Re: FPGA for PCI based servo control board

[John Sheahan <jrsheahan-at-optushome.com.au>]

On Wed, Apr 02, 2003 at 09:14:19AM -0500, jmkasunich-at-ra.rockwell.com wrote:
> 
> OK, maybe messages from somebody else got merged with yours
> in my head.

no worries. I need to learn to think before replying to email 
sometinmes, seems to be a not uncommon problem.
> 
> What kind of products are you designing?  When I think of
> fast comms, I think ethernet, ATM, T1 lines, ASDL, and
> so on.  That's still computer stuff, not industrial stuff.

that one was OC-48. a sonet fibre Coarse Wavelength Division M mux. 
OC48 is 2.488GHz (hence the virtex 2 pro).   
CWDM is using a bunch of different colour lasers and mirrors
and stuff to run 4 bidirectional OC-48 down the same fibre.
(I had nothing to do with the optics)
Low power, outside plant temperature ranges, small size.    
Telecomms at trunk rates are pretty different to PC.
volumes are lower, reliability more important than cost, 
longer lifetimes. Layout on the fats stuff is critical.
It has aspects of a few technologies.  Its industrial, but 
definitely not the power end of the industrial scale.

> 
> > for me - I'd be choosing modern fpgas that fit and route
> > well - with flexible io pinout - and are cheap.
> > which means low voltage core today.  The older stuff
> > is just horrible by comparison.
> 
> A lot of this is perspective.  I'm coming at this strictly
> from a hobby point of view.  I want simple circuits, parts
> that I can solder, and non-critical layout.  Ideally I'd
> be able to use a two layer board, since they are much
> cheaper in small quantities.  These priorities are completely
> different than those in commercial development of high tech
> products, where high clock speeds, low voltages, and fine
> pitch parts are the norm.
> 

Agree. my current projects have these constraints.
however for me a 'simple circuit' means few parts, easy-to-asemble.
Which sometimes means doing as much inside a fpga or a single-chip 
uP as practical.   


> Even the stuff I design at work is much lower tech than
> what you work on.  As I mentioned, we use 5V only.  Boards
> are four layer, not six or more.  

grin. the WDM board was 16 layer. (high speed routing 
with groundplane either side eats layers.  
Boards I'm drawing today are d.s.    it all depends on the product.

> The biggest part in our
> most recent design is 100 pins.  Much of it is analog,
> with 1206 and 0805 parts.  I know the bleeding edge folks
> are using 0402s and such, but for our kind of products
> it just doesn't make sense.

I cannot see well enough to consider 0402 for my home stuff.
elsewhere 0201 is being  evaluated. sigh. 
thruhole or 0804 is _much_ easier in low volume.

for anyone else reading this 0804 is a surface mount resistor or
capacitor package measuring about .08 inch long and about 
.04 inch wide.   Home-made-able

And I think this is the key statement for the project that started
this.  PLCC84 or 100qfp  (not tqfp) are perhaps practical upper limits
for hobby construction. Unless the device is on a daughter board.

> 
> Some designs are more partitionable (is that a word?)
> than others.  The CNC project has natural breaks, either
> by axis or by function.

Agreed.  My pain there was the additional limits of pin count and 
the limited routing withing the CPLDs.  A D/S board is limited 
enough by the time power is gridded enough.  A bigger CPLD 
was on TQFP (0.5mm legs).  3 smaller CPLD added 4 layers to the Pcb. 
Had to partition by channel, not function, to ease board route.

> 
> We definitely work in different worlds.  About 95% of a
> drive's electronics is interface.  The micro handles all
> of the logic, and includes RAM, flash, etc. on one chip
> (that's the 100 pin one).  The rest of the stuff on the
> boards is current and voltage feedback, IGBT gate drivers,
> fault detection, and user I/O, both analog and digital.
> Power supplies are a big part of it too.  All of the
> digital stuff is 5V, but we have +/-12V for analog,
> +/-20V for current sensors, isolated 24V for digital
> I/O, and 6 isolated supplies for the gate drives.

That is a fair summary. My 'logic' comments refer only to the 
inside of the uP.  (well, some of the analog IO could be included 
also, say with delta-sigma in logic) 
That could be partitioned into a SW and a  HW component.   
Wearing my hardware design hat, some things could be done better
inpure HW. (eg more accurate timing and generation) but very often
this is just overkill.


Back to the original thread - seems to me what is being considered 
here is moving some of the functionality of the micro to a 
FPGA running on PCI  inside a PC.   All the rest remains.

I personally find the inside of a PC a nasty environment to design
for. (mechanically, power, short life, crappy IO connector mounting)

I would leave the board features you describe together and use 
perhaps ethernet withoout the protocol stacks, to offload the 
functionality.  because I'm personally tired of parallel ports.  

then the bit inside the PC is a single cheapo NIC. 
The PC that runs my router changes more often than the rest of it.

I think this approach allows you to concentrate on the 'control' part 
with as little diversion as possible into the wonders of the PCI bus.
Having been involved in PCI interfacing. its more painful than ISA.

john