Newbie: Falling edge, what is the threshold? (Xilinx XC9572XL)

Hi,

I'm trying to find out when exactly the XC9572XL recognizes a falling edge.
It has a 3.3V supply, and the signal is a square wave between 0V and 5V
(this Xilinx is 5V input compatible). At what voltage would a falling edge
trigger?

I have looked at its datasheet but can't find anything that fits what I'm
seeing - I thought that maybe 0.8V was the falling edge threshold (since it
is the highest input low voltage), but from my testing it seems to happen a
*lot* earlier than that, maybe when the voltage drops to about 4.3V, but
that can't be correct I realize.

Apologies for what must seem such a basic question, but if anyone could
please clarify I would be extremely grateful, thank you.

James

Generally the point will lie between 0.8V(Vil) and 2.0V(Vih) which are the 
guaranteed limits. The XC9572XL has hysteresis on the I/O so level will be 
different on the way down to the on way up. Threshold will vary with 
temperature, voltage (and voltage noise) and with silicon batch so don't 
rely on anything other than guarantee limits.

John Adair
Enterpoint Ltd. - Pushing The PicoBlaze.
http://www.enterpoint.co.uk


"James A" <me@privacy.net> wrote in message 
news:qK0je.10223$yY4.9241@newsfe5-win.ntli.net...
> Hi,
>
> I'm trying to find out when exactly the XC9572XL recognizes a falling 
> edge.
> It has a 3.3V supply, and the signal is a square wave between 0V and 5V
> (this Xilinx is 5V input compatible). At what voltage would a falling edge
> trigger?
>
> I have looked at its datasheet but can't find anything that fits what I'm
> seeing - I thought that maybe 0.8V was the falling edge threshold (since 
> it
> is the highest input low voltage), but from my testing it seems to happen 
> a
> *lot* earlier than that, maybe when the voltage drops to about 4.3V, but
> that can't be correct I realize.
>
> Apologies for what must seem such a basic question, but if anyone could
> please clarify I would be extremely grateful, thank you.
>
> James
>
>
>
> 

Those 0.8V is the level, where it is guaranteed to detect a
low-level...

So every real device will switch somewhere in between "max. low" and
"min. high"

experience "normally" tells me (device independant):
   - take the supply voltage
   - divide it by 2
   - subtract a small portion 2 to 5%

=> this will be a reasonable switching level
  (hysteresis NOT taken into account)
   but you defentively can't rely on it !!

  The only thing you can rely on are those mentioned min/max values and
using
  a 'fast' transition  (whatever 'fast' means)

Jochen





Jochen

"John Adair" <removethisthenleavejea@replacewithcompanyname.co.uk> wrote in
message news:1116511092.18941.0@ersa.uk.clara.net...
> Generally the point will lie between 0.8V(Vil) and 2.0V(Vih) which are the
> guaranteed limits. The XC9572XL has hysteresis on the I/O so level will be
> different on the way down to the on way up. Threshold will vary with
> temperature, voltage (and voltage noise) and with silicon batch so don't
> rely on anything other than guarantee limits.
>
> John Adair
> Enterpoint Ltd. - Pushing The PicoBlaze.
> http://www.enterpoint.co.uk

Many thanks John.

- James

"Jochen" <JFrensch@HarmanBecker.com> wrote in message
news:1116511858.745274.202370@o13g2000cwo.googlegroups.com...
> Those 0.8V is the level, where it is guaranteed to detect a
> low-level...
>
> So every real device will switch somewhere in between "max. low" and
> "min. high"
>
> experience "normally" tells me (device independant):
>    - take the supply voltage
>    - divide it by 2
>    - subtract a small portion 2 to 5%
>
> => this will be a reasonable switching level
>   (hysteresis NOT taken into account)
>    but you defentively can't rely on it !!
>
>   The only thing you can rely on are those mentioned min/max values and
> using
>   a 'fast' transition  (whatever 'fast' means)
>
> Jochen
>

Thank you Jochen. I appreciate you passing on your experience too, as
someone with very little in this area! ;)

- James

If you want to measure the input threshold, just implement an inverting
data path from an input to an output, short the two pins together, and
decouple them to ground with a capacitor. You now have an oscillator
that switches around the threshold.
Peter Alfke, Xilinx

"Peter Alfke" <peter@xilinx.com> wrote in message
news:1116542484.664337.91630@z14g2000cwz.googlegroups.com...
> If you want to measure the input threshold, just implement an inverting
> data path from an input to an output, short the two pins together, and
> decouple them to ground with a capacitor. You now have an oscillator
> that switches around the threshold.
> Peter Alfke, Xilinx

Thanks Peter, that's a great suggestion!
- James

James A wrote:
> Hi,
> 
> I'm trying to find out when exactly the XC9572XL recognizes a falling edge.
> It has a 3.3V supply, and the signal is a square wave between 0V and 5V
> (this Xilinx is 5V input compatible). At what voltage would a falling edge
> trigger?
> 

Digital logic IO typically have a SINGLE threshold point despite the 
guaranteed values of Vil & Vih -- there is only one threshold point -- 
for TTL logic this has been around 1.5Vdc -- there are some logic which 
incorporates two threshold points for such as Schmitt Triggers called 
hysteresis ....

"JoeG" <JoeG@yahoo.com> wrote in message
news:1ifje.1131$mK.405@newssvr13.news.prodigy.com...
> James A wrote:
> > Hi,
> >
> > I'm trying to find out when exactly the XC9572XL recognizes a falling
edge.
> > It has a 3.3V supply, and the signal is a square wave between 0V and 5V
> > (this Xilinx is 5V input compatible). At what voltage would a falling
edge
> > trigger?
> >
>
> Digital logic IO typically have a SINGLE threshold point despite the
> guaranteed values of Vil & Vih -- there is only one threshold point -- 
> for TTL logic this has been around 1.5Vdc -- there are some logic which
> incorporates two threshold points for such as Schmitt Triggers called
> hysteresis ....

Thank you Joe. Would the supply being 3.3V (rather than 5V) have any bearing
on the 1.5V value?

- James

James A wrote:
> "JoeG" <JoeG@yahoo.com> wrote in message
> news:1ifje.1131$mK.405@newssvr13.news.prodigy.com...
> 
>>James A wrote:
>>
>>>Hi,
>>>
>>>I'm trying to find out when exactly the XC9572XL recognizes a falling
> 
> edge.
> 
>>>It has a 3.3V supply, and the signal is a square wave between 0V and 5V
>>>(this Xilinx is 5V input compatible). At what voltage would a falling
> 
> edge
> 
>>>trigger?
>>>
>>
>>Digital logic IO typically have a SINGLE threshold point despite the
>>guaranteed values of Vil & Vih -- there is only one threshold point -- 
>>for TTL logic this has been around 1.5Vdc -- there are some logic which
>>incorporates two threshold points for such as Schmitt Triggers called
>>hysteresis ....
> 
> 
> Thank you Joe. Would the supply being 3.3V (rather than 5V) have any bearing
> on the 1.5V value?
> 
> - James
> 
> 

Not much - it's still around 1.5Vdc ...