PIC18F8722 - New 2006 RC MPU, 4x more program memory!

[kmcclary]

Aha! I think I'm the first to notice this (no hits on a CDF search), so here goes... I now believe:

*** The 2006 Robot Controller has a new MPU, the PIC18F8722. ***

When compared to the PIC18F8520:
- FOUR times the program memory
... 128Kb (65K single-word instructions) vs 32Kb (16K single word instr)
- *Slightly* more SRAM
... 3936 bytes, vs 2048 in the 18F8520

In addition (I'm not sure if this matters, but it might to USART users):
- two EUSARTs (vs two USARTS, see spec sheets for differences)
- There are 2 CCP and 3 ECCP (vs 5 CCP only) PWM channels.
- Same amount of EEPROM, A/D, timers

You can look up both spec sheets at Microchip's Data Sheet Finder page:
http://microchip.com/stellent/idcplg...GE&nodeId=2046

Bottom line: MUCH bigger programs (FOUR times larger), and a tad
more RAM space! <yahoo!>

In addition, there is a new IFI loader (V12, vs V10) to support this on
the RC page that you'll need to download:
- http://www.ifirobotics.com/rc.shtml
> IFI_Loader v1.0.12 (zip, 1-3-2006)
> Must be used with the 2006 Robot Controller
> http://www.ifirobotics.com/docs/ifi-loader-v1-0-12.zip

FYI, This is how I figured it out... the INI file in the ZIP showed
PIC18F8722 support was added (over V10's "PIC18F442 & 18F8520
only" support.)

This probably means the KoP will have a new CBOT compiler this
year, to target the 18F8722.

I made a call, and heard rumor that the 18F8520 is no longer
in production (unverified), which would support a need to change
the MPU in the RC anyway. Given IFI's statement that you MUST
use the V12 loader for the 2006 controller, I'm pretty sure I'm right
about this.

Comments? Verification (or denials)?

- Keith

[phrontist]

What is a CBOT compiler?

[kmcclary]

Originally Posted by phrontist

What is a CBOT compiler?

What IFI calls their KoP distribution of the Microchip C compiler for the PIC18F series MPUs. Needed to compile your C source robot control programs into a form capable of downloading into the robot controller.

- Keith

[Eldarion]

Originally Posted by kmcclary

There are 2 CCP and 3 ECCP (vs 5 CCP only) PWM channels.

So does this mean we have less PWMs?

Too bad it's not any faster

[phrontist]

Originally Posted by kmcclary

What IFI calls their KoP distribution of the Microchip C compiler for the PIC18F series MPUs. Needed to compile your C source robot control programs into a form capable of downloading into the robot controller.

- Keith

Ah, I never noticed they called it that.

You theory is intruiging, but there is always the possibility that one of the other IFI systems (Issac?) is moving to that MPU.

Still, here's hoping it happens!

[Eldarion]

Originally Posted by phrontist

You theory is intruiging, but there is always the possibility that one of the other IFI systems (Issac?) is moving to that MPU.

Just out of curiosity, what chip do the VEX kits use? I have heard that the upgraded loader was changed simply so that VEX kits were supported.

[1075master]

Originally Posted by kmcclary

What IFI calls their KoP distribution of the Microchip C compiler for the PIC18F series MPUs. Needed to compile your C source robot control programs into a form capable of downloading into the robot controller.

- Keith

could this maybe explain my problem trying to open hex files in the new ifi loader?? could hex be changed to something else??

[billbo911]

Originally Posted by Eldarion

Just out of curiosity, what chip do the VEX kits use? I have heard that the upgraded loader was changed simply so that VEX kits were supported.

2005 IFI RC and Vex both use the same PIC 18F8520.
The IFI Loader version 1-0-10 works with the 2005 IFI RC (not sure about 2004 RC). Version 1-0-11 was created to work with Vex but also works with 2005 RC.
It sounds like version 1-0-12 works with the 2006 RC. I hope it is backward compatible with Vex and 2005 and 2004 RC's. That way we will only need one version of the IFI Loader on our PC/Laptops.

[Gdeaver]

I see there are 2 SPI - I2c moduals. This could open up many new devices.

[BrianBSL]

Originally Posted by Eldarion

So does this mean we have less PWMs?

Too bad it's not any faster

If this device was to be in the 06 RC, it wouldn't matter. The PWM's we use are generated by software and not by the CCP module.

[Manoel]

Originally Posted by BrianBSL

If this device was to be in the 06 RC, it wouldn't matter. The PWM's we use are generated by software and not by the CCP module.

Except PWMs 13-16, which could be user generated. They weren't used much, though, as an interrupt intensive code could really mess them.

If this rumor is true - and it certainly looks like it, great catch - this is an amazing thing, four times more program memory will probably be more than enough for our needs (or will it? ).

[Mike]

If this is truly our new chip, I am going to make a prediction that the scripting language will return and be waayy more powerful than last year.

[kmcclary]

Originally Posted by Manoel

Except PWMs 13-16, which could be user generated. They weren't used much, though, as an interrupt intensive code could really mess them.

If this rumor is true - and it certainly looks like it, great catch - this is an amazing thing, four times more program memory will probably be more than enough for our needs (or will it? ).

Hey Manoel! Long time no chat... Thanks. I was pretty excited to catch this one.

Let's hit all questions at once...

Number of PWMs - This doesn't affect the I/O at all.
IMHO, the best use of this upgrade would be for swapping out
the robot program MPU, enlarging storage space for the robot
program itself.

Besides, I was under the impression that most (if not all) of the PWMs
were done by the *other* processor in the RC (or did they do away
with the two processor design and simply load up the one MPU
with more tasks? I haven't had time to delve into the new design
as much as the old one. Been too busy starting teams...)

Backward compatibility - YES. The 1-0-12 loader simply
ADDS support for the 18F8722. IOW, this is *in addition* to the
other two, so this one loader should take care of the Vex and
2005 RC as well.

Opening hex file problem - No clue. It shouldn't affect it
(famous last words), but I haven't had time to install and run
the new loader yet myself to check it.

Scripting languages - Good question. Also, I'm wondering if this
will affect the new CoreChart graphical programming at all.
But I really feel that if we're getting into THAT this year, the
company SHOULD have been brought into the loop on this change.

- Keith

[Chaychay]

remeber how Kevin Waston kept hinting last year regarding "trig functions"....especially in his scripts. Maybe this new MCU has enough "power" to give us all the trig we need for angles and such, making autonomous much easier

[Eldarion]

Originally Posted by Chaychay

remeber how Kevin Waston kept hinting last year regarding "trig functions"....especially in his scripts. Maybe this new MCU has enough "power" to give us all the trig we need for angles and such, making autonomous much easier

But you NEVER want to use trancendental functions directly, especially not inside a loop. It is actually industry standard programming practice to use a lookup table when speed is most important.