Re: FPGA for PCI based servo control board

["John & Beth Kasunich" <JEKasunich-at-worldnet.att.net>]

> John Sheahan wrote:

> On Tue, Apr 01, 2003 at 09:26:19AM -0500, jmkasunich-at-ra.rockwell.com
wrote:
>
>> John Sheahan wrote:
>>
>>> On Tue, Apr 01, 2003 at 01:52:57AM -0500, Jon Elson wrote:
>>>>
>>>> That is NOT true 5 V I/O.  It is 5V TOLERANT I/O.
>>>
>>> true - but 5v pci only needs to be 5v tolerant.
>>> driving 3v3 is in specification.
>>>
>>> is there much 5v PCI any more?
>>>
>>
>> Repeat after me:
>> "The world is not a PC"
>> "The world is not a PC"
>
> please reread my comment, note the word 'PCI'.
> I have read and understood the VI curves in the PCI version 2.2
> specification.  The comment you were responding to was PCI related.

I apologise for my somewhat snarky response.  I am aware of the 
fact that the question was PCI related, but I was responding to
the fact that you seemed to consider PCI compatibility the sole
criterion for choosing the FPGA.

>> Just because you don't need 5V to talk to a PCI bus inside
>> a PC doesn't mean you don't need 5V.  The outputs of the
>> FPGA will have to drive opto-couplers, relay drivers, analog
>> filters, OPTO-22 modules, solid state relays, and all kinds
>> of other real world stuff.  For that stuff, 5V is FAR better
>> than 3.3V, and don't even mention 2.5V.

>
> Sure. I do this regularly.
> I have done this since TTL days. check out the ttl output specs
> apart from the 5v tolerant input part.
> one day. they are pretty similar to 3v3 output stages.

I know, that's why I mentioned HCT for a 3.3V to 5V buffer.
I use HC or AC for most logic stuff.  If the input is coming 
from TTL, then I use HCT or ACT.  I've never had to interface
to 3.3V, but I guessed that HCT would work there too.

I prefer the H versions to the A versions unless I really need
high drive and fast edges - keeps noise and crosstalk down.

> I tend to drive optos with active low outputs anyway. 

Same here.

> Your point is??

My point is that for an industrial type project, I'd stick with 5V
throughout the design, if given a choice.  I should have been
nicer about it.

>> On a positive note, I believe that you could use HCT series
>> buffer chips running on 5V, feed them a signal from a 3.3V
>> FPGA, and get a good 5V signal on the buffer outputs.  Could
>> get bulky if you have lots of output pins to buffer.

> that is the way most of the world does it, yes.
> HCT is still in full production at places like Phillips.
> john

It sure is.  I design VFDs and other industrial equipment.  ALL
of our stuff runs on 5V, even the newest designs.  We are using
some very recent Hitachi micros with built in PWM generators, 
A/Ds, and so on, and they are 5V parts.  Regardless of where
the PC world goes, 5V is still the universal standard in the 
industrial world, and I see no reason to change that.

John Kasunich