Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
14 years ago

Pre-Configuration power consumption issue

Hi

I'm currently investigating on the power consumption of my EP3SL50F780 Startix III device to have starting points for optimization. As I don't have reliable toggle information about the design, I measure the temperature on the device surface, subtract the ambient temperature and divide the result by 10.6 K/W (value the power analyzer assumes for a device without heat sink in still air), which should give a good guess of the power consumption of the currently loaded configuration.

To have a reference I did the same measurement on the unconfigured yet powered device, and notice the temperature to stabilize at a difference of 25.7 degrees, which points to a consumption of roughly 2400mW.

I can find no input-output clash so far. Further more, I tried the design once with nIO_pullup setting to high and low with no significant difference. So I assume that the current is not going through the default pullups.

Any hint where to search or how to tackle this issue?

Regards

3 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    14 years ago

    What configuration mode are you using? Is the Stratix trying to load a configuration when you measure the temperature?

    Do you have any programming/JTAG pin floating when the device is unconfigured?

    Do you have other sources of temperature nearby, such as power supplies, that would increase the temperature near the FPGA?

    Even if you don't have any shunt resistors on the power rails, there is an indirect method of measuring the supply current, if the DC/DC converters you are using have an external inductor. You can measure the average voltage across the inductor (with a low-pass filter and a voltmeter for example) and divide it by the DC resistance. It isn't very accurate, but it should tell you if you have an abnormal current on one of the supplies.
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    14 years ago

    > What configuration mode are you using?

    I use the FPP mode (Msel=000) and configuration is not being loaded when measuring.

    > Do you have any programming/JTAG pin floating when the device is unconfigured?

    I have a JTAG test board (with USB blaster programming access) interfacing to a connector on my board. If this test board is not connected, TDI, TCK and TMS are floating. However, the current is not different in both modes.

    In meantime I found out that I have fixed TRST at 2.5V (high) where VCCPGM is at 3.3V, but I don't think this may cause such high current. Do you have different opinion?

    > Even if you don't have any shunt resistors on the power rails...

    Actually I've got 0-Ohm bridges after the inductors which I may lift off for current measurement. This may serve as a starting point to an experiment of lifting of one after another connected device to find the "guilty guy". Right now, I just hoped that there might be non- or minimal-invasive methods, as those boards have a 4-digit BOM...

    Anyway, which power consumption (core, I/O) would you prior to programming such a device (right now found no numbers in the DC parameters)?
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    14 years ago

    I would expect a near-null current on Vccio (except if you have some pins shorted to ground), a slight current on the core voltages (static currents) and near 0 current on the pll supplies.

    You can une the early power estimator with an empty configuration, I think it should give you a good idea of what to expect.