Re: How big is my vhdl and am I approaching some size limitation on the chip.

On Sat, 21 Mar 2009 08:10:07 -0700 (PDT), jleslie48 <jon@jonathanleslie.com>
wrote:

>On Mar 21, 8:42 am, Brian Drummond <brian_drumm...@btconnect.com>
>wrote:
>> On Fri, 20 Mar 2009 17:56:39 -0700 (PDT), jleslie48 <j...@jonathanleslie.com>

>Hey everybody, thanks for all the good suggestions.
>
>1) So it is reasonable to conclude that the cold runner 3512 is way
>too small to even run a uart yes?

I would expect it can easily handle a UART - or several. 
But a UART and whatever else lies behind it - that could be another matter.

>2) is the board that Brian suggested, the raggedstone1 with the
>spartan XC3S1500 is big enough?

Seems well over 10x the chip you have now.
And there are XC3S2000 and 4000 if it isn't...


>
>Whats the deal withe the "PACKAGE" ( 676FCBGA)

well it's not something you'd hand solder...

>I see from AVNET that the XC3S1500 comes in lots of flavors:

>XC3S1500-4FG320C
>XC3S1500-4FG456C
>XC3S1500-4FG676C

3 packages, 2 speed grades (say 15-20% apart), leaded/lead-free,
commercial/industrial speed grades, but basically all the same chip.
Take your pick. (The larger 2000/4000 devices may only be available in the
largest package)

>3) The raggedstone1  has an added feature that I was told to consider,
>mounting into a PC. Initially I want to put it a stand alone box, so I
>will need to order the board, the PCI I/O header, and the Ocsillator
>and then I'm good to go yes?

I'd guess you can power it and use the headers for I/O even without the PCI
header. It's far from being the only option but it's an attractively large FPGA
for the money. (If the package, or hand soldering, is an issue, look at the DIP
or pin-grid-array replacements on the same site).

If you need to save every milliwatt, stay with the Coolrunner. 

Otherwise given the cost constraints you mentioned, I'd say you shouldn't be
wasting time trying to fit a too-small device. Is your software team trying to
keep their executable below 4K?

- Brian

On Mar 21, 8:39 pm, Brian Drummond <brian_drumm...@btconnect.com>
wrote:
> On Sat, 21 Mar 2009 08:10:07 -0700 (PDT), jleslie48 <j...@jonathanleslie.=
com>
> wrote:
>
> >On Mar 21, 8:42 am, Brian Drummond <brian_drumm...@btconnect.com>
> >wrote:
> >> On Fri, 20 Mar 2009 17:56:39 -0700 (PDT), jleslie48 <j...@jonathanlesl=
ie.com>
> >Hey everybody, thanks for all the good suggestions.
>
> >1) So it is reasonable to conclude that the cold runner 3512 is way
> >too small to even run a uart yes?
>
> I would expect it can easily handle a UART - or several.
> But a UART and whatever else lies behind it - that could be another matte=
r.
>
> >2) is the board that Brian suggested, the raggedstone1 with the
> >spartan XC3S1500 is big enough?
>
> Seems well over 10x the chip you have now.
> And there are XC3S2000 and 4000 if it isn't...
>
>
>
> >Whats the deal withe the "PACKAGE" ( 676FCBGA)
>
> well it's not something you'd hand solder...
>
> >I see from AVNET that the XC3S1500 comes in lots of flavors:
> >XC3S1500-4FG320C
> >XC3S1500-4FG456C
> >XC3S1500-4FG676C
>
> 3 packages, 2 speed grades (say 15-20% apart), leaded/lead-free,
> commercial/industrial speed grades, but basically all the same chip.
> Take your pick. (The larger 2000/4000 devices may only be available in th=
e
> largest package)
>
> >3) The raggedstone1  has an added feature that I was told to consider,
> >mounting into a PC. Initially I want to put it a stand alone box, so I
> >will need to order the board, the PCI I/O header, and the Ocsillator
> >and then I'm good to go yes?
>
> I'd guess you can power it and use the headers for I/O even without the P=
CI
> header. It's far from being the only option but it's an attractively larg=
e FPGA
> for the money. (If the package, or hand soldering, is an issue, look at t=
he DIP
> or pin-grid-array replacements on the same site).
>
> If you need to save every milliwatt, stay with the Coolrunner.
>
> Otherwise given the cost constraints you mentioned, I'd say you shouldn't=
 be
> wasting time trying to fit a too-small device. Is your software team tryi=
ng to
> keep their executable below 4K?
>
> - Brian


"Is your software team trying to  keep their executable below 4K?"

first off the software team is me.  For 20 years now I keep getting
suckered into these solo missions into [the companies] no man land...

I'm sorry are you referring to $4k or some size requirement of 4k???
We have yet to discuss a unit of measure for the size of the
executable, is that something else I'm overlooking?  Is that an Issue
with the enterpoint solutions?  I'm assuming I can load the program
into the enterpoint solutions and on powerup they start to
function...The coolrunner was able to do that, although I dont know
about any limitation to the executable size on the coolrunner that I'm
having trouble with other than the resource issues that started this
conversation.

The guys on this project have blown their budget for hardware already,
at least for this next deliverable, so any $$$ I spend (not counting
labor) are being watched carefully. Meantime they forgot this whole
set of functionality and now have to patch it in somehow.  So they
brought me in.  On the one card that is freed up, they have the
coolrunner chip, and as I was having so much fun with fpga-vhdl up
until a month ago, I figured I could use it and not have to add any
hardware components to the project.   this was a very attractive
solution to the PM (program manager)  The board is going to be re-spun
anyway to fix some cross talk issues, so breaking out a few of the
unused pins on the coolrunner for the added functionality wasn't
really going to impact the project at all.  So while I got the new
functionality to work on my virtex II chip, the hardware boys built me
a jury-rigged daughter board for the old coldrunner board:  they
actually hand soldered 30 gauge wire to the pins of the  coolrunner,
broke it out to a DB15 where I could attach a daughter board to get
the digital I/O onto a scope/leds/level shifters.   All was going well
until I popped the FF thing on the coolrunner.

So I Imagine I'm gonna either get it all to fit on the coolrunner, or
go with a new card, those cards from enterpoint look quite reasonable,
especially the really small ones. That solution is gonna add to my
footprint, the boss is gonna be pissed but hey, at least its no longer
a dead fish.  SWAP is more of an issue than even the money. Actually I
don't even care about Power, its just the size and weight.  Those
really little cards will be a lot easier to sell than the bigger
raggedstone1.

So back to the issue at had, the

1) Drigmorn1 has a XC3S500E-4CPG132C spartan-3 chip:
 Xilinx XC3S500E-4CPG132C
FPGA Spartan=AE-3E Family
500K Gates
10476 Cells
572MHz
Commercial 90nm (CMOS) Technology
1.2V 132-Pin CSBGA
	Package 	132CSBGA
	Family Name 	Spartan=AE-3E
	Device Logic Cells 	10476
	Device Logic Units 	1164
	Device System Gates 	500000
	Number of Registers 	9312
	Maximum Internal Frequency 	572 MHz
	Typical Operating Supply Voltage 	1.2 V
	Maximum Number of User I/Os 	92
	RAM Bits 	368640
	Re-programmability Support 	Yes


2) the Craignell has a (Xilinx XC3S500E) Spartan-3E chip as well,

3) the Raggedstone1 has a XC3S1500 chip:
 Xilinx XC3S1500-4FG320C
FPGA Spartan=AE-3 Family
1.5M Gates
29952 Cells
630MHz
Commercial 90nm Technology
1.2V
320-Pin FPBGA
	Package ???	320FPBGA ????
	Family Name 	Spartan=AE-3
	Device Logic Cells 	29952
	Device Logic Units 	3328
	Device System Gates 	1500000
	Number of Registers 	N/A
	Maximum Internal Frequency 	630 MHz
	Typical Operating Supply Voltage 	1.2 V
	Maximum Number of User I/Os 	221
	RAM Bits 	589824
	Re-programmability Support 	Yes

Which of these products will do the job and not be too small like the
coolrunner?

On Mar 22, 3:19=A0am, jleslie48 <j...@jonathanleslie.com> wrote:
>
> So back to the issue at had, the
>
> 1) Drigmorn1 has a XC3S500E-4CPG132C spartan-3 chip:
>
> 2) the Craignell has a (Xilinx XC3S500E) Spartan-3E chip as well,
>
> 3) the Raggedstone1 has a XC3S1500 chip:
>
> Which of these products will do the job and not be too small like the
> coolrunner?

IMHO any of these should be enough. A UART can be made very simple, or
it may have lots of bells and whistles.
The good thing is that you can run place and route and see what device
fits better.
You really should.
Of course, you are aware that you need the FPGA + some sort of device
to boot it.

Good luck.


Josep Duran

On Mar 22, 7:41 am, djj08230 <djj08...@gmail.com> wrote:
> On Mar 22, 3:19 am, jleslie48 <j...@jonathanleslie.com> wrote:
>
>
>
> > So back to the issue at had, the
>
> > 1) Drigmorn1 has a XC3S500E-4CPG132C spartan-3 chip:
>
> > 2) the Craignell has a (Xilinx XC3S500E) Spartan-3E chip as well,
>
> > 3) the Raggedstone1 has a XC3S1500 chip:
>
> > Which of these products will do the job and not be too small like the
> > coolrunner?
>
> IMHO any of these should be enough. A UART can be made very simple, or
> it may have lots of bells and whistles.
> The good thing is that you can run place and route and see what device
> fits better.
> You really should.
> Of course, you are aware that you need the FPGA + some sort of device
> to boot it.
>
> Good luck.
>
> Josep Duran

Good morning.

"Of course, you are aware that you need the FPGA + some sort of device
to boot it."

No I'm not aware, are you saying these devices can't be loaded with
the software so when they
power up they start running?

On Mar 22, 7:41 am, djj08230 <djj08...@gmail.com> wrote:
> On Mar 22, 3:19 am, jleslie48 <j...@jonathanleslie.com> wrote:
>
>
>
> > So back to the issue at had, the
>
> > 1) Drigmorn1 has a XC3S500E-4CPG132C spartan-3 chip:
>
> > 2) the Craignell has a (Xilinx XC3S500E) Spartan-3E chip as well,
>
> > 3) the Raggedstone1 has a XC3S1500 chip:
>
> > Which of these products will do the job and not be too small like the
> > coolrunner?
>
> IMHO any of these should be enough. A UART can be made very simple, or
> it may have lots of bells and whistles.
> The good thing is that you can run place and route and see what device
> fits better.
> You really should.
> Of course, you are aware that you need the FPGA + some sort of device
> to boot it.
>
> Good luck.
>
> Josep Duran

Now you got me all freaked out again.

I'm looking at this wonderful manual:
http://www.enterpoint.co.uk/techitips/Programming%20Raggedstone1%20User%20Guide.pdf

and its beautiful.  why xilinx can't make something like this is
ridiculous.  This is what is needed.

Anyway, it seems to me on page 23 of this guide it is programming a
PROM.  I read this as the raggedstone1 has built in a "some sort of
device to boot it."  as in the FPGA that is on the card will boot from
its own onboard PROM.  I'm also assuming that the PROM is big enough
so I can load my UART program into this PROM, power down the unit and
when I power it up the program starts up, starts talking on the UART,
blinking whatever LED's I tell it to blink, etc...

I ASSUMED the other two devices the craig and the drigmorn were
similiar, Can someone confirm or deny this?

On Mar 22, 1:44=A0pm, jleslie48 <j...@jonathanleslie.com> wrote:
> On Mar 22, 7:41 am, djj08230 <djj08...@gmail.com> wrote:
>
>
>
> > On Mar 22, 3:19 am, jleslie48 <j...@jonathanleslie.com> wrote:
>
> > > So back to the issue at had, the
>
> > > 1) Drigmorn1 has a XC3S500E-4CPG132C spartan-3 chip:
>
> > > 2) the Craignell has a (Xilinx XC3S500E) Spartan-3E chip as well,
>
> > > 3) the Raggedstone1 has a XC3S1500 chip:
>
> > > Which of these products will do the job and not be too small like the
> > > coolrunner?
>
> > IMHO any of these should be enough. A UART can be made very simple, or
> > it may have lots of bells and whistles.
> > The good thing is that you can run place and route and see what device
> > fits better.
> > You really should.
> > Of course, you are aware that you need the FPGA + some sort of device
> > to boot it.
>
> > Good luck.
>
> > Josep Duran
>
> Now you got me all freaked out again.
>
> I'm looking at this wonderful manual:http://www.enterpoint.co.uk/techitip=
s/Programming%20Raggedstone1%20Us...
>
> and its beautiful. =A0why xilinx can't make something like this is
> ridiculous. =A0This is what is needed.
>
> Anyway, it seems to me on page 23 of this guide it is programming a
> PROM. =A0I read this as the raggedstone1 has built in a "some sort of
> device to boot it." =A0as in the FPGA that is on the card will boot from
> its own onboard PROM. =A0I'm also assuming that the PROM is big enough
> so I can load my UART program into this PROM, power down the unit and
> when I power it up the program starts up, starts talking on the UART,
> blinking whatever LED's I tell it to blink, etc...
>
> I ASSUMED the other two devices the craig and the drigmorn were
> similiar, Can someone confirm or deny this?

I've been reading the thread, and I may have missed something. The
responses you get are from 'experts' so there really is nothing I can
add.

Raggedstone and the other products from ENTERPOINT are complete boards
that include the FPGA and all the support needed to make a fully
functional board. (May include other unneeded  functionality). If I
recall correctly, Raggedstone board actually includes an interface to
a PCI BUS.

I was under the impresion that you wanted to replace a Coolrunner for
an FPGA in order to get more gates. If this is so, just be aware that
the spartan-3 series usually need and external memory to load the
configuration at power on. You also need 3 different supplies 3.3v,
2.5v and 1.2v. (not sure if 3.3v is mandatory) And of course some sort
of external oscillator that you probably already have.
(as somenone already mentioned, the Spartan3AN has some flash memory
on chip)

I am also under the impresion that you are in a hurry. If this is so,
probably your best move would be get help from a professional external
source. Not that this is rocket science, but you should expect to be
bitten by the tools in the first few weeks.

I never meant to scare you, this is just my point of view.

Josep Duran

On Mar 22, 9:10 am, djj08230 <djj08...@gmail.com> wrote:
> On Mar 22, 1:44 pm, jleslie48 <j...@jonathanleslie.com> wrote:
>
>
>
> > On Mar 22, 7:41 am, djj08230 <djj08...@gmail.com> wrote:
>
> > > On Mar 22, 3:19 am, jleslie48 <j...@jonathanleslie.com> wrote:
>
> > > > So back to the issue at had, the
>
> > > > 1) Drigmorn1 has a XC3S500E-4CPG132C spartan-3 chip:
>
> > > > 2) the Craignell has a (Xilinx XC3S500E) Spartan-3E chip as well,
>
> > > > 3) the Raggedstone1 has a XC3S1500 chip:
>
> > > > Which of these products will do the job and not be too small like the
> > > > coolrunner?
>
> > > IMHO any of these should be enough. A UART can be made very simple, or
> > > it may have lots of bells and whistles.
> > > The good thing is that you can run place and route and see what device
> > > fits better.
> > > You really should.
> > > Of course, you are aware that you need the FPGA + some sort of device
> > > to boot it.
>
> > > Good luck.
>
> > > Josep Duran
>
> > Now you got me all freaked out again.
>
> > I'm looking at this wonderful manual:http://www.enterpoint.co.uk/techitips/Programming%20Raggedstone1%20Us...
>
> > and its beautiful.  why xilinx can't make something like this is
> > ridiculous.  This is what is needed.
>
> > Anyway, it seems to me on page 23 of this guide it is programming a
> > PROM.  I read this as the raggedstone1 has built in a "some sort of
> > device to boot it."  as in the FPGA that is on the card will boot from
> > its own onboard PROM.  I'm also assuming that the PROM is big enough
> > so I can load my UART program into this PROM, power down the unit and
> > when I power it up the program starts up, starts talking on the UART,
> > blinking whatever LED's I tell it to blink, etc...
>
> > I ASSUMED the other two devices the craig and the drigmorn were
> > similiar, Can someone confirm or deny this?
>
> I've been reading the thread, and I may have missed something. The
> responses you get are from 'experts' so there really is nothing I can
> add.
>
> Raggedstone and the other products from ENTERPOINT are complete boards
> that include the FPGA and all the support needed to make a fully
> functional board. (May include other unneeded  functionality). If I
> recall correctly, Raggedstone board actually includes an interface to
> a PCI BUS.
>
> I was under the impresion that you wanted to replace a Coolrunner for
> an FPGA in order to get more gates. If this is so, just be aware that
> the spartan-3 series usually need and external memory to load the
> configuration at power on. You also need 3 different supplies 3.3v,
> 2.5v and 1.2v. (not sure if 3.3v is mandatory) And of course some sort
> of external oscillator that you probably already have.
> (as somenone already mentioned, the Spartan3AN has some flash memory
> on chip)
>
> I am also under the impresion that you are in a hurry. If this is so,
> probably your best move would be get help from a professional external
> source. Not that this is rocket science, but you should expect to be
> bitten by the tools in the first few weeks.
>
> I never meant to scare you, this is just my point of view.
>
> Josep Duran

Ahh!!! ok whew!!!!

My first choice is to get it all to fit on the coolrunner: I'm not
sure it will, but I'm not convinced it won't either.  This is the best
path as it requires the least amount of re-engineering of the
hardware.

Replacing the coolrunner chip is probably the best solution, but it
will be a huge effort, and is not in the time constraints I have.  I'd
love to replace the coolrunner CHIP on the custom made interface
board, but that will not be something I can do quick.  I realize
that.  The entire board will have to be redone to match the new chip,
the traces all redone, etc, plus there is an old program written for
the coolrunner and if nothing else all its pins have to be moved to
the new chip, let alone any programming subtilties that might pop up.
No I would not attempt this without knowing full well what the
coolrunner was doing originally, and making sure the new chip meets
that needs.  That's not going to be a quick fix.

No, at this time, I'm just hoping to ADD an entire new card to do the
functionality that I need, That's the best I can hope for in my
timeframe.

The idea is to get one of the three (I'll probably buy all three just
to try them out.) to do my UART thing, and then just program the old
coldrunner with some new digital outputs and inputs to tell the new
board to start, stop, ack etc. An ugly patch job, but it will work, SO
LONG AS WHATEVER BOARD I USE BOOTS UP, AND RUNS THE NEW UARTS.

The boss will hate it, but at least it will work.  That's a whole lot
better than the paperweight he has now.  Next time he should of spec-
ed out the scope of the functionality a bit better.  Also the pred-
ecessors should of been a little more forward thinking of growth.  It
is ridiculous that they picked a $75 coolrunner and loaded it up to
70% of capacity.   I'm looking at the Spartan chips, at $20- $70 that
are 10-50x more powerful, and shaking my head as to why they went with
the coolrunner.  The guy that picked it is long gone.  My guess is
that he used it before, and knew it would do what he was asked to do.
My rule of thumb is for a first prototype is to never be more than 25%
of capacity.

When you have to a production run of more than 1000 units, then you
worry about right-sizing. Meantime the spartan would still have been
way cheaper anyway for all their "right-sizing" effort. Right-sizing
is only important if it saves you something.

I looked over the BOM for the box.  These knuckleheads managed to find
a $30 momentary switch to use as a reset switch on the box and then
tell me they saved money by right-sizing the chip...

>No I'm not aware, are you saying these devices can't be loaded with
>the software so when they
>power up they start running?

Why haven't you looked at the data sheet?  Yes, it's huge, but
you should have looked close enough to found the section
that describes loading it.


-- 
These are my opinions, not necessarily my employer's.  I hate spam.

On Mar 22, 3:00=A0pm, jleslie48 <j...@jonathanleslie.com> wrote:

> When you have to a production run of more than 1000 units, then you
> worry about right-sizing. Meantime the spartan would still have been
> way cheaper anyway for all their "right-sizing" effort. Right-sizing
> is only important if it saves you something.
>

Please, allow me to point in a completely different direction. As you
are a software guy, you have probably considered implementing the
UARTS in a uC. I think a small AVR (or PIC) could easily handle 10
serial channels, particullarly  at slow baud rates. Is there a reason
why you wouldn't follow this path ?
(I am talking about a complete software implementation here)

It is not about cost, but about development time.


Josep Dur=E0n

On Mar 22, 1:41 pm, hal-use...@ip-64-139-1-69.sjc.megapath.net (Hal
Murray) wrote:
> >No I'm not aware, are you saying these devices can't be loaded with
> >the software so when they
> >power up they start running?
>
> Why haven't you looked at the data sheet?  Yes, it's huge, but
> you should have looked close enough to found the section
> that describes loading it.
>
> --
> These are my opinions, not necessarily my employer's.  I hate spam.

I did and was comfortable that I saw I could load it up.  However when
Josep
stated:
"Of course, you are aware that you need the FPGA + some sort of device
to boot it. "

I thought I mis-understood something.  Josep is clearly warning me
that this needs some sort of device to boot it.  The last thing  I
need is to push a PO on my boss only to find out it isn't going to do
the job.

Plus I'm sorry these data sheets don't spell this stuff out.  They
seem to spend a lot of time telling you stuff so they can say "it was
in the data sheet" meantime the data sheet doesn't say the basics in
plain english.  for example in chapter 1, page one should be a heading
"FEATURES OF THIS BOARD "

1) xxx01 FF's
2) xxx02 macrocells
3) xxx03 pterms
4) onboard oscillator of xxx04 MHZ
5) slot for a different oscillator up to xxx05 MHZ
6) eprom for autoboot
7) xxx06-xxx07 input voltage range
8) peak amps:   xxx08 ma
9) typical amps running:  xxx09 ma
10) ) load your program from ISE Project manager into the onboard
prom.
11) load program using included xxx10 cable, or xxx11 cable (available
separately) or
     xxx12 cable from xilinx...
12) <<detailed photograph of top side all lettering easily readable>>
13) <<detailed photograph of back side all lettering easily readable>>
14) <<detailed photographs of all sides if connectors are at 90
degrees>>
15) << one or two isometric views with either a ruler or some scale
reference (a pen, computer CD or such >>
16) << picture/drawing of expected runtime deployment environment>>
17) << picture/drawing of expected programming envionment>>
18) list of what is included in the package
19) list of what is not included but required to program/run
20) list of what are the optional parts.
21) ...

Plus every board should come with a "hello world" program, that the
user should be able to follow a step-by-step
guide and get results and thus confirm that all is in working order.

I see that the raggedstone1 has really tried to get that "hello world"
program and step-by-step guide in place:
http://www.enterpoint.co.uk/techitips/Programming%20Raggedstone1%20User%20Guide.pdf

On the strength of that document alone they are going to get me as a
customer.  Its as good as I've seen, even though it fails to show the
cable connections for the board to the PC/ power supplies.    I only
wish my diligent board and ISE 10.1 project navigator had a document
like that when I purchased that way back.  If I had seen the
raggedstone1 then I would have bought 5 of those instead of the 1
diligent board.

That reminds me, I was looking at the waggle test for the drigmorn:
http://www.enterpoint.co.uk/component_replacements/drigmorn1_apps.html

in the user manual it said the the waggle test has a loopback for the
rs232 but I don't see th at in the .vhd code:
http://www.enterpoint.co.uk/component_replacements/DRIGMORN1_WAGGLE_TEST1.VHD

what's am I missing?