Maker Pro
Maker Pro

Stolen designs

D

David

Jan 1, 1970
0
John said:
The 68K family started in the early 70's and is still going strong.
The latest versions are RISC machines (all or mostly all hard-coded),
the Coldfire and (I think) Dragonball things. We still use the 68332,
with the nice CPU32 core, with cool stuff like a 64/32 hardware
divide.

John

The original Dragonball used a CPU32+ core (much like the 68332, but
with a full 32-bit databus) IIRC, combined with an LCD controller. A
later Dragonball used an ARM core (don't know why Motorola/Freescale
tried that one), and the current new Dragonfire has a ColdFire core.

Freescale refer to the ColdFire as a "variable instruction length RISC
architecture". It combines some CISC and some RISC aspects to give
something that I think is an ideal balance. The latest v4e cores have
full MMUs and FPUs if you want that sort of stuff.
 
R

Richard Henry

Jan 1, 1970
0
Mark Zenier said:
The Ultimate CISC, (unless
you count the iAPX-432).

What ever happened to the 432? All I ever got was the books.
 
Z

Zak

Jan 1, 1970
0
I'd have to say a functional equivalent isn't a rip off, though. I have seen
medical equipment companies come out with a functional equivalent of a
competetors device, like Vital Signs Monitors, that had very similar layouts
to where things hooked up, etc. They wanted it that way. But the innards
were 100% original (minus OEM module add ons made by a third party). Both
engineered their own system. The second group just wanted the layout to be
similar as it provided no learning curve for end users of the first device.
JMHO, but I could be wrong :(

Check out cars as well. After starting ouit different, stuff like pedals
were standardized very quickly. Som thing stook longer, but the wiper
hasn't always been on the right in every car.


Thomas
 
R

Richard Henry

Jan 1, 1970
0
Zak said:
Check out cars as well. After starting ouit different, stuff like pedals
were standardized very quickly. Som thing stook longer, but the wiper
hasn't always been on the right in every car.

My family has 5 cars. I have to sit and think for a while when I get in one
for things like parking brake release, seat adjustment, lights, wipers,
shifter --- no standards apparently.

And the worst is getting into a rental car at night when it is raining and
the rear window is fogged up.
 
A

Ancient_Hacker

Jan 1, 1970
0
Frank said:
I think it was the Tek 545 'scope that another company (Dumont? maybe
someone else) _copied_

I think those were legal copies. The feds wanted more scopes than Tek
could comfortably make. Of course the copies didnt have a lot of the
Tek quality details, but as usual the govt didnt care. Same thing
happened with the Collisn R-390A radios, which were farmed out to many
builders, including Helena Rubenstein.

As another example, there seem to be a bazillion mil-spec test leads
for sale. They seem to meet a whole slew of requirements--
foil-packed, many many replaceable test prods, good insulation. Every
requirement, except that they be usable: the probe ends tend to
unscrew themselves-- the alligator clips are too strongly sprung for
normal fingers, and the probe handles are hard to hold and too heavy.
Lots of specs met, except the useful ones.
 
J

John Larkin

Jan 1, 1970
0
I think it was the Tek 545 'scope that another company (Dumont? maybe
someone else) _copied_ (down to the number of turns on the power
transformer -- a hack-sawed Tek transformer was part of the evidence
IIRC), and sold to the US military. Decades later, Tek finally got
compensated for the loss of sales.

Lavoie, I think. Somebody else, Hickock maybe, made Tek clone plugins,
too.

John
 
J

John Larkin

Jan 1, 1970
0
Freescale (or Motorola before that) have wanted to EOL the 68332 for
years - it can't be made in their more modern lines, because the process
is to old. They tried to persuade customers to move to the MPC5xx line
(Power PC core, improved TPU), arranging prices to help, but the 68332
is as popular as ever.

They told me they were still out soliciting new business for the 332.
Why don't they just move it to a newer process? It's not that complex
by modern standards, and they wouldn't have to touch the microcode.
If you ever feel you want to move on, go for the ColdFire MCF523x line.
You get a lot more for your money (eTPU, Ethernet, 150 MHz, etc.), and
the ColdFire core is based on the same ISA as the 68332. Even
programming in assembly, there will not be many changes needed to your code.

Yes (grumble, snarl) I guess we'll have to do that one of these days.
We could use a little more speed now and then.

John
 
J

John Larkin

Jan 1, 1970
0
At least we can say that there was a point at which they did not care
about maintaining a reputation in the embedded market.

What their opinion on the subject today is, I don't know. It doesn't
matter to me because a direct lie they told me some years back got them on
my bad side and on such things I have a very long memory.

Intel is a notoriously agressive company, internally as well as to
vendors and customers.

John
 
J

Jim Thompson

Jan 1, 1970
0
[snip]
My family has 5 cars. I have to sit and think for a while when I get in one
for things like parking brake release, seat adjustment, lights, wipers,
shifter --- no standards apparently.

And the worst is getting into a rental car at night when it is raining and
the rear window is fogged up.

Ain't that the truth!

Particularly those items I have no familiarity with, like defrosters
;-)

Had an unfamiliar event happen in my own car this weekend.

Drove to San Diego to visit a granddaughter having spinal fusion
surgery.

On the way back it seemed to be getting hot in the car, so I manually
turned up the blower speed... no increase in air flow.

Finally dawned on me that San Diego is humid.

Opened up the circulation to outside air and blew snow out the vents
;-)

Then I ran over a piece of tread, from a truck tire, at 100MPH, just
east of Gila Bend on I8.

Not much choice, semi on my right, guard rail on the left :-(

Doesn't appear to have damaged anything... the Q45 has a nice skid
plate under the engine and tranny.

...Jim Thompson
 
J

John Larkin

Jan 1, 1970
0
The later (didn't really get out there until 1981) 68000 was more of
a 32 bit extension of the PDP-11, with two banks of 8 32 bit registers.
A much less complex architecture than VAX which had three address
instructions and a zillion address modes. The Ultimate CISC, (unless
you count the iAPX-432).

The 432 and the HP thing (3000 series) were sort of the pinnacle of
CISC microcoding, insanely macro instructions that did stuff like
polynomial expansion, threaded-list maintenance, frame management, all
that stuff. They turned out, generally, to be pig-slow.

Contrast that to Itanic, a modern VLIW risc architecture that's
pig-slow.

John
 
D

David

Jan 1, 1970
0
John said:
They told me they were still out soliciting new business for the 332.
Why don't they just move it to a newer process? It's not that complex
by modern standards, and they wouldn't have to touch the microcode.

The 68332 is an old design - moving it to a smaller process means a
great deal of re-work (compare it to moving your 68k assembly to a PPC
cpu). The ColdFire cores are all done as high-level synthesisable cores
which can be quickly implemented in different processes (such as if
you'd coded in C) - in fact, there are far more ColdFire cores produced
in ASICs than general purpose microcontrollers.

In a way, the MCF523x *is* the 68332 moved to a modern process.
Yes (grumble, snarl) I guess we'll have to do that one of these days.
We could use a little more speed now and then.

The 150 MHz 5234 will give you 10 to 20 times the speed, which is a
reasonable step up.
 
K

Keith

Jan 1, 1970
0
What ever happened to the 432? All I ever got was the books.
You got more than most. ;-) I was at the Grand Hyatt in NYC for
the iAPX432 unveiling. What a monster! (both *unfinished* hotel
and processor)
 
J

John Jardine.

Jan 1, 1970
0
I reverse-engineer stuff all the time. Likewise I collect schematics of
test gear etc. you never know when you might come across an idea worth
stealing :)

more often though, the trick is to figure out the underlying principles
of any "nice" circuit, then apply those principles to your thing.

Cheers
Terry

Ditto!.
I've just spotted a couple of neat features in a Wavetek 190, function
generator output amplifier and have already used ideas I've come across in
HP kit.
It's enormous fun to figure out someone else's hard work but subsequent use
of one of these 'not-invented-here' techniques always give me a bit of
guilt. S'pose I know deep down, it's stealing.
Best thing is for equipment makers not to allow their circuitry out of the
bulding. Won't stop determined copiers but'll stop people like me.
john
 
P

Phil Hobbs

Jan 1, 1970
0
John said:
The 432 and the HP thing (3000 series) were sort of the pinnacle of
CISC microcoding, insanely macro instructions that did stuff like
polynomial expansion, threaded-list maintenance, frame management, all
that stuff. They turned out, generally, to be pig-slow.

Contrast that to Itanic, a modern VLIW risc architecture that's
pig-slow.

John
I actually had a math coprocessor board for an HP 9816 lab computer that
had a 432 on it, iirc. It came with a nice compiler for HP Basic. My
instrument code just worked in those days--anyone else miss HP Basic for
lab code?

Cheers,

Phil Hobbs
 
J

John Larkin

Jan 1, 1970
0
I actually had a math coprocessor board for an HP 9816 lab computer that
had a 432 on it, iirc. It came with a nice compiler for HP Basic. My
instrument code just worked in those days--anyone else miss HP Basic for
lab code?

Nope. I use PowerBasic. It lets you do anything you want...

inline assembly

modern stuff like CASE and WHILE

dim an array AT an address

FOR loops that do useful things at 30 MHz

nice graphics

compiles a 4000-line program in, say, 0.1 seconds

John
 
T

Terry Given

Jan 1, 1970
0
John said:
Ditto!.
I've just spotted a couple of neat features in a Wavetek 190, function
generator output amplifier and have already used ideas I've come across in
HP kit.
It's enormous fun to figure out someone else's hard work but subsequent use
of one of these 'not-invented-here' techniques always give me a bit of
guilt. S'pose I know deep down, it's stealing.
Best thing is for equipment makers not to allow their circuitry out of the
bulding. Won't stop determined copiers but'll stop people like me.
john

I have yet to "rip off" anything, but I do collect schematics of old
stuff. And let it inspire me. sans guilt :)

Cheers
Terry
 
R

Rich Grise

Jan 1, 1970
0
They told me they were still out soliciting new business for the 332.
Why don't they just move it to a newer process? It's not that complex
by modern standards, and they wouldn't have to touch the microcode.

I thought, "If it's that simple, why not do it in a big FPGA or CPLD?"
So I did this:
http://www.google.com/search?q=68332+hdl+core
and one of the hits was this:
http://www.opencores.org/browse.cgi/by_category
and I kinda went, "Holy Yikes!" - there sure is a lot of stuff already
written - it's like free hardware design like free software, more
like free speech than free beers - you still have to buy the CPLD from
somebody. ;-)

Cheers!
Rich
 
J

Joel Kolstad

Jan 1, 1970
0
Rich Grise said:
http://www.opencores.org/browse.cgi/by_category
and I kinda went, "Holy Yikes!" - there sure is a lot of stuff already
written - it's like free hardware design like free software, more
like free speech than free beers - you still have to buy the CPLD from
somebody. ;-)

Yeah, and the support is also like free software: It may be great, it may be
non-existant... and this is a significant risk if you're going to build a
commercial product around it, since even with source code most mere mortals
aren't going to be able to just dive into, say, some WiMax core and fix some
obscure bug! (This reminds me of all those <$50 wireless routers out there
that run Linux... yeah, they work... mostly... but it's no surprise at all
that so many of them lock-up and/or reboot as soon as you start really
throwing a lot of data through them...)

---Joel
 
J

John Larkin

Jan 1, 1970
0
The NMOS 6800 was introduced at the bargain price of US $395 per chip
(back when that was not a bad week's wage) and required a nasty
dual-phase rail-to-rail (not TTL, it had to swing to within 250mV of
each rail fairly snappily) external clock circuit.

I designed the TANO Outpost 7 around the 6800, and I recall the cpu
costing around $15 once things got rolling. The later generation unit,
the Outpost 11, used a 6809 and occupied some tiny niche in computing
history, except that they did that version after they fired me.

I did write (in Juneau, Alaska) a decent RTOS for the 6800, including
a token-ring LAN, before anybody called it a token-ring LAN.

John
 
R

Robert Baer

Jan 1, 1970
0
Spehro said:
It's not necessarily 'piracy' in the sense of being illegal. I've had
several innovative designs copied, mostly by small garage shops who
could turn an apparent profit by cutting corners on tooling and
copying circuits directly. A couple others were new ideas for new
markets and the functionality was emulated, without being a direct
copy. That's actually more irritating. Oh, and I had an A****n
Instrument guy bitch to me at a trade show about a funcional
equivalent that I'd designed. Sorry, the customer asked for it. Seemed
they'd heard that before (probably because they charged really, really
high prices for something a good instrument design engineer could do
over a coffee, on a paper napkin).

If you publish schematics your designs will be copied whether or not
they are any good.


Best regards,
Spehro Pefhany
Not too sure about that; a few engineers looked at our original
Codatron design and said that it was not reproduceable..
We placed it in public domain; no bites.
 
Top