VCCAUX too high on a Spartan 3 design

Using linear adjustable regulators for VCCINT (1.25v), VCCIO (3.3v), and VCCAUX (2.5v). VCCINT and VCCIO are dead on, but VCCAUX is 2.72v, 2.88v, and 2.92v on the 3 boards I grabbed and measured. All boards seem to work just fine.

The regulator output voltage is controlled by just 2 resistors. When I changed one of the resistors to lover the voltage a bit, VCCAUX did not change. This leads me to believe that VCCAUX is somehow being "back" powered from the Xilinx chip. These voltages are present like this before the Xilinx chip has been programmed. I have not removed the regulator to measure current yet. Another thought was to put a shotkey diode between the regulator output and the load to see if the Xilinx really is powering VCCAUX, but I thought I'd post and see if anyone else has come across this issue. Half the I/O banks are 2.5v and half are 3.3v if that makes any difference.

Thanks - Dan

Reply to
Dan K
Loading thread data ...

It sounds like you're driving one of the dedicated configuration pins with a

3V3 signal. They should all be 2V5.

There's a Xilinx app note about 3V3 configration which suggests series resistors to limit current through the ESD diodes, and provision of current-sink capability on the 2V5 supply!

Reply to
Andrew Holme

I would load the Vccaux pin with a 100 Ohm resistor to ground. That adds 25 mA to the regulated current. Normall, this should hardly change the voltage, but if the regulator is back-fed, there will be a big voltage drop. Parallel resistors to ground are nice and easy, and do not cuase any irrepairable tdamage. Peter Alfke

Reply to
Peter Alfke

You can also drop the values of the two SET resistors, then you do not need to change the PCB design.

-jg

Reply to
Jim Granville

I should have expressed this better: Use the extra pull-down resistor just as an experiment to see whether the problem really is reverse current going into the regulator, or is something else. Just temporarily, just attach the resistor lightly, by hand if possible... Peter Alfke

Reply to
Peter Alfke

Unfortunately increasing the power dissipation with sinking/sourcing LDO on VCCAUX or external loads can't be the final solution. If there is a leackage current path from VCCIO to VCCAUX that must be founded and supressed somehow. Using series schottky between VCCAUX and FPGA will drop the output voltage to VCCAUX - (0.3V...0.7V) depending on load current and diode type.

Vasile

Reply to
vasile

Try adding a "dump" resistor from Vccaux to ground, and see if that pulls the voltage down.

John

Reply to
John Larkin

Reply to
Peter Alfke

before

to

You should make sure that no VCCAUX relative pins are hooked to VCCIO without a resistor otherwise any fix can blow up the esd diodes. Then, if you really need to save a couple of milliwatts, power the VCCCORE regulator from VCCAUX, this will suck the backrush of current right into another power supply instead of into a resistor. It is no coincidence that 1.2v LDO regulators are designed to run on 2.5v.

Cheers Jon Pry

Reply to
jonpry

I'm curious if the original poster has solved the problem in the indicated way above. I have a design with Stratix II which has the same behaviour.

thx, Vasile

Reply to
vasile

Yes, that was the problem. I had the M0, M1, and M2 pins connected to +3.3 v. As soon as they were disconnected from 3.3v and connected to 2.5v the

2.5v power supply was right on.

Dan

Reply to
Dan K

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.