Monday, June 19, 2017

LVDS to TTL Evaluation Board

Decided to find out what my options are for premanufactured evaluation boards for LVDS receivers so that I can verify the LVDS output of my FPGA board on a plain oscilloscope or logic analyzer.

I found several very nice boards, but none was quite right for my needs; at the time I was searching I was looking for for SMA connectors on both the inputs and outputs and having multiple channels.


I eventually settled on the fairly amazing 16-channel receiver board with the 100 psec slew spec that had headers for its I/O instead of SMA connectors. I decided to build it into a box with solder-cup BNC connectors to which I soldered twisted pair wires to go from the BNCs to the headers on the board. I was looking for an option for a panel-mount SMA connector that had flying leads that I could crimp header sockets onto but I never found one, and I didn't bother looking into solder-cup SMA connectors. There are lots of cheap SMA-BNC cable options to enable me to plug the SMA outputs from the FPGA board into my adapter box.

Here is the 16-channel board again: http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail&itemSeq=225947001&uq=636283012596540141
Here's the user's guide for the board: http://www.ti.com/lit/ug/sllu013/sllu013.pdf
Here's the datasheet for the chip: http://www.ti.com/lit/ds/symlink/sn65lvds386.pdf
I looked for the cheapest possible rear-mounted solder cup BNC connector, and found this one which is super nice for only $8. But multiply that by 48 for a full 16 channels in and out and it still becomes expensive: https://www.digikey.com/products/en?keywords=ACX1843-ND
As an alternative, here is the cheapest front mount BNC solder cup connector at only $3. Because this so far is my own money, I went with this one, coming up with the breakthrough that instead of having these connect to a full 32-row connector to the eval board, these BNCs would instead be wired to one-row connectors which I can plug into the header at whatever location I need; this allows me to feed the wires for this connector through the hole in the box from the front and slip the washer and nut on from the back. https://www.digikey.com/products/en?keywords=A32341-ND
Here is the one row, two column connector that I used for the interconnect from the header for each BNC:https://www.digikey.com/products/en?keywords=0050579002
When I first ordered the one-row connectors, I forgot to order the crimp pins. They're cheap. For no clear reason the manufacturer makes a dozen varieties, some more expensive than this one: https://www.digikey.com/product-detail/en/molex-llc/0016020086/WM2510CT-ND/467802
I looked at a lot of exciting box options, but finally settled on this dirt cheap Bud box to start with, largely because the result would be fully surrounded by metal which would give my prototype box improved noise resistance: https://www.digikey.com/products/en/boxes-enclosures-racks/boxes/594?k=AC-1418
With that box, you have to buy the bottom cover separately: https://www.digikey.com/products/en?keywords=BPA-1518
The evaluation board requires 3.3V. Fortunately it is easily possible to buy some 3.3V wall warts like this one: https://www.digikey.com/product-detail/en/cui-inc/SWI6-3.3-N-P5/102-3705-ND/5417822
This is the lovely 3-terminal chassis mount barrel connector that was used on the ASTM and which I also ordered for the Bud box: https://www.digikey.com/products/en?keywords=EJ502A
To connect the 3.3V to the board requires banana jacks. Fortunately there are dozens of (uninsulated) flat-topped banana plugs sold that you connect to using a ring terminal: https://www.digikey.com/products/en?keywords=108-0753-001




Some ideas that I may want to pursue later:

It is really easy to buy panels with 16 columns of RJ45 holes. Could these be used as is for BNC connectors? https://www.digikey.com/product-detail/en/assmann-wsw-components/A-PAN-16-MOD/AE10606-ND/2391722
https://www.digikey.com/product-detail/en/assmann-wsw-components/A-PAN-24-MOD/AE10607-ND/2391723