 26, 2017 at 16:37