cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
gauthier_auvray
Visitor
Visitor
1,406 Views
Registered: ‎02-19-2020

Artix7 power

Hi,

We have designed a video Ethernet board (Gb Ethernet) based on XC7A35T and XC7A50T. The current design uses XC7A50T CSG324.

We are facing a power issue. With no bitsream loaded, the board consumes 0.9W. When we load the bitstream the board consumes 3.8W and the internal temperature reaches 88°C (measured with XADC).

All power rails are OK (3.3V/1.8V/1.0V/1.5V)

The design includes a DDR3 controller @ 333Mhz, a triple speed Ethernet MAC and custom logic.

Everything works correctly except that the board is getting very hot!

The FPGA utilization is about 60% and the power estimation is 1.115W with a jonction temperature of 30.3°C

We have already checked the schematics and we observe the same behaviour with board equiped with XC7A35T and XC7A50T.

Any idea?

 

Regards

 

0 Kudos
27 Replies
bruce_karaffa
Scholar
Scholar
1,396 Views
Registered: ‎06-21-2017

The power estimation tool makes assumptions about switching activity, airflow and a heatsink.  Did you provide switching estimates, possibly from a simulation to override the default switching estimates?  Does your board have the airflow and heatsink that the power estimation tool assumes?  Did you take into account that other components on the board will draw more power if the FPGA is active, specifically the DDR3, but probably other components too? 

0 Kudos
joancab
Advisor
Advisor
1,395 Views
Registered: ‎05-11-2015

 

The 324 package is very small, so it doesn't surprise me it gets to 80C with a Gig Ethernet. 10 GbE I suppose.

What surprises me is that the power estimator guessed you 1W. My take is that something is missing there. If the board is dense, other hot spots (dcdc converters, etc) may "help" keeping the FPGA hot, if you have a thermal camera you can easily check that.

Other things to double check are pull-ups, terminations, any differential signal meant to be connected via DC-blocking series capacitor, etc.

0 Kudos
gauthier_auvray
Visitor
Visitor
1,367 Views
Registered: ‎02-19-2020

Hi joancab,

Ethernet speed is 1Gbps only (not 10G).

For DDR3 and IODELAY we use two clocks, one at 333Mhz and one at 200Mhz.

These are LVDS clocks connected to 1.5V banks so they are AC coupled.

I have attached the related schematics.

clocks.PNG
0 Kudos
gauthier_auvray
Visitor
Visitor
1,343 Views
Registered: ‎02-19-2020

FPGA input clock are DIFF_SSTL15 without internal termination.

0 Kudos
joancab
Advisor
Advisor
1,317 Views
Registered: ‎05-11-2015

The above is fine, that won't dissipate that much.

What power estimator do you mean, the excel spreadsheet Xilinx provides where you fill the details or the result from Vivado after implementation? The latter should be quite accurate.

0 Kudos
gauthier_auvray
Visitor
Visitor
1,284 Views
Registered: ‎02-19-2020

Vivado after implementation.

With following settings: no airflow, no heatsink, 4pf load, smal board, 8 to 11 layers we get 1.124W

In the design most of the consumption comes from the PHY of the DDR3 controller.

0 Kudos
bruce_karaffa
Scholar
Scholar
1,253 Views
Registered: ‎06-21-2017

Does the power estimation tool have an accurate estimate of switching rates?  What else on the board uses power?  The DDR3 will.  Do you have any DC/DC converters?  If they are 60% efficient, how much heat are they dissipating?  Any linear regulators?  If your board is getting hot from other components, it won't dissipate as much heat from the FPGA.  @joancab 's advice is good.  Get a infrared thermometer or camera and check all components.  If you can't reduce current draw, you must find a way to get rid of some heat.  A heatsink, airflow, a thermal pad to the chassis, something like that.

0 Kudos
gauthier_auvray
Visitor
Visitor
1,227 Views
Registered: ‎02-19-2020

Hi,

The toggle rate is set to its default value 12.5.

I Will check with a thermal camera but the hotter chip is the FPGA.

When everything is in reset state (FPGA Ethernet phy ddr3...) the board consumption is about 2W (0.9W without bitstream loaded).

0 Kudos
joancab
Advisor
Advisor
1,205 Views
Registered: ‎05-11-2015

"When everything is in reset state (FPGA Ethernet phy ddr3...) the board consumption is about 2W (0.9W without bitstream loaded)"

That makes me suspect there is something wrong with the DDR

0 Kudos
gauthier_auvray
Visitor
Visitor
1,192 Views
Registered: ‎02-19-2020

I agree with you. Here is our configuration:

  • DDR3 x16 @ 333Mhz
  • VTT and VTTREF generated by a dedicated component from 5V
  • DDR3 signals are connected to bank 34 and 35 powered by 1.5V (DCDC)

I have attache the schematics of the DDR3.

ddr3.PNG
0 Kudos
joancab
Advisor
Advisor
1,180 Views
Registered: ‎05-11-2015

 

Could it be you have there the 'L' part for 1V35 and you are supplying it with 1V5?

If the DDR clamps the inputs, even if powered at 1V5, then there would be a voltage mismatch between the FPGA and the DDR that would inevitably be dissipated somewhere

0 Kudos
gauthier_auvray
Visitor
Visitor
1,175 Views
Registered: ‎02-19-2020

We already checked that. The DDR3 chip is the correct one (without L).

joancab
Advisor
Advisor
1,160 Views
Registered: ‎05-11-2015

No parts from grey suppliers, all reputable and trustable?

0 Kudos
gauthier_auvray
Visitor
Visitor
1,156 Views
Registered: ‎02-19-2020

Parts purchased at Mouser.

0 Kudos
joancab
Advisor
Advisor
1,144 Views
Registered: ‎05-11-2015

Is it just one off doing that? Do you have other boards to try? (to discard any assembly fault)

 

0 Kudos
gauthier_auvray
Visitor
Visitor
1,127 Views
Registered: ‎02-19-2020

We have 5 boards:

  • 2 with XC7A35T
  • 3 with XC7A50T

All have the same behaviour.

0 Kudos
joancab
Advisor
Advisor
1,118 Views
Registered: ‎05-11-2015

Are the DDR and/ or GbE PHY chips hot as well or just the FPGA?

Could you tweak the design to disable either the Ethernet and the DDR to see if one of these (or none or both) are the culprit?

0 Kudos
gauthier_auvray
Visitor
Visitor
1,110 Views
Registered: ‎02-19-2020

We have monitored temperature with a thermal camera, the FPGA is clearly the hotter.

We have another board based on the same design (FPGA + DDR3) but without Ethernet that seems to present the same behaviour.

This board has the same FPGA (XC7A35T and same DDR3 memory); the internal logic is different.

0 Kudos
joancab
Advisor
Advisor
1,097 Views
Registered: ‎05-11-2015

 

Is the ddr phy generated with MIG? I would double check it, typically pins are swapped when routing the board so the IP needs to change accordingly, could that have happened?

0 Kudos
gauthier_auvray
Visitor
Visitor
1,084 Views
Registered: ‎02-19-2020

Yes DDR3 controller is the MIG from Xilinx. We have checked the fpga design pinout versus board routing and all match.

joancab
Advisor
Advisor
1,069 Views
Registered: ‎05-11-2015

 

That Altium symbol and footprint, generated in-house? double checked? sometimes devices have variations with slightly differing pinouts...

0 Kudos
gauthier_auvray
Visitor
Visitor
1,054 Views
Registered: ‎02-19-2020

I have checked the DDR3 symbol and it is OK.

0 Kudos
bruce_karaffa
Scholar
Scholar
897 Views
Registered: ‎06-21-2017

Can you measure the current on the individual power rails and see which one is higher than expected?

0 Kudos
joancab
Advisor
Advisor
892 Views
Registered: ‎05-11-2015

That may require one of those magnetic probes that go over the tracks that not everyone has...

0 Kudos
gauthier_auvray
Visitor
Visitor
818 Views
Registered: ‎02-19-2020

Unfortunatly we don't have this type of equipment.

0 Kudos
gauthier_auvray
Visitor
Visitor
795 Views
Registered: ‎02-19-2020

I measured the consumption at different levels:

  • No bitstream loaded : 43mA@24VDC
  • Ethernet Phy unreset and DDR3 in reset (no logic in FPGA except reset circuitry) : 70mA@24VDC
  • DDR3 unreset and Phy in reset (no logic in FPGA except reset circuitry) : 42mA@24VDC
  • Phy and DDR3 unreset (no logic in FPGA except reset circuitry) : 71mA@24VDC
  • Full operating design: 161mA@24VDC

Evrything works correctly. On the board we use 3.3V (IO), 1.5V (DDR3), 1.0V (Core), 1.8V (VCCAUX)

0 Kudos
gauthier_auvray
Visitor
Visitor
748 Views
Registered: ‎02-19-2020

Hi,

If we remove the MIG controller and associated logic we gain about 1.5w.

DDR3 is 1Gbit x16 at 333MHz 1.5V

0 Kudos