New ContributorQuartus and power domain
We actually one remaining pending issue with the Agilex chip that we can't get resolved and might need some help.
The basic description is that when comparing the power consumption of the Agilex 5 on our module, to the Agilex 5 on the dev kit, at same temperature, before any configuration, our Agilex 5 consumes about 0.5-1W more.
Besides this, all sub-systems seem to operate correctly (CPU, memory, FPGA, I/Os).
This additional power consumption is worrying though, and we would like to understand it before spinning the next revision.
Regarding the excess power on our board, it might also be a design issue, but we didn't find any hint yet despite multiple deep reviews and investigations.
We already spent several weeks looking at this in every way we could think about, so there is already quite a bit of things we tried out, such as:
- Measured the power rails, from the outside and from the inside as well, checking if there is any unexpected difference
- Measured temperature from the outside (thermal camera) as well as from the multiple internal sensors
- Reviewed power sequence, playing with various combinations to see if it made any difference
- Reviewed multiple times schematics against documentation and dev kits
The extra power is on the core rail.
Now please look at the attachement:
We collected side observations on an non-configured FPGA. Our board has a temperature-dependent power consumption. So we simply
switched the fan off for a short period and then switched it back on. The main power consumption is on the 0.8 V rail (U500 RAA210130).
A second interesting observation is the internal temperature of the Agilex: L0_0 is significantly hotter than the other three.
When the fan is switched back on, the power consumption returns to its initial value.
Very strange and unusual behavior. Could this somehow help set priorities and allow us to rank the most/least likely directions
for investigating the source of the elevated power consumption?
