XC9500XL keeper ?

[Jon Elson]

I just migrated a project from the XCV9500 to the XC9500XL, and am having
some problems with keepers on the used inputs. Looking at Xilinx' docs
on the 9500XL series, it looks like the keeper may be left on ALL
the inputs, not just UNUSED pins. You seem to have a choice between
a keeper and a pulldown (GND) option. Is there a way to turn off the
keeper on selected inputs? (Seems like the answer is no, but just
checking.) I have a reset pushbutton with a debounce RC and a
crystal oscillator that need to be tweaked to work OK with the keeper
in the circuit. I'm using web pack 10.1 in Linux, if that makes a
difference.

Thanks for any suggestions!

Jon

 

[Gabor]

On Friday, July 6, 2012 7:12:23 PM UTC-4, Jon Elson wrote:

I just migrated a project from the XCV9500 to the XC9500XL, and am having
some problems with keepers on the used inputs. Looking at Xilinx' docs
on the 9500XL series, it looks like the keeper may be left on ALL
the inputs, not just UNUSED pins. You seem to have a choice between
a keeper and a pulldown (GND) option. Is there a way to turn off the
keeper on selected inputs? (Seems like the answer is no, but just
checking.) I have a reset pushbutton with a debounce RC and a
crystal oscillator that need to be tweaked to work OK with the keeper
in the circuit. I'm using web pack 10.1 in Linux, if that makes a
difference.

Thanks for any suggestions!

Jon

It was my understanding that the Keeper vs Pulldown is a global
option, but between that and tristate can be selected pin by pin.
Used input pins normally have to have a keeper or pulldown constraint
in the .ucf file. With neither constraint the input should be high
impedance. You will get an error if you try to have keeper and
pulldown in the same project, but you should be able to have
keepers on just selected inputs. If you need to make individual
settings to unused pins, you need to add the pins to the design.

-- Gabor

 

[Jon Elson]

Gabor wrote:

It was my understanding that the Keeper vs Pulldown is a global
option, but between that and tristate can be selected pin by pin.
Used input pins normally have to have a keeper or pulldown constraint
in the .ucf file. With neither constraint the input should be high
impedance. You will get an error if you try to have keeper and
pulldown in the same project, but you should be able to have
keepers on just selected inputs. If you need to make individual
settings to unused pins, you need to add the pins to the design.
Well, that's the quirk! As far as I can tell, it has a keeper on

all USED inputs, and I am not specifying that - I think.
Digital inputs are not bothered by the keeper, but it is interfering
with the input pin of the clock oscillator and the reset RC
circuit. I seem to have the impedances of these circuits pulled down
low enough now that it is working, but I worry I may have reduced
the margin on those circuits. What I'd like would be a "no-keeper"
option in the UCF file, but I don't see any such thing.

Thanks,

Jon

 

[Michael Karas]

[This followup was posted to comp.arch.fpga and a copy was sent to the
cited author.]

In article <dN2dnSI6rMgiXGrSnZ2dnUVZ_jednZ2d@giganews.com>, elson@pico-
systems.com says...

Gabor wrote:



It was my understanding that the Keeper vs Pulldown is a global
option, but between that and tristate can be selected pin by pin.
Used input pins normally have to have a keeper or pulldown constraint
in the .ucf file. With neither constraint the input should be high
impedance. You will get an error if you try to have keeper and
pulldown in the same project, but you should be able to have
keepers on just selected inputs. If you need to make individual
settings to unused pins, you need to add the pins to the design.
Well, that's the quirk! As far as I can tell, it has a keeper on
all USED inputs, and I am not specifying that - I think.
Digital inputs are not bothered by the keeper, but it is interfering
with the input pin of the clock oscillator and the reset RC
circuit. I seem to have the impedances of these circuits pulled down
low enough now that it is working, but I worry I may have reduced
the margin on those circuits. What I'd like would be a "no-keeper"
option in the UCF file, but I don't see any such thing.

Thanks,

Jon

Maybe the message here is reset supervisor chip and oscillator.

--

Michael Karas
Carousel Design Solutions
http://www.carousel-design.com

 

[jt_eaton]

Maybe the message here is reset supervisor chip and oscillator.

I Agree.

You really have to be careful when your digital IC vendor tells you tha
the pads have "pullups and pulldowns". They don't, they have curren
sources and current sinks.

The PC board guys love to have them because they think that they don't hav
to put real ones on the PCA but if you look at the current spec they ar
usually speced to +100 -50 %. If you try to put any sort of analo
filtering on the pad then this range will make it hard to add an
hi-impedance circuitry.

If you can't disable them then you must us an external buffer

John Eaton


---------------------------------------
Posted through http://www.FPGARelated.com

 

[jt_eaton]

RC circuits on reset lines are very bad ideas anyway. They may work
well if the circuit if off for some amount of time, but if you get a
momentary glitch they can fail to reset at all leaving the circuit in
an undefined state because the power glitch disrupted the circuitry.

Rick

With todays chips there is also the problem with separate pad and core
voltages. The RC is on the pad supply and may create a reset that is long
gone before the core voltage can ramp up.

You have to have a voltage supervisor on ALL of the supplies.


John Eaton



---------------------------------------
Posted through http://www.FPGARelated.com

 

[rickman]

On Jul 8, 10:49 am, "jt_eaton"
<z3qmtr45@n_o_s_p_a_m.n_o_s_p_a_m.gmail.com> wrote:

Maybe the message here is reset supervisor chip and oscillator.

I Agree.

You really have to be careful when your digital IC vendor tells you that
the pads have "pullups and pulldowns". They don't, they have current
sources and current sinks.

The PC board guys love to have them because they think that they don't have
to put real ones on the PCA but if you look at the current spec they  are
usually speced to +100 -50 %.  If you try to put any sort of analog
filtering on the pad then this range will make it hard to add any
hi-impedance circuitry.

If you can't disable them then you must us an external buffer

John Eaton

RC circuits on reset lines are very bad ideas anyway. They may work
well if the circuit if off for some amount of time, but if you get a
momentary glitch they can fail to reset at all leaving the circuit in
an undefined state because the power glitch disrupted the circuitry.

Rick