FPGA Central

Hello ng,

what's the expected supply current of spartan-3 VCCAUX 2.5V ?

I saw in the datasheet, that it's quiescent current is in the range of
vccint.

Is it in general in the range of vccint ?

Does it depend mainly on the usage of the DCM'S ?

Is a 150mA regulator sufficient ?

Is an analog regulator preferred over a switching type ?

I want to use an X3S200 or X3S400 part.


Thanks for any info

MIKE

"M.Randelzhofer" <[email protected]> wrote in message
news:2n8safFtosabU1@uni-ber[email protected]..
> Hello ng,
>
> what's the expected supply current of spartan-3 VCCAUX 2.5V ?

The expected current on the VCCAUX supply depends on the specific device
used and the application. As described starting on page 15 of the following
document, the VCCAUX voltage input supplies the Digital Clock Managers
(DCMs), the dedicated configuration pins, the JTAG pins, the LVDS output
drivers, and forms an input reference voltage for some of the I/O standards.
http://www.xilinx.com/bvdocs/publications/ds099-4.pdf

The best method to determine power consumption is the Web Power Tool,
available online at the following link.
http://www.xilinx.com/power

This web-based calculator allows you to enter your expected operating
conditions. Based on the numbers of functions used, the device size, and
the operating frequency, the Web Power Tool provides an expected typical
power consumption for each supply.

An even more accurate caculator, called XPower, is built into the Xilinx ISE
software. However, XPower requires a completed FPGA design.

> I saw in the datasheet, that it's quiescent current is in the range of
> vccint.
>
> Is it in general in the range of vccint ?

At quiescent levels, VCCAUX and VCCINT are in the same general ballpark.
However, these are likely different in an operating application.

> Does it depend mainly on the usage of the DCM'S ?

Yes, the DCMs are powered by VCCAUX.

> Is a 150mA regulator sufficient ?

The answer depends on the type of application implemented in the FPGA. The
Web Power Tool can help you decide.

> Is an analog regulator preferred over a switching type ?

Either works. However, if you are using DCMs, especially in high
performance applications, be aware that the DCMs are powered by VCCAUX.
Noise on the VCCAUX supply adversely affects jitter on the DCM clock
outputs.
---------------------------------
Steven K. Knapp
Applications Manager, Xilinx Inc.
General Products Division
Spartan-3/II/IIE FPGAs
http://www.xilinx.com/spartan3
---------------------------------
Spartan-3: Make it Your ASIC

Hello Steve,

thanks for the information.
The power tool is a nice piece of SW, but fully unusable for my purposes.
I want to design a robust evaluation system.
Therefor i need a worst case maximum value for the currents for the
X3S200TQ144 and X3S400TQ144 devices.
It's not the exact value like 845,8765mA and 988,887mA i need for the given
voltages, but the general range.
I expect, there are special test patterns for your semiconductor end tests -
like all FF's toggling or shifting which needs a certain amount of current
on vccint or vccaux.
What's the average current in these 'all chip resources tests' at the
thermal limit ?

Ok, this question is rather weird because of the limit definitions, but i
hope you understand what i mean.
The Xilinx Spartan-2 Starter Kit (an excellent and low priced system) uses a
500mA regulator for the VCCAUX and a 1000mA regulator for VCCINT. Are these
ranges recommended from Xilinx, or a rough estimation of an experienced
designer ?

again thank you very much for your fast response and quality of information.

MIKE

Sorry, i mean the Spartan-3 Starter-Kit in the posting above.

MIKE

MIKE,
It's like asking the question how big a gas tank do I need on my sports car
to get 300 miles on a tank. It all depends on who's driving it and where. If
you really want worst case, you've gotta plug in the max toggle rate on all
the CLBs, FFs and BRAMs to the power consumption tool. But is anyone gonna
drive with their pedal to the metal constantly for an hour? Not even Michael
Schumacher does that.
Cheers, Syms.

"M.Randelzhofer" <[email protected]> wrote in message
news:[email protected]..
> Hello Steve,
>
> thanks for the information.
> The power tool is a nice piece of SW, but fully unusable for my purposes.
> I want to design a robust evaluation system.
> Therefor i need a worst case maximum value for the currents for the
> X3S200TQ144 and X3S400TQ144 devices.
> It's not the exact value like 845,8765mA and 988,887mA i need for the
given
> voltages, but the general range.
> I expect, there are special test patterns for your semiconductor end
tests -
> like all FF's toggling or shifting which needs a certain amount of current
> on vccint or vccaux.
> What's the average current in these 'all chip resources tests' at the
> thermal limit ?

The Web Power Tool will provide this information. If you believe the end
application will have all nodes switching, then you can model this in Web
Power Tool. However, most applications have a much smaller portion of the
device toggling.

However, you may be asking a different question. The performance of a
device is guaranteed up to a specific junction temperature--85C for the
commercial temperature range or 100C for the industrial.

Using the expected ambient temperature range, zero airflow, and the thermal
characteristics of the package, you can calculate the maximum possible power
that you can dissipate through the package and not violate the maximum
junction temperature.

For example, the Theta-JA of the XC3S200 in the TQ144 package is 34.1
degrees C per Watt. Assume that the ambient operating temperature is 25C.

25C + 34.1C/W (P) = 85C

Solving for P ...

P = 1.76 Watts maximum

> Ok, this question is rather weird because of the limit definitions, but i
> hope you understand what i mean.
> The Xilinx Spartan-2 Starter Kit (an excellent and low priced system) uses
a
> 500mA regulator for the VCCAUX and a 1000mA regulator for VCCINT. Are
these
> ranges recommended from Xilinx, or a rough estimation of an experienced
> designer ?

Although the Spartan-3 Starter Kit board has an XC3S200 in the FT256 ball
grid array package, the regulator is sized larger than actually required.
We used the Web Power Tool to generate a number of expected applications.
The board as presently defined meets the requirements of every XC3S200
design we could contemplate. It also meets most every XC3S400 design and
almost every XC3S1000 design, giving ourselves maximum flexibility with the
PCB design. Yes, you can create pathological designs that will break one or
more supply limits, but these require effort.

If you are interested, the schematics are available online.

Spartan-3 Starter Kit Board web site
http://www.xilinx.com/s3boards

Schematics (part of the Spartan-3 Starter Kit Board User Guide starting on
page 53)
http://www.xilinx.com/bvdocs/userguides/ug130.pdf

---------------------------------
Steven K. Knapp
Applications Manager, Xilinx Inc.
General Products Division
Spartan-3/II/IIE FPGAs
http://www.xilinx.com/spartan3
---------------------------------
Spartan-3: Make it Your ASIC