Interrupt RPM and Taylors Elapsed time on 18F4620

[Bruce]

Are your result figures correct? It shows the exact same result on the 877A for 6000
and 8000 RPM!

What do you get with something like this on both versions?

period.lowbyte = CCPR1L		' Store the captured value in
period.highbyte = CCPR1H	' period variable

SEROUT2 LCD,84, [Prefix,CursorPS,20,dec5 period]

[Tobias]

RPM and period numbers per chip
Rpm 4620 877a
5000 15090 14999
6000 12575 12499
7000 10777 10712
8000 9430 9373
9000 8383 8331
10000 7545 7499

[Bruce]

Do you have another 18F type besides a 4620 you can test this on?

That would tell you if it's a hardware bug on the 4620. This part has
a massive list of errata & odd problems with ECCP, Timer1, Timer3,
and more.

[Tobias]

I might have a 4520 floating around. Can you suggest a part number that will work with the boards I already have made? I will order it today if you have a suggestion.

[Bruce]

A 4520 should drop right in, but this one also has a boat-load of errata for CCP, timer, and
a bunch more. I would test it if you already have one before buying another PIC.

The 18F4431 is my own personal favorite, and I also like the older 18F452. The drop-in
replacement (your 4520) for the 452 has way too many bugs.

[Tobias]

Thanks it looks like a couple different config flags need to be changed to go with the 4431. The pinout appears to be the same.?.

[Bruce]

It's the same pinout. It also has a lot of extra features.

But - I would try that 4520 if you can find it.

Where are you getting your PICs from? I have samples, and would be
more than willing to shoot you an 18F4431 if you cover postage.

[Darrel Taylor]

RPM and period numbers per chip
Rpm 4620 877a
5000 15090 14999
6000 12575 12499
7000 10777 10712
8000 9430 9373
9000 8383 8331
10000 7545 7499

I think you need to clear the timer sooner. (Immediately after capture).

The program goes through the whole conversion to RPM with multiplications and DIV32's before clearing the timer.

The 18F's multiply faster because they have a hardware multiplier.
So it clears the timer quicker and ends up with bigger numbers.

The 4620 numbers are closer to reality, but it's still wrong.

Clear the timer immediately after a capture and both chips should read the same numbers (or really close).

hth,

[Bruce]

Hi Darrel,

Start:
      If Capture = 0 then ' ignores 1st capture
         Goto Start
      endif

      T1CON.0=1
      Capture = 0 
      
CaptureLoop:
If Capture = 0 then 
  Goto CaptureLoop
endif
T1Con.0=0  
period.LowByte=CCPR1L
period.HighByte=CCPR1H

With both PICs running at 20MHz, CCPR1L & CCPR1H should have the same values - since
it's all done in hardware with the capture module capturing the value of timer1 after the
4 rising edges.

And the value being captured is before the multiply.

[Darrel Taylor]

With both PICs running at 20MHz, CCPR1L & CCPR1H should have the same values - since it's all done in hardware with the capture module capturing the value of timer1 after the
4 rising edges.

Yes the captured value is stored in CCPR1L:H, but the CCP module does not clear Timer1. You have to do that. And it's not being cleared until after the math.

Then it goes back up to Loop: and clears it again, without any reference to the incoming signal. (actually, it is referenced to the last capture plus math time).

If you only clear it once, immediately after the capture, then it will be ready for the next capture.

The first capture is always wrong, and must be discarded.

Loop:
TMR1H = 0    'Clear 8-bit register                              
TMR1L = 0    'Clear 8-bit register
capture = 0  'Clear Timer
Start:
      If Capture = 0 then
         Goto Start
      endif

      T1CON.0=1
      Capture = 0 
      
CaptureLoop:
If Capture = 0 then 
  Goto CaptureLoop
endif
T1Con.0=0  
period.LowByte=CCPR1L
period.HighByte=CCPR1H

RPM = 10000
RPM = RPM * RPM ' 100,000,000
RPM = DIV32 period ' 100,000,000 / RevCount
RPM = RPM * 60 ' Per minute
RPM = DIV32 400
RPM = (RPM*5)

SEROUT2 LCD,84, [Prefix,CursorPS,20,"RPM ",dec5 RPM, " ", DEC CCPR1L, " ", dec CCPR1H]
  
Gosub ClearTimer1
 
Goto Loop

Added: I guess if Timer1 was Stopped in the ClearTimer1 routine (or at Loop: ), what you have would work better, but it will only capture every other 4-pulses. (8-pulses total)

[Darrel Taylor]

Oi vey,
I just saw ...

T1Con.0=0

Ignore what I just said, and I'm off to proteus to make sure before I open my mouth again.

[Bruce]

Cool. If I'm wrong, he gets a free PIC & you get the eBeer....

Like I said before, there could indeed be a timing difference between these points;

      T1CON.0=1
      Capture = 0

That would for sure be the cause.