12-20-2013 12:25 PM
I have developed a PCB with two XC6VLX550T-2FFG1759E and to power up the FPGAs I will be using DC Power supply's not power regulators.
-The power supply voltages that I will be applying to the device are as follows:
1) +1.0V core
2) +2.5V Aux together with +2.5V bank power
3) +1.5V to power FPGA DDR3 Banks as well as 16 DDR3 Memory chips.
4) I also have a prom and gig Ethernet IC's connected to both FPGAs...
-My first question is: At each of the said power supplies what should I see statically for an impedance? I am measuring 2.0ohms at the +1.0V to ground for both FPGAs...
_ My second question is: When I power up the FPGA's what should I see for current draw for each of the said voltage inputs?
-Lastly, When powering down all the devices does it matter which devices I power down first?
12-20-2013 01:43 PM
What you see for a static impedance depends on the voltage applied by the multimeter. Usually digital meters use very low voltage and I would expect to see a significantly larger resistance measures on Vccint. Other voltages may show lower resistance due to a large number of termination components, but this depends on termination schemes. However generally speaking the multimeter should measure DC resistance using a voltage that doesn't turn on the diodes inside the electronics and thus should not show such a low resistance unless you have some sort of assembly problems. I have to admit that I haven't worked with parts as large as yours, though so it may be possible that the static current draw due to leakage is enough to show up as 2 ohms for two parts that size. If you have a spare part that isn't mounted on a board, you could find a Vccint and Gnd ball and ohm it out to see. These balls should all connect to others of the same supply within the package so the resistance should be close to what you'd measure on a PC board.
As for static current draw, you could look at the device data sheet for static power. This will be highly dependent on temperature and vary significantly from part to part, but should always be lower or equal to the datasheet numbers if you have not loaded a configuratino bitstream into the part. Note that the current draw may be slightly high right after power on while the device is initializing, however using a multimeter you're not likely to catch that higher current number before the initialization completes.
For power supply sequencing you should also check the device data sheets. Some FPGA's don't have any hard requirement for power up or down sequence, but the data sheet will list a prefered order to minimize current draw. Also note that in your system other parts may also have a power sequencing requirement.
If you want to play it safe, you could use a current-limited power supply using a current-limit setting that matches the expected static power dissipation of your parts. Then if the supply doesn't get all the way up to 1V I would suspect assembly problems. Note that if you bring up only Vccint, the part will not start to self-initialize because it will remain in its internal power-on reset state. So you should not need to use a higher current limit to get past the initialization for this preliminary test.
12-20-2013 02:03 PM
Thanks for the reply..There is definitely no shorts on the board and I have determined that the digital millimeter that I used is measuring about .6 volts which is .4 volts away from spec voltage to power up the core. So, I think the 2.0 ohms is probably OK...
12-20-2013 02:19 PM
How did you measure the voltage on the multimeter? .6V at 2 Ohms would imply 300 milliamps - a lot of current for a multimeter. If you measure the voltage on the probes when they are not connected to the 2 ohm load, then I would believe that they have 0.6 volts on them. You want to measure the voltage on the multimeter probes while they are connected to the board and measuring 2 ohms, or use a 2 ohm resistor if you have one handy.
In any case if you don't think you have a short, I'd still recommend starting with a current-limited supply to reduce the risk of damage to the parts or the board laminate itself.
12-20-2013 02:38 PM
I have two millimeters connected together one measuring voltage and the other measuring ohms...
I assumed that - but you need to also make sure that both meters are connected to the board (i.e. the one measuring ohms should be reading about 2 ohms) while you make the voltage measurement. Otherwise the voltage measurement is meaningless.
12-20-2013 02:59 PM
Well I assume 0V is not quite zero, but your volt-meter is not sensitive enough to read the actual value. In any case this implies that there is something wrong with the board causing the low reading, possibly a short or improperly connected net.