Searching \ for '[OT] Opinion on ASICs' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: techref.massmind.org/techref/index.htm?key=opinion+asics
Search entire site for: 'Opinion on ASICs'.

Exact match. Not showing close matches.
PICList Thread
'[OT] Opinion on ASICs'
1998\12\09@005818 by Sean Breheny

face picon face
Hi all,

This is straying quite a bit OT as I am wont to do, but here goes:

It really seems to me that for many (at least) low power circuits, there is
a real trend toward more and more integration and less and less board-level
electronics, for obvious reasons. IF and WHEN do most of you feel that most
devices will consist of just one or more ASICs connected together with
almost no external components at all? Of course, I suppose that another
possibility may be that FPGAs and reconfigurable analog chips (consisting
of op-amp and switched cap filter arrays, or ADCs/DACs and a DSP processor)
may perform this function, instead of complete ASICs.

This has been a concern of mine for some time because I am fearful that
hobbists may eventually be left out of the loop if manufacturers go totally
ASIC for small-signal and digital applications and the production process
for ASICs is still too expensive for hobbists to get involved in.

Are my fears unfounded? I hope so.

Thanks,

Sean

+-------------------------------+
| Sean Breheny                  |
| Amateur Radio Callsign: KA3YXM|
| Electrical Engineering Student|
+-------------------------------+
Save lives, please look at http://www.all.org
Personal page: http://www.people.cornell.edu/pages/shb7
spam_OUTshb7TakeThisOuTspamcornell.edu  Phone(USA): (607) 253-0315 ICQ #: 3329174

1998\12\09@072422 by paulb

flavicon
face
Vance Gloster wrote, quoting Sean Breheny:

>> This has been a concern of mine for some time because I am fearful
>> that hobbists may eventually be left out of the loop if manufacturers
>> go totally ASIC for small-signal and digital applications

> Yes, and no.  The bad news is the manufacturers are going totally
> ASIC.  The good news is the PIC is the ultimate ASIC platform.  PIC
> code is MUUUUUCH cheaper than doing even a semicustom, much less a
> full custom chip.

 We haven't fared badly.  For over thirty years since the IC became
commonplace, at every advance in technology, and equally with hardware
*and* software, it seemed that serious production is so complex that
no-one without the resources of a large company could ever afford to
develop a custom design either for personal use or market.

 This list for one seems to give the lie to that argument.  You
wouldn't go out and build yourself a TV or a CD player, a satellite
receiver or a Pentium motherboard and those devices are certainly highly
integrated because they are high-volume consumer appliances.

 These things however no more represent or fulfil the whole of people's
requirements than does MS Word 10 running under Windoze 98 or whatever.
AAMOF, except for the "soma" sub-culture, these things really have very
*little* real impact on our lives (well, maybe I should be careful about
discounting the impact of an e-mail machine on mine!).

 It would be sad indeed if we could be satisfied with these appliances,
but of course there will always be small markets, and indeed a *lot* of
them.  These niches will require inexpensive development systems like
the ICP and Flash PICs, and I'm sure there will be enough of them to
make catering for this market worthwhile for a long time to come.

 Something else comes to mind.  ASICs actually have a severe limitation
called "feature creep".  If you put *everything* into one ASIC, or
indeed mask-programmed MCU, next week the marketing department (or maybe
the service department?) decides it's all wrong, needs something else.
Using a collection of simpler parts on the PCB allows *both* the
manufacturer and the aftermarket experimenter to "hack" the design and
tweak it.
--
 Cheers,
       Paul B.

1998\12\09@092955 by Harrison Cooper

flavicon
face
take a look and compare any PC product.  What use to be a bunch of
'jellybean' logic, and maybe a custom chip..is now maybe one or two custom
chips.

I took nearly all the simple logic functions in my latest redesign, and did
put them all in FPGA's.  Bottom line of cutting costs drives us that
way...and also the increase functionality of designs, while keeping the same
board size.

I think both Altera and Xilinx are trying to hit the small user markets with
some of thier products.

1998\12\09@102241 by Reginald Neale

flavicon
face
Sean said:

>
>This is straying quite a bit OT as I am wont to do, but here goes:
>
>It really seems to me that for many (at least) low power circuits, there is
>a real trend toward more and more integration and less and less board-level
>electronics, for obvious reasons. IF and WHEN do most of you feel that most
>devices will consist of just one or more ASICs connected together with
>almost no external components at all? Of course, I suppose that another
>possibility may be that FPGAs and reconfigurable analog chips (consisting
>of op-amp and switched cap filter arrays, or ADCs/DACs and a DSP processor)
>may perform this function, instead of complete ASICs.
>
>This has been a concern of mine for some time because I am fearful that
>hobbists may eventually be left out of the loop if manufacturers go totally
>ASIC for small-signal and digital applications and the production process
>for ASICs is still too expensive for hobbists to get involved in.
>
>Are my fears unfounded? I hope so.
>

If I understand your point correctly, I think they are. IMO ASICs aren't
going to take over the world, or even consumer applications. Application
flexibility always has value, ESPECIALLY at high levels of integration.
Yesterday someone on this list noted that their old app notes for a
quadrature encoder called for the use of a hardware debouncer chip; they
had never even heard of it. That approach is dinosaur technology now, along
with half of your logic handbook's discrete function logic chips. Largely
displaced by fifty-cent microcontrollers, which are proliferating at least
as fast as ASICs.

Low-cost controllers, programmers and development systems are the
designer's best friend.

Reg Neale

1998\12\09@103653 by Harrison Cooper

flavicon
face
There was an interesting article about Xilinx new Virtex chips, being used
in remote apps...namely satellites.  They can be reconfigured over a data
link, be it RF or hardwire.  ASIC's in their true form are not going to take
over the world, but I do believe that FPGA's will.....and already are taking
the place of ASIC's in smaller quantities....

1998\12\09@111633 by Sean Breheny

face picon face
Hi all,

Thanks for all the replies.

My main concern was that non-reconfigureable ASICs would dominate and
therefore leave the hobbist no real way to build useful stuff. I can see
from your responses that this is probably not true. However, most of your
responses revolved around full digital FPGAs. What about analog stuff? What
do you feel will happen there? Will it continue to be implemented in
several chips? Go ASIC? Be replaced by DSP? Be implemented by
reconfigureable analog chips? Or a combination of these?

I love digital electronics, especially microcontrollers. However, I still
love analog/mixed signal stuff. I have also bee long concerned by the
apparent fact that very few people design at the transistor level anymore
would harm the industry by leaving hobbists(the best source of engineers,
IMHO) ignorant of and uninterested in electronics below the level of
configuring a customizable IC. I can already see this here at my school.
>From what I can tell, out of my EE class of about 2,000 students, very few
seem to have an interest in electronics at this level. I could be wrong,
and hope I am. Also, think of how often the more experienced members of
this list have to explain to people that one or two discreet transistors
does what they need(and how such a circuit works!), instead of a 14 pin IC.

Of course, intergration is a very good thing. We wouldn't have PICs without
it! But I just feel that in order for the design of ICs to change and for a
good understanding of how to interface to ICs, we need engineers to
understand what's going on inside, and it seems to me that you only get
this by doing some actual design once in a while involving electronics at
this more basic level.

I'd love additional comments if you can stand this discussion any longer :-)

Thanks,

Sean


+-------------------------------+
| Sean Breheny                  |
| Amateur Radio Callsign: KA3YXM|
| Electrical Engineering Student|
+-------------------------------+
Save lives, please look at http://www.all.org
Personal page: http://www.people.cornell.edu/pages/shb7
.....shb7KILLspamspam@spam@cornell.edu  Phone(USA): (607) 253-0315 ICQ #: 3329174

1998\12\09@113136 by Peter L. Peres

picon face
On Wed, 9 Dec 1998, Sean Breheny wrote:

> It really seems to me that for many (at least) low power circuits, there is
> a real trend toward more and more integration and less and less board-level
> electronics, for obvious reasons. IF and WHEN do most of you feel that most
> devices will consist of just one or more ASICs connected together with
> almost no external components at all? Of course, I suppose that another

About 2 years ago imho. Take a peek inside a top of the line miniature
digital camcorder some day for oohs and aahs, also look at what is on your
graphics card in the PC, same for frame grabber, integrated peripherals on
motherboard etc etc.

> possibility may be that FPGAs and reconfigurable analog chips (consisting
> of op-amp and switched cap filter arrays, or ADCs/DACs and a DSP processor)
> may perform this function, instead of complete ASICs.

Most complex applications are 'mixed design' ASICs that include analog and
digital parts, have been for some time now in higher end consumer
electronics with enough volume.

> This has been a concern of mine for some time because I am fearful that
> hobbists may eventually be left out of the loop if manufacturers go totally
> ASIC for small-signal and digital applications and the production process
> for ASICs is still too expensive for hobbists to get involved in.

imho nearly all the technology available to large manufacturers until now,
will move into garage startups etc. FPGAs and other developments are
leading there, and it works. Everyone knows about Synario and the famous
download cable, and FPGAs cost $10-15 in ones. I do not dare predict what
industry will be manufacturing with by then. Custom atoms ? Not far off.

I don't think that hobbyists will ever run out of hardware. If for no
other reason, then because they will certainly start making semiconductors
in the basement. There are weekly postings in newsgroups from people who
want to build their own LED/transistor/CPU etc. They mostly make it
eventually, or so it seems from followups.

Peter

1998\12\09@114549 by Sean Breheny

face picon face
Hi Peter,

At 06:07 PM 12/9/98 +0000, you wrote:
>I don't think that hobbyists will ever run out of hardware. If for no
>other reason, then because they will certainly start making semiconductors
>in the basement. There are weekly postings in newsgroups from people who
>want to build their own LED/transistor/CPU etc. They mostly make it
>eventually, or so it seems from followups.

Actually, I have wanted to try that for quite some time now, but thought
that it was completely futile to attempt. Have any pointers to info about
it? Could you tell me the names of the newsgroups you mention?

Thanks,

Sean


+-------------------------------+
| Sean Breheny                  |
| Amateur Radio Callsign: KA3YXM|
| Electrical Engineering Student|
+-------------------------------+
Save lives, please look at http://www.all.org
Personal page: http://www.people.cornell.edu/pages/shb7
shb7spamKILLspamcornell.edu  Phone(USA): (607) 253-0315 ICQ #: 3329174

1998\12\09@122057 by goflo

flavicon
face
Peter L. Peres wrote:
FPGAs and other developments are
> leading there, and it works. Everyone knows about Synario and the famous
> download cable...

Everyone but me. Care to enlighten me?
I'm looking for a place to start in pgm-able logic,

Regards, Jack

1998\12\09@144551 by Peter L. Peres

picon face
On Wed, 9 Dec 1998 .....gofloKILLspamspam.....pacbell.net wrote:

> Peter L. Peres wrote:
>  FPGAs and other developments are
> > leading there, and it works. Everyone knows about Synario and the famous
> > download cable...
>
> Everyone but me. Care to enlighten me?
> I'm looking for a place to start in pgm-able logic,

http://www.lattice.com

and follow the links. Also, I think that one should look at the online
VHDL courses / demos (use a web search engine to find them) for a
taster, and purchase a proper book for proper learning. Hope this helps,

Peter

1998\12\09@150229 by Harold Hallikainen

picon face
On Wed, 9 Dec 1998 00:56:25 -0500 Sean Breheny <EraseMEshb7spam_OUTspamTakeThisOuTCORNELL.EDU> writes:
>
>
>This has been a concern of mine for some time because I am fearful
>that hobbists may eventually be left out of the loop if manufacturers go

>totally ASIC for small-signal and digital applications and the
production
>process  for ASICs is still too expensive for hobbists to get involved
in.
>
>Are my fears unfounded? I hope so.
>


       I think ASICs only make sense on large volume products.  There
are still plenty of smaller volume products where programmable micros
(OTP PICs) and FPGAs (or PALs or whatever) would still make a lot of
sense.  I got started in all this (after learning vacuum tube
electronics) by wire wrapping 7400 series logic and 741 op amps about 30
years ago.  You can still buy those parts.
       So...  ASICs (or mask ROM micros) will take over large volume
products, but programmable and small scale integration stuff should be
around for smaller volume products and hobbyists.

Just my opinion..


Harold




Harold Hallikainen
haroldspamspam_OUThallikainen.com
Hallikainen & Friends, Inc.
See the FCC Rules at http://hallikainen.com/FccRules and comments filed
in LPFM proceeding at http://hallikainen.com/lpfm

___________________________________________________________________
You don't need to buy Internet access to use free Internet e-mail.
Get completely free e-mail from Juno at http://www.juno.com/getjuno.html
or call Juno at (800) 654-JUNO [654-5866]

1998\12\09@215535 by goflo

flavicon
face
Peter L. Peres wrote:

> Er, that URL was supposed to be:
> http://www.latticesemi.com
> sorry it's late here,

Thanks. Looks good.

Jack

1998\12\10@065040 by g.daniel.invent.design

flavicon
face
In my life time I just expect to see more and more functions included on
chip, ie better performance A2D,EEprom,counters,comparators,FPGA,on
board crystal /resonator, serial buffering, larger RAM,ROM etc.

ASICS require major applications to justify the cost, usually this also
means a lot of cost cutting also by chopping redundant functions, these
get stuck in a limited range of product/s and so do not usually threaten
any other market.

regards,
graham.

Sean Breheny wrote:
{Quote hidden}

1998\12\10@113128 by Martin McCormick

flavicon
face
       To me, it has always been a good news/bad news situation.  I
remember as a boy the ease with which one could work on vacuum-tube
equipment and also the frequency with which one needed to work on it.

       As am amateur radio operator who is also blind, I remember the
common practice of connecting a voltage controlled oscillator, itself
made of discrete transistors or even tubes to the meter or meters that
showed anode current and grid drive for the final power amplifier
stage.  The idea was to listen for a peak in the tone frequency or to
indirectly read the numerical value of the meter by comparing the tone
generated by the voltage across the meter movement against a
calibrated source with a pointer and raised dots around the dial.
When the pitch of the meter matched the pitch as heard from the
calibrated test dial, then we could "read" the meter by checking the
position of the pointer on the scale.

       Even with discrete transistors, one could always find some tap
point in the circuit to get a desired signal.  There are even some
construction articles which would give speech to older digital
multimeters or frequency counters by connecting the speech controller
to the segment drivers, if accessible, or even the segment lines and
also to the strobes.  The result would be messy, but did work for some
people.

       The bad news is that things are so well integrated today that
there aren't many if any practical places to tap signals that wouldn't
physically ruin the display.  The good news is that one can now buy
digital meters and frequency counters that have computer ports.  This
is the way to go.  I only hope that through-hole components and
wire-wrap technology stays around for a few more years.  On a personal
level, I can handle the size of components and pin spacing with this
form of technology with no real problem but that surface mount stuff
just feels like blobs of plastic with slightly rough edges.:-)  One
can take a scribe or pin which I sometimes call my Braille microscope
and tell something about the number of leads or the presence of an
index mark, but I don't know any practical way to build or repair
surface mount equipment.

       The only good news there is that some chips have matching
sockets that can be put in to a wire-wrap nest.

       These are the good old days of tinkering.  Now, if I can just
get a decent operating system on my computers that wasn't designed by
total idiots.  I think I am going Linux for there is still hope with
it.

Martin McCormick WB5AGZ  Stillwater, OK
OSU Center for Computing and Information Services Data Communications Group

1998\12\11@123325 by Marc

flavicon
face
> wire-wrap technology stays around for a few more years.  On a personal
> level, I can handle the size of components and pin spacing with this
> form of technology with no real problem but that surface mount stuff
> just feels like blobs of plastic with slightly rough edges.:-)  One

There exist test clips that can be clipped onto those popular 1.27mm
SOICs. On the upper end of the clip they have 2.54mm spacing for
cable connection. (Farnell, RS Components, etc)

More... (looser matching)
- Last day of these posts
- In 1998 , 1999 only
- Today
- New search...