Author Topic: Xilinx coolrunner startup current  (Read 4063 times)

0 Members and 1 Guest are viewing this topic.

Offline kubeek

  • Contributor
  • Posts: 24
  • Country: cz
Xilinx coolrunner startup current
« on: December 16, 2013, 08:42:04 pm »
Hi, I plan to use the coolrunner II CPLD in a very low power system, and I am not sure if the datasheet is not missing the startup current, as other companies state this number (could be beacuse their chips are flash based and the coolrunner seems to be eeprom based, but you never know).
Does anyone use any chip from this family, and could you confirm that the startup current is the same as the running current of the chip?
 

Offline Kohanbash

  • Regular Contributor
  • *
  • Posts: 175
  • Country: us
    • Robots for Roboticists
Re: Xilinx coolrunner startup current
« Reply #1 on: December 16, 2013, 09:00:23 pm »
It depends primarily on your clock rate and the number of pins that get set on startup.

Xilinx has power equations for their devices. The part of the inrush current you need to determine is the dynamic portion
http://www.xilinx.com/support/documentation/application_notes/xapp317.pdf

If you use xpower within ISE you can get a more accurate number

Robots for Roboticists Blog - http://robotsforroboticists.com/
 

Offline kubeek

  • Contributor
  • Posts: 24
  • Country: cz
Re: Xilinx coolrunner startup current
« Reply #2 on: December 16, 2013, 09:07:20 pm »
Well since the pins / gates that get set are being set only for a tiny fraction of time, shouldn't the startup / inrush current be almost negligible (or high,  but for a tiny amount of time which the supply caps should easily take care of)?

Or at least, the startup current should be less or comparable to the operating current when the chip is running?
I used the xpower from ise, but it states only quiescent and running current. It's the hunderds of miliamps that fpgas devour during configuration that I am worried about. The thing will be powered from a rfid card reader, so my total budget is around 2mA at 1.8V.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 21752
  • Country: nl
    • NCT Developments
Re: Xilinx coolrunner startup current
« Reply #3 on: December 16, 2013, 09:46:53 pm »
There should be some appnotes from Xilinx. Rush-in current used to be a major problem with Xilinx devices but the Coolrunner series was originally developed by Philips (now NXP) so things could be entirely differen. One way to make sure is getting a development board and monitor the current with a scope and a current shunt.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Kohanbash

  • Regular Contributor
  • *
  • Posts: 175
  • Country: us
    • Robots for Roboticists
Re: Xilinx coolrunner startup current
« Reply #4 on: December 17, 2013, 01:16:37 am »
Well since the pins / gates that get set are being set only for a tiny fraction of time, shouldn't the startup / inrush current be almost negligible (or high,  but for a tiny amount of time which the supply caps should easily take care of)?

Or at least, the startup current should be less or comparable to the operating current when the chip is running?
I used the xpower from ise, but it states only quiescent and running current. It's the hunderds of miliamps that fpgas devour during configuration that I am worried about. The thing will be powered from a rfid card reader, so my total budget is around 2mA at 1.8V.

I am not positive but yes it should be some small amount for some small amount of time.

The inrush current of a CPLD is typically less than an FPGA (I know there are many different sizes, etc... for each). One key reason for this  is that it is preconfigured (non-volatile) and does not need to be configured at startup like an FPGA (volatile).
Robots for Roboticists Blog - http://robotsforroboticists.com/
 

Offline marshallh

  • Supporter
  • ****
  • Posts: 1462
  • Country: us
    • retroactive
Re: Xilinx coolrunner startup current
« Reply #5 on: December 17, 2013, 01:39:34 am »
classical CPLDs have a very high power consumption rate, which is linearly proportional to the number of gates.
Newer CPLDs are just stripped down FPGAs and they do have startup inrush, they load bitstream from an onboard eeprom.

Coolrunner II belongs to the former type. If you want to improve power usage consider iCE40.
Verilog tips
BGA soldering intro

11:37 <@ktemkin> c4757p: marshall has transcended communications media
11:37 <@ktemkin> He speaks protocols directly.
 

Offline kubeek

  • Contributor
  • Posts: 24
  • Country: cz
Re: Xilinx coolrunner startup current
« Reply #6 on: December 17, 2013, 09:14:47 am »
classical CPLDs have a very high power consumption rate, which is linearly proportional to the number of gates.
Do you mean number of gates being flipped at a time, or number of gates on the die regardless of usage? I hope you mean  the first one, since my design is quite slow but needs lots of logic for a FSM, and that is how it looks from the datasheet.
 

Offline Kohanbash

  • Regular Contributor
  • *
  • Posts: 175
  • Country: us
    • Robots for Roboticists
Re: Xilinx coolrunner startup current
« Reply #7 on: December 17, 2013, 02:01:48 pm »
It is related to the number of gates used. If you look at the app sheet I posted earlier you can see the equation.
Robots for Roboticists Blog - http://robotsforroboticists.com/
 

Offline kubeek

  • Contributor
  • Posts: 24
  • Country: cz
Re: Xilinx coolrunner startup current
« Reply #8 on: December 17, 2013, 03:21:32 pm »
I read the equation beforeand know that this sort of relationship is true for almost all cmos devices, but stil I wanted to make sure of the meaning of that sentence.
Seems the only way to know for sure is to buy the chip and measure both insrush and dynamic current with the firmware loaded.
 

Offline lorth

  • Contributor
  • Posts: 46
  • Country: us
Re: Xilinx coolrunner startup current
« Reply #9 on: December 18, 2013, 02:26:01 pm »
The inrush current of a CPLD is typically less than an FPGA (I know there are many different sizes, etc... for each). One key reason for this  is that it is preconfigured (non-volatile) and does not need to be configured at startup like an FPGA (volatile).

Careful! as marshallh says, The CoolRunner IS, and should be considered despite Marketing, a Volatile device, since it is SRAM based. There is a non-volatile memory on-chip with the configuration, but the device needs to be programmed at start-up. As you can see in the XAPP388 (page 1):

"Real Digital technology combines a nonvolatile programming cell with a volatile one. Most  digital designers are familiar with shadow memories, where one memory array tracks another one [...]. With Real Digital, the nonvolatile cell is a version of EPROM that is electrically  programmable and the volatile cell is similar to an SRAM cell, but not organized with the same random architecture of an SRAM. [...]"

The only true non-volatile FPGA/CPLD that I know are the ones from MicroSemi, which are Flash based and not SRAM based. (Have you checked the Igloo nano family?)

You might want to check this: http://www.xilinx.com/support/documentation/application_notes/xapp389.pdf

Also, be careful with the power estimators... they do Estimate,... Get a uCurrent Gold (handy the new BW...) and a test the platform...

Quote
Do you mean number of gates being flipped at a time, or number of gates on the die regardless of usage? I hope you mean  the first one, since my design is quite slow but needs lots of logic for a FSM, and that is how it looks from the datasheet.
Both, you will have a dynamic and a static power. One does depend on switching frequency, the other we could say it depend with the number of gates.
« Last Edit: December 18, 2013, 02:34:54 pm by lorth »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf