Fixing Displayport to VGA adapter

Projects /  Repairs Original post date: 12/06/2016

I got the DisplayPort to VGA adapter from China today. It was DOA.  May be I always try to buy the cheapest ones and don't like pay 5X markups for local reseller.

Would you rather see all my blog page filled with boring unboxing?

It was cheap enough ($3.60 CAD) that I have decided to open it up and take a look inside.

There are 5 wires that are not connected. Ooops!

I was expecting insulation displacement connector because of the wire gauge (32awg?) and the way those plastic thingy there, but it only has solder pads. Aside from a few balls of solders left on the plastic, the soldering job on the connector side was actually good.

The chip is a "Realtek RTD2166", but I can't find any datasheet for it. I google for Displayport to VGA eval board and found a NXP document without a schematic. Their datasheet for PTN3356 part on the eval board has enough info to go on.

Basically, there are 4 high speed differential lanes on the Displayport and 1 aux pair. For a DisplayPort to VGA dongle, only 2 of the lanes are used and the Aux is used for sideband I2C over AUX that is used for monitor configuration. (They probably use the same cable for different dongles e.g. HDMI that would require the lanes.)

I probed all the connections for the cable and connector to figure the connections.

According to the wiki page, that dangling red wire (pin 13) should be connected to GND. That's what prevent the dongle to work. I didn't connect pin 14.

I untangled the wiring and resoldered them. I cut off the ends of the 2 unused pairs so that they don't short.  The insulation on those wires are low temperature ones, so don't set the soldering temperature too high.

The silk screen has pin numbers on it, but they do not match the pinout on wiki page. The differential pairs are pretty obvious however.

I applied a generous amount of hot glue around the soldered joints to act as relief strain.

Now it works.

I gambled a bit by ordering another dongle right away thinking I could get the same batch. After waiting for close to 3 months, it finally arrived - DOA again.

The PCB looks different and uses a different chip. There seems to be a switch mode supply to provide regulation. They don't bother to pretend to route the high speed lanes as differential pairs. :( The silk screen this time has the right pin numbering.

I melted the obligatory hot glue to have a look and broke a few connections in the process.

I cleaned off the mess, re-stripped and soldered the wires based on the silk screen pin numbers and my scribbled down notes.

I can't say I would agree with some of their original connections. No matter how I looked at it, the AUX CH (used for EEPROM) wasn't connected correctly as CONFIG1 was used as AUX CH(p). I connect the pin 13 (CONFIG1) to the pad for GND on pin 2. I moved the (only) ground connection to pin 19 right next to the DP_Power on pin 20 to minimize the loop to the 2 decoupling caps C4/C20. I left a bit of the foil shields on the ML_Lane pairs.

I got it working.