hi C,
Looking at that TERM.txt, I agree with 'jjw' it looks wrong.?
Why are you printing the HB and LB bytes separately instead of a WORD value.?
E

Hi E,
I was happy to see the 2xBYTEs appear, but will look at WORD later.
In you CODE did you send 2xBYTEs or 1xWORD?
I was under the impression that the combining of the 2x BYTEs would be in the MASTER.
C

 

hi C,
The pulses per revolution is 360, which variant of the QEI method are you using.?
E
You can only send one 8 Bit Byte per SPI exchange, but a value of 359 requires 2 Bytes, so its sent as HB and then LB, the program OR's to the Bytes to make a 16Bit WORD.

eg: say brg=359 deci,
then
HB= 0x01 , LB = 0x67
So WORD = 0x0167 == 359 decimal

 

hi C,
The pulses per revolution is 360, which variant of the QEI method are you using.?
E
You can only send one 8 Bit Byte per SPI exchange, but a value of 359 requires 2 Bytes, so its sent as HB and then LB, the program OR's to the Bytes to make a 16Bit WORD.

eg: say brg=359 deci,
then
HB= 0x01 , LB = 0x67
So WORD = 0x0167 == 359 decimal

Hi E,
This all should be somewhere in my files, as we've done it all before.
I'm remembering POSCNT and MAXCNT. Things will dawn on me, I'll report back
C.

 

hi C,
The pulses per revolution is 360, which variant of the QEI method are you using.?
E
You can only send one 8 Bit Byte per SPI exchange, but a value of 359 requires 2 Bytes, so its sent as HB and then LB, the program OR's to the Bytes to make a 16Bit WORD.

eg: say brg=359 deci,
then
HB= 0x01 , LB = 0x67
So WORD = 0x0167 == 359 decimal

Hi E,
How does Oshonsoft do [ ORs ]?
C

 

OR not needed.
X.HB= encoderhb // highbyte from encoder
X.LB= encoderlb // lowbyte "
X is the angle 0-359

 

hi,
Let brg be a 16 bit WORD
When brg = 0, thats 0000 0000, 0000 0000 ie 16 bit locations

Now say brg.HB = 0x01 thats 0000 0001
and
say brg.LB = 0x67 thats 0110 0111

So the compiler places
brg.HB in the 8 high bit locations of that WORD ie: 0000 0001 0000 0000
brg.LB in the 8 low bit locations of that WORD ie: 0000 0001 0110 0111

WORD is now 0x0167 == 359 decimal

You should really know how to construct a WORD by now.

E

 

hi jjw,
I used the WORD 'OR' so that 'C' might better understand what is happening with the HB and LB bytes.

I am surprised that he is having problems at this stage in building a WORD from HB LB Bytes.

E

 

hi jjw,
I used the WORD 'OR' so that 'C' might better understand what is happening with the HB and LB bytes.

I am surprised that he is having problems at this stage in building a WORD from HB LB Bytes.

E

Hi E,
I simply don't retain programming protocols.
If somethings not working as in this case, I look up old examples, then copy/paste, or ask in a forum.
I have many skills, that I'm natural at, I won't list them, but computing isn't one of them.
So either look up old examples, or ask in a forum, it's my only way. As long as someone's out there, I'll carry on.
Enjoy your computing expertise

So, the problem here:
I'm trying POSCNT tests, and I haven't yet figured out POSCNTL or POSCNTLB or POSCNTL = POSCNTL = POSCNTLB etc.
I tried [ poscnt = POSCNTH Or POSCNTL ] but as you have just explained, this didn't work.
I think I've got it now, let's see.
C

 

hi jjw,
I used the WORD 'OR' so that 'C' might better understand what is happening with the HB and LB bytes.

I am surprised that he is having problems at this stage in building a WORD from HB LB Bytes.

E

This was discussed about 1.5 years ago in reading the bmp280.

 

This was discussed about 1.5 years ago in reading the bmp280.

Hi J,
Yes, I know, and here: https://www.electro-tech-online.com/threads/gps-controlled-nmea-data-following-tracker.146733/page-2 even longer ago.
Embarrassing isn't it?

Actually I was thrown by the use of [ OR ]. This must have reminded me of a different context, probably in a calculation somewhere in the past, and Oshonsoft has a different way of implementing it, I think.
C

 

Hi,
This is the SLAVE only.
Using the 4X Option for QEI, that I calculate to be 360X4 = 1440 - 1 =1439.
I set MAXCNT to 1439.

Now the SLAVE READout looks ok to me.

Here are the program and the TERM view, showing 1REV CW.

The master READout still shows odd number, I'll investigate.
C

 

Hi,
Slight update on the SLAVE program.
C

 

Hi,
Just found an interesting problem. The MASTER READings are 50% of the SLAVE WRITEings.
Here's a TXT: (I switch between the MASTER/SLAVE UARTS., so you see when I change)
I'll keep checking, and post the programs again, when I've sorted it out, or I'll ask.

C

 

Hi,
Following from #513, here is a TERM view plus an analysis image showing that the SLAVE output is the same as the SPI READings, which tells me that the corruption is in the MASTER.

Is it possible that timing is off, SHIFTING the MASTER input 1x BIT to the right?
C.

 

Hi,
I've programming the PICs to see where the '1/2ing error' started, and it appears to have been there since getting sensible results, see #497 TERM view. You can see [ 128 ] showing in the results, which is actually %00000000 shifted right 1xBIT, by in this case a [ 1 ], but it could be a [ 0 ] which makes it difficult to spot.
In the case above, TERM is showing the shift, but lately the TERM showing correct BYTEs, and the shift doesn't show till in the MASTER. It seems to me like to tight clock tolerances?

Could a pull down resistor on the MISO track help?

Does anyone know how this can happen?

EDIT: I'm getting a feint recollection of clock sampling 1/2 way along the CLK? Can someone remind me if they know what I mean please?
EDIT EDIT: Could it be something to do with SSPSTAT BIT7?
C.

 

Hi,
I changed both SSPSTAT.SMP BITs to [ 1 ]
Here's the result: Turning the Incremental encoder 1xREV CW
Hopefully, I've stumbled on the answer to the [ 1/2ing ] problem, but can someone explain or verify what I've done please?
C

 

hi C,
It is easier to follow if you print the WORD #QEIAZI
Showing it this way is a pain to follow the actual values... From SLAVE QEIAZILB= 2 QEIAZIHB= 84
Lets see the values as decimal values.
E

 

hi C,
It is easier to follow if you print the WORD #QEIAZI
Showing it this way is a pain to follow the actual values... From SLAVE QEIAZILB= 2 QEIAZIHB= 84
Lets see the values as decimal values.
E

Hi E,
I had to change round the LB and HB.
Here is the result:
Plus both programs.
C

 

hi,
Is appears to be counting in 0.25deg increments.
If you made the QEIAZI a Single you could resolve to 0.25deg, if required.

E

 

hi,
Is appears to be counting in 0.25deg increments.
If you made the QEIAZI a Single you could resolve to 0.25deg, if required.

E

Hi E,
I used what I thought to be the best choice here: [ 110 = QEI enabled in 4x Update mode; position counter is reset on period match (POSCNT = MAXCNT) ] in SLAVE.

Once in MASTER how do I divide it down to DEGREES please?

When I changed the [ SMP ] BIT setting, what did I do?
Do you think this could have been the problem?
C