2

I use the Magewell Eco Capture Quad SDI M.2 (a small form factor quad-channel) capture device which keeps overheating after a few minutes (up to 100 C). I've read that this might be due to cameras not being connected and SDI inputs unused. Since I have only a single SDI camera, I was thinking if I can try terminating inputs with BNC 50 or 75 Ohm caps.

Is this a common practice for SDI devices?

Random
  • 121
  • 3

1 Answers1

1

The SDI input will already have a 75 ohm to ground and no connection or signal should not affect device power consumption.

Given the board has a significant heatsink yet only rated to draw about 8 W, something appears to have gone wrong during installation.
enter image description here ESD handling precautions must be adhered to. Consult with distributor for an RMA.

Power consumption should not increase with no input signal connection.

The SDI input should match TI's standard circuit design.

enter image description here

enter image description here

There is a latchup effect of shorting out the CMOS chip if a transient occurs outside the supply range. It should be only a temporary fault if no burning smell or smoke occurs. ;) *See above caution during installation.*

Tony Stewart EE75
  • 1
  • 3
  • 54
  • 182
  • "Given the board has a significant heatsink yet only rated to draw about 8 W, something appears to have gone wrong during installation." Can you expand? I'm consulting with Magewell, they get 80C with a single input on an actively cooled card. I'm also thinking about M.2 adapter I'm using - the only PCIe to M.2 M-key adapters I found are for NVMe SSDs and it looks like their design might vary – Random Feb 21 '22 at 02:53
  • Is adequate active cooling being used in your case? – Tony Stewart EE75 Feb 21 '22 at 03:21
  • At this stage, I'm trying to figure out with the vendor what is the definition of an adequate working temperature. There was no guidance on using active cooling till their last support ticket response so I need to revisit it. – Random Feb 21 '22 at 04:19
  • 1
    The core temp should be < 85'C and I recall reading it was stated to as 79'C core temp without any conditions as if it didn't matter for input signals.. But there must be either convection air flow or active air flow which can make a huge difference in heatsink temp rise especially if 1m/s air velocity thru the sink. It is poorly defined and reliability depends on minimizing temp rise. For CPU core, I would expect < 75'C core temp in order to minimize rise in slew rate of signals which I would expect to be more critical. – Tony Stewart EE75 Feb 21 '22 at 05:57
  • Thanks, I'll work on the active cooling then – Random Feb 21 '22 at 06:09
  • I meant minimize rise of slew time of CPU clock edges or "reduction of slew rate".. CMOS runs slower with rising temp due to increased RdsOn into a fixed capacitance. – Tony Stewart EE75 Feb 21 '22 at 14:32