Maximum Current utilized by Spartan-3

I am trying to use a Spartan 3 in my design.

The power requirements and current requirement are going to vary
according to applications. According to a TI website for Sparatan 3,
the VCCINT (1.5v) has the current range from 300mA to 10A, and for
VCCO(3.3V) the current range is from 50mA to 3A. For VCCAUX(2.5V) the
current is maxed at 300mA.

Has anyone used such high currents when using a Spartan 3.

I will be implementing soft-cores and will be utilizing most of the
core. So I am expecting a heavy load on the FPGA. I just wanted to be
able to estimate the current that would be required, so I can choose
the correct Voltage Regulators. Many of the voltage regulator specify
the current limits as 1-2.5A, which may not meet the requirement for
upper limit of the current requirement for Spartan -3.

I did post a previous topic regarding chossing a voltage regulator,but
I wanted to be more specific on the issue at large.

Thanks for your help.

-Yaju

Yaju N wrote:
> I will be implementing soft-cores and will be utilizing most of the
> core. So I am expecting a heavy load on the FPGA. I just wanted to be

Which Spartan-3?  A -1500 part will consume a lot more current than the 
-200 part for designs which use "most of the core."

What clock frequency?  Check out the Xilinx power utility (online). 
What type of utilization?  SRL functionality consumes more power than 
standard LUT.

2.5v rail should be easier to estimate if you're only using it for 
VCCAUX.  (I think) it's only used for the DCMs and configuration logic.

Play with the power utility to determine ballpark figures or even upper 
bounds.  It'll give you a general feel for how the thing will be sucking 
down current.

    Jake

Yaju,

Use our Power Prediction web tool to esitmate your actual power.

The 'problem' with FPGAs, is that they will use whatever power you tell 
them to:  so we (and TI) have no idea how much power they will need to 
do your job (after all, it is your job to do, and we don't know what 
that is).

http://www.xilinx.com/cgi-bin/power_tool/power_Spartan3

Austin

Yaju N wrote:
> I am trying to use a Spartan 3 in my design.
> 
> The power requirements and current requirement are going to vary
> according to applications. According to a TI website for Sparatan 3,
> the VCCINT (1.5v) has the current range from 300mA to 10A, and for
> VCCO(3.3V) the current range is from 50mA to 3A. For VCCAUX(2.5V) the
> current is maxed at 300mA.
> 
> Has anyone used such high currents when using a Spartan 3.
> 
> I will be implementing soft-cores and will be utilizing most of the
> core. So I am expecting a heavy load on the FPGA. I just wanted to be
> able to estimate the current that would be required, so I can choose
> the correct Voltage Regulators. Many of the voltage regulator specify
> the current limits as 1-2.5A, which may not meet the requirement for
> upper limit of the current requirement for Spartan -3.
> 
> I did post a previous topic regarding chossing a voltage regulator,but
> I wanted to be more specific on the issue at large.
> 
> Thanks for your help.
> 
> -Yaju
> 

Yaju, remember, this is CMOS which (until recently) consumed no static
power at all. The size of the chip and the size of the design does not
matter, the power comes from clocking the logic. And the current or
power is proportional to the clock rate.

Things have become more complicated in 130 and 90 nm technology, where
we have a significant amount of leakage current and power that is
independent of utiliation and clock frequency, but increases very much
with higher teperature.

Except for the leakage current, the power is design-dependent, i.e.
your design!
Peter Alfke, Xilinx Applications

I am using a Spartan 3 the PQ 208 package. The maximum input current
available to the board is going to be 2.5A. I am just concerned whether
that will be good enough. I am planning on using a 50Mhz oscillator on
the board, similar to the one one the Spartan 3 demo board.

I am not sure what my application is going to be. I am currently
developing the board as a part of my Master's research. So I will be
experiementing with different designs on the board. I just wanted to
finalise the board so that the FPGA would be a ready to go product,
without having to worry about the power and current.

The online estimation tool requires some very specific information
about the prospective design or application which I may not be able to
accurately predict at this point. But I guess I will give it a try.

I guess what I would be looking for if I have missed an "CRITICAL"
current or power specs, which might render my board useless if I
implement some amazingly futuristic design or application.

Is power and current specification a very important issue at all, as I
feel I might be speding too much time on something trivial? I am
planning on just throwing in the new TPS7xxx buck regulator hoping it
will work fine. Other than that I guess my advisor suggests sticking to
the Linear Dropout Regulators as used in those Demo Board, since they
have been in use for a while and hopefully they work fine.

Yaju
y a j u at b y u edu

The easiest for you would be to "play" with an existing evaluation
board, where you can easily find out the power consumption of different
designs. Implement a long shift register with a toggling first
flip-flop ( relatively high power) or a counter  ( less power per bit).
You should be able to do a lot of logic with 2.5 A and 50 MHz...
Peter Alfke

Yaju N wrote:
> I am using a Spartan 3 the PQ 208 package. The maximum input current
> available to the board is going to be 2.5A. I am just concerned whether
> that will be good enough. I am planning on using a 50Mhz oscillator on
> the board, similar to the one one the Spartan 3 demo board.
> 
> I am not sure what my application is going to be. I am currently
> developing the board as a part of my Master's research. So I will be
> experiementing with different designs on the board. I just wanted to
> finalise the board so that the FPGA would be a ready to go product,
> without having to worry about the power and current.
> 
> The online estimation tool requires some very specific information
> about the prospective design or application which I may not be able to
> accurately predict at this point. But I guess I will give it a try.
> 
> I guess what I would be looking for if I have missed an "CRITICAL"
> current or power specs, which might render my board useless if I
> implement some amazingly futuristic design or application.
> 
> Is power and current specification a very important issue at all, as I
> feel I might be speding too much time on something trivial? I am
> planning on just throwing in the new TPS7xxx buck regulator hoping it
> will work fine. Other than that I guess my advisor suggests sticking to
> the Linear Dropout Regulators as used in those Demo Board, since they
> have been in use for a while and hopefully they work fine.
> 
> Yaju
> y a j u at b y u edu
> 


If this is just a one-off or very low volume device that doesn't require 
  a high-efficiency power source, why wouldn't you use an LDO?  They're 
easy!  Although switchers, etc, are getting easier and more reliable, 
they're more complicated -- just in component count alone.

Also-- what's wrong with off-the-shelf eval boards?

    Jake

>If this is just a one-off or very low volume device that doesn't require
>  a high-efficiency power source, why wouldn't you use an LDO?  They're
>easy!  Although switchers, etc, are getting easier and more reliable,
>they're more complicated -- just in component count alone.

Beware.  Modern LDO regulators have restrictions on the ESR of the
filter caps.  Too low or too high and they oscillate.

I'm far from a wizard on this topic.  But I got burned several years
ago so it's on my hot-list of things to check carefully and then still
be suspicious.

I think older non-LDO type linear regulators are easier to work with.
But they often don't go down to 1.2V.

-- 
The suespammers.org mail server is located in California.  So are all my
other mailboxes.  Please do not send unsolicited bulk e-mail or unsolicited
commercial e-mail to my suespammers.org address or any of my other addresses.
These are my opinions, not necessarily my employer's.  I hate spam.

On Sun, 27 Feb 2005 22:07:24 -0600, hmurray@suespammers.org (Hal Murray) wrote:

>>If this is just a one-off or very low volume device that doesn't require
>>  a high-efficiency power source, why wouldn't you use an LDO?  They're
>>easy!  Although switchers, etc, are getting easier and more reliable,
>>they're more complicated -- just in component count alone.
>
>Beware.  Modern LDO regulators have restrictions on the ESR of the
>filter caps.  Too low or too high and they oscillate.
>
>I'm far from a wizard on this topic.  But I got burned several years
>ago so it's on my hot-list of things to check carefully and then still
>be suspicious.
>
>I think older non-LDO type linear regulators are easier to work with.
>But they often don't go down to 1.2V.

Also, bear in mind that older devices may not have the load-transient response performance that
modern high-speed logic may need.  

hmurray@suespammers.org (Hal Murray) writes:
> Beware.  Modern LDO regulators have restrictions on the ESR of the
> filter caps.  Too low or too high and they oscillate.
[...]
> I think older non-LDO type linear regulators are easier to work with.
> But they often don't go down to 1.2V.

Even LDOs that go to 1.2V are fairly uncommon.  National has one, the
LP3881ES-1.2 ($2.58 from Digikey in quantity 100).  That seems quite
expensive, but it is well documented.  The typical application shows
4.7uF tantalums for the input and output filters, says that 4.7uF is the
recommended minimum for both the input and output filter capacitors, and
gives a range of output capacitor ESRs for which it is stable (dependent
on load current).  They point out that aluminum electrolytics have high
ESR below 10C, and often have their ESR specified only at low
frequencies.  Ceramic capacitors have too low an ESR for stability.

Someone else in this newsgroup brought to my attention the very
inexpensive Sharp 1.2V LDO, PQ012FZ ($0.66 from Digikey in quantity
100), but the data sheet doesn't describe the requirements for the
capacitance and ESR of the filter caps.  The test circuit shown in the
data sheet has a 0.33uF on the input, and a 100uF 50V on the output, but
no ESR or specific capacitor type is mentioned.

Eric