Searching \ for '[EE] Programmable logic' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=programmable+logic
Search entire site for: 'Programmable logic'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Programmable logic'
2010\08\20@130111 by Denny Esterline

picon face
Hi all,
I have a project redesinging a little specialty industial controller.
It was orginaly designed around about a half dozen relays, including
some "Agastat" adjustable time delay relays. Unsurprisingly, some of
these are being obsoleted and I've been asked to do the redesign. So I
guess it's a good problem to have.

Before it's suggested, yes, I _could_ handle it with a minor PIC, I
may even protoype it with one anyway to verify the rest of the circuit
and to quickly prove the logic. Due to some arcane rules, there's a
ton of red tape to get a microprossesor based product installed at
some of the end user sites. It's agravating, but that's the world I
live in.

So, I've done a cocktail napkin design and I figure I need about 30-40
logic gates and four timers to duplicate all thier functions, of
course, that's before the inevitable "feature creap" begins. Certainly
in the realm of something that I can do with discreet parts, but I'm
fairly sure it's time to look at some of the low end programmable
logic parts.

I've been chewing through the some of the apparent options, Xilinx,
Altera, Atmel, and it's a little daunting. Not only do I need to get
my arms around the development side of it, but I need to understand
the production level tools as well. And picking a part that can handle
my I/O count (13 and 12) and amount of logic is straitforward, until I
notice some of the suppliers are obsoleting many of the parts I was
initialy looking at. Even when some of the parts aren't being
obsoleted, I'm seeing the dev tools being obsoleted, so I'm forced to
wonder how much longer the parts will be around.

I know that learning VHDL/Verilog will be neccesary at some point in
my life, but here in the beginnning I'd prefer something simpler,
especialy for such a simple application. Schematic capture is an
option, as would be some type of logical expresion system (output X =
input A AND input B OR input C)

So, I'm seeking suggestions for a manufacturer / product family / dev
tools / production tools, and any pointers to good reference material
to shorten the learning curve would also be appreciated. My criteria
in descending order of importance:

Longevity of part (no obsolecence next week)
Short learning curve
Available dev tools
Available production tools
Complexity of support electronics
Cost of production tools
Cost of dev tools
Cost of part


You'll notice package size and power are not on the list, The
equipment is line powered and controls a 1kW motor, a few more mA for
the logic is fairly irrelevant. Part cost is last, because it's fairly
low volume. At about 100 pcs / year, an expensive toolchain could
easily cost more than the sum of the chips.


Thanks
-Denn

2010\08\20@142518 by Carl Denk

flavicon
face
You say "Programmable Logic", add one more word, "Controller" >> PLC , and your application points to that. There are numerous out there, you don't say what country you are in, there are numerous manufacturers and models out there. Many are heavy duty industrial quality, UL, CE, and other approvals. Allen Bradley Modicon is probably the most popular. Most of the auto manufacturers use them to run very complicated machines, but there are small units available. The PLC was the replacement for relay logic, and the relay ladder diagram programming, understandable by many trade electricians is still used today. Some have a high level language like Basic available to handle analog and some digital logic. I would think this would easiest to implement and maintain. Just plug a laptop with their software, and can quickly change things.

As a hobbyist, I have a standby generator that is managed by a PLC. This includes starting, running, and shutting down on gasoline, propane, or natural gas, switching under load between fuels depending what is available. I use a TRI-PLC T100MD+ with MD-EXP4040 expansion I/O, and a MD-HMI display keyboard. They have many other models available, some much more basic, and some more advanced with Ethernet capability.
http://www.tri-plc.com

On 8/20/2010 1:01 PM, Denny Esterline wrote:
{Quote hidden}

>

2010\08\20@143034 by Oli Glaser

flavicon
face


--------------------------------------------------
From: "Denny Esterline" <spam_OUTdesterlineTakeThisOuTspamgmail.com>
Sent: Friday, August 20, 2010 6:01 PM
To: "Microcontroller discussion list - Public." <.....piclistKILLspamspam@spam@mit.edu>
Subject: [EE] Programmable logic

{Quote hidden}

I am in the process of getting up to speed with FPGAs/Verilog etc at the moment. I agree it seems very daunting at first...
I persevered though as it's the way forward for a lot of things and they are very capable for not too much cost nowadays, and much easier than standard logic once you get going (just like an electronic breadboard)

I went for Actel ProASIC3 to start off with as they are pretty fast (350MHz), flash based (not SRAM, so no need for external PROM, better code security and live at power up) and the USB JTAG programmer (FlashPro4) is nice and cheap (~£40) They are quite cheap too - <£3 for 15K, 30K, ~£4 for 60K gate (which can run a small core with a few peripherals) ~£8 for 250K gate which can do an 8051 with lots of other stuff. Up to parts with onboard ARM cores and millions of gates for hundreds of pounds (3M gates is top I think)

The IDE, Libero is pretty good (although as I haven't tried the others I have nothing to compare to) and you can use ready rolled logic gates, peripherals, soft cores etc, with components written yourself in Verilog/VHDL just by dragging stuff onto a canvas and wiring it all together. I think Libero is free for all but the highest end chips (and you have to pay for some IP cores) Synthesis, Place and Route, Simulation (ModelSim) and Programming are all controlled (or launched) from within Libero.
The ProASIC3 dev board costs ~£1000 but I just made my own as I didn't need 90% of the features so I wasn't paying that kind of money. I read the Datasheets thoroughly and used the dev board schematics as a reference. My board works fine and I have managed to write and test a few of my own Verilog tests (flashing LEDs, shift registers, UART etc) and use the ABC soft core okay - in fact it's all going a bit too smoothly at the moment :-)
The downside with Actel is less support (Forums, web examples etc) than Xilinx or Altera, but in general FPGAs/CPLDs are quite a bit harder to get into than MCUs with a lot less support. Having said that as long as you spend (quite a bit of) time reading the extensive but well written datasheets you *should* be okay, at least getting started.
With Verilog if you know C (or Ada) then the syntax is similar, but obviously you can't think sequentially when describing hardware, so you need to get your mind into a different mode with HDL. I'm reading about 4 or 5 books in parallel and I found "FPGA prototyping by Verilog examples" quite useful, but there are a few good books out there - "The Complete Verilog Book" or another good reference is useful also.
With Schematic Capture, Libero has something called View Draw which I believe can do this, though I have not used it yet - I have just used Verilog and wired stuff up on the Smart Design Canvas.
Of course Xilinx and Altera are the two biggest players and have equivalent stuff (Actel, Lattice, Atmel etc - they all have pretty similar capabilities/IDEs though) Xilinx and others are mostly SRAM based though so will probably need more external parts to use (I think Xilinx do have some Flash based offerings though)

Hope this helps a bit, if you do decide to try Actel and need any other info (basic connections, voltages etc) just let me know.


2010\08\20@160024 by Denny Esterline

picon face
On Fri, Aug 20, 2010 at 11:25 AM, Carl Denk <cdenkspamKILLspamwindstream.net> wrote:
> You say "Programmable Logic", add one more word, "Controller" >> PLC ,
> and your application points to that. There are numerous out there, you
<snip>

Uhh, errr, thanks for the suggestion, but no.
I'm actually quite familliar with PLCs, one of our other product lines
has consumed about 250 of them the last couple years. This is
definitly not an application I want a PLC for.
I used the term "programmable logic" as a genaric term for PALs, GALs,
PLAs, PLDs, SPLDs, CPLDs, FPGAs etc. Seems like every manufacturer has
a silghtly different name for them, or several at various levels of
capabilities

2010\08\20@162631 by Isaac Marino Bavaresco

flavicon
face
Em 20/8/2010 14:01, Denny Esterline escreveu:
{Quote hidden}

Xilinx has lots of active parts in the range you need (small CPLDs).
Their (free) toll-chain is "Xilinx ISE Design Suite" and isvery active also..
For programming you could build a "Xilinx Parallel Cable III", very
simple, with schematic available and still supported by the tool-chain.
Uses parallel port Though.

Altera also has many suitable devices, their free tool-chain is the
"Quartus II Web Edition Software", actively maintained. Their simplest
download cable is the "Byte Blaster", that can be built easily.


For language, "Xilinx ISE" still supports the ancient "ABEL" language,
very easy to learn, much like the  "some type of logical expresion
system (output X = input A AND input B OR input C" you mentioned.


Best regards,

Isaac
__________________________________________________
Fale com seus amigos  de graça com o novo Yahoo! Messenger http://br.messenger.yahoo.com/

2010\08\21@041605 by RussellMc

face picon face
> Before it's suggested, yes, I _could_ handle it with a minor PIC, I
> may even protoype it with one anyway to verify the rest of the circuit
> and to quickly prove the logic. Due to some arcane rules, there's a
> ton of red tape to get a microprossesor based product installed at
> some of the end user sites. It's agravating, but that's the world I
> live in.

Presumably you and the red tape producers have asked the questions:

- What do we want this to do or achieve that a properly designed
processor based solution cannot achieve?

- Is the proposed solution really free of the issues which causes us
to reject a properly designed processor based solution?

- Why do we believe that?

- Do we genuinely believe that answer?

- Iterate the prior two questions until the answer to the prior
question is "Of course not! Are you mad? This is just "political"
expediency!"

- Do it, but properly.

If you are handling crucial human safety and/or big dollars (eg
nuclear station shutdown sequences, launchpad umbilical detachment,
pyro sdafing, interlocks to life threatening system, Tokamac run up /
shut down cycling  etc) do people really think that the "programming"
which goes into a logic replacement system, and/or the risk of
marginal conditions (undetected race condition, lock into forbidden
state, go mad on glitch, ...) really make it inherently safer than a
properly designed, fail safe, redundant processor based system? Do you
believe their answwer? Iterate ... :-) .



 Russel

2010\08\21@061131 by Justin Richards

face picon face
> and to quickly prove the logic. Due to some arcane rules, there's a
> ton of red tape to get a microprossesor based product installed at
> some of the end user sites. It's agravating, but that's the world I
> live in.
>
I am not familiar with FPGA etc but I thought they could be programmed
to behave like a micro so wouldnt the red tape people consider it so

2010\08\21@064554 by Michael Watterson

face picon face
 On 21/08/2010 11:11, Justin Richards wrote:
>> and to quickly prove the logic. Due to some arcane rules, there's a
>> ton of red tape to get a microprossesor based product installed at
>> some of the end user sites. It's agravating, but that's the world I
>> live in.
>>
> I am not familiar with FPGA etc but I thought they could be programmed
> to behave like a micro so wouldnt the red tape people consider it so.
Only in the sense that you can make a Micro out of loads of gates.
There are "off the shelf" "soft cores", design files to add FPGA simulated CPU core. Then that has to read a program at boot time.
But you can synthesise a non-CPU design. How bug or error (not same thing) free depends on the type of design approach.

Proving mathematical correctness of a design is different  for a CPU executing a program. Not usually possible with C implementations. Some proofs achieved with sequential Occam programs.

Bigger problems:
1) Is the Specification correct?
2) Does the Design satisfy the Specification?
3) How can you decide?

Does the actual program or FPGA synthesis implementation correctly implement the Design?
Then at "Run Time" what kind of errors can occur?
Can they be detected...

Many worms. C and Assembler are not good choices for LARGE systems that must be error free.

You can't "test out" about more than 1/2 the bugs. Low amount of bugs / errors is a result of good design not good testing.

Verilog might look "C" like. It's not at all. In that sense VHDL or ABEL might be a safer starting point for FPGA or CPLD than Verilog for long time C programmers

2010\08\21@065738 by Oli Glaser

flavicon
face
> So, I've done a cocktail napkin design and I figure I need about 30-40
> logic gates and four timers to duplicate all thier functions, of
> course, that's before the inevitable "feature creap" begins. Certainly
> in the realm of something that I can do with discreet parts, but I'm
> fairly sure it's time to look at some of the low end programmable
> logic parts.

I didn't notice this part before - if you only need 30-40 gates then maybe it's worth thinking about a SPLD from Atmel or something similar. They have a few hundred gates, are very cheap, and can be programmed in CUPL (IIRC) which I believe is pretty easy to learn.
Something like this maybe:
http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=6810415

I did get some myself but never used them as I decided that I needed more gates etc so went for the higher end stuff. I'm pretty sure they would be fine, but maybe someone else who has actually used them could confirm that these would do the job okay.
I'm sure it would be a lot easier this way than trying to get started with the big ones, especially if it's just a one off.
However, if you want to have the benefits that FPGAs can bring for the future also, I'd take the more involved route (FPGAs from Xilinx, Altera, Actel etc) and gain some valuable tools/knowledge.

2010\08\21@080355 by Isaac Marino Bavaresco

flavicon
face
Em 21/8/2010 07:57, Oli Glaser escreveu:
{Quote hidden}

CPLDs are better for small designs because they don't boot from an
external memory like most FPGAs.

There are CPLDs with macrocell counts from a few tens to a few hundreds.
A macrocell is much more than simply "one gate".

A CPLD with 64 or 72 macrocells may fulfill the OP's needs, and they can
be found for around $1 to $3.


Isaac

__________________________________________________
Fale com seus amigos  de graça com o novo Yahoo! Messenger http://br.messenger.yahoo.com/

2010\08\21@085421 by Oli Glaser

flavicon
face


--------------------------------------------------
From: "Isaac Marino Bavaresco" <.....isaacbavarescoKILLspamspam.....yahoo.com.br>
Sent: Saturday, August 21, 2010 1:04 PM
To: "Microcontroller discussion list - Public." <EraseMEpiclistspam_OUTspamTakeThisOuTmit.edu>
Subject: Re: [EE] Programmable logic

> CPLDs are better for small designs because they don't boot from an
> external memory like most FPGAs.

True, that's why I suggested the Actel parts though if the OP does want to go for an FPGA, as they are mostly based on Flash memory.

> There are CPLDs with macrocell counts from a few tens to a few hundreds.
> A macrocell is much more than simply "one gate".

Yes, as I understand it, a macrocell can be different for different vendors, though usually something like a 3 or 4 input LUT with a mux and flip flop, enabling them to perform a range of different functions. The macrocell count is maybe the best thing to look at when deciding whether the chip can do the job. On the FGPA side, I think the Actel ProASIC3 15K gate has 128 macrocells, whereas the Atmel AVF20V8B SPLD only has 300 gates equivalent to 8 macrocells.

> A CPLD with 64 or 72 macrocells may fulfill the OP's needs, and they can
> be found for around $1 to $3.

If this is the case then maybe the 15K gate, 128 macrocell Actel part (£2..77 from Mouser) mentioned above may do okay, and requires no external memory. Also, once the tools are acquired and the hang got of them, then there is a lot more power available later on.
Or use a CPLD of some sort as you say...

2010\08\21@094251 by Isaac Marino Bavaresco

flavicon
face
Em 21/8/2010 09:54, Oli Glaser escreveu:
{Quote hidden}

The Xilinx XC9572XL-10VQG64C is a 10ns 3.3V CPLD with 72 macrocells, 52
I/O pins in a TQFP-64 package and costs just $2.40 at Digikey.

Altera's EPM3064ATC44-10N is a 10ns 3.3V CPLD with 64 macrocells, 34
I/Os in a TQFP-44, costs $2.55

There are faster models at a higher cost. I don't know the voltage the
OP needs, but there are 5V versions available.


Isaac

__________________________________________________
Fale com seus amigos  de graça com o novo Yahoo! Messenger http://br.messenger.yahoo.com/

2010\08\21@101020 by RussellMc

face picon face
> If this is the case then maybe the 15K gate, 128 macrocell Actel part (£2.77
> from Mouser) mentioned above may do okay, and requires no external memory..
> Also, once the tools are acquired and the hang got of them, then there is a
> lot more power available later on.
> Or use a CPLD of some sort as you say...

What are the considerations and differences that will make a device
like this acceptable and red tape free when a micro-controller is not?
Are these differences meaningful in reality rather than just bureaucratically?
Does the use of a device like this have issues that a microcontroller
doesn't, and are they liable to be significant?
How long is a piece of string?


                  R

2010\08\21@105818 by Michael Watterson

face picon face
 On 21/08/2010 15:09, RussellMc wrote:
>> If this is the case then maybe the 15K gate, 128 macrocell Actel part (£2.77
>> from Mouser) mentioned above may do okay, and requires no external memory.
>> Also, once the tools are acquired and the hang got of them, then there is a
>> lot more power available later on.
>> Or use a CPLD of some sort as you say...
> What are the considerations and differences that will make a device
> like this acceptable and red tape free when a micro-controller is not?
> Are these differences meaningful in reality rather than just bureaucratically?
> Does the use of a device like this have issues that a microcontroller
> doesn't, and are they liable to be significant?
> How long is a piece of string?
>
>
>
It depends :)

Consider a well defined state machine implemented with an EPROM and a Latch..

Then consider the same functional specification (with a HW timer interrupt for added amusement) on a 16F in C by someone with little experience of the mysteries of C, 16F/PIC architecture and a naive faith in C libraries and such things as "printf" (abhorred by /Barjne Stroustrup )/

It depends on what tools and methodologies are used to create a CPLD or FPGA design. Good Digital Logic designers in my experience put in less bugs than good Programmers. Most Programmers are not good.

If a CPLD is created by someone with little HW design experience it might be worse than a PIC programmed in C. It partly depends on the nature of the application too.

These are not easy questions to give answers to. If a Company historically had people that designed with Relay Logic and Mechanical Sequencers and some SSI gates latterly there may be project management issues with Software approach until suitable staff is in place.

A change to CPLD or FPGA may be higher risk changing from SSI 7400 series and relays than PIC micro (even with C) as a different approach to implementation and a completely new tool set is needed. It's not HW design in classic sense (you'd be mad to use Schematic Mode entry rather than Verilog or VHDL) nor SW design in the conventional programming sense.  It's probably easier  for an experienced "Beginner" HW engineer to PIC (using JAL, Assembler or C) to implement a simple relay logic board than for same HW engineer to do it with CPLD or FPGA.  It's kind of easier for me to do DSP on an FPGA than translate an existing Logic board.  I have a Spartan 3 starter kit with all the tools for FPGA and CPLD (the board has both, plus HW ethernet, 2 x 16 ch LCD, RS232, VGA, rotary encoder, LEDs, switches...). Design from scratch in VHDL is far easier than migrate existing  HW design.

The Coolrunner CPLD and Spartan 3E FPGA have a much harder learning curve than ARM, x86, Z80 or PIC for a HW engineer.

If you ignore 16F and go 18F + JAL (Most the 18pin, 28pin and 40pin 18Fs work in 16F development boards, the USB versions only requiring a connector and 220uF ceramic cap soldered between two pins) then the learning curve is very gentle and short.


How long is a piece of string?


I don't know. It depends on the application and the expertise of the Company. The ban might equally be valid for CPLD and FPGA. Or More so. A change to unfamiliar technology for a Company needs to be tried out in parallel to commercial development to evaluated the Specification,  Design,   verification & test, training & Expertise and Management issues.

2010\08\21@113649 by RussellMc

face picon face
>>> If this is the case then maybe the 15K gate, 128 macrocell Actel part ....

>> What are the considerations and differences that will make a device
>> like this acceptable and red tape free when a micro-controller is not?

> Consider a well defined state machine implemented with an EPROM and a Latch.

> Then consider the same functional specification (with a HW timer
> interrupt for added amusement) on a 16F in C by someone with little
> experience of the mysteries of C, 16F/PIC architecture and a naive faith
> in C libraries and such things as "printf" (abhorred by /Barjne
> Stroustrup )/

Or
:-)
Consider a PIC 18F solution, designed en toto  by eg O. Lathrop Esq &
Co and programmed by them in assembler, compared with a
programmable-hardware solution designed and implemented bya competent
EE who has no CPLD / FPGA / Verilog / VHDL experience (as is probably
the case here). Without casting any aspersions whatsoever on Denny's
abilities in any field that he considers himself competent in, the
choice seems extremely clear unless there is an extraordinarily
unusual set of prerequisites. If the programmable hardware solution
was designed and implemented by someone with vast expertise the best
solution may well differ.



2010\08\21@114904 by Michael Watterson

face picon face
 On 21/08/2010 16:36, RussellMc wrote:
{Quote hidden}

I think we have the same hymn book.

Who is doing the design is often more important than technology

2010\08\21@115030 by John Gardner

picon face
Why C reduces reliability seems intuitive :)  but no one
has really addressed the why's of say, ladder-logic vs.
asm+micro...

Is it simply the reduced number of possible states of a given
clutch of parts/instructions?

Jac

2010\08\21@125132 by Sean Breheny

face picon face
Hi Denny,

Why would you not consider a PLC for this application? I agree with
Carl that it seems like a good fit.

Also, what do you mean by "consumed"? Do you mean that you have had to
replace lots of PLCs due to failure?

Sean


On Fri, Aug 20, 2010 at 4:00 PM, Denny Esterline <KILLspamdesterlineKILLspamspamgmail.com> wrote:
{Quote hidden}

>

2010\08\21@132741 by Carl Denk

flavicon
face
The link I provided does have a free software tutorial available, here's the link:
http://www.tri-plc.com/trilogiintro.htm
click both on examples and versions. This would be an excellent way to learn about PLC's.

As for failures, I have had none on the PLC side hardware side, and mine are connected right next to an ATS (Automatic transfer switch) that with a large clunk switches maybe 100 amp load from utility power to generator and back again many times. The Tri-PLC is only one brand out there, I have been satisfied with their hardware and support, check out their forum that tech support actively participates. If you want to be very bullet proof, I would suggest the Modicon. I have been around 100's of them (not working with them) in a heavy industrial plant, and only rarely did I see one changed out.

They are available in numerous sizes, from a few I/O's, timers and counters, and I consider mine a moderate sized with more than 80 digital I/O, 8 ADC, pulse measurement (I use to measure generator frequency (60 Hz)), 64 timers, 64 counters.  If I could pick it up easily from nothing, it can't be that hard. Besides many times, just with different programming, same or similar unit can handle much different tasks.

Of course if you want job security, by being the only one to maintain, this wouldn't be it! And it is possible to copy protect the software.

On 8/21/2010 12:51 PM, Sean Breheny wrote:
{Quote hidden}

>> -

2010\08\22@161225 by Denny Esterline

picon face
On Sat, Aug 21, 2010 at 1:15 AM, RussellMc <RemoveMEapptechnzspamTakeThisOuTgmail.com> wrote:
{Quote hidden}

A lot of other people have commented further on this, but what it
comes down to is this: Yes, one of the end users will be nuclear power
stations. This product is nowhere near the "nuclear" part of the
facility, but all the same red tape applies. We can produce a
microprocessor version, there is a path to getting the software
accepted, the process is so onerous that it makes the project an
economical non-starter.

To be more specific, this product exists today and there have been
several hundred sold over the past 10 years or so. Like I said, it's
currently built around a half dozen octal base relays. There are
several compelling reasons to redesign this product, the most acute
being one of the critical parts being obsoleted from the manufacturer
and no direct cross being available. I have worked out a replacement
design based on discreet logic. It's up to 35 gates (11 logic chips),
four 555 timers and a couple comparators. While I'm certain this will
duplicate all of the original functionality, refresh the supply chain
and save about $500 per unit in fabrication costs (wiring all those
octal bases is surprisingly expensive), I'm not convinced discreet
logic the best tool to throw at the problem. I'm very much aware of
the old adage - "if the only tool you have is a hammer, every problem
looks like a nail". To continue the metaphor, I was hoping to learn
more about these new-fangled screwdrivers I keep hearing about.

I looked at the XC9500 series from Xilinx, they certainly look capable
of the task and the price is plenty reasonable, but then I noticed
that the eval boards for the XC9500 series have been obsoleted.
Where's that leave me, are the chips next on the chopping block?
Then there's Atmel, I'm not seeing eval boards / starter kits for
their lower end parts either. And while I don't know about Atmel's
programmable logic division, we all know their reputation for
discontinuing microcontrollers seemingly at random.

-Denny

2010\08\22@170631 by M.L.

flavicon
face
On Sun, Aug 22, 2010 at 4:12 PM, Denny Esterline <desterlineEraseMEspam.....gmail.com> wrote:
>
> To be more specific, this product exists today and there have been
> several hundred sold over the past 10 years or so. Like I said, it's
> currently built around a half dozen octal base relays. There are
> several compelling reasons to redesign this product, the most acute
> being one of the critical parts being obsoleted from the manufacturer
> and no direct cross being available. I have worked out a replacement
> design based on discreet logic. It's up to 35 gates (11 logic chips),
> four 555 timers and a couple comparators. While I'm certain this will
> duplicate all of the original functionality, refresh the supply chain
> and save about $500 per unit in fabrication costs (wiring all those
> octal bases is surprisingly expensive), I'm not convinced discreet
> logic the best tool to throw at the problem. I'm very much aware of
> the old adage - "if the only tool you have is a hammer, every problem
> looks like a nail". To continue the metaphor, I was hoping to learn
> more about these new-fangled screwdrivers I keep hearing about.
>


I'm coming into this late in the thread.
If you haven't seen this yet then you might be interested:
<http://digilentinc.com/Products/Detail.cfm?NavPath=2,400,798&Prod=CMOD>

Verbatim from Digilent Inc.:
"-All C-MOD boards are compatible with the free Xilinx WebPack tools
-Design can easily be ported between C-MODs using different CPLD device families
-Once programmed, CPLD designs are non-volatile
All available user I/O signals are brought out to DIP pins"


"C-Mod boards combine a Xilinx CPLD, a JTAG programming port, and
power supply circuits in a convenient 600-mil, 40-pin DIP package.
C-Mods are ideally suited for breadboard or other prototype circuit
designs where the use of small surface mount packages is impractical."


--
Martin K

2010\08\22@185801 by YES NOPE9

flavicon
face
You may want to take a look at this FPGA vendor
http://www.knjn.com/ShopBoards_USB2.html  <=== online shopping
http://www.knjn.com/ShopBoards_RS232.html

http://www.fpga4fun.com/  <=== informational site

Best
Gu

2010\08\23@000446 by RussellMc

face picon face
{Quote hidden}

OK. That answers most of the above questions - it's not necessarily
better or safer etc but it's achievable this way.

FPGA despite their many merits are probably excluded if they load from
external memory as some people may think this makes them
indistinguishable from being programmable - regardless of the
arguments for or against, it's not an argument you want to have to get
into.

CPLD looks good - almost certainly able to meet your needs and is
essentially as  self contained as a block of discrete parts  logic.

Ladder logic may shorten the learning process and also allow
comfortable comparison with the old solution for lookers on. "This is
just like the relay solution because ..."

Martin's  suggested
digilentinc.com/Products/Detail.cfm?NavPath=2,400,798&Prod=CMOD
looks like a good place to start looking for rolling-your-own product
in small volume.





 Russell

2010\08\23@013443 by Wouter van Ooijen

face picon face
> A lot of other people have commented further on this, but what it
> comes down to is this: Yes, one of the end users will be nuclear power
> stations. This product is nowhere near the "nuclear" part of the
> facility, but all the same red tape applies. We can produce a
> microprocessor version, there is a path to getting the software
> accepted, the process is so onerous that it makes the project an
> economical non-starter.

ROFL! So up to today no red-taper has the idea to extend the rules for software to extend programmable logic?

--
Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

2010\08\23@030258 by Denny Esterline

picon face
On Sun, Aug 22, 2010 at 10:34 PM, Wouter van Ooijen <EraseMEwouterspamvoti.nl> wrote:
>> A lot of other people have commented further on this, but what it
>> comes down to is this: Yes, one of the end users will be nuclear power
>> stations. This product is nowhere near the "nuclear" part of the
>> facility, but all the same red tape applies. We can produce a
>> microprocessor version, there is a path to getting the software
>> accepted, the process is so onerous that it makes the project an
>> economical non-starter.
>
> ROFL! So up to today no red-taper has the idea to extend the rules for
> software to extend programmable logic?
>
> --
>
> Wouter van Ooijen
>

I don't understand all of the details, I suspect that buried somewhere
in the bowels of the bureaucraptic system is some brain dead paper
pusher that has a form with a check box labeled "does this product
contain a microproccessor" with instructions to verify some other
mountain of paper if it's checked.
As long as I don't have to check the box, I'm golden.

-Denn

2010\08\23@031337 by Denny Esterline

picon face
> FPGA despite their many merits are probably excluded if they load from
> external memory as some people may think this makes them
> indistinguishable from being programmable - regardless of the
> arguments for or against, it's not an argument you want to have to get
> into.
>

Probably, but beyond that, I wonder if a full FPGA isn't a bit of
overkill to replace <40 logic gates?

{Quote hidden}

Digilent's definitely one of the ones I'm looking at, incedently
they're the ones that (recently?) obsoleted their Xlilinx XC9500 dev
board I mentioned.

-Denny

2010\08\23@032129 by Wouter van Ooijen

face picon face
> I don't understand all of the details, I suspect that buried somewhere
> in the bowels of the bureaucraptic system is some brain dead paper
> pusher that has a form with a check box labeled "does this product
> contain a microproccessor" with instructions to verify some other
> mountain of paper if it's checked.
> As long as I don't have to check the box, I'm golden.

I hope you live far far away from the plant that is operated in this way...

Maybe you can get away with including a macroprocessor ;) Or a microsequencer! (ROM + register)

--
Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

2010\08\23@041001 by RussellMc

face picon face
> > pusher that has a form with a check box labeled "does this product
> > contain a microproccessor" with instructions to verify some other
> > mountain of paper if it's checked.

Those who live by the sword die by the sword.

If it DID contain one, no matter how hard you tried to explain that it
was triply redundant, watch dog protected, logic verified and tye code
checked by experts, you'd STILL have top go through the formal
procedures. They may share attributes with other systems, but a
microprocessor isn't a FPGA or a CPLD etc. Right? OK - rules are rules
and you have to follow them to the letter.

Now go and ask suitably qualified experts what a microprocessor IS.
Ask them what a microcontroller IS. Is there a clear distinction? (You
may have to choose your experts :-) ). If the product really really
really has nothing to do with things nuclear or safety or security or
.... you may wish to invoke this
rules-must-be-followed-to-the-letter-distinction. But probably not.
BUT it just may be that the red tape handler is also able to be
persuaded that a uC and a uP ARE sufficiently different, especially in
cases where it  really really really doesn't matter, to actually
approve your request when all facts are made know.

eg "Does this device contain a microprocessor".
"No, but see explanatory note below ...).

I suspect this is not a useful avenue, but ... :-)



2010\08\23@050857 by Chris Roper

picon face
Or you may get away with replacing the part, that is no longer
available, without having to say  you redesigned with a
Microprocessor. Take a look at this:

http://www.divelbiss.com/Products/CatFamDetails.asp?ProdCatID=1&ProdFamID=23

On 23 August 2010 10:09, RussellMc <RemoveMEapptechnzEraseMEspamEraseMEgmail.com> wrote:
>> > pusher that has a form with a check box labeled "does this product
>> > contain a microproccessor" with instructions to verify some other
>> > mountain of paper if it's checked

2010\08\23@092928 by Xiaofan Chen

face picon face
On Mon, Aug 23, 2010 at 4:12 AM, Denny Esterline <RemoveMEdesterlinespam_OUTspamKILLspamgmail.com> wrote:

> I looked at the XC9500 series from Xilinx, they certainly look capable
> of the task and the price is plenty reasonable, but then I noticed
> that the eval boards for the XC9500 series have been obsoleted.
> Where's that leave me, are the chips next on the chopping block?

XC9500 is definitely old. I am not so sure if they are going
to obsolete them soon. Xilinx seems to keep parts rather long
as well. But then the price could be up significantly. I've
seen US$26 part in one BOM for a very old Xilinx FPGA.
Redesigning it does not make economic sense over the
years due to the low quantity.


-- Xiaofa

2010\08\23@094552 by RussellMc

face picon face
>From what he said, PLC is also a verboten word.



                Russel

2010\08\23@101719 by Wouter van Ooijen

face picon face
RussellMc wrote:
>>From what he said, PLC is also a verboten word.

Not so much PLC itself, but the most likely (only reasonable?) implementation.

--
Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

2010\08\23@104406 by Sean Breheny

face picon face
Hmm. There are safety-certified PLCs available, although they are
several thousand USD. Seems to me that modern FPGAs are just as able
to have subtle programming errors and failure modes as
microcontrollers.

Sean


On Mon, Aug 23, 2010 at 9:45 AM, RussellMc <RemoveMEapptechnzTakeThisOuTspamspamgmail.com> wrote:
> >From what he said, PLC is also a verboten word.
>
>
>
>                 Russell
>

2010\08\23@122529 by Charles Craft

picon face

What is "pyro sdafing" ?


-----Original Message-----
{Quote hidden}

2010\08\23@133237 by RussellMc

face picon face
> What is "pyro sdafing" ?

A typo :-).

BUT  you'll find "pyro safing" in the precise context that I intended
it in the following references
It means "making sure that things that are meant to go woosh and/or
bang at intended times don't do so at unintended times".

http://www.nasaspaceflight.com/2008/12/endeavour-to-visit-three-states/

http://www.nasaspaceflight.com/2009/05/endeavour-rollaround-atlantis-departing-california/

http://lsda.jsc.nasa.gov/books/apollo/s5ch1.htm

___________________

Here you get 20+ uses of "safing" including the glorious -

                          Comment: How is the “Drag Chute Pyro Safing” in the
                          Safing Sequence different than the “Drag
Chute Safing”
                          during the initial Safing Processes?

That's something like "How much wood would a wood chuck chuck if a
woodchuck could chuck wood?"
And actually uses "pyro safing" in context :-).

    http://www.abovetopsecret.com/forum/thread261344/pg2

2010\08\27@141230 by Herbert Graf

picon face
On Fri, 2010-08-20 at 13:00 -0700, Denny Esterline wrote:
> On Fri, Aug 20, 2010 at 11:25 AM, Carl Denk <cdenkSTOPspamspamspam_OUTwindstream.net> wrote:
> > You say "Programmable Logic", add one more word, "Controller" >> PLC ,
> > and your application points to that. There are numerous out there, you
> <snip>
>
> Uhh, errr, thanks for the suggestion, but no.
> I'm actually quite familliar with PLCs, one of our other product lines
> has consumed about 250 of them the last couple years. This is
> definitly not an application I want a PLC for.
> I used the term "programmable logic" as a genaric term for PALs, GALs,
> PLAs, PLDs, SPLDs, CPLDs, FPGAs etc. Seems like every manufacturer has
> a silghtly different name for them, or several at various levels of
> capabilities.

The Xilinx 9536 (it's a CPLD, given your application going to an FPGA
isn't worth it) should fit the bill (if you need more logic, the 9572
has twice the capacity with the same footprint if you choose your
package right).

The Xilinx webpack is a free dev environment that gives you all you need
to get going.

I've never used "schematic capture" type solutions for programmable
logic, so I can't comment there.

If you know C, then verilog will feel familiar. I'd steer clear of VHDL,
it can be very "interesting" in the way it does things, and is more of a
hill to climb.

TTYL

2010\08\27@142715 by jim

flavicon
face

I have used schematic capture for these very CPLD's in the past.  It's
a simple process.  Just draw the
schematic using the schematic capture app part of the Zilinx
environment, compile it, and burn it into your
part.  You have to do a little bit of house keeping before you get that
far though such as determining what
inputs and outputs you want, giving each a name, etc., etc., but it
isn't difficult at all.  At least not
with the stand alone package. I have not used the web version, so I
can't speak for it.    The last time I
looked, (some time ago), you could still download the older stand alone
packages.  If the web version don't
work for you, you could try the standalone package.


Regards,

Jim

> ---{Original Message removed}

2010\08\27@165920 by Herbert Graf

picon face
On Fri, 2010-08-27 at 11:27 -0700, spamBeGonejimSTOPspamspamEraseMEjpes.com wrote:
> I have used schematic capture for these very CPLD's in the past.  It's
> a simple process.  Just draw the
>  schematic using the schematic capture app part of the Zilinx
> environment, compile it, and burn it into your
>  part.  You have to do a little bit of house keeping before you get that
> far though such as determining what
>  inputs and outputs you want, giving each a name, etc., etc., but it
> isn't difficult at all.  At least not
>  with the stand alone package. I have not used the web version, so I
> can't speak for it.    The last time I
>  looked, (some time ago), you could still download the older stand alone
> packages.  If the web version don't
>  work for you, you could try the standalone package.

Just to clarify:

the webpack isn't a "web version". It's simply a version of the
standalone software that you install from the web, it's still an app
running on your PC.

It's pretty much identical to ISE, you're just limited in the size of
parts it works for (for pretty much all hobbyists out there the limit is
more then high enough to never worry about).

TTYL

2010\08\27@170607 by Denny Esterline

picon face
> The Xilinx 9536 (it's a CPLD, given your application going to an FPGA
> isn't worth it) should fit the bill (if you need more logic, the 9572
> has twice the capacity with the same footprint if you choose your
> package right).

That one one of the first ones I considered. But I see Digilent has
discontinued it's dev tools for the xc9500 series and judging by part
pricing, I suspect Xilinx is pushing new designs towards the
CoolRunner II parts (<$1.50 in singles). My new dev kit should be here
Monday.



{Quote hidden}

For this, I'm willing to try schematic capture, it is only about 40
gates after all. But I'm sure I need to learn some HDL at some point.

-Denn

2010\08\27@180816 by jim

flavicon
face

Herb,

I see.  Thanks for the clarification.  I haven't used that version
before, so I didn't know what it was.
I assumed from the name that is was an online version.  But thinking
about it now, I should have known
it wasn't an online version.  Oh well, live and learn.  
However, if it is basically the same as the ISE, then schematic capture
entry of the logic needed will
be relatively easy.  It's basically a matter of placing logic symbols
and connecting them to achieve the
function desired.  
Once one gains proficiency with CPLD's, working with FPGA's is more
easily understood.  And FPGA's can  hold many times more complex functions than a CPLD.

And if you work with Mixed Signal FPGA's, you can create your own Mixed
Signal ASIC type circuits and
chipset(s).

Regards,

Jim



> ---{Original Message removed}

2010\08\27@180910 by Dwayne Reid

flavicon
face
At 03:06 PM 8/27/2010, Denny Esterline wrote:
> > The Xilinx 9536 (it's a CPLD, given your application going to an FPGA
> > isn't worth it) should fit the bill (if you need more logic, the 9572
> > has twice the capacity with the same footprint if you choose your
> > package right).

I'm just back from holidays and am thus just catching up with this thread, but I have a slightly off-the-wall suggestion to offer.

The guys at Dangerous Prototypes and The Gadget Factory have designed a kick-ass logic analyzer based on a Xilinx FPGA.  Its available from Seeed Studios for less than US $45, including shipping.  Note that the price does NOT include the cables and clips if you plan to order one for use in its intended purpose.

The board contains two large (ish) chips: the PIC18F24J50 (USB interface) and the Spartan3E XC3S250E FPGA.  It also contains the input buffer chip, boot ROM and 2 or 3 PSU regulators.  The regulators can supply at least 150mA - might be more.

16 of the FPGA i/o pins are fed by a 5V tolerant input buffer and are thus input only.  Another 16 i/o lines come out to header pin holes and can be input or output.  The PIC is used only for loading the boot ROM that the FPGA reads on power-up and is therefore not a factor in your 'no micros' policy - the FPGA doesn't even have to know that the PIC exists (but it can).

This board is a dandy FPGA development platform and is tidy enough that you can use the board as-is in your final product.  All that you need to do is to supply the input and output signal conditioning.  The on-board headers make that easy.

Some links:
<www.seeedstudio.com/depot/open-workbench-logic-sniffer-p-612.html?cPath=61_68>
<http://dangerousprototypes.com/docs/Open_Bench_Logic_Sniffer>

The development tools are free downloads.

Hope this helps!

dwayne

-- Dwayne Reid   <KILLspamdwaynerspamBeGonespamplanet.eon.net>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
(780) 489-3199 voice          (780) 487-6397 fax
http://www.trinity-electronics.com
Custom Electronics Design and Manufacturing

2010\08\27@194330 by Michael Watterson

face picon face
 On 27/08/2010 19:12, Herbert Graf wrote:
>
> If you know C, then verilog will feel familiar. I'd steer clear of VHDL,
> it can be very "interesting" in the way it does things, and is more of a
> hill to climb.
>
> TTYL
>
That's exactly why he should learn VHDL and then Verilog. The Familarity of Verilog to a C programmer is a trap.
It's not a program in the normal C sense at all. But a way of specifing behaviour of a totally parallel system.

2010\08\27@194718 by Michael Watterson

face picon face
 On 27/08/2010 22:06, Denny Esterline wrote:
{Quote hidden}

Coolrunner II is a good choice.

Smaller systems can be done by schematic capture. But it's worth learning VHDL *and* Verilog (even if eventually you only use one of them) if either you get stuck or you are doing anything larger.

2010\08\28@122719 by Herbert Graf

picon face
On Sat, 2010-08-28 at 00:43 +0100, Michael Watterson wrote:
> On 27/08/2010 19:12, Herbert Graf wrote:
> >
> > If you know C, then verilog will feel familiar. I'd steer clear of VHDL,
> > it can be very "interesting" in the way it does things, and is more of a
> > hill to climb.
> >
> > TTYL
> >
> That's exactly why he should learn VHDL and then Verilog. The Familarity
> of Verilog to a C programmer is a trap.
> It's not a program in the normal C sense at all. But a way of specifing
> behaviour of a totally parallel system.

Very true, but to be frank I find it EASIER to write "horrible" code in
VHDL then in Verilog, perhaps it's just a personal way of thinking, but
I just find writing good code easier in Verilog.

That said, VHDL is on the downhill slope, more and more companies are
switching to exclusively use Verilog (my company completely switched
over 3 years ago). While a ton of code is out there in VHDL, the newest
development is all Verilog.

Even today, some code is now being written in system verilog. The
benefit there is system verilog is a superset of verilog, so learning it
isn't too bad (although it is VERY different in some ways, my first look
at system verilog felt like I was living on a different planet).

I stand by my recommendation: bypass VHDL, it's archane in syntax, it's
being used less and less, and Verilog is easier to learn (for most).

TTYL

2010\08\28@141227 by Michael Watterson

face picon face
 On 28/08/2010 17:26, Herbert Graf wrote:
{Quote hidden}

Though VHDL isn't arcane to people with experience of Modula-2 and ADA co-routines and syntax and of experience of Occam.

I think USA and US companies are more Verilog and other folks VHDL. But even if you are going to use Verilog it will help in understanding to do some VHDL tutorials.
Also of course you can use both in the same project, especially if some off the shelf core is in the opposite language to the one you are writing in.

2010\08\28@194608 by Herbert Graf

picon face
On Sat, 2010-08-28 at 19:12 +0100, Michael Watterson wrote:
> Also of course you can use both in the same project, especially if some
> off the shelf core is in the opposite language to the one you are
> writing in.

You often can, but I'd STRONGLY recommend not mixing languages,
especially those new to HDL.

While most tools "support" mixed language mode, the most time consuming
tool related problems I've had dealt with the mixing of code. What the
vendors SAY works isn't always what ends up working. As recently as two
weeks ago I hit a bug in the Xilinx toolkit when trying to call verilog
code from a VHDL tb.

If you're using a chunk of IP that is only available in the "other"
language then I suppose you don't have a choice. Personally, I just skip
any IP that isn't of the language I need and either find one that is in
the language I need, or write it myself (sometimes writing IP yourself
takes less time then figuring out how to integrate some IP and debug
it's bugs).

TTYL

2010\08\29@012018 by Denny Esterline

picon face
On Sat, Aug 28, 2010 at 4:46 PM, Herbert Graf <EraseMEhkgrafspamEraseMEgmail.com> wrote:
{Quote hidden}

Wow, all good stuff. Definitely useful to those just entering the
arena (like me).

Help me understand another aspect of this. In doing actual production
with these parts, how are you handling the final programming of the
parts? You doing it? Your assembler? What tools are being used? etc.

Thanks,
-Denn

2010\08\31@112131 by Herbert Graf

picon face
On Sat, 2010-08-28 at 22:20 -0700, Denny Esterline wrote:
> Wow, all good stuff. Definitely useful to those just entering the
> arena (like me).
>
> Help me understand another aspect of this. In doing actual production
> with these parts, how are you handling the final programming of the
> parts? You doing it? Your assembler? What tools are being used? etc.

Hello Denny,
the work I've done doesn't reach production since I'm more in the
prototype, R&D and verification side of the equation.

For actual production the options are pretty similar as with PICs:

1) Get the manu to program the part
2) Get the assembly house to program the part
3) Program the part yourself
4) Make the design "self programming", i.e. many USB controllers out
there have a "boot strap" type mode to load the firmware of the USB
chips microcontroller from the host PC; adding a step where that
microcontroller programs the FPGA isn't too hard to do.
5) Put an SD or CF card slot on your board (along with the the right
chip, i.e. SystemACE from Xilinx) and supply the SD or CF card with your
product.

All these options depend on volume and cost, if your volume is small
then #3 is probably easier, if you volume is massive #1 or #2 is
probably the best choice, if your interface is of the right structure #4
might be best, and if cost is no object #5 might be the right choice.

TTYL

More... (looser matching)
- Last day of these posts
- In 2010 , 2011 only
- Today
- New search...