Searching \ for '[PIC] PIC18F252 vs. dsPIC30F3013 Shootout' 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/microchip/ios.htm?key=spi
Search entire site for: 'PIC18F252 vs. dsPIC30F3013 Shootout'.

Exact match. Not showing close matches.
PICList Thread
'[PIC] PIC18F252 vs. dsPIC30F3013 Shootout'
2005\08\28@133910 by davedilatush

picon face
Greetings all,

I'm looking for some guidance in choosing between a PIC18F device
and a dsPIC30F device for a battery-powered application that's in
a very early stage of development-- specifically, any solid
information on relative current consumption at equal levels of
throughput.  The current-hogginess of dsPICs has been discussed
several times on the list in recent weeks, and I don't want to
make a design decision I'll end up regretting: battery life is
important.

This is a digital signal processing application which will be
very math-intensive: each sample (at a 10 KHz sampling rate) will
involve an A/D conversion, a 32x32 unsigned multiply, and several
dozen additions, subtractions, copies and bit shift operations,
all with 64-bit words.  There is very little byte-level
processing or bit-flipping involved in the application, and what
little of it there is, is done at background level (i.e., not at
sampling rate) and is very non-time-critical.  I've done some
exploratory coding and simulation with MPSIM, enough to determine
that a PIC18F252 running at 32 MHz (8 MHz crystal with 4X PLL)
would be adequate for the job.

However, at a quick glance, this would seem an ideal application
for a dsPIC30F3013: with the dsPIC's longer word length, and
consequently higher throughput-per-instruction in math
operations, I would expect to be able to reduce the clock
frequency significantly (maybe by a factor of two?) relative to
what I'd need for the PIC18F-- and perhaps even more, if I can
make good use of the dsPIC's DSP hardware/software resources.

So my concern (prompted by recent PICLIST discussions) is, is
dsPIC current consumption so wretched that a dsPIC30F3013 would
be likely to draw more current than a PIC18F here, even though
clocked at a much lower rate?

My vague, uninformed, totally off-the-wall WAG impression is that
the current consumption of the two approaches will probably be
approximately equal-- and if it is, I'll probably choose the
dsPIC30F simply because I'm eager to begin using these devices
and this project would be as good an opportunity as any.  But if
one approach is likely to give dramatically better battery life
than the other, then that's going to have to be the deciding
factor.

Anybody have any guidance on this?  It's going to be several
weeks before I can conduct a side-by-side "shootout" between the
two processors myself, and until then I'm just guessing.

Thanks in advance,

Dave D.

2005\08\28@153237 by olin piclist

face picon face
Dave Dilatush wrote:
> My vague, uninformed, totally off-the-wall WAG impression is that
> the current consumption of the two approaches will probably be
> approximately equal

I haven't looked at the number with that in mind either, but my WAG is that
they are about equal at the same clock rate.  In other words, they are
probably close in joules/instruction.  If you can use the wider word and
special math capabilities of the dsPIC at that clock rate, then I expect
(but again, you should sit down and do the math) that the dsPIC takes less
energy to crank the same computation.  In other words less joules/equation.

I would look for one of the newer dsPICs that come with an internal RC
oscillator and clock switching modes.  You can run it maxxed out when there
is something to do, then scale back the clock or got to sleep when you're
done.  That way you only pay power for the computation acutally performed.

I also don't understand why people are knocking the power consumption of the
dsPIC.  It might draw 180mA at 5V when running at 120Mhz, but scale that
back to the 40MHz a 18F can do and you get 60mA.  That's not out of line for
an 18F at max speed.  And then considering these instructions have the
potential for doing a lot more than the 18F instructions, it sounds like the
dsPIC is a bargain power wise.  Where's the problem?


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

2005\08\29@001231 by Chen Xiao Fan

face
flavicon
face
Maybe it is because of low power ARMs out there or other low
power chips. Maybe it is because people do not want to spend
money on a SMPS. Maybe it is because people are so much used
to the less than 10mA current consumption of little PICs.

Or maybe it is because of the so much negative information
regarding the Pentium 4 and Power PC. It is said that Microsoft
and Sony will use water cooling for XBox 2 and PS3.

Anyway, all these are maybes.

Regards,
Xiaofan

{Original Message removed}

2005\08\30@070956 by davedilatush

picon face
Thanks, Olin.  Looks like it's time to do some testing and
generate some hard numbers for comparison.

As soon as I can get far enough up on the dsPIC learning curve to
write code that takes full advantage of its capabilities, I'll
run some side-by-side tests and post the results.

Dave D.

Olin Lathrop wrote...
{Quote hidden}

More... (looser matching)
- Last day of these posts
- In 2005 , 2006 only
- Today
- New search...