cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Highlighted
Explorer
Explorer
1,280 Views
Registered: ‎04-20-2010

Unintentional back-power Spartan-6 through IO?

Jump to solution

I'm having a situation where it seems the LX9 I'm using is being back-powered by LVDS33 outputs (HDMI out).  Is this possible?

 

I have an LX9 that is an HDMI source.  This is a bare minimum interface, so only the four LVDS33 pairs, their associated grounds, and the required 5V power to the source are connected.  I also drive the DONE pin and have an LED on it as an indicator that the FPGA is powered and configured.

 

When I power up, things work fine, I get video on the monitor, etc.  However, when I remove power from the FPGA board, the DONE LED goes to about half the dimness, and the picture on the monitor does not go blank, but rather starts to degrade over time.  If I unplug the HDMI connector the DONE LED goes out immediately and the monitor goes blank.  I can also leave the cable plugged in, and over time (about 20 seconds or so) the DONE LED will finally fade out and the monitor will go blank.

 

If I apply power to the FPGA in this weird state, it will not reconfigure.  Also, without the monitor connected the board powers on and off just fine.

 

The HDMI spec indicates that the sink it to have pull-ups on the LVDS lines, so there is certainly the possibility to have voltage there.

 

I have never seen a device back-powered via its IO before.  Is this a possibility on the LX9?

 

Thanks,

Matthew

 

0 Kudos
1 Solution

Accepted Solutions
Highlighted
Mentor
Mentor
1,620 Views
Registered: ‎02-24-2014

oh yes.   Every IO pin has a protection diode to the Vcco bus, and this produces a back channel into that bus with a .7V drop.  If your LX9 board derives the core voltage (Vccint) from the Vcco bus, then the core gets backpowered too (likely, from your discription).     There's really no way to prevent this backpowering, since the protection diodes are hardwired to the IO pads.

 

You might want to consider a power "good" circuit that asserts PGM_B low when power fails, forcing an FPGA reset.   This may or not prevent the backpower freeze symptom you are seeing.    Other boards with HDMI outputs use HDMI drivers as a buffer, which prevents this.

Don't forget to close a thread when possible by accepting a post as a solution.

View solution in original post

3 Replies
Highlighted
Mentor
Mentor
1,621 Views
Registered: ‎02-24-2014

oh yes.   Every IO pin has a protection diode to the Vcco bus, and this produces a back channel into that bus with a .7V drop.  If your LX9 board derives the core voltage (Vccint) from the Vcco bus, then the core gets backpowered too (likely, from your discription).     There's really no way to prevent this backpowering, since the protection diodes are hardwired to the IO pads.

 

You might want to consider a power "good" circuit that asserts PGM_B low when power fails, forcing an FPGA reset.   This may or not prevent the backpower freeze symptom you are seeing.    Other boards with HDMI outputs use HDMI drivers as a buffer, which prevents this.

Don't forget to close a thread when possible by accepting a post as a solution.

View solution in original post

Highlighted
Xilinx Employee
Xilinx Employee
1,247 Views
Registered: ‎11-03-2016
To me it makes a whole lot more sense that you're getting back-powered through your 50ohm pull-ups (present in TMDS, not in LVDS) to Vcco than through the protection diode.
0 Kudos
Explorer
Explorer
1,237 Views
Registered: ‎04-20-2010

Thanks for the info!  I have also now found some documentation from ON Semi and TI that explain this exact problem.  Hopefully one of their HDMI protection ICs will solve this problem since I don't have room for a full-on buffer.  I'll also check into the possibility of the power-good circuit.

 

My VCCO (3V3) and VCCINT (1V2) are generated separately by independent LDOs, each being supplied by the host computer's 5V supply.

 

@yannickl

Yes, I did mean TMDS not LVDS, thanks.  The pull-ups are not on my side (the source) they are on the sink (monitor) side, so I can't do anything about them; they are part of the HDMI spec as well if I'm reading the docs correctly.

0 Kudos